id	sid	tid	token	lemma	pos
bracis-33605	1	1	reinforcement	reinforcement	NOUN
bracis-33605	1	2	learning	learn	VERB
bracis-33605	1	3	with	with	ADP
bracis-33605	1	4	 	 	SPACE
bracis-33605	1	5	utility	utility	NOUN
bracis-33605	1	6	-	-	PUNCT
bracis-33605	1	7	based	base	VERB
bracis-33605	1	8	semantic	semantic	NOUN
bracis-33605	1	9	for	for	ADP
bracis-33605	1	10	 	 	SPACE
bracis-33605	1	11	goals	goal	NOUN
bracis-33605	1	12	|	|	ADV
bracis-33605	1	13	springer	springer	NOUN
bracis-33605	1	14	nature	nature	PROPN
bracis-33605	1	15	link	link	PROPN
bracis-33605	1	16	(	(	PUNCT
bracis-33605	1	17	formerly	formerly	ADV
bracis-33605	1	18	springerlink	springerlink	NOUN
bracis-33605	1	19	)	)	PUNCT
bracis-33605	1	20	skip	skip	VERB
bracis-33605	1	21	to	to	ADP
bracis-33605	1	22	main	main	ADJ
bracis-33605	1	23	content	content	NOUN
bracis-33605	1	24	advertisement	advertisement	NOUN
bracis-33605	1	25	log	log	NOUN
bracis-33605	1	26	in	in	ADP
bracis-33605	1	27	menu	menu	NOUN
bracis-33605	1	28	find	find	VERB
bracis-33605	1	29	a	a	DET
bracis-33605	1	30	journal	journal	NOUN
bracis-33605	1	31	publish	publish	VERB
bracis-33605	1	32	with	with	ADP
bracis-33605	1	33	us	we	PRON
bracis-33605	1	34	track	track	VERB
bracis-33605	1	35	your	your	PRON
bracis-33605	1	36	research	research	NOUN
bracis-33605	1	37	search	search	NOUN
bracis-33605	1	38	cart	cart	NOUN
bracis-33605	1	39	home	home	NOUN
bracis-33605	1	40	intelligent	intelligent	ADJ
bracis-33605	1	41	systems	system	NOUN
bracis-33605	1	42	conference	conference	NOUN
bracis-33605	1	43	paper	paper	NOUN
bracis-33605	1	44	reinforcement	reinforcement	NOUN
bracis-33605	1	45	learning	learn	VERB
bracis-33605	1	46	with	with	ADP
bracis-33605	1	47	 	 	SPACE
bracis-33605	1	48	utility	utility	NOUN
bracis-33605	1	49	-	-	PUNCT
bracis-33605	1	50	based	base	VERB
bracis-33605	1	51	semantic	semantic	NOUN
bracis-33605	1	52	for	for	ADP
bracis-33605	1	53	 	 	SPACE
bracis-33605	1	54	goals	goal	NOUN
bracis-33605	1	55	conference	conference	NOUN
bracis-33605	1	56	paper	paper	NOUN
bracis-33605	1	57	first	first	ADV
bracis-33605	1	58	online	online	ADV
bracis-33605	1	59	:	:	PUNCT
bracis-33605	1	60	30	30	NUM
bracis-33605	1	61	january	january	NOUN
bracis-33605	1	62	2025	2025	NUM
bracis-33605	1	63	pp	pp	PROPN
bracis-33605	1	64	354–369	354–369	NUM
bracis-33605	1	65	cite	cite	VERB
bracis-33605	1	66	this	this	DET
bracis-33605	1	67	conference	conference	NOUN
bracis-33605	1	68	paper	paper	NOUN
bracis-33605	1	69	access	access	NOUN
bracis-33605	1	70	provided	provide	VERB
bracis-33605	1	71	by	by	ADP
bracis-33605	1	72	university	university	PROPN
bracis-33605	1	73	of	of	ADP
bracis-33605	1	74	notre	notre	PROPN
bracis-33605	1	75	dame	dame	PROPN
bracis-33605	1	76	hesburgh	hesburgh	PROPN
bracis-33605	1	77	library	library	PROPN
bracis-33605	1	78	download	download	PROPN
bracis-33605	1	79	book	book	NOUN
bracis-33605	1	80	pdf	pdf	PROPN
bracis-33605	1	81	download	download	NOUN
bracis-33605	1	82	book	book	NOUN
bracis-33605	1	83	epub	epub	PROPN
bracis-33605	1	84	intelligent	intelligent	ADJ
bracis-33605	1	85	systems	system	NOUN
bracis-33605	1	86	(	(	PUNCT
bracis-33605	1	87	bracis	bracis	NOUN
bracis-33605	1	88	2024	2024	NUM
bracis-33605	1	89	)	)	PUNCT
bracis-33605	1	90	reinforcement	reinforcement	NOUN
bracis-33605	1	91	learning	learning	NOUN
bracis-33605	1	92	with	with	ADP
bracis-33605	1	93	 	 	SPACE
bracis-33605	1	94	utility	utility	NOUN
bracis-33605	1	95	-	-	PUNCT
bracis-33605	1	96	based	base	VERB
bracis-33605	1	97	semantic	semantic	NOUN
bracis-33605	1	98	for	for	ADP
bracis-33605	1	99	 	 	SPACE
bracis-33605	1	100	goals	goal	NOUN
bracis-33605	1	101	download	download	VERB
bracis-33605	1	102	book	book	NOUN
bracis-33605	1	103	pdf	pdf	PROPN
bracis-33605	1	104	download	download	NOUN
bracis-33605	1	105	book	book	PROPN
bracis-33605	1	106	epub	epub	PROPN
bracis-33605	1	107	christian	christian	PROPN
bracis-33605	1	108	delgado	delgado	VERB
bracis-33605	1	109	polar	polar	ADJ
bracis-33605	1	110	  	  	SPACE
bracis-33605	1	111	orcid	orcid	NOUN
bracis-33605	1	112	:	:	PUNCT
bracis-33605	1	113	orcid.org/0000-0001-6123-64969,10	orcid.org/0000-0001-6123-64969,10	PROPN
bracis-33605	1	114	,	,	PUNCT
bracis-33605	1	115	karina	karina	PROPN
bracis-33605	1	116	valdivia	valdivia	PROPN
bracis-33605	1	117	delgado	delgado	PROPN
bracis-33605	1	118	  	  	SPACE
bracis-33605	1	119	orcid	orcid	PROPN
bracis-33605	1	120	:	:	PUNCT
bracis-33605	1	121	orcid.org/0000-0002-9120-89879	orcid.org/0000-0002-9120-89879	PROPN
bracis-33605	1	122	&	&	CCONJ
bracis-33605	1	123	valdinei	valdinei	PROPN
bracis-33605	1	124	freire	freire	PROPN
bracis-33605	1	125	  	  	SPACE
bracis-33605	1	126	orcid	orcid	NOUN
bracis-33605	1	127	:	:	PUNCT
bracis-33605	1	128	orcid.org/0000-0003-0330-39319	orcid.org/0000-0003-0330-39319	PROPN
bracis-33605	1	129	  	  	SPACE
bracis-33605	1	130	part	part	NOUN
bracis-33605	1	131	of	of	ADP
bracis-33605	1	132	the	the	DET
bracis-33605	1	133	book	book	NOUN
bracis-33605	1	134	series	series	NOUN
bracis-33605	1	135	:	:	PUNCT
bracis-33605	1	136	lecture	lecture	NOUN
bracis-33605	1	137	notes	note	NOUN
bracis-33605	1	138	in	in	ADP
bracis-33605	1	139	computer	computer	NOUN
bracis-33605	1	140	science	science	NOUN
bracis-33605	1	141	(	(	PUNCT
bracis-33605	1	142	(	(	PUNCT
bracis-33605	1	143	lnai	lnai	ADJ
bracis-33605	1	144	,	,	PUNCT
bracis-33605	1	145	volume	volume	NOUN
bracis-33605	1	146	15413	15413	NUM
bracis-33605	1	147	)	)	PUNCT
bracis-33605	1	148	)	)	PUNCT
bracis-33605	1	149	included	include	VERB
bracis-33605	1	150	in	in	ADP
bracis-33605	1	151	the	the	DET
bracis-33605	1	152	following	follow	VERB
bracis-33605	1	153	conference	conference	NOUN
bracis-33605	1	154	series	series	NOUN
bracis-33605	1	155	:	:	PUNCT
bracis-33605	1	156	brazilian	brazilian	ADJ
bracis-33605	1	157	conference	conference	NOUN
bracis-33605	1	158	on	on	ADP
bracis-33605	1	159	intelligent	intelligent	ADJ
bracis-33605	1	160	systems	system	NOUN
bracis-33605	1	161	378	378	NUM
bracis-33605	1	162	accesses	access	VERB
bracis-33605	1	163	abstract	abstract	ADJ
bracis-33605	1	164	stochastic	stochastic	ADJ
bracis-33605	1	165	shortest	short	ADJ
bracis-33605	1	166	path	path	NOUN
bracis-33605	1	167	problems	problem	NOUN
bracis-33605	1	168	(	(	PUNCT
bracis-33605	1	169	ssps	ssps	NOUN
bracis-33605	1	170	)	)	PUNCT
bracis-33605	1	171	are	be	AUX
bracis-33605	1	172	markov	markov	NOUN
bracis-33605	1	173	decision	decision	NOUN
bracis-33605	1	174	processes	process	NOUN
bracis-33605	1	175	with	with	ADP
bracis-33605	1	176	goal	goal	NOUN
bracis-33605	1	177	states	state	NOUN
bracis-33605	1	178	and	and	CCONJ
bracis-33605	1	179	the	the	DET
bracis-33605	1	180	problem	problem	NOUN
bracis-33605	1	181	is	be	AUX
bracis-33605	1	182	to	to	PART
bracis-33605	1	183	find	find	VERB
bracis-33605	1	184	policies	policy	NOUN
bracis-33605	1	185	to	to	PART
bracis-33605	1	186	achieve	achieve	VERB
bracis-33605	1	187	the	the	DET
bracis-33605	1	188	goal	goal	NOUN
bracis-33605	1	189	with	with	ADP
bracis-33605	1	190	the	the	DET
bracis-33605	1	191	lowest	low	ADJ
bracis-33605	1	192	possible	possible	ADJ
bracis-33605	1	193	expected	expect	VERB
bracis-33605	1	194	cost	cost	NOUN
bracis-33605	1	195	.	.	PUNCT
bracis-33605	2	1	something	something	PRON
bracis-33605	2	2	common	common	ADJ
bracis-33605	2	3	in	in	ADP
bracis-33605	2	4	this	this	DET
bracis-33605	2	5	type	type	NOUN
bracis-33605	2	6	of	of	ADP
bracis-33605	2	7	problem	problem	NOUN
bracis-33605	2	8	is	be	AUX
bracis-33605	2	9	the	the	DET
bracis-33605	2	10	existence	existence	NOUN
bracis-33605	2	11	of	of	ADP
bracis-33605	2	12	states	state	NOUN
bracis-33605	2	13	from	from	ADP
bracis-33605	2	14	which	which	PRON
bracis-33605	2	15	it	it	PRON
bracis-33605	2	16	is	be	AUX
bracis-33605	2	17	not	not	PART
bracis-33605	2	18	possible	possible	ADJ
bracis-33605	2	19	to	to	PART
bracis-33605	2	20	reach	reach	VERB
bracis-33605	2	21	the	the	DET
bracis-33605	2	22	goal	goal	NOUN
bracis-33605	2	23	,	,	PUNCT
bracis-33605	2	24	these	these	DET
bracis-33605	2	25	states	state	NOUN
bracis-33605	2	26	are	be	AUX
bracis-33605	2	27	called	call	VERB
bracis-33605	2	28	dead	dead	ADJ
bracis-33605	2	29	-	-	PUNCT
bracis-33605	2	30	end	end	NOUN
bracis-33605	2	31	.	.	PUNCT
bracis-33605	3	1	in	in	ADP
bracis-33605	3	2	this	this	DET
bracis-33605	3	3	case	case	NOUN
bracis-33605	3	4	,	,	PUNCT
bracis-33605	3	5	it	it	PRON
bracis-33605	3	6	is	be	AUX
bracis-33605	3	7	important	important	ADJ
bracis-33605	3	8	to	to	PART
bracis-33605	3	9	have	have	VERB
bracis-33605	3	10	methods	method	NOUN
bracis-33605	3	11	that	that	PRON
bracis-33605	3	12	consider	consider	VERB
bracis-33605	3	13	the	the	DET
bracis-33605	3	14	importance	importance	NOUN
bracis-33605	3	15	,	,	PUNCT
bracis-33605	3	16	not	not	PART
bracis-33605	3	17	only	only	ADV
bracis-33605	3	18	of	of	ADP
bracis-33605	3	19	cost	cost	NOUN
bracis-33605	3	20	,	,	PUNCT
bracis-33605	3	21	but	but	CCONJ
bracis-33605	3	22	also	also	ADV
bracis-33605	3	23	of	of	ADP
bracis-33605	3	24	the	the	DET
bracis-33605	3	25	probability	probability	NOUN
bracis-33605	3	26	of	of	ADP
bracis-33605	3	27	achieving	achieve	VERB
bracis-33605	3	28	the	the	DET
bracis-33605	3	29	goal	goal	NOUN
bracis-33605	3	30	.	.	PUNCT
bracis-33605	4	1	in	in	ADP
bracis-33605	4	2	reinforcement	reinforcement	NOUN
bracis-33605	4	3	learning	learning	NOUN
bracis-33605	4	4	(	(	PUNCT
bracis-33605	4	5	rl	rl	NOUN
bracis-33605	4	6	)	)	PUNCT
bracis-33605	4	7	,	,	PUNCT
bracis-33605	4	8	the	the	DET
bracis-33605	4	9	problem	problem	NOUN
bracis-33605	4	10	of	of	ADP
bracis-33605	4	11	making	make	VERB
bracis-33605	4	12	this	this	DET
bracis-33605	4	13	trade	trade	NOUN
bracis-33605	4	14	-	-	PUNCT
bracis-33605	4	15	off	off	NOUN
bracis-33605	4	16	between	between	ADP
bracis-33605	4	17	cost	cost	NOUN
bracis-33605	4	18	to	to	ADP
bracis-33605	4	19	goal	goal	NOUN
bracis-33605	4	20	and	and	CCONJ
bracis-33605	4	21	probability	probability	NOUN
bracis-33605	4	22	to	to	ADP
bracis-33605	4	23	goal	goal	NOUN
bracis-33605	4	24	has	have	AUX
bracis-33605	4	25	been	be	AUX
bracis-33605	4	26	little	little	ADV
bracis-33605	4	27	studied	study	VERB
bracis-33605	4	28	and	and	CCONJ
bracis-33605	4	29	the	the	DET
bracis-33605	4	30	common	common	ADJ
bracis-33605	4	31	strategy	strategy	NOUN
bracis-33605	4	32	to	to	PART
bracis-33605	4	33	deal	deal	VERB
bracis-33605	4	34	with	with	ADP
bracis-33605	4	35	ssps	ssps	NOUN
bracis-33605	4	36	with	with	ADP
bracis-33605	4	37	dead	dead	ADJ
bracis-33605	4	38	-	-	PUNCT
bracis-33605	4	39	ends	end	NOUN
bracis-33605	4	40	is	be	AUX
bracis-33605	4	41	using	use	VERB
bracis-33605	4	42	discounts	discount	NOUN
bracis-33605	4	43	.	.	PUNCT
bracis-33605	5	1	in	in	ADP
bracis-33605	5	2	some	some	DET
bracis-33605	5	3	works	work	NOUN
bracis-33605	5	4	,	,	PUNCT
bracis-33605	5	5	penalties	penalty	NOUN
bracis-33605	5	6	are	be	AUX
bracis-33605	5	7	used	use	VERB
bracis-33605	5	8	on	on	ADP
bracis-33605	5	9	exploration	exploration	NOUN
bracis-33605	5	10	in	in	ADP
bracis-33605	5	11	ssps	ssps	NOUN
bracis-33605	5	12	with	with	ADP
bracis-33605	5	13	dead	dead	ADJ
bracis-33605	5	14	-	-	PUNCT
bracis-33605	5	15	ends	end	NOUN
bracis-33605	5	16	.	.	PUNCT
bracis-33605	6	1	however	however	ADV
bracis-33605	6	2	,	,	PUNCT
bracis-33605	6	3	using	use	VERB
bracis-33605	6	4	discounts	discount	NOUN
bracis-33605	6	5	and	and	CCONJ
bracis-33605	6	6	penalties	penalty	NOUN
bracis-33605	6	7	can	can	AUX
bracis-33605	6	8	lead	lead	VERB
bracis-33605	6	9	to	to	ADP
bracis-33605	6	10	errors	error	NOUN
bracis-33605	6	11	in	in	ADP
bracis-33605	6	12	finding	find	VERB
bracis-33605	6	13	a	a	DET
bracis-33605	6	14	policy	policy	NOUN
bracis-33605	6	15	with	with	ADP
bracis-33605	6	16	a	a	DET
bracis-33605	6	17	desired	desire	VERB
bracis-33605	6	18	trade	trade	NOUN
bracis-33605	6	19	-	-	PUNCT
bracis-33605	6	20	off	off	NOUN
bracis-33605	6	21	.	.	PUNCT
bracis-33605	7	1	the	the	DET
bracis-33605	7	2	gubs	gubs	ADJ
bracis-33605	7	3	criterion	criterion	NOUN
bracis-33605	7	4	(	(	PUNCT
bracis-33605	7	5	goals	goal	NOUN
bracis-33605	7	6	with	with	ADP
bracis-33605	7	7	utility	utility	NOUN
bracis-33605	7	8	-	-	PUNCT
bracis-33605	7	9	based	base	VERB
bracis-33605	7	10	semantic	semantic	NOUN
bracis-33605	7	11	)	)	PUNCT
bracis-33605	7	12	considers	consider	VERB
bracis-33605	7	13	this	this	DET
bracis-33605	7	14	type	type	NOUN
bracis-33605	7	15	of	of	ADP
bracis-33605	7	16	trade	trade	NOUN
bracis-33605	7	17	-	-	PUNCT
bracis-33605	7	18	off	off	NOUN
bracis-33605	7	19	without	without	ADP
bracis-33605	7	20	using	use	VERB
bracis-33605	7	21	discounts	discount	NOUN
bracis-33605	7	22	or	or	CCONJ
bracis-33605	7	23	penalties	penalty	NOUN
bracis-33605	7	24	,	,	PUNCT
bracis-33605	7	25	has	have	VERB
bracis-33605	7	26	good	good	ADJ
bracis-33605	7	27	semantics	semantic	NOUN
bracis-33605	7	28	based	base	VERB
bracis-33605	7	29	on	on	ADP
bracis-33605	7	30	the	the	DET
bracis-33605	7	31	expected	expect	VERB
bracis-33605	7	32	utility	utility	NOUN
bracis-33605	7	33	theory	theory	NOUN
bracis-33605	7	34	and	and	CCONJ
bracis-33605	7	35	has	have	AUX
bracis-33605	7	36	been	be	AUX
bracis-33605	7	37	used	use	VERB
bracis-33605	7	38	to	to	PART
bracis-33605	7	39	solve	solve	VERB
bracis-33605	7	40	ssps	ssps	VERB
bracis-33605	7	41	with	with	ADP
bracis-33605	7	42	dead	dead	ADJ
bracis-33605	7	43	-	-	PUNCT
bracis-33605	7	44	end	end	NOUN
bracis-33605	7	45	states	state	NOUN
bracis-33605	7	46	in	in	ADP
bracis-33605	7	47	the	the	DET
bracis-33605	7	48	planning	planning	NOUN
bracis-33605	7	49	area	area	NOUN
bracis-33605	7	50	.	.	PUNCT
bracis-33605	8	1	thus	thus	ADV
bracis-33605	8	2	,	,	PUNCT
bracis-33605	8	3	in	in	ADP
bracis-33605	8	4	this	this	DET
bracis-33605	8	5	work	work	NOUN
bracis-33605	8	6	,	,	PUNCT
bracis-33605	8	7	the	the	DET
bracis-33605	8	8	gubs	gubs	ADJ
bracis-33605	8	9	criterion	criterion	NOUN
bracis-33605	8	10	is	be	AUX
bracis-33605	8	11	used	use	VERB
bracis-33605	8	12	to	to	PART
bracis-33605	8	13	propose	propose	VERB
bracis-33605	8	14	the	the	DET
bracis-33605	8	15	first	first	ADJ
bracis-33605	8	16	two	two	NUM
bracis-33605	8	17	algorithms	algorithm	NOUN
bracis-33605	8	18	for	for	SCONJ
bracis-33605	8	19	rl	rl	NOUN
bracis-33605	8	20	to	to	PART
bracis-33605	8	21	do	do	VERB
bracis-33605	8	22	a	a	DET
bracis-33605	8	23	trade	trade	NOUN
bracis-33605	8	24	-	-	PUNCT
bracis-33605	8	25	off	off	NOUN
bracis-33605	8	26	between	between	ADP
bracis-33605	8	27	probability	probability	NOUN
bracis-33605	8	28	to	to	ADP
bracis-33605	8	29	goal	goal	NOUN
bracis-33605	8	30	and	and	CCONJ
bracis-33605	8	31	cost	cost	NOUN
bracis-33605	8	32	to	to	ADP
bracis-33605	8	33	goal	goal	NOUN
bracis-33605	8	34	,	,	PUNCT
bracis-33605	8	35	without	without	ADP
bracis-33605	8	36	using	use	VERB
bracis-33605	8	37	discounts	discount	NOUN
bracis-33605	8	38	or	or	CCONJ
bracis-33605	8	39	penalties	penalty	NOUN
bracis-33605	8	40	:	:	PUNCT
bracis-33605	8	41	the	the	DET
bracis-33605	8	42	q	q	NOUN
bracis-33605	8	43	-	-	PUNCT
bracis-33605	8	44	learning	learning	NOUN
bracis-33605	8	45	-	-	PUNCT
bracis-33605	8	46	gubs	gub	NOUN
bracis-33605	8	47	and	and	CCONJ
bracis-33605	8	48	q	q	NOUN
bracis-33605	8	49	-	-	PUNCT
bracis-33605	8	50	learning	learning	NOUN
bracis-33605	8	51	-	-	PUNCT
bracis-33605	8	52	egubs+cmax	egubs+cmax	PROPN
bracis-33605	8	53	.	.	PUNCT
bracis-33605	9	1	theoretical	theoretical	ADJ
bracis-33605	9	2	and	and	CCONJ
bracis-33605	9	3	experimental	experimental	ADJ
bracis-33605	9	4	results	result	NOUN
bracis-33605	9	5	show	show	VERB
bracis-33605	9	6	that	that	SCONJ
bracis-33605	9	7	the	the	DET
bracis-33605	9	8	proposed	propose	VERB
bracis-33605	9	9	algorithms	algorithm	NOUN
bracis-33605	9	10	make	make	VERB
bracis-33605	9	11	this	this	DET
bracis-33605	9	12	trade	trade	NOUN
bracis-33605	9	13	-	-	PUNCT
bracis-33605	9	14	off	off	NOUN
bracis-33605	9	15	according	accord	VERB
bracis-33605	9	16	to	to	ADP
bracis-33605	9	17	the	the	DET
bracis-33605	9	18	configuration	configuration	NOUN
bracis-33605	9	19	of	of	ADP
bracis-33605	9	20	the	the	DET
bracis-33605	9	21	gubs	gub	NOUN
bracis-33605	9	22	’	'	PUNCT
bracis-33605	9	23	parameters	parameter	NOUN
bracis-33605	9	24	(	(	PUNCT
bracis-33605	9	25	code	code	PROPN
bracis-33605	9	26	enabled	enable	VERB
bracis-33605	9	27	on	on	ADP
bracis-33605	9	28	https://github.com/qlearninggubs/code	https://github.com/qlearninggubs/code	NOUN
bracis-33605	9	29	)	)	PUNCT
bracis-33605	9	30	.	.	PUNCT
bracis-33605	10	1	access	access	NOUN
bracis-33605	10	2	provided	provide	VERB
bracis-33605	10	3	by	by	ADP
bracis-33605	10	4	university	university	PROPN
bracis-33605	10	5	of	of	ADP
bracis-33605	10	6	notre	notre	PROPN
bracis-33605	10	7	dame	dame	PROPN
bracis-33605	10	8	hesburgh	hesburgh	PROPN
bracis-33605	10	9	library	library	PROPN
bracis-33605	10	10	.	.	PUNCT
bracis-33605	11	1	download	download	PROPN
bracis-33605	11	2	conference	conference	NOUN
bracis-33605	11	3	paper	paper	NOUN
bracis-33605	11	4	pdf	pdf	NOUN
bracis-33605	11	5	similar	similar	ADJ
bracis-33605	11	6	content	content	NOUN
bracis-33605	11	7	being	be	AUX
bracis-33605	11	8	viewed	view	VERB
bracis-33605	11	9	by	by	ADP
bracis-33605	11	10	others	other	NOUN
bracis-33605	11	11	$	$	SYM
bracis-33605	11	12	$	$	SYM
bracis-33605	11	13	\alpha	\alpha	ADP
bracis-33605	11	14	$	$	SYM
bracis-33605	11	15	$	$	SYM
bracis-33605	11	16	-mcmp	-mcmp	NOUN
bracis-33605	11	17	:	:	PUNCT
bracis-33605	11	18	trade	trade	NOUN
bracis-33605	11	19	-	-	PUNCT
bracis-33605	11	20	offs	off	NOUN
bracis-33605	11	21	between	between	ADP
bracis-33605	11	22	probability	probability	NOUN
bracis-33605	11	23	and	and	CCONJ
bracis-33605	11	24	 	 	SPACE
bracis-33605	11	25	cost	cost	NOUN
bracis-33605	11	26	in	in	ADP
bracis-33605	11	27	 	 	SPACE
bracis-33605	11	28	ssps	ssps	VERB
bracis-33605	11	29	with	with	ADP
bracis-33605	11	30	 	 	SPACE
bracis-33605	11	31	the	the	DET
bracis-33605	11	32	 	 	SPACE
bracis-33605	11	33	mcmp	mcmp	NOUN
bracis-33605	11	34	criterion	criterion	NOUN
bracis-33605	11	35	chapter	chapter	NOUN
bracis-33605	11	36	©	©	PROPN
bracis-33605	11	37	2023	2023	NUM
bracis-33605	11	38	trading	trading	NOUN
bracis-33605	11	39	utility	utility	NOUN
bracis-33605	11	40	and	and	CCONJ
bracis-33605	11	41	uncertainty	uncertainty	NOUN
bracis-33605	11	42	:	:	PUNCT
bracis-33605	11	43	applying	apply	VERB
bracis-33605	11	44	the	the	DET
bracis-33605	11	45	value	value	NOUN
bracis-33605	11	46	of	of	ADP
bracis-33605	11	47	information	information	NOUN
bracis-33605	11	48	to	to	PART
bracis-33605	11	49	resolve	resolve	VERB
bracis-33605	11	50	the	the	DET
bracis-33605	11	51	exploration	exploration	NOUN
bracis-33605	11	52	–	–	PUNCT
bracis-33605	11	53	exploitation	exploitation	NOUN
bracis-33605	11	54	dilemma	dilemma	NOUN
bracis-33605	11	55	in	in	ADP
bracis-33605	11	56	reinforcement	reinforcement	NOUN
bracis-33605	11	57	learning	learning	NOUN
bracis-33605	11	58	chapter	chapter	NOUN
bracis-33605	11	59	©	©	PROPN
bracis-33605	11	60	2021	2021	NUM
bracis-33605	11	61	a	a	DET
bracis-33605	11	62	linear	linear	ADJ
bracis-33605	11	63	online	online	ADV
bracis-33605	11	64	guided	guide	VERB
bracis-33605	11	65	policy	policy	NOUN
bracis-33605	11	66	search	search	NOUN
bracis-33605	11	67	algorithm	algorithm	NOUN
bracis-33605	11	68	chapter	chapter	NOUN
bracis-33605	11	69	©	©	PROPN
bracis-33605	11	70	2017	2017	NUM
bracis-33605	11	71	1	1	NUM
bracis-33605	11	72	introduction	introduction	NOUN
bracis-33605	11	73	stochastic	stochastic	NOUN
bracis-33605	11	74	shortest	short	ADJ
bracis-33605	11	75	path	path	NOUN
bracis-33605	11	76	problems	problem	NOUN
bracis-33605	11	77	(	(	PUNCT
bracis-33605	11	78	ssps	ssps	NOUN
bracis-33605	11	79	)	)	PUNCT
bracis-33605	11	80	are	be	AUX
bracis-33605	11	81	markov	markov	NOUN
bracis-33605	11	82	decision	decision	NOUN
bracis-33605	11	83	processes	process	NOUN
bracis-33605	11	84	(	(	PUNCT
bracis-33605	11	85	mdps	mdps	NOUN
bracis-33605	11	86	)	)	PUNCT
bracis-33605	11	87	with	with	ADP
bracis-33605	11	88	a	a	DET
bracis-33605	11	89	set	set	NOUN
bracis-33605	11	90	of	of	ADP
bracis-33605	11	91	goal	goal	NOUN
bracis-33605	11	92	states	state	NOUN
bracis-33605	11	93	.	.	PUNCT
bracis-33605	12	1	in	in	ADP
bracis-33605	12	2	these	these	DET
bracis-33605	12	3	problems	problem	NOUN
bracis-33605	12	4	,	,	PUNCT
bracis-33605	12	5	the	the	DET
bracis-33605	12	6	objective	objective	NOUN
bracis-33605	12	7	is	be	AUX
bracis-33605	12	8	to	to	PART
bracis-33605	12	9	find	find	VERB
bracis-33605	12	10	a	a	DET
bracis-33605	12	11	policy	policy	NOUN
bracis-33605	12	12	with	with	ADP
bracis-33605	12	13	the	the	DET
bracis-33605	12	14	lowest	low	ADJ
bracis-33605	12	15	expected	expect	VERB
bracis-33605	12	16	cumulative	cumulative	ADJ
bracis-33605	12	17	cost	cost	NOUN
bracis-33605	12	18	to	to	PART
bracis-33605	12	19	achieve	achieve	VERB
bracis-33605	12	20	these	these	DET
bracis-33605	12	21	goals	goal	NOUN
bracis-33605	12	22	.	.	PUNCT
bracis-33605	13	1	in	in	ADP
bracis-33605	13	2	many	many	ADJ
bracis-33605	13	3	cases	case	NOUN
bracis-33605	13	4	,	,	PUNCT
bracis-33605	13	5	these	these	DET
bracis-33605	13	6	problems	problem	NOUN
bracis-33605	13	7	 	 	SPACE
bracis-33605	13	8	have	have	VERB
bracis-33605	13	9	states	state	NOUN
bracis-33605	13	10	from	from	ADP
bracis-33605	13	11	which	which	PRON
bracis-33605	13	12	it	it	PRON
bracis-33605	13	13	is	be	AUX
bracis-33605	13	14	not	not	PART
bracis-33605	13	15	possible	possible	ADJ
bracis-33605	13	16	to	to	PART
bracis-33605	13	17	reach	reach	VERB
bracis-33605	13	18	the	the	DET
bracis-33605	13	19	goal	goal	NOUN
bracis-33605	13	20	,	,	PUNCT
bracis-33605	13	21	these	these	DET
bracis-33605	13	22	states	state	NOUN
bracis-33605	13	23	are	be	AUX
bracis-33605	13	24	called	call	VERB
bracis-33605	13	25	dead	dead	ADJ
bracis-33605	13	26	-	-	PUNCT
bracis-33605	13	27	ends	end	NOUN
bracis-33605	13	28	.	.	PUNCT
bracis-33605	14	1	in	in	ADP
bracis-33605	14	2	problems	problem	NOUN
bracis-33605	14	3	with	with	ADP
bracis-33605	14	4	dead	dead	ADJ
bracis-33605	14	5	-	-	PUNCT
bracis-33605	14	6	ends	end	VERB
bracis-33605	14	7	the	the	DET
bracis-33605	14	8	criterion	criterion	NOUN
bracis-33605	14	9	of	of	ADP
bracis-33605	14	10	the	the	DET
bracis-33605	14	11	lowest	low	ADJ
bracis-33605	14	12	expected	expect	VERB
bracis-33605	14	13	cost	cost	NOUN
bracis-33605	14	14	becomes	become	VERB
bracis-33605	14	15	ill	ill	ADV
bracis-33605	14	16	-	-	PUNCT
bracis-33605	14	17	defined	define	VERB
bracis-33605	14	18	when	when	SCONJ
bracis-33605	14	19	there	there	PRON
bracis-33605	14	20	is	be	VERB
bracis-33605	14	21	no	no	DET
bracis-33605	14	22	a	a	DET
bracis-33605	14	23	proper	proper	ADJ
bracis-33605	14	24	policy	policy	NOUN
bracis-33605	14	25	(	(	PUNCT
bracis-33605	14	26	one	one	NOUN
bracis-33605	14	27	that	that	PRON
bracis-33605	14	28	reaches	reach	VERB
bracis-33605	14	29	the	the	DET
bracis-33605	14	30	goal	goal	NOUN
bracis-33605	14	31	with	with	ADP
bracis-33605	14	32	probability	probability	NOUN
bracis-33605	14	33	1	1	NUM
bracis-33605	14	34	)	)	PUNCT
bracis-33605	15	1	[	[	X
bracis-33605	15	2	6	6	NUM
bracis-33605	15	3	]	]	PUNCT
bracis-33605	15	4	.	.	PUNCT
bracis-33605	16	1	some	some	DET
bracis-33605	16	2	works	work	NOUN
bracis-33605	16	3	try	try	VERB
bracis-33605	16	4	to	to	PART
bracis-33605	16	5	solve	solve	VERB
bracis-33605	16	6	ssps	ssps	VERB
bracis-33605	16	7	with	with	ADP
bracis-33605	16	8	dead	dead	ADJ
bracis-33605	16	9	-	-	PUNCT
bracis-33605	16	10	end	end	NOUN
bracis-33605	16	11	,	,	PUNCT
bracis-33605	16	12	some	some	PRON
bracis-33605	16	13	of	of	ADP
bracis-33605	16	14	them	they	PRON
bracis-33605	16	15	seek	seek	VERB
bracis-33605	16	16	to	to	PART
bracis-33605	16	17	find	find	VERB
bracis-33605	16	18	the	the	DET
bracis-33605	16	19	policy	policy	NOUN
bracis-33605	16	20	with	with	ADP
bracis-33605	16	21	the	the	DET
bracis-33605	16	22	maximum	maximum	ADJ
bracis-33605	16	23	probability	probability	NOUN
bracis-33605	16	24	of	of	ADP
bracis-33605	16	25	reaching	reach	VERB
bracis-33605	16	26	the	the	DET
bracis-33605	16	27	goal	goal	NOUN
bracis-33605	16	28	(	(	PUNCT
bracis-33605	16	29	maxprob	maxprob	ADV
bracis-33605	16	30	)	)	PUNCT
bracis-33605	17	1	[	[	X
bracis-33605	17	2	7	7	X
bracis-33605	17	3	]	]	PUNCT
bracis-33605	17	4	or	or	CCONJ
bracis-33605	17	5	give	give	VERB
bracis-33605	17	6	greater	great	ADJ
bracis-33605	17	7	preference	preference	NOUN
bracis-33605	17	8	to	to	ADP
bracis-33605	17	9	probability	probability	NOUN
bracis-33605	17	10	and	and	CCONJ
bracis-33605	17	11	consider	consider	VERB
bracis-33605	17	12	cost	cost	NOUN
bracis-33605	17	13	as	as	ADP
bracis-33605	17	14	a	a	DET
bracis-33605	17	15	secondary	secondary	ADJ
bracis-33605	17	16	aspect	aspect	NOUN
bracis-33605	17	17	(	(	PUNCT
bracis-33605	17	18	lexicographic	lexicographic	ADJ
bracis-33605	17	19	criteria	criterion	NOUN
bracis-33605	17	20	)	)	PUNCT
bracis-33605	18	1	[	[	X
bracis-33605	18	2	6	6	NUM
bracis-33605	18	3	,	,	PUNCT
bracis-33605	18	4	14	14	NUM
bracis-33605	18	5	]	]	PUNCT
bracis-33605	18	6	.	.	PUNCT
bracis-33605	19	1	however	however	ADV
bracis-33605	19	2	,	,	PUNCT
bracis-33605	19	3	it	it	PRON
bracis-33605	19	4	is	be	AUX
bracis-33605	19	5	not	not	PART
bracis-33605	19	6	always	always	ADV
bracis-33605	19	7	recommended	recommend	VERB
bracis-33605	19	8	to	to	PART
bracis-33605	19	9	find	find	VERB
bracis-33605	19	10	policies	policy	NOUN
bracis-33605	19	11	with	with	ADP
bracis-33605	19	12	the	the	DET
bracis-33605	19	13	highest	high	ADJ
bracis-33605	19	14	probability	probability	NOUN
bracis-33605	19	15	of	of	ADP
bracis-33605	19	16	achieving	achieve	VERB
bracis-33605	19	17	the	the	DET
bracis-33605	19	18	goal	goal	NOUN
bracis-33605	19	19	as	as	SCONJ
bracis-33605	19	20	they	they	PRON
bracis-33605	19	21	may	may	AUX
bracis-33605	19	22	have	have	VERB
bracis-33605	19	23	high	high	ADJ
bracis-33605	19	24	costs	cost	NOUN
bracis-33605	19	25	.	.	PUNCT
bracis-33605	20	1	sometimes	sometimes	ADV
bracis-33605	20	2	it	it	PRON
bracis-33605	20	3	is	be	AUX
bracis-33605	20	4	better	well	ADJ
bracis-33605	20	5	to	to	PART
bracis-33605	20	6	make	make	VERB
bracis-33605	20	7	a	a	DET
bracis-33605	20	8	trade	trade	NOUN
bracis-33605	20	9	-	-	PUNCT
bracis-33605	20	10	off	off	NOUN
bracis-33605	20	11	between	between	ADP
bracis-33605	20	12	probability	probability	NOUN
bracis-33605	20	13	and	and	CCONJ
bracis-33605	20	14	cost	cost	NOUN
bracis-33605	20	15	,	,	PUNCT
bracis-33605	20	16	for	for	ADP
bracis-33605	20	17	example	example	NOUN
bracis-33605	20	18	,	,	PUNCT
bracis-33605	20	19	decreasing	decrease	VERB
bracis-33605	20	20	the	the	DET
bracis-33605	20	21	probability	probability	NOUN
bracis-33605	20	22	a	a	DET
bracis-33605	20	23	little	little	ADJ
bracis-33605	20	24	to	to	PART
bracis-33605	20	25	get	get	VERB
bracis-33605	20	26	a	a	DET
bracis-33605	20	27	large	large	ADJ
bracis-33605	20	28	decrease	decrease	NOUN
bracis-33605	20	29	in	in	ADP
bracis-33605	20	30	cost	cost	NOUN
bracis-33605	20	31	or	or	CCONJ
bracis-33605	20	32	increasing	increase	VERB
bracis-33605	20	33	the	the	DET
bracis-33605	20	34	cost	cost	NOUN
bracis-33605	20	35	a	a	DET
bracis-33605	20	36	little	little	ADJ
bracis-33605	20	37	to	to	PART
bracis-33605	20	38	get	get	VERB
bracis-33605	20	39	a	a	DET
bracis-33605	20	40	large	large	ADJ
bracis-33605	20	41	increase	increase	NOUN
bracis-33605	20	42	in	in	ADP
bracis-33605	20	43	probability	probability	NOUN
bracis-33605	20	44	,	,	PUNCT
bracis-33605	20	45	this	this	DET
bracis-33605	20	46	characteristic	characteristic	NOUN
bracis-33605	20	47	is	be	AUX
bracis-33605	20	48	called	call	VERB
bracis-33605	20	49	infinite	infinite	ADJ
bracis-33605	20	50	-	-	PUNCT
bracis-33605	20	51	infinitesimal	infinitesimal	ADJ
bracis-33605	20	52	trade	trade	NOUN
bracis-33605	20	53	-	-	PUNCT
bracis-33605	20	54	off	off	NOUN
bracis-33605	20	55	[	[	X
bracis-33605	20	56	1	1	NUM
bracis-33605	20	57	]	]	PUNCT
bracis-33605	20	58	.	.	PUNCT
bracis-33605	21	1	an	an	DET
bracis-33605	21	2	example	example	NOUN
bracis-33605	21	3	of	of	ADP
bracis-33605	21	4	this	this	DET
bracis-33605	21	5	type	type	NOUN
bracis-33605	21	6	of	of	ADP
bracis-33605	21	7	infinite	infinite	ADJ
bracis-33605	21	8	-	-	PUNCT
bracis-33605	21	9	infinitesimal	infinitesimal	ADJ
bracis-33605	21	10	trade	trade	NOUN
bracis-33605	21	11	-	-	PUNCT
bracis-33605	21	12	off	off	NOUN
bracis-33605	21	13	is	be	AUX
bracis-33605	21	14	illustrated	illustrate	VERB
bracis-33605	21	15	in	in	ADP
bracis-33605	21	16	the	the	DET
bracis-33605	21	17	river	river	NOUN
bracis-33605	21	18	problem	problem	NOUN
bracis-33605	21	19	.	.	PUNCT
bracis-33605	22	1	the	the	DET
bracis-33605	22	2	river	river	NOUN
bracis-33605	22	3	problem	problem	NOUN
bracis-33605	22	4	[	[	X
bracis-33605	22	5	4	4	X
bracis-33605	22	6	]	]	PUNCT
bracis-33605	22	7	considers	consider	VERB
bracis-33605	22	8	a	a	DET
bracis-33605	22	9	\(n_x	\(n_x	NOUN
bracis-33605	22	10	\times	\times	ADP
bracis-33605	22	11	n_y\	n_y\	NOUN
bracis-33605	22	12	)	)	PUNCT
bracis-33605	22	13	grid	grid	NOUN
bracis-33605	22	14	.	.	PUNCT
bracis-33605	23	1	the	the	DET
bracis-33605	23	2	agent	agent	NOUN
bracis-33605	23	3	must	must	AUX
bracis-33605	23	4	cross	cross	VERB
bracis-33605	23	5	a	a	DET
bracis-33605	23	6	river	river	NOUN
bracis-33605	23	7	from	from	ADP
bracis-33605	23	8	one	one	NUM
bracis-33605	23	9	river	river	NOUN
bracis-33605	23	10	bank	bank	NOUN
bracis-33605	23	11	starting	start	VERB
bracis-33605	23	12	from	from	ADP
bracis-33605	23	13	an	an	DET
bracis-33605	23	14	initial	initial	ADJ
bracis-33605	23	15	position	position	NOUN
bracis-33605	23	16	\(s_0\	\(s_0\	NOUN
bracis-33605	23	17	)	)	PUNCT
bracis-33605	23	18	to	to	ADP
bracis-33605	23	19	a	a	DET
bracis-33605	23	20	destination	destination	NOUN
bracis-33605	23	21	position	position	NOUN
bracis-33605	23	22	on	on	ADP
bracis-33605	23	23	the	the	DET
bracis-33605	23	24	other	other	ADJ
bracis-33605	23	25	bank	bank	NOUN
bracis-33605	23	26	.	.	PUNCT
bracis-33605	24	1	there	there	PRON
bracis-33605	24	2	is	be	VERB
bracis-33605	24	3	a	a	DET
bracis-33605	24	4	bridge	bridge	NOUN
bracis-33605	24	5	at	at	ADP
bracis-33605	24	6	position	position	NOUN
bracis-33605	24	7	\(y	\(y	PRON
bracis-33605	24	8	=	=	SYM
bracis-33605	24	9	n_y\	n_y\	NOUN
bracis-33605	24	10	)	)	PUNCT
bracis-33605	24	11	.	.	PUNCT
bracis-33605	25	1	the	the	DET
bracis-33605	25	2	agent	agent	NOUN
bracis-33605	25	3	’s	’s	PART
bracis-33605	25	4	actions	action	NOUN
bracis-33605	25	5	are	be	AUX
bracis-33605	25	6	to	to	PART
bracis-33605	25	7	move	move	VERB
bracis-33605	25	8	in	in	ADP
bracis-33605	25	9	the	the	DET
bracis-33605	25	10	four	four	NUM
bracis-33605	25	11	cardinal	cardinal	ADJ
bracis-33605	25	12	directions	direction	NOUN
bracis-33605	25	13	.	.	PUNCT
bracis-33605	26	1	the	the	DET
bracis-33605	26	2	actions	action	NOUN
bracis-33605	26	3	on	on	ADP
bracis-33605	26	4	the	the	DET
bracis-33605	26	5	river	river	NOUN
bracis-33605	26	6	’s	’s	PART
bracis-33605	26	7	bank	bank	NOUN
bracis-33605	26	8	(	(	PUNCT
bracis-33605	26	9	\(x	\(x	NOUN
bracis-33605	26	10	=	=	SYM
bracis-33605	26	11	1\	1\	NUM
bracis-33605	26	12	)	)	PUNCT
bracis-33605	26	13	or	or	CCONJ
bracis-33605	26	14	\(x	\(x	NOUN
bracis-33605	26	15	=	=	SYM
bracis-33605	26	16	n_x\	n_x\	PROPN
bracis-33605	26	17	)	)	PUNCT
bracis-33605	26	18	)	)	PUNCT
bracis-33605	26	19	and	and	CCONJ
bracis-33605	26	20	the	the	DET
bracis-33605	26	21	bridge	bridge	NOUN
bracis-33605	26	22	are	be	AUX
bracis-33605	26	23	deterministic	deterministic	ADJ
bracis-33605	26	24	.	.	PUNCT
bracis-33605	27	1	the	the	DET
bracis-33605	27	2	actions	action	NOUN
bracis-33605	27	3	are	be	AUX
bracis-33605	27	4	probabilistic	probabilistic	ADJ
bracis-33605	27	5	within	within	ADP
bracis-33605	27	6	the	the	DET
bracis-33605	27	7	river	river	NOUN
bracis-33605	27	8	.	.	PUNCT
bracis-33605	28	1	in	in	ADP
bracis-33605	28	2	the	the	DET
bracis-33605	28	3	river	river	NOUN
bracis-33605	28	4	,	,	PUNCT
bracis-33605	28	5	there	there	PRON
bracis-33605	28	6	is	be	VERB
bracis-33605	28	7	a	a	DET
bracis-33605	28	8	probability	probability	NOUN
bracis-33605	28	9	p	p	NOUN
bracis-33605	28	10	of	of	ADP
bracis-33605	28	11	following	follow	VERB
bracis-33605	28	12	down	down	ADP
bracis-33605	28	13	the	the	DET
bracis-33605	28	14	river	river	NOUN
bracis-33605	28	15	,	,	PUNCT
bracis-33605	28	16	with	with	ADP
bracis-33605	28	17	the	the	DET
bracis-33605	28	18	risk	risk	NOUN
bracis-33605	28	19	of	of	ADP
bracis-33605	28	20	reaching	reach	VERB
bracis-33605	28	21	the	the	DET
bracis-33605	28	22	waterfall	waterfall	NOUN
bracis-33605	28	23	located	locate	VERB
bracis-33605	28	24	at	at	ADP
bracis-33605	28	25	\(y	\(y	NOUN
bracis-33605	28	26	=	=	SYM
bracis-33605	28	27	1\	1\	NUM
bracis-33605	28	28	)	)	PUNCT
bracis-33605	28	29	,	,	PUNCT
bracis-33605	28	30	where	where	SCONJ
bracis-33605	28	31	,	,	PUNCT
bracis-33605	28	32	the	the	DET
bracis-33605	28	33	agent	agent	NOUN
bracis-33605	28	34	becomes	become	VERB
bracis-33605	28	35	trapped	trap	VERB
bracis-33605	28	36	or	or	CCONJ
bracis-33605	28	37	killed	kill	VERB
bracis-33605	28	38	(	(	PUNCT
bracis-33605	28	39	dead	dead	ADJ
bracis-33605	28	40	-	-	PUNCT
bracis-33605	28	41	end	end	NOUN
bracis-33605	28	42	)	)	PUNCT
bracis-33605	28	43	.	.	PUNCT
bracis-33605	29	1	the	the	DET
bracis-33605	29	2	optimal	optimal	ADJ
bracis-33605	29	3	solution	solution	NOUN
bracis-33605	29	4	for	for	ADP
bracis-33605	29	5	the	the	DET
bracis-33605	29	6	lexicographical	lexicographical	ADJ
bracis-33605	29	7	criteria	criterion	NOUN
bracis-33605	29	8	is	be	AUX
bracis-33605	29	9	to	to	PART
bracis-33605	29	10	walk	walk	VERB
bracis-33605	29	11	toward	toward	ADP
bracis-33605	29	12	the	the	DET
bracis-33605	29	13	bridge	bridge	NOUN
bracis-33605	29	14	and	and	CCONJ
bracis-33605	29	15	traverse	traverse	VERB
bracis-33605	29	16	it	it	PRON
bracis-33605	29	17	since	since	SCONJ
bracis-33605	29	18	this	this	DET
bracis-33605	29	19	policy	policy	NOUN
bracis-33605	29	20	has	have	VERB
bracis-33605	29	21	a	a	DET
bracis-33605	29	22	probability	probability	NOUN
bracis-33605	29	23	1	1	NUM
bracis-33605	29	24	to	to	PART
bracis-33605	29	25	reach	reach	VERB
bracis-33605	29	26	the	the	DET
bracis-33605	29	27	goal	goal	NOUN
bracis-33605	29	28	,	,	PUNCT
bracis-33605	29	29	even	even	ADV
bracis-33605	29	30	if	if	SCONJ
bracis-33605	29	31	the	the	DET
bracis-33605	29	32	cost	cost	NOUN
bracis-33605	29	33	is	be	AUX
bracis-33605	29	34	very	very	ADV
bracis-33605	29	35	high	high	ADJ
bracis-33605	29	36	.	.	PUNCT
bracis-33605	30	1	in	in	ADP
bracis-33605	30	2	this	this	DET
bracis-33605	30	3	problem	problem	NOUN
bracis-33605	30	4	,	,	PUNCT
bracis-33605	30	5	for	for	ADP
bracis-33605	30	6	example	example	NOUN
bracis-33605	30	7	,	,	PUNCT
bracis-33605	30	8	a	a	DET
bracis-33605	30	9	solution	solution	NOUN
bracis-33605	30	10	with	with	ADP
bracis-33605	30	11	a	a	DET
bracis-33605	30	12	probability	probability	NOUN
bracis-33605	30	13	0.9999	0.9999	PRON
bracis-33605	30	14	should	should	AUX
bracis-33605	30	15	be	be	AUX
bracis-33605	30	16	accepted	accept	VERB
bracis-33605	30	17	if	if	SCONJ
bracis-33605	30	18	we	we	PRON
bracis-33605	30	19	get	get	VERB
bracis-33605	30	20	a	a	DET
bracis-33605	30	21	large	large	ADJ
bracis-33605	30	22	decrease	decrease	NOUN
bracis-33605	30	23	in	in	ADP
bracis-33605	30	24	cost	cost	NOUN
bracis-33605	30	25	.	.	PUNCT
bracis-33605	31	1	in	in	ADP
bracis-33605	31	2	reinforcement	reinforcement	NOUN
bracis-33605	31	3	learning	learning	NOUN
bracis-33605	31	4	(	(	PUNCT
bracis-33605	31	5	rl	rl	NOUN
bracis-33605	31	6	)	)	PUNCT
bracis-33605	31	7	,	,	PUNCT
bracis-33605	31	8	most	most	ADJ
bracis-33605	31	9	of	of	ADP
bracis-33605	31	10	the	the	DET
bracis-33605	31	11	theoretical	theoretical	ADJ
bracis-33605	31	12	works	work	NOUN
bracis-33605	31	13	focus	focus	VERB
bracis-33605	31	14	on	on	ADP
bracis-33605	31	15	finite	finite	NOUN
bracis-33605	31	16	-	-	NOUN
bracis-33605	31	17	horizon	horizon	NOUN
bracis-33605	31	18	and	and	CCONJ
bracis-33605	31	19	infinite	infinite	NOUN
bracis-33605	31	20	-	-	PUNCT
bracis-33605	31	21	horizon	horizon	NOUN
bracis-33605	31	22	or	or	CCONJ
bracis-33605	31	23	on	on	ADP
bracis-33605	31	24	loop	loop	NOUN
bracis-33605	31	25	-	-	PUNCT
bracis-33605	31	26	free	free	ADJ
bracis-33605	31	27	ssp	ssp	NOUN
bracis-33605	31	28	[	[	X
bracis-33605	31	29	12	12	NUM
bracis-33605	31	30	]	]	PUNCT
bracis-33605	31	31	.	.	PUNCT
bracis-33605	32	1	also	also	ADV
bracis-33605	32	2	,	,	PUNCT
bracis-33605	32	3	the	the	DET
bracis-33605	32	4	problem	problem	NOUN
bracis-33605	32	5	of	of	ADP
bracis-33605	32	6	making	make	VERB
bracis-33605	32	7	trade	trade	NOUN
bracis-33605	32	8	-	-	PUNCT
bracis-33605	32	9	off	off	NOUN
bracis-33605	32	10	has	have	AUX
bracis-33605	32	11	been	be	AUX
bracis-33605	32	12	little	little	ADJ
bracis-33605	32	13	studied	study	VERB
bracis-33605	32	14	for	for	ADP
bracis-33605	32	15	ssps	ssps	NOUN
bracis-33605	32	16	with	with	ADP
bracis-33605	32	17	dead	dead	ADJ
bracis-33605	32	18	-	-	PUNCT
bracis-33605	32	19	ends	end	NOUN
bracis-33605	32	20	.	.	PUNCT
bracis-33605	33	1	additionally	additionally	ADV
bracis-33605	33	2	,	,	PUNCT
bracis-33605	33	3	a	a	DET
bracis-33605	33	4	common	common	ADJ
bracis-33605	33	5	strategy	strategy	NOUN
bracis-33605	33	6	to	to	PART
bracis-33605	33	7	follow	follow	VERB
bracis-33605	33	8	with	with	ADP
bracis-33605	33	9	ssps	ssps	NOUN
bracis-33605	33	10	with	with	ADP
bracis-33605	33	11	dead	dead	ADJ
bracis-33605	33	12	-	-	PUNCT
bracis-33605	33	13	end	end	NOUN
bracis-33605	33	14	states	state	NOUN
bracis-33605	33	15	in	in	ADP
bracis-33605	33	16	rl	rl	PROPN
bracis-33605	33	17	is	be	AUX
bracis-33605	33	18	using	use	VERB
bracis-33605	33	19	discounts	discount	NOUN
bracis-33605	33	20	.	.	PUNCT
bracis-33605	34	1	geibel	geibel	NOUN
bracis-33605	34	2	and	and	CCONJ
bracis-33605	34	3	wysotzki	wysotzki	ADJ
bracis-33605	34	4	[	[	PUNCT
bracis-33605	34	5	5	5	NUM
bracis-33605	34	6	]	]	PUNCT
bracis-33605	34	7	consider	consider	VERB
bracis-33605	34	8	an	an	DET
bracis-33605	34	9	mdp	mdp	NOUN
bracis-33605	34	10	with	with	ADP
bracis-33605	34	11	goal	goal	NOUN
bracis-33605	34	12	and	and	CCONJ
bracis-33605	34	13	error	error	NOUN
bracis-33605	34	14	states	state	NOUN
bracis-33605	34	15	(	(	PUNCT
bracis-33605	34	16	undesirable	undesirable	ADJ
bracis-33605	34	17	or	or	CCONJ
bracis-33605	34	18	dangerous	dangerous	ADJ
bracis-33605	34	19	states	state	NOUN
bracis-33605	34	20	)	)	PUNCT
bracis-33605	34	21	.	.	PUNCT
bracis-33605	35	1	the	the	DET
bracis-33605	35	2	proposed	propose	VERB
bracis-33605	35	3	rl	rl	NOUN
bracis-33605	35	4	solution	solution	NOUN
bracis-33605	35	5	is	be	AUX
bracis-33605	35	6	based	base	VERB
bracis-33605	35	7	on	on	ADP
bracis-33605	35	8	weighting	weight	VERB
bracis-33605	35	9	the	the	DET
bracis-33605	35	10	cost	cost	NOUN
bracis-33605	35	11	and	and	CCONJ
bracis-33605	35	12	the	the	DET
bracis-33605	35	13	probability	probability	NOUN
bracis-33605	35	14	of	of	ADP
bracis-33605	35	15	entering	enter	VERB
bracis-33605	35	16	error	error	NOUN
bracis-33605	35	17	states	state	NOUN
bracis-33605	35	18	.	.	PUNCT
bracis-33605	36	1	this	this	DET
bracis-33605	36	2	work	work	NOUN
bracis-33605	36	3	uses	use	VERB
bracis-33605	36	4	discount	discount	NOUN
bracis-33605	36	5	factor	factor	NOUN
bracis-33605	36	6	.	.	PUNCT
bracis-33605	37	1	park	park	NOUN
bracis-33605	38	1	[	[	X
bracis-33605	38	2	8	8	NUM
bracis-33605	38	3	]	]	PUNCT
bracis-33605	38	4	considers	consider	VERB
bracis-33605	38	5	an	an	DET
bracis-33605	38	6	mdp	mdp	NOUN
bracis-33605	38	7	with	with	ADP
bracis-33605	38	8	goal	goal	NOUN
bracis-33605	38	9	and	and	CCONJ
bracis-33605	38	10	fail	fail	VERB
bracis-33605	38	11	states	state	NOUN
bracis-33605	38	12	.	.	PUNCT
bracis-33605	39	1	the	the	DET
bracis-33605	39	2	proposed	propose	VERB
bracis-33605	39	3	q	q	ADJ
bracis-33605	39	4	-	-	PUNCT
bracis-33605	39	5	learning	learning	NOUN
bracis-33605	39	6	-	-	PUNCT
bracis-33605	39	7	like	like	ADJ
bracis-33605	39	8	algorithm	algorithm	NOUN
bracis-33605	39	9	records	record	NOUN
bracis-33605	39	10	past	past	ADP
bracis-33605	39	11	failure	failure	NOUN
bracis-33605	39	12	experiences	experience	NOUN
bracis-33605	39	13	and	and	CCONJ
bracis-33605	39	14	uses	use	VERB
bracis-33605	39	15	them	they	PRON
bracis-33605	39	16	to	to	PART
bracis-33605	39	17	guide	guide	VERB
bracis-33605	39	18	the	the	DET
bracis-33605	39	19	exploration	exploration	NOUN
bracis-33605	39	20	,	,	PUNCT
bracis-33605	39	21	this	this	DET
bracis-33605	39	22	work	work	NOUN
bracis-33605	39	23	[	[	X
bracis-33605	39	24	8	8	NUM
bracis-33605	39	25	]	]	PUNCT
bracis-33605	39	26	also	also	ADV
bracis-33605	39	27	uses	use	VERB
bracis-33605	39	28	a	a	DET
bracis-33605	39	29	discount	discount	NOUN
bracis-33605	39	30	factor	factor	NOUN
bracis-33605	39	31	.	.	PUNCT
bracis-33605	40	1	using	use	VERB
bracis-33605	40	2	a	a	DET
bracis-33605	40	3	discount	discount	NOUN
bracis-33605	40	4	factor	factor	NOUN
bracis-33605	40	5	could	could	AUX
bracis-33605	40	6	have	have	VERB
bracis-33605	40	7	an	an	DET
bracis-33605	40	8	undesirable	undesirable	ADJ
bracis-33605	40	9	effect	effect	NOUN
bracis-33605	40	10	of	of	ADP
bracis-33605	40	11	biasing	bias	VERB
bracis-33605	40	12	importance	importance	NOUN
bracis-33605	40	13	towards	towards	ADP
bracis-33605	40	14	short	short	ADJ
bracis-33605	40	15	-	-	PUNCT
bracis-33605	40	16	term	term	NOUN
bracis-33605	40	17	behavior	behavior	NOUN
bracis-33605	40	18	and	and	CCONJ
bracis-33605	40	19	thus	thus	ADV
bracis-33605	40	20	diminishing	diminish	VERB
bracis-33605	40	21	the	the	DET
bracis-33605	40	22	incentive	incentive	NOUN
bracis-33605	40	23	to	to	PART
bracis-33605	40	24	eventually	eventually	ADV
bracis-33605	40	25	achieve	achieve	VERB
bracis-33605	40	26	the	the	DET
bracis-33605	40	27	goal	goal	NOUN
bracis-33605	40	28	state	state	NOUN
bracis-33605	41	1	[	[	X
bracis-33605	41	2	1	1	NUM
bracis-33605	41	3	,	,	PUNCT
bracis-33605	41	4	12	12	NUM
bracis-33605	41	5	]	]	PUNCT
bracis-33605	41	6	.	.	PUNCT
bracis-33605	42	1	in	in	ADP
bracis-33605	42	2	some	some	DET
bracis-33605	42	3	works	work	NOUN
bracis-33605	42	4	[	[	X
bracis-33605	42	5	12	12	NUM
bracis-33605	42	6	]	]	PUNCT
bracis-33605	42	7	,	,	PUNCT
bracis-33605	42	8	penalties	penalty	NOUN
bracis-33605	42	9	are	be	AUX
bracis-33605	42	10	used	use	VERB
bracis-33605	42	11	in	in	ADP
bracis-33605	42	12	the	the	DET
bracis-33605	42	13	exploration	exploration	NOUN
bracis-33605	42	14	problem	problem	NOUN
bracis-33605	42	15	for	for	ADP
bracis-33605	42	16	ssps	ssps	NOUN
bracis-33605	42	17	with	with	ADP
bracis-33605	42	18	dead	dead	ADJ
bracis-33605	42	19	-	-	PUNCT
bracis-33605	42	20	ends	end	NOUN
bracis-33605	42	21	.	.	PUNCT
bracis-33605	43	1	however	however	ADV
bracis-33605	43	2	,	,	PUNCT
bracis-33605	43	3	using	use	VERB
bracis-33605	43	4	penalties	penalty	NOUN
bracis-33605	43	5	can	can	AUX
bracis-33605	43	6	lead	lead	VERB
bracis-33605	43	7	to	to	ADP
bracis-33605	43	8	errors	error	NOUN
bracis-33605	43	9	in	in	ADP
bracis-33605	43	10	finding	find	VERB
bracis-33605	43	11	a	a	DET
bracis-33605	43	12	policy	policy	NOUN
bracis-33605	43	13	with	with	ADP
bracis-33605	43	14	a	a	DET
bracis-33605	43	15	desired	desire	VERB
bracis-33605	43	16	trade	trade	NOUN
bracis-33605	43	17	-	-	PUNCT
bracis-33605	43	18	off	off	NOUN
bracis-33605	43	19	too	too	ADV
bracis-33605	43	20	,	,	PUNCT
bracis-33605	43	21	as	as	SCONJ
bracis-33605	43	22	was	be	AUX
bracis-33605	43	23	formally	formally	ADV
bracis-33605	43	24	demonstrated	demonstrate	VERB
bracis-33605	43	25	by	by	ADP
bracis-33605	43	26	[	[	X
bracis-33605	43	27	1	1	NUM
bracis-33605	43	28	]	]	PUNCT
bracis-33605	43	29	.	.	PUNCT
bracis-33605	44	1	the	the	DET
bracis-33605	44	2	gubs	gubs	ADJ
bracis-33605	44	3	criterion	criterion	NOUN
bracis-33605	44	4	(	(	PUNCT
bracis-33605	44	5	goals	goal	NOUN
bracis-33605	44	6	with	with	ADP
bracis-33605	44	7	utility	utility	NOUN
bracis-33605	44	8	-	-	PUNCT
bracis-33605	44	9	based	base	VERB
bracis-33605	44	10	semantic	semantic	NOUN
bracis-33605	44	11	)	)	PUNCT
bracis-33605	45	1	[	[	X
bracis-33605	45	2	4	4	X
bracis-33605	45	3	]	]	PUNCT
bracis-33605	45	4	was	be	AUX
bracis-33605	45	5	proposed	propose	VERB
bracis-33605	45	6	to	to	PART
bracis-33605	45	7	make	make	VERB
bracis-33605	45	8	a	a	DET
bracis-33605	45	9	correct	correct	ADJ
bracis-33605	45	10	infinite	infinite	ADJ
bracis-33605	45	11	-	-	PUNCT
bracis-33605	45	12	infinitesimal	infinitesimal	ADJ
bracis-33605	45	13	trade	trade	NOUN
bracis-33605	45	14	-	-	PUNCT
bracis-33605	45	15	off	off	NOUN
bracis-33605	45	16	in	in	ADP
bracis-33605	45	17	planning	planning	NOUN
bracis-33605	45	18	problems	problem	NOUN
bracis-33605	45	19	.	.	PUNCT
bracis-33605	46	1	this	this	DET
bracis-33605	46	2	criterion	criterion	NOUN
bracis-33605	46	3	allows	allow	VERB
bracis-33605	46	4	us	we	PRON
bracis-33605	46	5	to	to	PART
bracis-33605	46	6	find	find	VERB
bracis-33605	46	7	solutions	solution	NOUN
bracis-33605	46	8	for	for	ADP
bracis-33605	46	9	ssps	ssps	NOUN
bracis-33605	46	10	with	with	ADP
bracis-33605	46	11	dead	dead	ADJ
bracis-33605	46	12	-	-	PUNCT
bracis-33605	46	13	end	end	NOUN
bracis-33605	46	14	,	,	PUNCT
bracis-33605	46	15	with	with	ADP
bracis-33605	46	16	a	a	DET
bracis-33605	46	17	parameterization	parameterization	NOUN
bracis-33605	46	18	with	with	ADP
bracis-33605	46	19	meaning	meaning	NOUN
bracis-33605	46	20	based	base	VERB
bracis-33605	46	21	on	on	ADP
bracis-33605	46	22	the	the	DET
bracis-33605	46	23	expected	expect	VERB
bracis-33605	46	24	utility	utility	NOUN
bracis-33605	46	25	theory	theory	NOUN
bracis-33605	46	26	.	.	PUNCT
bracis-33605	47	1	thus	thus	ADV
bracis-33605	47	2	,	,	PUNCT
bracis-33605	47	3	in	in	ADP
bracis-33605	47	4	this	this	DET
bracis-33605	47	5	work	work	NOUN
bracis-33605	47	6	,	,	PUNCT
bracis-33605	47	7	the	the	DET
bracis-33605	47	8	gubs	gubs	ADJ
bracis-33605	47	9	criterion	criterion	NOUN
bracis-33605	47	10	is	be	AUX
bracis-33605	47	11	used	use	VERB
bracis-33605	47	12	to	to	PART
bracis-33605	47	13	propose	propose	VERB
bracis-33605	47	14	the	the	DET
bracis-33605	47	15	first	first	ADJ
bracis-33605	47	16	two	two	NUM
bracis-33605	47	17	algorithms	algorithm	NOUN
bracis-33605	47	18	for	for	SCONJ
bracis-33605	47	19	rl	rl	NOUN
bracis-33605	47	20	to	to	PART
bracis-33605	47	21	do	do	VERB
bracis-33605	47	22	a	a	DET
bracis-33605	47	23	trade	trade	NOUN
bracis-33605	47	24	-	-	PUNCT
bracis-33605	47	25	off	off	NOUN
bracis-33605	47	26	between	between	ADP
bracis-33605	47	27	probability	probability	NOUN
bracis-33605	47	28	to	to	ADP
bracis-33605	47	29	goal	goal	NOUN
bracis-33605	47	30	and	and	CCONJ
bracis-33605	47	31	cost	cost	NOUN
bracis-33605	47	32	to	to	ADP
bracis-33605	47	33	goal	goal	NOUN
bracis-33605	47	34	,	,	PUNCT
bracis-33605	47	35	without	without	ADP
bracis-33605	47	36	using	use	VERB
bracis-33605	47	37	discounts	discount	NOUN
bracis-33605	47	38	or	or	CCONJ
bracis-33605	47	39	penalties	penalty	NOUN
bracis-33605	47	40	,	,	PUNCT
bracis-33605	47	41	that	that	PRON
bracis-33605	47	42	are	be	AUX
bracis-33605	47	43	:	:	PUNCT
bracis-33605	47	44	the	the	DET
bracis-33605	47	45	q	q	NOUN
bracis-33605	47	46	-	-	PUNCT
bracis-33605	47	47	learning	learning	NOUN
bracis-33605	47	48	-	-	PUNCT
bracis-33605	47	49	gubs	gub	NOUN
bracis-33605	47	50	and	and	CCONJ
bracis-33605	47	51	the	the	DET
bracis-33605	47	52	q	q	ADJ
bracis-33605	47	53	-	-	PUNCT
bracis-33605	47	54	learning	learning	NOUN
bracis-33605	47	55	-	-	PUNCT
bracis-33605	47	56	egubs+cmax	egubs+cmax	PROPN
bracis-33605	47	57	.	.	PUNCT
bracis-33605	48	1	additionally	additionally	ADV
bracis-33605	48	2	,	,	PUNCT
bracis-33605	48	3	to	to	ADP
bracis-33605	48	4	the	the	DET
bracis-33605	48	5	best	good	ADJ
bracis-33605	48	6	of	of	ADP
bracis-33605	48	7	our	our	PRON
bracis-33605	48	8	knowledge	knowledge	NOUN
bracis-33605	48	9	,	,	PUNCT
bracis-33605	48	10	the	the	DET
bracis-33605	48	11	proposed	propose	VERB
bracis-33605	48	12	algorithms	algorithm	NOUN
bracis-33605	48	13	are	be	AUX
bracis-33605	48	14	the	the	DET
bracis-33605	48	15	first	first	ADJ
bracis-33605	48	16	to	to	PART
bracis-33605	48	17	use	use	VERB
bracis-33605	48	18	the	the	DET
bracis-33605	48	19	expected	expect	VERB
bracis-33605	48	20	utility	utility	NOUN
bracis-33605	48	21	theory	theory	NOUN
bracis-33605	48	22	to	to	PART
bracis-33605	48	23	make	make	VERB
bracis-33605	48	24	this	this	DET
bracis-33605	48	25	trade	trade	NOUN
bracis-33605	48	26	-	-	PUNCT
bracis-33605	48	27	off	off	NOUN
bracis-33605	48	28	in	in	ADP
bracis-33605	48	29	rl	rl	NOUN
bracis-33605	48	30	,	,	PUNCT
bracis-33605	48	31	which	which	PRON
bracis-33605	48	32	allows	allow	VERB
bracis-33605	48	33	more	more	ADV
bracis-33605	48	34	precise	precise	ADJ
bracis-33605	48	35	value	value	NOUN
bracis-33605	48	36	criteria	criterion	NOUN
bracis-33605	48	37	to	to	PART
bracis-33605	48	38	be	be	AUX
bracis-33605	48	39	designed	design	VERB
bracis-33605	48	40	.	.	PUNCT
bracis-33605	49	1	2	2	NUM
bracis-33605	49	2	problem	problem	NOUN
bracis-33605	49	3	formalism	formalism	NOUN
bracis-33605	49	4	and	and	CCONJ
bracis-33605	49	5	 	 	SPACE
bracis-33605	49	6	background	background	NOUN
bracis-33605	49	7	an	an	DET
bracis-33605	49	8	stochastic	stochastic	ADJ
bracis-33605	49	9	shortest	short	ADJ
bracis-33605	49	10	path	path	NOUN
bracis-33605	49	11	(	(	PUNCT
bracis-33605	49	12	ssp	ssp	NOUN
bracis-33605	49	13	)	)	PUNCT
bracis-33605	49	14	is	be	AUX
bracis-33605	49	15	defined	define	VERB
bracis-33605	49	16	by	by	ADP
bracis-33605	49	17	a	a	DET
bracis-33605	49	18	tuple	tuple	NOUN
bracis-33605	49	19	\(m=	\(m=	PROPN
bracis-33605	50	1	\langle	\langle	PROPN
bracis-33605	50	2	s	s	PROPN
bracis-33605	50	3	,	,	PUNCT
bracis-33605	50	4	a	a	PRON
bracis-33605	50	5	,	,	PUNCT
bracis-33605	50	6	p	p	NOUN
bracis-33605	50	7	,	,	PUNCT
bracis-33605	50	8	c,\mathcal	c,\mathcal	ADJ
bracis-33605	50	9	{	{	PUNCT
bracis-33605	50	10	g	g	NOUN
bracis-33605	50	11	}	}	PUNCT
bracis-33605	50	12	\rangle	\rangle	PROPN
bracis-33605	50	13	\	\	NOUN
bracis-33605	50	14	)	)	PUNCT
bracis-33605	50	15	,	,	PUNCT
bracis-33605	50	16	where	where	SCONJ
bracis-33605	50	17	:	:	PUNCT
bracis-33605	50	18	s	s	X
bracis-33605	50	19	is	be	AUX
bracis-33605	50	20	a	a	DET
bracis-33605	50	21	discrete	discrete	ADJ
bracis-33605	50	22	and	and	CCONJ
bracis-33605	50	23	finite	finite	ADJ
bracis-33605	50	24	set	set	NOUN
bracis-33605	50	25	of	of	ADP
bracis-33605	50	26	completely	completely	ADV
bracis-33605	50	27	observed	observed	ADJ
bracis-33605	50	28	states	state	NOUN
bracis-33605	50	29	that	that	PRON
bracis-33605	50	30	model	model	VERB
bracis-33605	50	31	the	the	DET
bracis-33605	50	32	world	world	NOUN
bracis-33605	50	33	,	,	PUNCT
bracis-33605	50	34	a	a	PRON
bracis-33605	50	35	is	be	AUX
bracis-33605	50	36	a	a	DET
bracis-33605	50	37	finite	finite	ADJ
bracis-33605	50	38	set	set	NOUN
bracis-33605	50	39	of	of	ADP
bracis-33605	50	40	actions	action	NOUN
bracis-33605	50	41	,	,	PUNCT
bracis-33605	50	42	\(p	\(p	PROPN
bracis-33605	50	43	:	:	PUNCT
bracis-33605	50	44	s	s	VERB
bracis-33605	50	45	\times	\time	VERB
bracis-33605	50	46	a	a	DET
bracis-33605	50	47	\times	\time	NOUN
bracis-33605	50	48	s	s	VERB
bracis-33605	50	49	\rightarrow	\rightarrow	PROPN
bracis-33605	50	50	[	[	X
bracis-33605	50	51	0,1]\	0,1]\	NOUN
bracis-33605	50	52	)	)	PUNCT
bracis-33605	50	53	is	be	AUX
bracis-33605	50	54	a	a	DET
bracis-33605	50	55	probabilistic	probabilistic	ADJ
bracis-33605	50	56	transition	transition	NOUN
bracis-33605	50	57	function	function	NOUN
bracis-33605	50	58	,	,	PUNCT
bracis-33605	50	59	where	where	SCONJ
bracis-33605	50	60	,	,	PUNCT
bracis-33605	50	61	\(p(s	\(p(s	PROPN
bracis-33605	50	62	,	,	PUNCT
bracis-33605	50	63	a	a	DET
bracis-33605	50	64	,	,	PUNCT
bracis-33605	50	65	s	s	NOUN
bracis-33605	50	66	'	'	PUNCT
bracis-33605	50	67	)	)	PUNCT
bracis-33605	51	1	=	=	SYM
bracis-33605	51	2	pr(s_{t+1	pr(s_{t+1	PROPN
bracis-33605	51	3	}	}	PUNCT
bracis-33605	51	4	=	=	PUNCT
bracis-33605	51	5	s'|s_t	s'|s_t	PUNCT
bracis-33605	52	1	=	=	SYM
bracis-33605	52	2	s	s	X
bracis-33605	52	3	,	,	PUNCT
bracis-33605	52	4	a_t	a_t	PUNCT
bracis-33605	52	5	=	=	SYM
bracis-33605	52	6	a)\	a)\	X
bracis-33605	52	7	)	)	PUNCT
bracis-33605	52	8	is	be	AUX
bracis-33605	52	9	the	the	DET
bracis-33605	52	10	probability	probability	NOUN
bracis-33605	52	11	of	of	ADP
bracis-33605	52	12	transition	transition	NOUN
bracis-33605	52	13	from	from	ADP
bracis-33605	52	14	a	a	DET
bracis-33605	52	15	state	state	NOUN
bracis-33605	52	16	\(s	\(s	ADJ
bracis-33605	52	17	\in	\in	PROPN
bracis-33605	52	18	s\	s\	NOUN
bracis-33605	52	19	)	)	PUNCT
bracis-33605	52	20	to	to	ADP
bracis-33605	52	21	a	a	DET
bracis-33605	52	22	state	state	NOUN
bracis-33605	52	23	\(s	\(s	PROPN
bracis-33605	52	24	'	'	PART
bracis-33605	52	25	\in	\in	NOUN
bracis-33605	52	26	s\	s\	NOUN
bracis-33605	52	27	)	)	PUNCT
bracis-33605	52	28	after	after	ADP
bracis-33605	52	29	executing	execute	VERB
bracis-33605	52	30	an	an	DET
bracis-33605	52	31	action	action	NOUN
bracis-33605	52	32	\(a	\(a	NOUN
bracis-33605	52	33	\in	\in	PROPN
bracis-33605	52	34	a\	a\	PROPN
bracis-33605	52	35	)	)	PUNCT
bracis-33605	52	36	,	,	PUNCT
bracis-33605	52	37	\(c	\(c	NOUN
bracis-33605	52	38	:	:	PUNCT
bracis-33605	52	39	s	s	AUX
bracis-33605	52	40	\times	\time	VERB
bracis-33605	52	41	a	a	DET
bracis-33605	52	42	\rightarrow	\rightarrow	PROPN
bracis-33605	52	43	\mathcal	\mathcal	PROPN
bracis-33605	52	44	{	{	PUNCT
bracis-33605	52	45	r}\	r}\	NUM
bracis-33605	52	46	)	)	PUNCT
bracis-33605	52	47	is	be	AUX
bracis-33605	52	48	a	a	DET
bracis-33605	52	49	cost	cost	NOUN
bracis-33605	52	50	function	function	NOUN
bracis-33605	52	51	for	for	ADP
bracis-33605	52	52	each	each	DET
bracis-33605	52	53	action	action	NOUN
bracis-33605	52	54	\(a	\(a	PROPN
bracis-33605	52	55	\in	\in	PROPN
bracis-33605	52	56	a\	a\	PROPN
bracis-33605	52	57	)	)	PUNCT
bracis-33605	52	58	in	in	ADP
bracis-33605	52	59	a	a	DET
bracis-33605	52	60	state	state	NOUN
bracis-33605	52	61	\(s	\(s	ADJ
bracis-33605	52	62	\in	\in	PROPN
bracis-33605	52	63	s\	s\	NOUN
bracis-33605	52	64	)	)	PUNCT
bracis-33605	52	65	,	,	PUNCT
bracis-33605	52	66	\(\mathcal	\(\mathcal	ADJ
bracis-33605	52	67	{	{	PUNCT
bracis-33605	52	68	g	g	PROPN
bracis-33605	52	69	}	}	PUNCT
bracis-33605	52	70	\subset	\subset	NOUN
bracis-33605	52	71	s\	s\	NOUN
bracis-33605	52	72	)	)	PUNCT
bracis-33605	52	73	is	be	AUX
bracis-33605	52	74	the	the	DET
bracis-33605	52	75	set	set	NOUN
bracis-33605	52	76	of	of	ADP
bracis-33605	52	77	goal	goal	NOUN
bracis-33605	52	78	states	state	NOUN
bracis-33605	52	79	,	,	PUNCT
bracis-33605	52	80	i.e.	i.e.	X
bracis-33605	52	81	\(g	\(g	X
bracis-33605	52	82	\in	\in	PROPN
bracis-33605	52	83	\mathcal	\mathcal	PROPN
bracis-33605	52	84	{	{	PUNCT
bracis-33605	52	85	g}\	g}\	PROPN
bracis-33605	52	86	)	)	PUNCT
bracis-33605	52	87	,	,	PUNCT
bracis-33605	52	88	such	such	ADJ
bracis-33605	52	89	that	that	SCONJ
bracis-33605	52	90	\(c(g	\(c(g	PROPN
bracis-33605	52	91	,	,	PUNCT
bracis-33605	52	92	a)=0\	a)=0\	NUM
bracis-33605	52	93	)	)	PUNCT
bracis-33605	52	94	,	,	PUNCT
bracis-33605	52	95	\(p(g	\(p(g	PROPN
bracis-33605	52	96	,	,	PUNCT
bracis-33605	52	97	a	a	PRON
bracis-33605	52	98	,	,	PUNCT
bracis-33605	52	99	g)=1\	g)=1\	NOUN
bracis-33605	52	100	)	)	PUNCT
bracis-33605	52	101	,	,	PUNCT
bracis-33605	52	102	\(\forall	\(\forall	ADV
bracis-33605	52	103	a	a	DET
bracis-33605	52	104	\in	\in	PROPN
bracis-33605	52	105	a\	a\	NOUN
bracis-33605	52	106	)	)	PUNCT
bracis-33605	52	107	and	and	CCONJ
bracis-33605	52	108	\(c(s	\(c(s	PROPN
bracis-33605	52	109	,	,	PUNCT
bracis-33605	52	110	a	a	PRON
bracis-33605	52	111	)	)	PUNCT
bracis-33605	52	112	>	>	X
bracis-33605	52	113	0\	0\	PROPN
bracis-33605	52	114	)	)	PUNCT
bracis-33605	52	115	,	,	PUNCT
bracis-33605	52	116	\(\forall	\(\forall	ADV
bracis-33605	52	117	s	s	VERB
bracis-33605	52	118	\in	\in	PROPN
bracis-33605	52	119	s	s	PART
bracis-33605	52	120	\setminus	\setminus	PROPN
bracis-33605	52	121	\mathcal	\mathcal	PROPN
bracis-33605	52	122	{	{	PUNCT
bracis-33605	52	123	g}\	g}\	PROPN
bracis-33605	52	124	)	)	PUNCT
bracis-33605	52	125	.	.	PUNCT
bracis-33605	53	1	in	in	ADP
bracis-33605	53	2	a	a	DET
bracis-33605	53	3	state	state	NOUN
bracis-33605	53	4	\(s_t	\(s_t	PROPN
bracis-33605	53	5	\in	\in	PROPN
bracis-33605	53	6	s\	s\	NOUN
bracis-33605	53	7	)	)	PUNCT
bracis-33605	53	8	of	of	ADP
bracis-33605	53	9	the	the	DET
bracis-33605	53	10	environment	environment	NOUN
bracis-33605	53	11	at	at	ADP
bracis-33605	53	12	time	time	NOUN
bracis-33605	53	13	t	t	PROPN
bracis-33605	53	14	,	,	PUNCT
bracis-33605	53	15	the	the	DET
bracis-33605	53	16	agent	agent	NOUN
bracis-33605	53	17	executes	execute	VERB
bracis-33605	53	18	an	an	DET
bracis-33605	53	19	action	action	NOUN
bracis-33605	53	20	\(a_t	\(a_t	ADV
bracis-33605	53	21	\in	\in	PROPN
bracis-33605	53	22	a\	a\	PROPN
bracis-33605	53	23	)	)	PUNCT
bracis-33605	53	24	that	that	PRON
bracis-33605	53	25	has	have	VERB
bracis-33605	53	26	a	a	DET
bracis-33605	53	27	cost	cost	NOUN
bracis-33605	53	28	\(c_t	\(c_t	NOUN
bracis-33605	53	29	=	=	ADJ
bracis-33605	53	30	c(s_t	c(s_t	NOUN
bracis-33605	53	31	,	,	PUNCT
bracis-33605	53	32	a_t)\	a_t)\	NOUN
bracis-33605	53	33	)	)	PUNCT
bracis-33605	53	34	and	and	CCONJ
bracis-33605	53	35	probabilistic	probabilistic	ADJ
bracis-33605	53	36	effects	effect	NOUN
bracis-33605	53	37	generating	generate	VERB
bracis-33605	53	38	a	a	DET
bracis-33605	53	39	future	future	ADJ
bracis-33605	53	40	state	state	NOUN
bracis-33605	53	41	\(s_{t+1}\	\(s_{t+1}\	NUM
bracis-33605	53	42	)	)	PUNCT
bracis-33605	53	43	as	as	ADP
bracis-33605	53	44	a	a	DET
bracis-33605	53	45	function	function	NOUN
bracis-33605	53	46	of	of	ADP
bracis-33605	53	47	\(p(s_t	\(p(s_t	PROPN
bracis-33605	53	48	,	,	PUNCT
bracis-33605	53	49	a_t	a_t	X
bracis-33605	53	50	,	,	PUNCT
bracis-33605	53	51	s_{t+1})\	s_{t+1})\	NOUN
bracis-33605	53	52	)	)	PUNCT
bracis-33605	53	53	.	.	PUNCT
bracis-33605	54	1	every	every	DET
bracis-33605	54	2	interaction	interaction	NOUN
bracis-33605	54	3	that	that	PRON
bracis-33605	54	4	occurs	occur	VERB
bracis-33605	54	5	can	can	AUX
bracis-33605	54	6	be	be	AUX
bracis-33605	54	7	summarized	summarize	VERB
bracis-33605	54	8	in	in	ADP
bracis-33605	54	9	a	a	DET
bracis-33605	54	10	history	history	NOUN
bracis-33605	54	11	of	of	ADP
bracis-33605	54	12	length	length	NOUN
bracis-33605	54	13	t	t	PROPN
bracis-33605	54	14	represented	represent	VERB
bracis-33605	54	15	by	by	ADP
bracis-33605	54	16	\(h_t=\{s_1,a_1,c_1,s_2,a_2,c_2,s_3,	\(h_t=\{s_1,a_1,c_1,s_2,a_2,c_2,s_3,	NOUN
bracis-33605	54	17	...	...	PUNCT
bracis-33605	54	18	,s_{t-1},a	,s_{t-1},a	PUNCT
bracis-33605	55	1	_	_	PRON
bracis-33605	55	2	{	{	PUNCT
bracis-33605	55	3	t-1},c_{t-1},s_t\}\	t-1},c_{t-1},s_t\}\	NOUN
bracis-33605	55	4	)	)	PUNCT
bracis-33605	55	5	.	.	PUNCT
bracis-33605	56	1	the	the	DET
bracis-33605	56	2	set	set	NOUN
bracis-33605	56	3	of	of	ADP
bracis-33605	56	4	all	all	DET
bracis-33605	56	5	historical	historical	ADJ
bracis-33605	56	6	possibilities	possibility	NOUN
bracis-33605	56	7	is	be	AUX
bracis-33605	56	8	\(\mathcal	\(\mathcal	ADJ
bracis-33605	56	9	{	{	PUNCT
bracis-33605	56	10	h}=(s	h}=(s	PROPN
bracis-33605	56	11	\times	\time	VERB
bracis-33605	56	12	a	a	DET
bracis-33605	56	13	\times	\time	NOUN
bracis-33605	56	14	\mathcal	\mathcal	PROPN
bracis-33605	56	15	{	{	PUNCT
bracis-33605	56	16	r}_{>0})^	r}_{>0})^	PROPN
bracis-33605	56	17	*	*	VERB
bracis-33605	56	18	\times	\times	ADP
bracis-33605	56	19	s\	s\	NOUN
bracis-33605	56	20	)	)	PUNCT
bracis-33605	56	21	.	.	PUNCT
bracis-33605	57	1	in	in	ADP
bracis-33605	57	2	an	an	DET
bracis-33605	57	3	ssp	ssp	NOUN
bracis-33605	57	4	,	,	PUNCT
bracis-33605	57	5	the	the	DET
bracis-33605	57	6	process	process	NOUN
bracis-33605	57	7	follows	follow	VERB
bracis-33605	57	8	a	a	DET
bracis-33605	57	9	history	history	NOUN
bracis-33605	57	10	\(h_t\	\(h_t\	NOUN
bracis-33605	57	11	)	)	PUNCT
bracis-33605	57	12	that	that	PRON
bracis-33605	57	13	ends	end	VERB
bracis-33605	57	14	in	in	ADP
bracis-33605	57	15	a	a	DET
bracis-33605	57	16	goal	goal	NOUN
bracis-33605	57	17	state	state	NOUN
bracis-33605	57	18	\(g	\(g	PROPN
bracis-33605	57	19	\in	\in	PROPN
bracis-33605	57	20	\mathcal	\mathcal	PROPN
bracis-33605	57	21	{	{	PUNCT
bracis-33605	57	22	g}\	g}\	PROPN
bracis-33605	57	23	)	)	PUNCT
bracis-33605	57	24	.	.	PUNCT
bracis-33605	58	1	the	the	DET
bracis-33605	58	2	solution	solution	NOUN
bracis-33605	58	3	for	for	ADP
bracis-33605	58	4	an	an	DET
bracis-33605	58	5	ssp	ssp	NOUN
bracis-33605	58	6	is	be	AUX
bracis-33605	58	7	a	a	DET
bracis-33605	58	8	policy	policy	NOUN
bracis-33605	58	9	\(\pi	\(\pi	ADP
bracis-33605	58	10	\	\	NOUN
bracis-33605	58	11	)	)	PUNCT
bracis-33605	58	12	that	that	PRON
bracis-33605	58	13	can	can	AUX
bracis-33605	58	14	be	be	AUX
bracis-33605	58	15	stationary	stationary	ADJ
bracis-33605	58	16	(	(	PUNCT
bracis-33605	58	17	\(\pi	\(\pi	NOUN
bracis-33605	58	18	:	:	PUNCT
bracis-33605	58	19	s	s	PROPN
bracis-33605	58	20	\rightarrow	\rightarrow	PROPN
bracis-33605	58	21	a\	a\	PROPN
bracis-33605	58	22	)	)	PUNCT
bracis-33605	58	23	)	)	PUNCT
bracis-33605	58	24	where	where	SCONJ
bracis-33605	58	25	,	,	PUNCT
bracis-33605	58	26	each	each	DET
bracis-33605	58	27	state	state	NOUN
bracis-33605	58	28	s	s	PART
bracis-33605	58	29	is	be	AUX
bracis-33605	58	30	mapped	map	VERB
bracis-33605	58	31	on	on	ADP
bracis-33605	58	32	an	an	DET
bracis-33605	58	33	action	action	NOUN
bracis-33605	58	34	a	a	PRON
bracis-33605	58	35	or	or	CCONJ
bracis-33605	58	36	not	not	PART
bracis-33605	58	37	stationary	stationary	ADJ
bracis-33605	58	38	(	(	PUNCT
bracis-33605	58	39	\(\pi	\(\pi	NOUN
bracis-33605	58	40	:	:	PUNCT
bracis-33605	58	41	\mathcal	\mathcal	ADJ
bracis-33605	58	42	{	{	PUNCT
bracis-33605	58	43	h	h	NOUN
bracis-33605	58	44	}	}	PUNCT
bracis-33605	58	45	\rightarrow	\rightarrow	PROPN
bracis-33605	58	46	a\	a\	NOUN
bracis-33605	58	47	)	)	PUNCT
bracis-33605	58	48	)	)	PUNCT
bracis-33605	58	49	where	where	SCONJ
bracis-33605	58	50	,	,	PUNCT
bracis-33605	58	51	a	a	DET
bracis-33605	58	52	historical	historical	ADJ
bracis-33605	58	53	\(h_t\	\(h_t\	NOUN
bracis-33605	58	54	)	)	PUNCT
bracis-33605	58	55	is	be	AUX
bracis-33605	58	56	mapped	map	VERB
bracis-33605	58	57	at	at	ADP
bracis-33605	58	58	time	time	NOUN
bracis-33605	58	59	t	t	NOUN
bracis-33605	58	60	in	in	ADP
bracis-33605	58	61	an	an	DET
bracis-33605	58	62	action	action	NOUN
bracis-33605	58	63	\(a_t\	\(a_t\	NUM
bracis-33605	58	64	)	)	PUNCT
bracis-33605	58	65	.	.	PUNCT
bracis-33605	59	1	this	this	DET
bracis-33605	59	2	policy	policy	NOUN
bracis-33605	59	3	must	must	AUX
bracis-33605	59	4	be	be	AUX
bracis-33605	59	5	optimal	optimal	ADJ
bracis-33605	59	6	\(\pi	\(\pi	PROPN
bracis-33605	59	7	^*\	^*\	X
bracis-33605	59	8	)	)	PUNCT
bracis-33605	59	9	under	under	ADP
bracis-33605	59	10	some	some	DET
bracis-33605	59	11	criteria	criterion	NOUN
bracis-33605	59	12	.	.	PUNCT
bracis-33605	60	1	a	a	DET
bracis-33605	60	2	policy	policy	NOUN
bracis-33605	60	3	is	be	AUX
bracis-33605	60	4	called	call	VERB
bracis-33605	60	5	proper	proper	ADJ
bracis-33605	60	6	if	if	SCONJ
bracis-33605	60	7	the	the	DET
bracis-33605	60	8	process	process	NOUN
bracis-33605	60	9	achieves	achieve	VERB
bracis-33605	60	10	a	a	DET
bracis-33605	60	11	goal	goal	NOUN
bracis-33605	60	12	state	state	NOUN
bracis-33605	60	13	with	with	ADP
bracis-33605	60	14	probability	probability	NOUN
bracis-33605	60	15	1	1	NUM
bracis-33605	60	16	when	when	SCONJ
bracis-33605	60	17	it	it	PRON
bracis-33605	60	18	is	be	AUX
bracis-33605	60	19	followed	follow	VERB
bracis-33605	60	20	,	,	PUNCT
bracis-33605	60	21	formally	formally	ADV
bracis-33605	60	22	,	,	PUNCT
bracis-33605	60	23	\(\lim	\(\lim	PUNCT
bracis-33605	61	1	_	_	PRON
bracis-33605	61	2	{	{	PUNCT
bracis-33605	61	3	t\rightarrow	t\rightarrow	ADJ
bracis-33605	61	4	\infty	\infty	PROPN
bracis-33605	61	5	}	}	PUNCT
bracis-33605	61	6	pr(s_t	pr(s_t	ADJ
bracis-33605	61	7	\in	\in	PROPN
bracis-33605	61	8	\mathcal	\mathcal	PROPN
bracis-33605	61	9	{	{	PUNCT
bracis-33605	61	10	g}|\	g}|\	NOUN
bracis-33605	61	11	\pi	\pi	PROPN
bracis-33605	61	12	,	,	PUNCT
bracis-33605	61	13	s_0)=1\	s_0)=1\	PROPN
bracis-33605	61	14	)	)	PUNCT
bracis-33605	61	15	where	where	SCONJ
bracis-33605	61	16	,	,	PUNCT
bracis-33605	61	17	\(s_0\	\(s_0\	PROPN
bracis-33605	61	18	)	)	PUNCT
bracis-33605	61	19	is	be	AUX
bracis-33605	61	20	the	the	DET
bracis-33605	61	21	initial	initial	ADJ
bracis-33605	61	22	state	state	NOUN
bracis-33605	61	23	of	of	ADP
bracis-33605	61	24	the	the	DET
bracis-33605	61	25	process	process	NOUN
bracis-33605	61	26	.	.	PUNCT
bracis-33605	62	1	the	the	DET
bracis-33605	62	2	optimal	optimal	ADJ
bracis-33605	62	3	value	value	NOUN
bracis-33605	62	4	function	function	NOUN
bracis-33605	62	5	\(v^*\	\(v^*\	NOUN
bracis-33605	62	6	)	)	PUNCT
bracis-33605	62	7	for	for	ADP
bracis-33605	62	8	an	an	DET
bracis-33605	62	9	ssp	ssp	NOUN
bracis-33605	62	10	is	be	AUX
bracis-33605	62	11	defined	define	VERB
bracis-33605	62	12	by	by	ADP
bracis-33605	62	13	the	the	DET
bracis-33605	62	14	bellman	bellman	NOUN
bracis-33605	62	15	equation	equation	NOUN
bracis-33605	62	16	\(v^{*}(s	\(v^{*}(s	NOUN
bracis-33605	62	17	)	)	PUNCT
bracis-33605	63	1	=	=	NOUN
bracis-33605	63	2	\min	\min	NOUN
bracis-33605	63	3	_	_	PUNCT
bracis-33605	63	4	{	{	PUNCT
bracis-33605	63	5	a	a	DET
bracis-33605	63	6	\in	\in	PROPN
bracis-33605	63	7	a	a	DET
bracis-33605	63	8	}	}	PUNCT
bracis-33605	63	9	\	\	NOUN
bracis-33605	63	10	{	{	PUNCT
bracis-33605	63	11	c(s	c(	NOUN
bracis-33605	63	12	,	,	PUNCT
bracis-33605	63	13	a	a	PRON
bracis-33605	63	14	)	)	PUNCT
bracis-33605	63	15	+	+	NUM
bracis-33605	63	16	\displaystyle	\displaystyle	PROPN
bracis-33605	63	17	\sum	\sum	NOUN
bracis-33605	63	18	_	_	PUNCT
bracis-33605	63	19	{	{	PUNCT
bracis-33605	63	20	s	s	NOUN
bracis-33605	63	21	'	'	CCONJ
bracis-33605	63	22	\in	\in	PROPN
bracis-33605	63	23	s}p(s	s}p(s	NOUN
bracis-33605	63	24	,	,	PUNCT
bracis-33605	63	25	a	a	PRON
bracis-33605	63	26	,	,	PUNCT
bracis-33605	63	27	s')v^{*}(s	s')v^{*}(s	NOUN
bracis-33605	63	28	'	'	PUNCT
bracis-33605	63	29	)	)	PUNCT
bracis-33605	63	30	\}\	\}\	NOUN
bracis-33605	63	31	)	)	PUNCT
bracis-33605	63	32	.	.	PUNCT
bracis-33605	64	1	the	the	DET
bracis-33605	64	2	optimal	optimal	ADJ
bracis-33605	64	3	policy	policy	NOUN
bracis-33605	64	4	\(\pi	\(\pi	ADP
bracis-33605	64	5	^*(s)\	^*(s)\	NUM
bracis-33605	64	6	)	)	PUNCT
bracis-33605	64	7	for	for	ADP
bracis-33605	64	8	state	state	NOUN
bracis-33605	64	9	s	s	PART
bracis-33605	64	10	is	be	AUX
bracis-33605	64	11	\(\pi	\(\pi	ADJ
bracis-33605	64	12	^*(s	^*(s	PUNCT
bracis-33605	64	13	)	)	PUNCT
bracis-33605	64	14	=	=	PUNCT
bracis-33605	65	1	\arg	\arg	ADP
bracis-33605	65	2	\min	\min	PROPN
bracis-33605	65	3	_	_	PROPN
bracis-33605	65	4	{	{	PUNCT
bracis-33605	65	5	a	a	DET
bracis-33605	65	6	\in	\in	PROPN
bracis-33605	65	7	a}\{c(s	a}\{c(s	PROPN
bracis-33605	65	8	,	,	PUNCT
bracis-33605	65	9	a)+\displaystyle	a)+\displaystyle	PROPN
bracis-33605	65	10	\sum	\sum	NOUN
bracis-33605	65	11	_	_	PUNCT
bracis-33605	65	12	{	{	PUNCT
bracis-33605	65	13	s	s	NOUN
bracis-33605	65	14	'	'	CCONJ
bracis-33605	65	15	\in	\in	PROPN
bracis-33605	65	16	s}p(s	s}p(s	NOUN
bracis-33605	65	17	,	,	PUNCT
bracis-33605	65	18	a	a	DET
bracis-33605	65	19	,	,	PUNCT
bracis-33605	65	20	s')v^{*}(s')\}\	s')v^{*}(s')\}\	NOUN
bracis-33605	65	21	)	)	PUNCT
bracis-33605	65	22	.	.	PUNCT
bracis-33605	66	1	a	a	DET
bracis-33605	66	2	state	state	NOUN
bracis-33605	66	3	s	s	PART
bracis-33605	66	4	is	be	AUX
bracis-33605	66	5	called	call	VERB
bracis-33605	66	6	dead	dead	ADJ
bracis-33605	66	7	-	-	PUNCT
bracis-33605	66	8	end	end	NOUN
bracis-33605	66	9	state	state	NOUN
bracis-33605	66	10	when	when	SCONJ
bracis-33605	66	11	the	the	DET
bracis-33605	66	12	probability	probability	NOUN
bracis-33605	66	13	of	of	ADP
bracis-33605	66	14	reaching	reach	VERB
bracis-33605	66	15	a	a	DET
bracis-33605	66	16	goal	goal	NOUN
bracis-33605	66	17	state	state	NOUN
bracis-33605	66	18	from	from	ADP
bracis-33605	66	19	s	s	PROPN
bracis-33605	66	20	is	be	AUX
bracis-33605	66	21	0	0	NUM
bracis-33605	66	22	.	.	PUNCT
bracis-33605	67	1	formally	formally	ADV
bracis-33605	67	2	,	,	PUNCT
bracis-33605	67	3	\(\lim	\(\lim	PUNCT
bracis-33605	68	1	_	_	PRON
bracis-33605	68	2	{	{	PUNCT
bracis-33605	68	3	t\rightarrow	t\rightarrow	ADJ
bracis-33605	68	4	\infty	\infty	PROPN
bracis-33605	68	5	}	}	PUNCT
bracis-33605	68	6	pr(s_t	pr(s_t	ADJ
bracis-33605	68	7	\in	\in	PROPN
bracis-33605	68	8	\mathcal	\mathcal	PROPN
bracis-33605	68	9	{	{	PUNCT
bracis-33605	68	10	g}|\pi	g}|\pi	NOUN
bracis-33605	68	11	,	,	PUNCT
bracis-33605	68	12	s_0	s_0	PROPN
bracis-33605	68	13	\in	\in	PROPN
bracis-33605	68	14	\mathcal	\mathcal	PROPN
bracis-33605	68	15	{	{	PUNCT
bracis-33605	68	16	d})=0\	d})=0\	NOUN
bracis-33605	68	17	)	)	PUNCT
bracis-33605	68	18	.	.	PUNCT
bracis-33605	69	1	\(\mathcal	\(\mathcal	ADJ
bracis-33605	69	2	{	{	PUNCT
bracis-33605	69	3	d	d	NOUN
bracis-33605	69	4	}	}	PUNCT
bracis-33605	69	5	\subset	\subset	PROPN
bracis-33605	69	6	s	s	PART
bracis-33605	69	7	\	\	PROPN
bracis-33605	69	8	)	)	PUNCT
bracis-33605	69	9	is	be	AUX
bracis-33605	69	10	the	the	DET
bracis-33605	69	11	set	set	NOUN
bracis-33605	69	12	of	of	ADP
bracis-33605	69	13	dead	dead	ADJ
bracis-33605	69	14	-	-	PUNCT
bracis-33605	69	15	end	end	NOUN
bracis-33605	69	16	states	state	NOUN
bracis-33605	69	17	.	.	PUNCT
bracis-33605	70	1	conventional	conventional	ADJ
bracis-33605	70	2	ssps	ssps	AUX
bracis-33605	70	3	do	do	AUX
bracis-33605	70	4	not	not	PART
bracis-33605	70	5	consider	consider	VERB
bracis-33605	70	6	dead	dead	ADJ
bracis-33605	70	7	-	-	PUNCT
bracis-33605	70	8	end	end	NOUN
bracis-33605	70	9	.	.	PUNCT
bracis-33605	71	1	2.1	2.1	NUM
bracis-33605	71	2	criteria	criterion	NOUN
bracis-33605	71	3	for	for	ADP
bracis-33605	71	4	 	 	SPACE
bracis-33605	71	5	ssps	ssps	VERB
bracis-33605	71	6	with	with	ADP
bracis-33605	71	7	 	 	SPACE
bracis-33605	71	8	dead	dead	ADJ
bracis-33605	71	9	-	-	PUNCT
bracis-33605	71	10	ends	end	NOUN
bracis-33605	71	11	in	in	ADP
bracis-33605	71	12	environments	environment	NOUN
bracis-33605	71	13	with	with	ADP
bracis-33605	71	14	dead	dead	ADJ
bracis-33605	71	15	-	-	PUNCT
bracis-33605	71	16	end	end	NOUN
bracis-33605	71	17	,	,	PUNCT
bracis-33605	71	18	the	the	DET
bracis-33605	71	19	agent	agent	NOUN
bracis-33605	71	20	must	must	AUX
bracis-33605	71	21	decide	decide	VERB
bracis-33605	71	22	between	between	ADP
bracis-33605	71	23	finding	find	VERB
bracis-33605	71	24	a	a	DET
bracis-33605	71	25	policy	policy	NOUN
bracis-33605	71	26	with	with	ADP
bracis-33605	71	27	a	a	DET
bracis-33605	71	28	low	low	ADJ
bracis-33605	71	29	expected	expect	VERB
bracis-33605	71	30	total	total	ADJ
bracis-33605	71	31	cost	cost	NOUN
bracis-33605	71	32	\(\overline{c}_g(s)\	\(\overline{c}_g(s)\	NOUN
bracis-33605	71	33	)	)	PUNCT
bracis-33605	71	34	from	from	ADP
bracis-33605	71	35	an	an	DET
bracis-33605	71	36	initial	initial	ADJ
bracis-33605	71	37	state	state	NOUN
bracis-33605	71	38	s	s	ADJ
bracis-33605	71	39	until	until	ADP
bracis-33605	71	40	the	the	DET
bracis-33605	71	41	goal	goal	NOUN
bracis-33605	71	42	state	state	NOUN
bracis-33605	71	43	or	or	CCONJ
bracis-33605	71	44	a	a	DET
bracis-33605	71	45	policy	policy	NOUN
bracis-33605	71	46	with	with	ADP
bracis-33605	71	47	the	the	DET
bracis-33605	71	48	maximum	maximum	ADJ
bracis-33605	71	49	probability	probability	NOUN
bracis-33605	71	50	of	of	ADP
bracis-33605	71	51	achieving	achieve	VERB
bracis-33605	71	52	the	the	DET
bracis-33605	71	53	goal	goal	NOUN
bracis-33605	71	54	\(p_g(s)\	\(p_g(s)\	ADJ
bracis-33605	71	55	)	)	PUNCT
bracis-33605	71	56	.	.	PUNCT
bracis-33605	72	1	several	several	ADJ
bracis-33605	72	2	criteria	criterion	NOUN
bracis-33605	72	3	have	have	AUX
bracis-33605	72	4	been	be	AUX
bracis-33605	72	5	proposed	propose	VERB
bracis-33605	72	6	to	to	PART
bracis-33605	72	7	solve	solve	VERB
bracis-33605	72	8	ssps	ssps	VERB
bracis-33605	72	9	with	with	ADP
bracis-33605	72	10	dead	dead	ADJ
bracis-33605	72	11	-	-	PUNCT
bracis-33605	72	12	end	end	NOUN
bracis-33605	72	13	some	some	PRON
bracis-33605	72	14	of	of	ADP
bracis-33605	72	15	which	which	PRON
bracis-33605	72	16	prioritize	prioritize	VERB
bracis-33605	72	17	the	the	DET
bracis-33605	72	18	probability	probability	NOUN
bracis-33605	72	19	\(p_g(s)\	\(p_g(s)\	ADJ
bracis-33605	72	20	)	)	PUNCT
bracis-33605	73	1	[	[	X
bracis-33605	73	2	7	7	NUM
bracis-33605	73	3	,	,	PUNCT
bracis-33605	73	4	13,14,15	13,14,15	NUM
bracis-33605	73	5	]	]	PUNCT
bracis-33605	73	6	.	.	PUNCT
bracis-33605	74	1	instead	instead	ADV
bracis-33605	74	2	of	of	ADP
bracis-33605	74	3	prioritizing	prioritize	VERB
bracis-33605	74	4	probability	probability	NOUN
bracis-33605	74	5	,	,	PUNCT
bracis-33605	74	6	an	an	DET
bracis-33605	74	7	alternative	alternative	NOUN
bracis-33605	74	8	is	be	AUX
bracis-33605	74	9	to	to	PART
bracis-33605	74	10	make	make	VERB
bracis-33605	74	11	a	a	DET
bracis-33605	74	12	trade	trade	NOUN
bracis-33605	74	13	-	-	PUNCT
bracis-33605	74	14	off	off	NOUN
bracis-33605	74	15	between	between	ADP
bracis-33605	74	16	cost	cost	NOUN
bracis-33605	74	17	to	to	ADP
bracis-33605	74	18	goal	goal	NOUN
bracis-33605	74	19	or	or	CCONJ
bracis-33605	74	20	probability	probability	NOUN
bracis-33605	74	21	to	to	ADP
bracis-33605	74	22	goal	goal	NOUN
bracis-33605	74	23	.	.	PUNCT
bracis-33605	75	1	the	the	DET
bracis-33605	75	2	criterion	criterion	NOUN
bracis-33605	75	3	fsspude	fsspude	NOUN
bracis-33605	75	4	[	[	X
bracis-33605	75	5	6	6	NUM
bracis-33605	75	6	]	]	PUNCT
bracis-33605	75	7	extends	extend	VERB
bracis-33605	75	8	the	the	DET
bracis-33605	75	9	original	original	ADJ
bracis-33605	75	10	ssp	ssp	NOUN
bracis-33605	75	11	with	with	ADP
bracis-33605	75	12	an	an	DET
bracis-33605	75	13	artificial	artificial	ADJ
bracis-33605	75	14	action	action	NOUN
bracis-33605	75	15	q	q	NOUN
bracis-33605	75	16	and	and	CCONJ
bracis-33605	75	17	a	a	DET
bracis-33605	75	18	penalty	penalty	NOUN
bracis-33605	75	19	d.	d.	NOUN
bracis-33605	75	20	when	when	SCONJ
bracis-33605	75	21	q	q	NOUN
bracis-33605	75	22	is	be	AUX
bracis-33605	75	23	used	use	VERB
bracis-33605	75	24	,	,	PUNCT
bracis-33605	75	25	the	the	DET
bracis-33605	75	26	process	process	NOUN
bracis-33605	75	27	transitions	transition	NOUN
bracis-33605	75	28	to	to	ADP
bracis-33605	75	29	a	a	DET
bracis-33605	75	30	goal	goal	NOUN
bracis-33605	75	31	state	state	NOUN
bracis-33605	75	32	and	and	CCONJ
bracis-33605	75	33	the	the	DET
bracis-33605	75	34	agent	agent	NOUN
bracis-33605	75	35	pays	pay	VERB
bracis-33605	75	36	the	the	DET
bracis-33605	75	37	cost	cost	NOUN
bracis-33605	75	38	d	d	NOUN
bracis-33605	75	39	,	,	PUNCT
bracis-33605	75	40	therefore	therefore	ADV
bracis-33605	75	41	\(p(s	\(p(s	PROPN
bracis-33605	75	42	,	,	PUNCT
bracis-33605	75	43	q	q	NOUN
bracis-33605	75	44	,	,	PUNCT
bracis-33605	75	45	g	g	NOUN
bracis-33605	75	46	)	)	PUNCT
bracis-33605	75	47	=	=	SYM
bracis-33605	75	48	1\	1\	NUM
bracis-33605	75	49	)	)	PUNCT
bracis-33605	75	50	,	,	PUNCT
bracis-33605	75	51	\(\forall	\(\forall	ADV
bracis-33605	75	52	s	s	VERB
bracis-33605	75	53	\in	\in	ADJ
bracis-33605	75	54	s\	s\	NOUN
bracis-33605	75	55	)	)	PUNCT
bracis-33605	75	56	and	and	CCONJ
bracis-33605	75	57	for	for	ADP
bracis-33605	75	58	a	a	DET
bracis-33605	75	59	goal	goal	NOUN
bracis-33605	75	60	state	state	NOUN
bracis-33605	75	61	g	g	NOUN
bracis-33605	75	62	;	;	PUNCT
bracis-33605	75	63	and	and	CCONJ
bracis-33605	75	64	\(c(s	\(c(s	PROPN
bracis-33605	75	65	,	,	PUNCT
bracis-33605	75	66	q	q	NOUN
bracis-33605	75	67	)	)	PUNCT
bracis-33605	75	68	=	=	SYM
bracis-33605	75	69	d\	d\	NOUN
bracis-33605	75	70	)	)	PUNCT
bracis-33605	75	71	,	,	PUNCT
bracis-33605	75	72	\(\forall	\(\forall	ADV
bracis-33605	75	73	s	s	VERB
bracis-33605	75	74	\in	\in	ADJ
bracis-33605	75	75	s\	s\	NOUN
bracis-33605	75	76	)	)	PUNCT
bracis-33605	75	77	.	.	PUNCT
bracis-33605	76	1	then	then	ADV
bracis-33605	76	2	,	,	PUNCT
bracis-33605	76	3	the	the	DET
bracis-33605	76	4	objective	objective	NOUN
bracis-33605	76	5	is	be	AUX
bracis-33605	76	6	to	to	PART
bracis-33605	76	7	find	find	VERB
bracis-33605	76	8	a	a	DET
bracis-33605	76	9	policy	policy	NOUN
bracis-33605	76	10	that	that	PRON
bracis-33605	76	11	minimizes	minimize	VERB
bracis-33605	76	12	the	the	DET
bracis-33605	76	13	expected	expect	VERB
bracis-33605	76	14	accumulated	accumulate	VERB
bracis-33605	76	15	cost	cost	NOUN
bracis-33605	76	16	\(\overline{c}_g(s	\(\overline{c}_g(s	NOUN
bracis-33605	76	17	)	)	PUNCT
bracis-33605	77	1	=	=	VERB
bracis-33605	77	2	\lim	\lim	NOUN
bracis-33605	77	3	_	_	PRON
bracis-33605	77	4	{	{	PUNCT
bracis-33605	77	5	t\rightarrow	t\rightarrow	ADJ
bracis-33605	77	6	\infty	\infty	PROPN
bracis-33605	77	7	}	}	PUNCT
bracis-33605	77	8	e[\sum	e[\sum	VERB
bracis-33605	77	9	^{t-1}_{t=0	^{t-1}_{t=0	ADJ
bracis-33605	77	10	}	}	PUNCT
bracis-33605	77	11	c(s_t	c(s_t	PROPN
bracis-33605	77	12	,	,	PUNCT
bracis-33605	77	13	a_t)|	a_t)|	ADJ
bracis-33605	77	14	\pi	\pi	NOUN
bracis-33605	77	15	,	,	PUNCT
bracis-33605	77	16	s_0]\	s_0]\	NOUN
bracis-33605	77	17	)	)	PUNCT
bracis-33605	77	18	.	.	PUNCT
bracis-33605	78	1	thus	thus	ADV
bracis-33605	78	2	,	,	PUNCT
bracis-33605	78	3	fsspude	fsspude	NOUN
bracis-33605	78	4	can	can	AUX
bracis-33605	78	5	give	give	VERB
bracis-33605	78	6	more	more	ADJ
bracis-33605	78	7	or	or	CCONJ
bracis-33605	78	8	less	less	ADJ
bracis-33605	78	9	importance	importance	NOUN
bracis-33605	78	10	to	to	PART
bracis-33605	78	11	avoid	avoid	VERB
bracis-33605	78	12	dead	dead	ADJ
bracis-33605	78	13	-	-	PUNCT
bracis-33605	78	14	end	end	NOUN
bracis-33605	78	15	by	by	ADP
bracis-33605	78	16	choosing	choose	VERB
bracis-33605	78	17	an	an	DET
bracis-33605	78	18	appropriate	appropriate	ADJ
bracis-33605	78	19	d.	d.	NOUN
bracis-33605	78	20	the	the	DET
bracis-33605	78	21	discounted	discount	VERB
bracis-33605	78	22	cost	cost	NOUN
bracis-33605	78	23	criterion	criterion	NOUN
bracis-33605	78	24	uses	use	VERB
bracis-33605	78	25	the	the	DET
bracis-33605	78	26	discount	discount	NOUN
bracis-33605	78	27	factor	factor	NOUN
bracis-33605	78	28	to	to	PART
bracis-33605	78	29	allow	allow	VERB
bracis-33605	78	30	policies	policy	NOUN
bracis-33605	78	31	with	with	ADP
bracis-33605	78	32	a	a	DET
bracis-33605	78	33	probability	probability	NOUN
bracis-33605	78	34	of	of	ADP
bracis-33605	78	35	reaching	reach	VERB
bracis-33605	78	36	the	the	DET
bracis-33605	78	37	goal	goal	NOUN
bracis-33605	78	38	less	less	ADJ
bracis-33605	78	39	than	than	ADP
bracis-33605	78	40	1	1	NUM
bracis-33605	78	41	,	,	PUNCT
bracis-33605	78	42	to	to	PART
bracis-33605	78	43	converge	converge	VERB
bracis-33605	78	44	[	[	X
bracis-33605	78	45	13	13	NUM
bracis-33605	78	46	]	]	PUNCT
bracis-33605	78	47	.	.	PUNCT
bracis-33605	79	1	the	the	DET
bracis-33605	79	2	objective	objective	NOUN
bracis-33605	79	3	is	be	AUX
bracis-33605	79	4	to	to	PART
bracis-33605	79	5	minimize	minimize	VERB
bracis-33605	79	6	the	the	DET
bracis-33605	79	7	expected	expect	VERB
bracis-33605	79	8	accumulated	accumulate	VERB
bracis-33605	79	9	cost	cost	NOUN
bracis-33605	79	10	\(\overline{c}_g(s	\(\overline{c}_g(s	NOUN
bracis-33605	79	11	)	)	PUNCT
bracis-33605	80	1	=	=	VERB
bracis-33605	80	2	\lim	\lim	NOUN
bracis-33605	80	3	_	_	PRON
bracis-33605	80	4	{	{	PUNCT
bracis-33605	80	5	t\rightarrow	t\rightarrow	ADJ
bracis-33605	80	6	\infty	\infty	PROPN
bracis-33605	80	7	}	}	PUNCT
bracis-33605	80	8	e[\sum	e[\sum	VERB
bracis-33605	80	9	^{t-1}_{t=0	^{t-1}_{t=0	ADJ
bracis-33605	80	10	}	}	PUNCT
bracis-33605	80	11	\gamma	\gamma	ADP
bracis-33605	80	12	^t	^t	PROPN
bracis-33605	80	13	c	c	PROPN
bracis-33605	80	14	(	(	PUNCT
bracis-33605	80	15	s_t	s_t	NOUN
bracis-33605	80	16	,	,	PUNCT
bracis-33605	80	17	a_t)|	a_t)|	ADJ
bracis-33605	80	18	\pi	\pi	NOUN
bracis-33605	80	19	,	,	PUNCT
bracis-33605	80	20	s_0]\	s_0]\	NOUN
bracis-33605	80	21	)	)	PUNCT
bracis-33605	80	22	.	.	PUNCT
bracis-33605	81	1	the	the	DET
bracis-33605	81	2	penalty	penalty	NOUN
bracis-33605	81	3	d	d	AUX
bracis-33605	81	4	used	use	VERB
bracis-33605	81	5	in	in	ADP
bracis-33605	81	6	fsspude	fsspude	NOUN
bracis-33605	81	7	and	and	CCONJ
bracis-33605	81	8	the	the	DET
bracis-33605	81	9	discount	discount	NOUN
bracis-33605	81	10	factor	factor	NOUN
bracis-33605	81	11	used	use	VERB
bracis-33605	81	12	in	in	ADP
bracis-33605	81	13	discounted	discount	VERB
bracis-33605	81	14	cost	cost	NOUN
bracis-33605	81	15	criterion	criterion	NOUN
bracis-33605	81	16	depend	depend	VERB
bracis-33605	81	17	on	on	ADP
bracis-33605	81	18	the	the	DET
bracis-33605	81	19	ssp	ssp	NOUN
bracis-33605	81	20	.	.	PUNCT
bracis-33605	82	1	because	because	SCONJ
bracis-33605	82	2	it	it	PRON
bracis-33605	82	3	is	be	AUX
bracis-33605	82	4	necessary	necessary	ADJ
bracis-33605	82	5	to	to	PART
bracis-33605	82	6	know	know	VERB
bracis-33605	82	7	the	the	DET
bracis-33605	82	8	structure	structure	NOUN
bracis-33605	82	9	of	of	ADP
bracis-33605	82	10	the	the	DET
bracis-33605	82	11	ssp	ssp	NOUN
bracis-33605	82	12	,	,	PUNCT
bracis-33605	82	13	both	both	DET
bracis-33605	82	14	values	value	NOUN
bracis-33605	82	15	are	be	AUX
bracis-33605	82	16	difficult	difficult	ADJ
bracis-33605	82	17	to	to	PART
bracis-33605	82	18	determine	determine	VERB
bracis-33605	82	19	to	to	PART
bracis-33605	82	20	make	make	VERB
bracis-33605	82	21	the	the	DET
bracis-33605	82	22	desired	desire	VERB
bracis-33605	82	23	trade	trade	NOUN
bracis-33605	82	24	-	-	PUNCT
bracis-33605	82	25	off	off	NOUN
bracis-33605	82	26	between	between	ADP
bracis-33605	82	27	probability	probability	NOUN
bracis-33605	82	28	and	and	CCONJ
bracis-33605	82	29	cost	cost	NOUN
bracis-33605	82	30	[	[	X
bracis-33605	82	31	1	1	NUM
bracis-33605	82	32	]	]	PUNCT
bracis-33605	82	33	.	.	PUNCT
bracis-33605	83	1	besides	besides	SCONJ
bracis-33605	83	2	,	,	PUNCT
bracis-33605	83	3	these	these	DET
bracis-33605	83	4	criteria	criterion	NOUN
bracis-33605	83	5	do	do	AUX
bracis-33605	83	6	not	not	PART
bracis-33605	83	7	have	have	VERB
bracis-33605	83	8	the	the	DET
bracis-33605	83	9	property	property	NOUN
bracis-33605	83	10	\(\alpha	\(\alpha	NOUN
bracis-33605	83	11	-\)strong	-\)strong	ADP
bracis-33605	83	12	probability	probability	NOUN
bracis-33605	83	13	-	-	PUNCT
bracis-33605	83	14	to	to	ADP
bracis-33605	83	15	-	-	PUNCT
bracis-33605	83	16	goal	goal	NOUN
bracis-33605	83	17	for	for	ADP
bracis-33605	83	18	\(\alpha	\(\alpha	NOUN
bracis-33605	83	19	>	>	X
bracis-33605	83	20	0\	0\	PROPN
bracis-33605	83	21	)	)	PUNCT
bracis-33605	84	1	[	[	X
bracis-33605	84	2	1	1	NUM
bracis-33605	84	3	]	]	PUNCT
bracis-33605	84	4	.	.	PUNCT
bracis-33605	85	1	definition	definition	NOUN
bracis-33605	85	2	1	1	NUM
bracis-33605	85	3	(	(	PUNCT
bracis-33605	85	4	\(\alpha	\(\alpha	NOUN
bracis-33605	85	5	-\	-\	NOUN
bracis-33605	85	6	)	)	PUNCT
bracis-33605	85	7	strong	strong	ADJ
bracis-33605	85	8	probability	probability	NOUN
bracis-33605	85	9	-	-	PUNCT
bracis-33605	85	10	to	to	ADP
bracis-33605	85	11	-	-	PUNCT
bracis-33605	85	12	goal	goal	NOUN
bracis-33605	85	13	[	[	X
bracis-33605	85	14	1	1	NUM
bracis-33605	85	15	]	]	NUM
bracis-33605	85	16	)	)	PUNCT
bracis-33605	85	17	.	.	PUNCT
bracis-33605	86	1	for	for	ADP
bracis-33605	86	2	\(0	\(0	PROPN
bracis-33605	86	3	\le	\le	PROPN
bracis-33605	86	4	\alpha	\alpha	PROPN
bracis-33605	86	5	\le	\le	PROPN
bracis-33605	86	6	1\	1\	NUM
bracis-33605	86	7	)	)	PUNCT
bracis-33605	86	8	,	,	PUNCT
bracis-33605	86	9	a	a	DET
bracis-33605	86	10	criterion	criterion	NOUN
bracis-33605	86	11	has	have	VERB
bracis-33605	86	12	\(\alpha	\(\alpha	SYM
bracis-33605	86	13	-\)strong	-\)strong	NUM
bracis-33605	86	14	probability	probability	NOUN
bracis-33605	86	15	-	-	PUNCT
bracis-33605	86	16	to	to	ADP
bracis-33605	86	17	-	-	PUNCT
bracis-33605	86	18	goal	goal	NOUN
bracis-33605	86	19	if	if	SCONJ
bracis-33605	86	20	for	for	ADP
bracis-33605	86	21	every	every	DET
bracis-33605	86	22	ssp	ssp	NOUN
bracis-33605	86	23	and	and	CCONJ
bracis-33605	86	24	for	for	ADP
bracis-33605	86	25	any	any	DET
bracis-33605	86	26	pair	pair	NOUN
bracis-33605	86	27	of	of	ADP
bracis-33605	86	28	policies	policy	NOUN
bracis-33605	86	29	\(\pi	\(\pi	PROPN
bracis-33605	86	30	,	,	PUNCT
bracis-33605	86	31	\pi	\pi	PROPN
bracis-33605	86	32	'	'	PUNCT
bracis-33605	86	33	\in	\in	PROPN
bracis-33605	86	34	\varpi	\varpi	PROPN
bracis-33605	86	35	\	\	PROPN
bracis-33605	86	36	)	)	PUNCT
bracis-33605	86	37	,	,	PUNCT
bracis-33605	86	38	the	the	DET
bracis-33605	86	39	following	follow	VERB
bracis-33605	86	40	condition	condition	NOUN
bracis-33605	86	41	is	be	AUX
bracis-33605	86	42	true	true	ADJ
bracis-33605	86	43	:	:	PUNCT
bracis-33605	87	1	\(\pi	\(\pi	PROPN
bracis-33605	87	2	\succ	\succ	PROPN
bracis-33605	87	3	\pi	\pi	PROPN
bracis-33605	87	4	'	'	PUNCT
bracis-33605	87	5	\longrightarrow	\longrightarrow	PROPN
bracis-33605	87	6	\frac{p^{\pi	\frac{p^{\pi	PROPN
bracis-33605	87	7	}	}	PUNCT
bracis-33605	87	8	_	_	PUNCT
bracis-33605	87	9	g(s_0)}{p^{\pi	g(s_0)}{p^{\pi	NOUN
bracis-33605	87	10	'	'	NUM
bracis-33605	87	11	}	}	PUNCT
bracis-33605	87	12	_	_	PUNCT
bracis-33605	87	13	g(s_0	g(s_0	ADJ
bracis-33605	87	14	)	)	PUNCT
bracis-33605	87	15	}	}	PUNCT
bracis-33605	87	16	\ge	\ge	PROPN
bracis-33605	87	17	\alpha	\alpha	ADJ
bracis-33605	87	18	\	\	PROPN
bracis-33605	87	19	)	)	PUNCT
bracis-33605	87	20	.	.	PUNCT
bracis-33605	88	1	where	where	SCONJ
bracis-33605	88	2	\(\pi	\(\pi	NOUN
bracis-33605	88	3	\succ	\succ	PROPN
bracis-33605	88	4	\pi	\pi	PROPN
bracis-33605	88	5	'	'	PUNCT
bracis-33605	88	6	\	\	NOUN
bracis-33605	88	7	)	)	PUNCT
bracis-33605	88	8	means	mean	VERB
bracis-33605	88	9	that	that	SCONJ
bracis-33605	88	10	the	the	DET
bracis-33605	88	11	decision	decision	NOUN
bracis-33605	88	12	-	-	PUNCT
bracis-33605	88	13	maker	maker	NOUN
bracis-33605	88	14	prefers	prefer	VERB
bracis-33605	88	15	\(\pi	\(\pi	ADP
bracis-33605	88	16	\	\	NOUN
bracis-33605	88	17	)	)	PUNCT
bracis-33605	88	18	to	to	ADP
bracis-33605	88	19	\(\pi	\(\pi	VERB
bracis-33605	88	20	'	'	PUNCT
bracis-33605	88	21	\	\	NOUN
bracis-33605	88	22	)	)	PUNCT
bracis-33605	88	23	.	.	PUNCT
bracis-33605	89	1	2.2	2.2	NUM
bracis-33605	89	2	goals	goal	NOUN
bracis-33605	89	3	with	with	ADP
bracis-33605	89	4	 	 	SPACE
bracis-33605	89	5	utility	utility	NOUN
bracis-33605	89	6	-	-	PUNCT
bracis-33605	89	7	based	base	VERB
bracis-33605	89	8	semantic	semantic	ADJ
bracis-33605	89	9	-	-	PUNCT
bracis-33605	89	10	gubs	gub	NOUN
bracis-33605	89	11	a	a	DET
bracis-33605	89	12	criterion	criterion	NOUN
bracis-33605	89	13	that	that	PRON
bracis-33605	89	14	solves	solve	NOUN
bracis-33605	89	15	ssps	ssps	VERB
bracis-33605	89	16	with	with	ADP
bracis-33605	89	17	dead	dead	ADJ
bracis-33605	89	18	-	-	PUNCT
bracis-33605	89	19	end	end	NOUN
bracis-33605	89	20	with	with	ADP
bracis-33605	89	21	a	a	DET
bracis-33605	89	22	clear	clear	ADJ
bracis-33605	89	23	trade	trade	NOUN
bracis-33605	89	24	-	-	PUNCT
bracis-33605	89	25	off	off	NOUN
bracis-33605	89	26	between	between	ADP
bracis-33605	89	27	cost	cost	NOUN
bracis-33605	89	28	and	and	CCONJ
bracis-33605	89	29	probability	probability	NOUN
bracis-33605	89	30	of	of	ADP
bracis-33605	89	31	achieving	achieve	VERB
bracis-33605	89	32	the	the	DET
bracis-33605	89	33	goal	goal	NOUN
bracis-33605	89	34	is	be	AUX
bracis-33605	89	35	the	the	DET
bracis-33605	89	36	 	 	SPACE
bracis-33605	89	37	gubs	gub	NOUN
bracis-33605	89	38	[	[	X
bracis-33605	89	39	4	4	NUM
bracis-33605	89	40	]	]	PUNCT
bracis-33605	89	41	criterion	criterion	NOUN
bracis-33605	89	42	.	.	PUNCT
bracis-33605	90	1	gubs	gub	NOUN
bracis-33605	90	2	is	be	AUX
bracis-33605	90	3	based	base	VERB
bracis-33605	90	4	on	on	ADP
bracis-33605	90	5	utility	utility	NOUN
bracis-33605	90	6	theory	theory	NOUN
bracis-33605	90	7	and	and	CCONJ
bracis-33605	90	8	evaluates	evaluate	VERB
bracis-33605	90	9	policies	policy	NOUN
bracis-33605	90	10	based	base	VERB
bracis-33605	90	11	on	on	ADP
bracis-33605	90	12	:	:	PUNCT
bracis-33605	90	13	$	$	SYM
bracis-33605	90	14	$	$	SYM
bracis-33605	90	15	\begin{aligned	\begin{aligne	VERB
bracis-33605	90	16	}	}	PUNCT
bracis-33605	90	17	\begin{aligned	\begin{aligne	VERB
bracis-33605	90	18	}	}	PUNCT
bracis-33605	90	19	v^\pi	v^\pi	X
bracis-33605	90	20	(	(	PUNCT
bracis-33605	90	21	s	s	NOUN
bracis-33605	90	22	)	)	PUNCT
bracis-33605	90	23	&	&	CCONJ
bracis-33605	90	24	=	=	NOUN
bracis-33605	90	25	\lim	\lim	PROPN
bracis-33605	90	26	_	_	PUNCT
bracis-33605	90	27	{	{	PUNCT
bracis-33605	90	28	t	t	NOUN
bracis-33605	90	29	\rightarrow	\rightarrow	PROPN
bracis-33605	90	30	\infty	\infty	PROPN
bracis-33605	90	31	}	}	PUNCT
bracis-33605	90	32	e[u(c_t,\beta	e[u(c_t,\beta	X
bracis-33605	90	33	_	_	NOUN
bracis-33605	90	34	t)|\pi	t)|\pi	NOUN
bracis-33605	90	35	,	,	PUNCT
bracis-33605	90	36	s_0	s_0	PROPN
bracis-33605	90	37	=	=	SYM
bracis-33605	90	38	s	s	X
bracis-33605	90	39	]	]	X
bracis-33605	90	40	,	,	PUNCT
bracis-33605	90	41	\\	\\	NOUN
bracis-33605	90	42	u(c_t,\beta	u(c_t,\beta	PROPN
bracis-33605	90	43	_	_	PROPN
bracis-33605	90	44	t	t	PROPN
bracis-33605	90	45	)	)	PUNCT
bracis-33605	90	46	&	&	CCONJ
bracis-33605	90	47	=	=	SYM
bracis-33605	90	48	u(c_t	u(c_t	PROPN
bracis-33605	90	49	)	)	PUNCT
bracis-33605	90	50	+	+	PUNCT
bracis-33605	91	1	k_g	k_g	X
bracis-33605	91	2	\beta	\beta	PROPN
bracis-33605	91	3	_	_	PROPN
bracis-33605	91	4	t	t	PROPN
bracis-33605	91	5	,	,	PUNCT
bracis-33605	91	6	\end{aligned	\end{aligne	VERB
bracis-33605	91	7	}	}	PUNCT
bracis-33605	91	8	\end{aligned}$$	\end{aligned}$$	X
bracis-33605	91	9	(	(	PUNCT
bracis-33605	91	10	1	1	NUM
bracis-33605	91	11	)	)	PUNCT
bracis-33605	92	1	where	where	SCONJ
bracis-33605	92	2	,	,	PUNCT
bracis-33605	92	3	u	u	NOUN
bracis-33605	92	4	(	(	PUNCT
bracis-33605	92	5	.	.	PUNCT
bracis-33605	92	6	)	)	PUNCT
bracis-33605	92	7	is	be	AUX
bracis-33605	92	8	a	a	DET
bracis-33605	92	9	utility	utility	NOUN
bracis-33605	92	10	function	function	NOUN
bracis-33605	92	11	,	,	PUNCT
bracis-33605	92	12	\(c_t	\(c_t	NOUN
bracis-33605	92	13	=	=	SYM
bracis-33605	92	14	\sum	\sum	NOUN
bracis-33605	92	15	^{t-1}_{t=0}c(s_t	^{t-1}_{t=0}c(s_t	PUNCT
bracis-33605	92	16	,	,	PUNCT
bracis-33605	92	17	a_t)\	a_t)\	NOUN
bracis-33605	92	18	)	)	PUNCT
bracis-33605	93	1	is	be	AUX
bracis-33605	93	2	the	the	DET
bracis-33605	93	3	accumulated	accumulate	VERB
bracis-33605	93	4	cost	cost	NOUN
bracis-33605	93	5	in	in	ADP
bracis-33605	93	6	time	time	NOUN
bracis-33605	93	7	t	t	PROPN
bracis-33605	93	8	,	,	PUNCT
bracis-33605	93	9	\(k_g	\(k_g	PROPN
bracis-33605	93	10	\	\	NOUN
bracis-33605	93	11	)	)	PUNCT
bracis-33605	93	12	is	be	AUX
bracis-33605	93	13	the	the	DET
bracis-33605	93	14	reward	reward	NOUN
bracis-33605	93	15	for	for	ADP
bracis-33605	93	16	reaching	reach	VERB
bracis-33605	93	17	a	a	DET
bracis-33605	93	18	goal	goal	NOUN
bracis-33605	93	19	state	state	NOUN
bracis-33605	93	20	,	,	PUNCT
bracis-33605	93	21	and	and	CCONJ
bracis-33605	94	1	\(\beta	\(\beta	ADP
bracis-33605	94	2	_	_	PUNCT
bracis-33605	94	3	t\	t\	VERB
bracis-33605	94	4	)	)	PUNCT
bracis-33605	94	5	is	be	AUX
bracis-33605	94	6	1	1	NUM
bracis-33605	94	7	if	if	SCONJ
bracis-33605	94	8	a	a	DET
bracis-33605	94	9	goal	goal	NOUN
bracis-33605	94	10	is	be	AUX
bracis-33605	94	11	reached	reach	VERB
bracis-33605	94	12	.	.	PUNCT
bracis-33605	95	1	unlike	unlike	ADP
bracis-33605	95	2	the	the	DET
bracis-33605	95	3	discounted	discount	VERB
bracis-33605	95	4	cost	cost	NOUN
bracis-33605	95	5	criterion	criterion	NOUN
bracis-33605	95	6	and	and	CCONJ
bracis-33605	95	7	fsspude	fsspude	NOUN
bracis-33605	95	8	,	,	PUNCT
bracis-33605	95	9	in	in	ADP
bracis-33605	95	10	which	which	PRON
bracis-33605	95	11	the	the	DET
bracis-33605	95	12	choice	choice	NOUN
bracis-33605	95	13	of	of	ADP
bracis-33605	95	14	penalty	penalty	NOUN
bracis-33605	95	15	and	and	CCONJ
bracis-33605	95	16	discount	discount	NOUN
bracis-33605	95	17	factor	factor	NOUN
bracis-33605	95	18	depend	depend	VERB
bracis-33605	95	19	on	on	ADP
bracis-33605	95	20	the	the	DET
bracis-33605	95	21	ssp	ssp	NOUN
bracis-33605	95	22	,	,	PUNCT
bracis-33605	95	23	the	the	DET
bracis-33605	95	24	value	value	NOUN
bracis-33605	95	25	\(k_g\	\(k_g\	NOUN
bracis-33605	95	26	)	)	PUNCT
bracis-33605	95	27	do	do	AUX
bracis-33605	95	28	not	not	PART
bracis-33605	95	29	depend	depend	VERB
bracis-33605	95	30	on	on	ADP
bracis-33605	95	31	the	the	DET
bracis-33605	95	32	ssp	ssp	NOUN
bracis-33605	95	33	.	.	PUNCT
bracis-33605	96	1	additionally	additionally	ADV
bracis-33605	96	2	,	,	PUNCT
bracis-33605	96	3	gubs	gub	NOUN
bracis-33605	96	4	has	have	VERB
bracis-33605	96	5	the	the	DET
bracis-33605	96	6	\(\alpha	\(\alpha	NOUN
bracis-33605	96	7	-\)strong	-\)strong	PROPN
bracis-33605	96	8	probability	probability	NOUN
bracis-33605	96	9	-	-	PUNCT
bracis-33605	96	10	to	to	ADP
bracis-33605	96	11	-	-	PUNCT
bracis-33605	96	12	goal	goal	NOUN
bracis-33605	96	13	property	property	NOUN
bracis-33605	96	14	[	[	X
bracis-33605	96	15	1	1	NUM
bracis-33605	96	16	]	]	PUNCT
bracis-33605	96	17	.	.	PUNCT
bracis-33605	97	1	some	some	DET
bracis-33605	97	2	planning	planning	NOUN
bracis-33605	97	3	algorithms	algorithm	NOUN
bracis-33605	97	4	have	have	AUX
bracis-33605	97	5	been	be	AUX
bracis-33605	97	6	proposed	propose	VERB
bracis-33605	97	7	to	to	PART
bracis-33605	97	8	calculate	calculate	VERB
bracis-33605	97	9	the	the	DET
bracis-33605	97	10	gubs	gubs	ADJ
bracis-33605	97	11	criterion	criterion	NOUN
bracis-33605	97	12	,	,	PUNCT
bracis-33605	97	13	they	they	PRON
bracis-33605	97	14	are	be	AUX
bracis-33605	97	15	:	:	PUNCT
bracis-33605	97	16	gubs	gubs	ADJ
bracis-33605	97	17	-	-	PUNCT
bracis-33605	97	18	vi	vi	NOUN
bracis-33605	98	1	[	[	X
bracis-33605	98	2	4	4	NUM
bracis-33605	98	3	]	]	PUNCT
bracis-33605	98	4	,	,	PUNCT
bracis-33605	98	5	egubs	egub	NOUN
bracis-33605	98	6	-	-	PUNCT
bracis-33605	98	7	vi	vi	NOUN
bracis-33605	99	1	[	[	X
bracis-33605	99	2	3	3	NUM
bracis-33605	99	3	]	]	PUNCT
bracis-33605	99	4	and	and	CCONJ
bracis-33605	99	5	egubs	egub	NOUN
bracis-33605	99	6	-	-	PUNCT
bracis-33605	99	7	ao	ao	NOUN
bracis-33605	99	8	*	*	PUNCT
bracis-33605	100	1	[	[	X
bracis-33605	100	2	1	1	NUM
bracis-33605	100	3	]	]	PUNCT
bracis-33605	100	4	.	.	PUNCT
bracis-33605	101	1	the	the	DET
bracis-33605	101	2	gubs	gubs	ADJ
bracis-33605	101	3	-	-	PUNCT
bracis-33605	101	4	vi	vi	NOUN
bracis-33605	101	5	algorithm	algorithm	NOUN
bracis-33605	101	6	,	,	PUNCT
bracis-33605	101	7	which	which	PRON
bracis-33605	101	8	is	be	AUX
bracis-33605	101	9	based	base	VERB
bracis-33605	101	10	on	on	ADP
bracis-33605	101	11	the	the	DET
bracis-33605	101	12	value	value	NOUN
bracis-33605	101	13	iteration	iteration	NOUN
bracis-33605	101	14	(	(	PUNCT
bracis-33605	101	15	vi	vi	NOUN
bracis-33605	101	16	)	)	PUNCT
bracis-33605	101	17	algorithm	algorithm	NOUN
bracis-33605	101	18	solves	solve	VERB
bracis-33605	101	19	gubs	gub	NOUN
bracis-33605	101	20	for	for	ADP
bracis-33605	101	21	discrete	discrete	ADJ
bracis-33605	101	22	costs	cost	NOUN
bracis-33605	101	23	and	and	CCONJ
bracis-33605	101	24	considers	consider	VERB
bracis-33605	101	25	as	as	ADP
bracis-33605	101	26	a	a	DET
bracis-33605	101	27	parameter	parameter	NOUN
bracis-33605	101	28	the	the	DET
bracis-33605	101	29	maximum	maximum	ADJ
bracis-33605	101	30	accumulated	accumulate	VERB
bracis-33605	101	31	cost	cost	NOUN
bracis-33605	101	32	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	101	33	)	)	PUNCT
bracis-33605	101	34	for	for	ADP
bracis-33605	101	35	every	every	DET
bracis-33605	101	36	state	state	NOUN
bracis-33605	101	37	,	,	PUNCT
bracis-33605	101	38	including	include	VERB
bracis-33605	101	39	dead	dead	ADJ
bracis-33605	101	40	-	-	PUNCT
bracis-33605	101	41	end	end	NOUN
bracis-33605	101	42	states	state	NOUN
bracis-33605	101	43	,	,	PUNCT
bracis-33605	101	44	which	which	PRON
bracis-33605	101	45	will	will	AUX
bracis-33605	101	46	have	have	VERB
bracis-33605	101	47	a	a	DET
bracis-33605	101	48	cost	cost	NOUN
bracis-33605	101	49	of	of	ADP
bracis-33605	101	50	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	101	51	)	)	PUNCT
bracis-33605	101	52	.	.	PUNCT
bracis-33605	102	1	as	as	SCONJ
bracis-33605	102	2	the	the	DET
bracis-33605	102	3	algorithm	algorithm	NOUN
bracis-33605	102	4	needs	need	VERB
bracis-33605	102	5	to	to	PART
bracis-33605	102	6	calculate	calculate	VERB
bracis-33605	102	7	the	the	DET
bracis-33605	102	8	utility	utility	NOUN
bracis-33605	102	9	based	base	VERB
bracis-33605	102	10	on	on	ADP
bracis-33605	102	11	the	the	DET
bracis-33605	102	12	accumulated	accumulate	VERB
bracis-33605	102	13	cost	cost	NOUN
bracis-33605	102	14	,	,	PUNCT
bracis-33605	102	15	the	the	DET
bracis-33605	102	16	policy	policy	NOUN
bracis-33605	102	17	found	find	VERB
bracis-33605	102	18	is	be	AUX
bracis-33605	102	19	non	non	ADJ
bracis-33605	102	20	-	-	ADJ
bracis-33605	102	21	stationary	stationary	ADJ
bracis-33605	102	22	.	.	PUNCT
bracis-33605	103	1	to	to	PART
bracis-33605	103	2	be	be	AUX
bracis-33605	103	3	able	able	ADJ
bracis-33605	103	4	to	to	PART
bracis-33605	103	5	use	use	VERB
bracis-33605	103	6	vi	vi	PROPN
bracis-33605	103	7	,	,	PUNCT
bracis-33605	103	8	extended	extend	VERB
bracis-33605	103	9	states	state	NOUN
bracis-33605	103	10	must	must	AUX
bracis-33605	103	11	be	be	AUX
bracis-33605	103	12	used	use	VERB
bracis-33605	103	13	with	with	ADP
bracis-33605	103	14	the	the	DET
bracis-33605	103	15	cost	cost	NOUN
bracis-33605	103	16	accumulated	accumulate	VERB
bracis-33605	103	17	so	so	ADV
bracis-33605	103	18	far	far	ADV
bracis-33605	103	19	.	.	PUNCT
bracis-33605	104	1	the	the	DET
bracis-33605	104	2	formulation	formulation	NOUN
bracis-33605	104	3	with	with	ADP
bracis-33605	104	4	discrete	discrete	ADJ
bracis-33605	104	5	cost	cost	NOUN
bracis-33605	104	6	and	and	CCONJ
bracis-33605	104	7	extended	extended	ADJ
bracis-33605	104	8	states	state	NOUN
bracis-33605	104	9	for	for	ADP
bracis-33605	104	10	gubs	gub	NOUN
bracis-33605	104	11	is	be	AUX
bracis-33605	104	12	presented	present	VERB
bracis-33605	104	13	next	next	ADV
bracis-33605	104	14	.	.	PUNCT
bracis-33605	105	1	definition	definition	NOUN
bracis-33605	105	2	2	2	NUM
bracis-33605	105	3	(	(	PUNCT
bracis-33605	105	4	discrete	discrete	ADJ
bracis-33605	105	5	cost	cost	NOUN
bracis-33605	105	6	[	[	X
bracis-33605	105	7	4	4	NUM
bracis-33605	105	8	]	]	NUM
bracis-33605	105	9	)	)	PUNCT
bracis-33605	105	10	.	.	PUNCT
bracis-33605	106	1	an	an	DET
bracis-33605	106	2	ssp	ssp	NOUN
bracis-33605	106	3	with	with	ADP
bracis-33605	106	4	dead	dead	ADJ
bracis-33605	106	5	-	-	PUNCT
bracis-33605	106	6	end	end	NOUN
bracis-33605	106	7	can	can	AUX
bracis-33605	106	8	be	be	AUX
bracis-33605	106	9	restated	restate	VERB
bracis-33605	106	10	into	into	ADP
bracis-33605	106	11	an	an	DET
bracis-33605	106	12	mdp	mdp	NOUN
bracis-33605	106	13	with	with	ADP
bracis-33605	106	14	bounded	bounded	ADJ
bracis-33605	106	15	discrete	discrete	ADJ
bracis-33605	106	16	costs	cost	NOUN
bracis-33605	106	17	\(m=	\(m=	PROPN
bracis-33605	107	1	\langle	\langle	PROPN
bracis-33605	107	2	x	x	SYM
bracis-33605	107	3	,	,	PUNCT
bracis-33605	107	4	a	a	DET
bracis-33605	107	5	,	,	PUNCT
bracis-33605	107	6	p',c',r'_g	p',c',r'_g	NOUN
bracis-33605	107	7	\rangle	\rangle	PROPN
bracis-33605	107	8	\	\	NOUN
bracis-33605	107	9	)	)	PUNCT
bracis-33605	107	10	,	,	PUNCT
bracis-33605	107	11	where	where	SCONJ
bracis-33605	107	12	,	,	PUNCT
bracis-33605	107	13	\(x	\(x	NOUN
bracis-33605	107	14	=	=	SYM
bracis-33605	107	15	{	{	PUNCT
bracis-33605	107	16	s	s	NOUN
bracis-33605	107	17	}	}	PUNCT
bracis-33605	107	18	\times	\time	NOUN
bracis-33605	107	19	\mathbb	\mathbb	PROPN
bracis-33605	107	20	{	{	PUNCT
bracis-33605	107	21	n}\	n}\	NOUN
bracis-33605	107	22	)	)	PUNCT
bracis-33605	107	23	is	be	AUX
bracis-33605	107	24	an	an	DET
bracis-33605	107	25	extended	extended	ADJ
bracis-33605	107	26	state	state	NOUN
bracis-33605	107	27	space	space	NOUN
bracis-33605	107	28	,	,	PUNCT
bracis-33605	107	29	which	which	PRON
bracis-33605	107	30	are	be	AUX
bracis-33605	107	31	the	the	DET
bracis-33605	107	32	states	state	NOUN
bracis-33605	107	33	with	with	ADP
bracis-33605	107	34	the	the	DET
bracis-33605	107	35	accumulated	accumulate	VERB
bracis-33605	107	36	cost	cost	NOUN
bracis-33605	107	37	so	so	ADV
bracis-33605	107	38	far	far	ADV
bracis-33605	107	39	,	,	PUNCT
bracis-33605	107	40	\(p'(x	\(p'(x	NOUN
bracis-33605	107	41	,	,	PUNCT
bracis-33605	107	42	a	a	PRON
bracis-33605	107	43	,	,	PUNCT
bracis-33605	107	44	x')=p(s	x')=p(s	NUM
bracis-33605	107	45	,	,	PUNCT
bracis-33605	107	46	a	a	DET
bracis-33605	107	47	,	,	PUNCT
bracis-33605	107	48	s')\	s')\	NOUN
bracis-33605	107	49	)	)	PUNCT
bracis-33605	107	50	where	where	SCONJ
bracis-33605	107	51	,	,	PUNCT
bracis-33605	107	52	\(x	\(x	NOUN
bracis-33605	107	53	=	=	SYM
bracis-33605	107	54	(	(	PUNCT
bracis-33605	107	55	s	s	PROPN
bracis-33605	107	56	,	,	PUNCT
bracis-33605	107	57	c)\	c)\	NOUN
bracis-33605	107	58	)	)	PUNCT
bracis-33605	107	59	and	and	CCONJ
bracis-33605	107	60	\(x'=(s',c+c(s	\(x'=(s',c+c(	NOUN
bracis-33605	107	61	,	,	PUNCT
bracis-33605	107	62	a))\	a))\	PROPN
bracis-33605	107	63	)	)	PUNCT
bracis-33605	107	64	,	,	PUNCT
bracis-33605	107	65	\(x	\(x	NOUN
bracis-33605	107	66	,	,	PUNCT
bracis-33605	107	67	x	x	NOUN
bracis-33605	107	68	'	'	CCONJ
bracis-33605	107	69	\in	\in	PROPN
bracis-33605	107	70	x\	x\	NOUN
bracis-33605	107	71	)	)	PUNCT
bracis-33605	107	72	and	and	CCONJ
bracis-33605	107	73	c	c	PROPN
bracis-33605	107	74	is	be	AUX
bracis-33605	107	75	the	the	DET
bracis-33605	107	76	accumulated	accumulate	VERB
bracis-33605	107	77	cost	cost	NOUN
bracis-33605	107	78	,	,	PUNCT
bracis-33605	107	79	\(c'(x	\(c'(x	NOUN
bracis-33605	107	80	,	,	PUNCT
bracis-33605	107	81	a)=u	a)=u	X
bracis-33605	107	82	(	(	PUNCT
bracis-33605	107	83	c+c(s	c+c(s	X
bracis-33605	107	84	,	,	PUNCT
bracis-33605	107	85	a))-u(c)\)footnote	a))-u(c)\)footnote	NOUN
bracis-33605	107	86	1	1	NUM
bracis-33605	107	87	for	for	ADP
bracis-33605	107	88	a	a	DET
bracis-33605	107	89	utility	utility	NOUN
bracis-33605	107	90	function	function	NOUN
bracis-33605	107	91	u	u	NOUN
bracis-33605	107	92	(	(	PUNCT
bracis-33605	107	93	.	.	PUNCT
bracis-33605	107	94	)	)	PUNCT
bracis-33605	107	95	,	,	PUNCT
bracis-33605	107	96	and	and	CCONJ
bracis-33605	107	97	\(r'_g	\(r'_g	NUM
bracis-33605	107	98	=	=	SYM
bracis-33605	107	99	k_g\	k_g\	X
bracis-33605	107	100	)	)	PUNCT
bracis-33605	107	101	is	be	AUX
bracis-33605	107	102	the	the	DET
bracis-33605	107	103	terminal	terminal	ADJ
bracis-33605	107	104	reward	reward	NOUN
bracis-33605	107	105	for	for	ADP
bracis-33605	107	106	the	the	DET
bracis-33605	107	107	goal	goal	NOUN
bracis-33605	107	108	states	state	NOUN
bracis-33605	107	109	with	with	ADP
bracis-33605	107	110	value	value	NOUN
bracis-33605	107	111	\(k_g	\(k_g	PROPN
bracis-33605	107	112	\in	\in	PROPN
bracis-33605	107	113	\mathbb	\mathbb	PROPN
bracis-33605	107	114	{	{	PUNCT
bracis-33605	107	115	r}\	r}\	NUM
bracis-33605	107	116	)	)	PUNCT
bracis-33605	107	117	.	.	PUNCT
bracis-33605	108	1	the	the	DET
bracis-33605	108	2	gubs	gubs	ADJ
bracis-33605	108	3	-	-	PUNCT
bracis-33605	108	4	vi	vi	NOUN
bracis-33605	108	5	algorithm	algorithm	NOUN
bracis-33605	108	6	uses	use	VERB
bracis-33605	108	7	definition	definition	NOUN
bracis-33605	108	8	2	2	NUM
bracis-33605	108	9	to	to	PART
bracis-33605	108	10	calculate	calculate	VERB
bracis-33605	108	11	the	the	DET
bracis-33605	108	12	q	q	NOUN
bracis-33605	108	13	-	-	PUNCT
bracis-33605	108	14	value	value	NOUN
bracis-33605	108	15	:	:	PUNCT
bracis-33605	108	16	$	$	SYM
bracis-33605	108	17	$	$	SYM
bracis-33605	108	18	\begin{aligned	\begin{aligne	VERB
bracis-33605	108	19	}	}	PUNCT
bracis-33605	108	20	\small	\small	ADP
bracis-33605	108	21	q(s	q(s	PROPN
bracis-33605	108	22	,	,	PUNCT
bracis-33605	108	23	c	c	X
bracis-33605	108	24	,	,	PUNCT
bracis-33605	108	25	a	a	PRON
bracis-33605	108	26	)	)	PUNCT
bracis-33605	108	27	\leftarrow	\leftarrow	NOUN
bracis-33605	108	28	c'(s	c'(s	NOUN
bracis-33605	108	29	,	,	PUNCT
bracis-33605	108	30	a	a	PRON
bracis-33605	108	31	)	)	PUNCT
bracis-33605	108	32	+	+	NUM
bracis-33605	108	33	\displaystyle	\displaystyle	PROPN
bracis-33605	108	34	\sum	\sum	NOUN
bracis-33605	108	35	_	_	PUNCT
bracis-33605	108	36	{	{	PUNCT
bracis-33605	108	37	s	s	X
bracis-33605	108	38	'	'	PUNCT
bracis-33605	108	39	\in	\in	PROPN
bracis-33605	108	40	s	s	PART
bracis-33605	108	41	\cup	\cup	NOUN
bracis-33605	108	42	\mathcal	\mathcal	PROPN
bracis-33605	108	43	{	{	PUNCT
bracis-33605	108	44	g	g	NOUN
bracis-33605	108	45	}	}	PUNCT
bracis-33605	108	46	}	}	PUNCT
bracis-33605	108	47	p(s	p(s	PROPN
bracis-33605	108	48	,	,	PUNCT
bracis-33605	108	49	a	a	PRON
bracis-33605	108	50	,	,	PUNCT
bracis-33605	108	51	s')v(s',c+c(s	s')v(s',c+c(s	ADJ
bracis-33605	108	52	,	,	PUNCT
bracis-33605	108	53	a	a	DET
bracis-33605	108	54	)	)	PUNCT
bracis-33605	108	55	)	)	PUNCT
bracis-33605	108	56	.	.	PUNCT
bracis-33605	109	1	\end{aligned}$$	\end{aligned}$$	X
bracis-33605	109	2	(	(	PUNCT
bracis-33605	109	3	2	2	X
bracis-33605	109	4	)	)	PUNCT
bracis-33605	109	5	a	a	DET
bracis-33605	109	6	special	special	ADJ
bracis-33605	109	7	case	case	NOUN
bracis-33605	109	8	of	of	ADP
bracis-33605	109	9	the	the	DET
bracis-33605	109	10	gubs	gubs	ADJ
bracis-33605	109	11	criterion	criterion	NOUN
bracis-33605	109	12	is	be	AUX
bracis-33605	109	13	the	the	DET
bracis-33605	109	14	exponential	exponential	ADJ
bracis-33605	109	15	gubs	gub	NOUN
bracis-33605	109	16	(	(	PUNCT
bracis-33605	109	17	egubs	egub	NOUN
bracis-33605	109	18	)	)	PUNCT
bracis-33605	109	19	criterion	criterion	NOUN
bracis-33605	109	20	.	.	PUNCT
bracis-33605	110	1	in	in	ADP
bracis-33605	110	2	this	this	DET
bracis-33605	110	3	case	case	NOUN
bracis-33605	110	4	,	,	PUNCT
bracis-33605	110	5	the	the	DET
bracis-33605	110	6	utility	utility	NOUN
bracis-33605	110	7	function	function	NOUN
bracis-33605	110	8	is	be	AUX
bracis-33605	110	9	an	an	DET
bracis-33605	110	10	exponential	exponential	ADJ
bracis-33605	110	11	function	function	NOUN
bracis-33605	110	12	used	use	VERB
bracis-33605	110	13	in	in	ADP
bracis-33605	110	14	risk	risk	NOUN
bracis-33605	110	15	sensitive	sensitive	ADJ
bracis-33605	110	16	ssps	ssps	NOUN
bracis-33605	110	17	(	(	PUNCT
bracis-33605	110	18	risk	risk	NOUN
bracis-33605	110	19	sensitive	sensitive	ADJ
bracis-33605	110	20	ssp	ssp	ADJ
bracis-33605	110	21	rs	rs	NOUN
bracis-33605	110	22	-	-	PUNCT
bracis-33605	110	23	ssp	ssp	NOUN
bracis-33605	110	24	)	)	PUNCT
bracis-33605	111	1	[	[	X
bracis-33605	111	2	9	9	NUM
bracis-33605	111	3	]	]	PUNCT
bracis-33605	111	4	,	,	PUNCT
bracis-33605	111	5	in	in	ADP
bracis-33605	111	6	which	which	PRON
bracis-33605	111	7	the	the	DET
bracis-33605	111	8	risk	risk	NOUN
bracis-33605	111	9	factor	factor	NOUN
bracis-33605	111	10	\(\lambda	\(\lambda	NOUN
bracis-33605	111	11	\	\	NOUN
bracis-33605	111	12	)	)	PUNCT
bracis-33605	111	13	,	,	PUNCT
bracis-33605	111	14	that	that	PRON
bracis-33605	111	15	models	model	VERB
bracis-33605	111	16	the	the	DET
bracis-33605	111	17	agent	agent	NOUN
bracis-33605	111	18	’s	’s	PART
bracis-33605	111	19	risk	risk	NOUN
bracis-33605	111	20	attitude	attitude	NOUN
bracis-33605	111	21	,	,	PUNCT
bracis-33605	111	22	is	be	AUX
bracis-33605	111	23	negative	negative	ADJ
bracis-33605	111	24	.	.	PUNCT
bracis-33605	112	1	the	the	DET
bracis-33605	112	2	egubs	egub	NOUN
bracis-33605	112	3	is	be	AUX
bracis-33605	112	4	defined	define	VERB
bracis-33605	112	5	in	in	ADP
bracis-33605	112	6	terms	term	NOUN
bracis-33605	112	7	of	of	ADP
bracis-33605	112	8	the	the	DET
bracis-33605	112	9	lexicographic	lexicographic	ADJ
bracis-33605	112	10	criterion	criterion	NOUN
bracis-33605	112	11	.	.	PUNCT
bracis-33605	113	1	definition	definition	NOUN
bracis-33605	113	2	3	3	NUM
bracis-33605	113	3	(	(	PUNCT
bracis-33605	113	4	risk	risk	NOUN
bracis-33605	113	5	-	-	PUNCT
bracis-33605	113	6	sensitive	sensitive	ADJ
bracis-33605	113	7	lexicographic	lexicographic	ADJ
bracis-33605	113	8	criterion	criterion	NOUN
bracis-33605	114	1	[	[	X
bracis-33605	114	2	1	1	NUM
bracis-33605	114	3	,	,	PUNCT
bracis-33605	114	4	3	3	NUM
bracis-33605	114	5	]	]	NUM
bracis-33605	114	6	)	)	PUNCT
bracis-33605	114	7	.	.	PUNCT
bracis-33605	115	1	the	the	DET
bracis-33605	115	2	risk	risk	NOUN
bracis-33605	115	3	-	-	PUNCT
bracis-33605	115	4	sensitive	sensitive	ADJ
bracis-33605	115	5	lexicographic	lexicographic	ADJ
bracis-33605	115	6	criterion	criterion	NOUN
bracis-33605	115	7	considers	consider	VERB
bracis-33605	115	8	:	:	PUNCT
bracis-33605	115	9	if	if	SCONJ
bracis-33605	115	10	\(p_g^{\pi	\(p_g^{\pi	VERB
bracis-33605	115	11	}	}	PUNCT
bracis-33605	115	12	(	(	PUNCT
bracis-33605	115	13	s	s	NOUN
bracis-33605	115	14	)	)	PUNCT
bracis-33605	115	15	>	>	X
bracis-33605	115	16	p_g^{\pi	p_g^{\pi	PROPN
bracis-33605	115	17	'	'	NUM
bracis-33605	115	18	}	}	PUNCT
bracis-33605	115	19	(	(	PUNCT
bracis-33605	115	20	s)\	s)\	NOUN
bracis-33605	115	21	)	)	PUNCT
bracis-33605	115	22	,	,	PUNCT
bracis-33605	115	23	then	then	ADV
bracis-33605	115	24	\(\pi	\(\pi	PROPN
bracis-33605	115	25	\succ	\succ	PROPN
bracis-33605	115	26	\pi	\pi	PROPN
bracis-33605	115	27	'	'	PUNCT
bracis-33605	115	28	\	\	NOUN
bracis-33605	115	29	)	)	PUNCT
bracis-33605	115	30	;	;	PUNCT
bracis-33605	115	31	or	or	CCONJ
bracis-33605	115	32	if	if	SCONJ
bracis-33605	115	33	\(p_g^{\pi	\(p_g^{\pi	NOUN
bracis-33605	115	34	}	}	PUNCT
bracis-33605	115	35	(	(	PUNCT
bracis-33605	115	36	s	s	NOUN
bracis-33605	115	37	)	)	PUNCT
bracis-33605	115	38	=	=	SYM
bracis-33605	115	39	p_g^{\pi	p_g^{\pi	PROPN
bracis-33605	115	40	'	'	NUM
bracis-33605	115	41	}	}	PUNCT
bracis-33605	115	42	(	(	PUNCT
bracis-33605	115	43	s)\	s)\	NOUN
bracis-33605	115	44	)	)	PUNCT
bracis-33605	115	45	and	and	CCONJ
bracis-33605	115	46	\(v^{\pi	\(v^{\pi	NOUN
bracis-33605	115	47	}	}	PUNCT
bracis-33605	115	48	_	_	PUNCT
bracis-33605	115	49	{	{	PUNCT
bracis-33605	115	50	\lambda	\lambda	PROPN
bracis-33605	115	51	}	}	PUNCT
bracis-33605	115	52	>	>	PUNCT
bracis-33605	115	53	v^{\pi	v^{\pi	PROPN
bracis-33605	115	54	'	'	NUM
bracis-33605	115	55	}	}	PUNCT
bracis-33605	115	56	_	_	X
bracis-33605	115	57	{	{	PUNCT
bracis-33605	115	58	\lambda	\lambda	PROPN
bracis-33605	115	59	}	}	PUNCT
bracis-33605	115	60	\	\	PROPN
bracis-33605	115	61	)	)	PUNCT
bracis-33605	115	62	,	,	PUNCT
bracis-33605	115	63	then	then	ADV
bracis-33605	115	64	\(\pi	\(\pi	PROPN
bracis-33605	115	65	\succ	\succ	PROPN
bracis-33605	115	66	\pi	\pi	PROPN
bracis-33605	115	67	'	'	PUNCT
bracis-33605	115	68	\	\	NOUN
bracis-33605	115	69	)	)	PUNCT
bracis-33605	115	70	.	.	PUNCT
bracis-33605	116	1	the	the	DET
bracis-33605	116	2	egubs	egub	NOUN
bracis-33605	116	3	-	-	PUNCT
bracis-33605	116	4	vi	vi	NOUN
bracis-33605	116	5	algorithm	algorithm	NOUN
bracis-33605	116	6	[	[	X
bracis-33605	116	7	3	3	X
bracis-33605	116	8	]	]	X
bracis-33605	116	9	solves	solve	VERB
bracis-33605	116	10	the	the	DET
bracis-33605	116	11	egubs	egub	NOUN
bracis-33605	116	12	criterion	criterion	NOUN
bracis-33605	116	13	exactly	exactly	ADV
bracis-33605	116	14	.	.	PUNCT
bracis-33605	117	1	egubs	egub	NOUN
bracis-33605	117	2	-	-	PUNCT
bracis-33605	117	3	vi	vi	NOUN
bracis-33605	117	4	calls	call	VERB
bracis-33605	117	5	the	the	DET
bracis-33605	117	6	risk	risk	NOUN
bracis-33605	117	7	-	-	PUNCT
bracis-33605	117	8	sensitive	sensitive	ADJ
bracis-33605	117	9	-	-	PUNCT
bracis-33605	117	10	lexicographic	lexicographic	ADJ
bracis-33605	117	11	-	-	PUNCT
bracis-33605	117	12	vi	vi	NOUN
bracis-33605	117	13	algorithm	algorithm	NOUN
bracis-33605	117	14	[	[	X
bracis-33605	117	15	3	3	X
bracis-33605	117	16	]	]	PUNCT
bracis-33605	117	17	which	which	PRON
bracis-33605	117	18	receives	receive	VERB
bracis-33605	117	19	an	an	DET
bracis-33605	117	20	ssp	ssp	NOUN
bracis-33605	117	21	and	and	CCONJ
bracis-33605	117	22	a	a	DET
bracis-33605	117	23	\(\lambda	\(\lambda	NOUN
bracis-33605	117	24	\	\	NOUN
bracis-33605	117	25	)	)	PUNCT
bracis-33605	117	26	and	and	CCONJ
bracis-33605	117	27	calculates	calculate	VERB
bracis-33605	117	28	the	the	DET
bracis-33605	117	29	optimal	optimal	ADJ
bracis-33605	117	30	policy	policy	NOUN
bracis-33605	117	31	\(\pi	\(\pi	PROPN
bracis-33605	117	32	^*_l\	^*_l\	NUM
bracis-33605	117	33	)	)	PUNCT
bracis-33605	117	34	,	,	PUNCT
bracis-33605	117	35	the	the	DET
bracis-33605	117	36	optimal	optimal	ADJ
bracis-33605	117	37	value	value	NOUN
bracis-33605	117	38	\(v	\(v	VERB
bracis-33605	117	39	_	_	PRON
bracis-33605	117	40	{	{	PUNCT
bracis-33605	117	41	\lambda	\lambda	PROPN
bracis-33605	117	42	}	}	PUNCT
bracis-33605	117	43	\	\	NOUN
bracis-33605	117	44	)	)	PUNCT
bracis-33605	117	45	and	and	CCONJ
bracis-33605	117	46	the	the	DET
bracis-33605	117	47	maximum	maximum	ADJ
bracis-33605	117	48	probability	probability	NOUN
bracis-33605	117	49	for	for	ADP
bracis-33605	117	50	the	the	DET
bracis-33605	117	51	goal	goal	NOUN
bracis-33605	117	52	\(p_g\	\(p_g\	PROPN
bracis-33605	117	53	)	)	PUNCT
bracis-33605	117	54	for	for	ADP
bracis-33605	117	55	the	the	DET
bracis-33605	117	56	lexicographic	lexicographic	ADJ
bracis-33605	117	57	criterion	criterion	NOUN
bracis-33605	117	58	described	describe	VERB
bracis-33605	117	59	in	in	ADP
bracis-33605	117	60	definition	definition	NOUN
bracis-33605	117	61	3	3	NUM
bracis-33605	117	62	.	.	PUNCT
bracis-33605	118	1	we	we	PRON
bracis-33605	118	2	will	will	AUX
bracis-33605	118	3	use	use	VERB
bracis-33605	118	4	risk	risk	NOUN
bracis-33605	118	5	-	-	PUNCT
bracis-33605	118	6	sensitive	sensitive	ADJ
bracis-33605	118	7	-	-	PUNCT
bracis-33605	118	8	lexicographic	lexicographic	ADJ
bracis-33605	118	9	-	-	PUNCT
bracis-33605	118	10	vi	vi	NOUN
bracis-33605	118	11	as	as	ADP
bracis-33605	118	12	part	part	NOUN
bracis-33605	118	13	of	of	ADP
bracis-33605	118	14	one	one	NUM
bracis-33605	118	15	of	of	ADP
bracis-33605	118	16	the	the	DET
bracis-33605	118	17	algorithms	algorithm	NOUN
bracis-33605	118	18	proposed	propose	VERB
bracis-33605	118	19	in	in	ADP
bracis-33605	118	20	this	this	DET
bracis-33605	118	21	work	work	NOUN
bracis-33605	118	22	.	.	PUNCT
bracis-33605	119	1	the	the	DET
bracis-33605	119	2	egubs	egub	NOUN
bracis-33605	119	3	-	-	PUNCT
bracis-33605	119	4	ao	ao	PROPN
bracis-33605	119	5	*	*	PROPN
bracis-33605	119	6	algorithm	algorithm	NOUN
bracis-33605	119	7	optimizes	optimize	VERB
bracis-33605	119	8	the	the	DET
bracis-33605	119	9	search	search	NOUN
bracis-33605	119	10	for	for	ADP
bracis-33605	119	11	the	the	DET
bracis-33605	119	12	solution	solution	NOUN
bracis-33605	119	13	considering	consider	VERB
bracis-33605	119	14	a	a	DET
bracis-33605	119	15	finite	finite	NOUN
bracis-33605	119	16	,	,	PUNCT
bracis-33605	119	17	acyclic	acyclic	ADJ
bracis-33605	119	18	hypergraph	hypergraph	VERB
bracis-33605	119	19	,	,	PUNCT
bracis-33605	119	20	whose	whose	DET
bracis-33605	119	21	nodes	node	NOUN
bracis-33605	119	22	are	be	AUX
bracis-33605	119	23	augmented	augment	VERB
bracis-33605	119	24	states	state	NOUN
bracis-33605	119	25	(	(	PUNCT
bracis-33605	119	26	s	s	X
bracis-33605	119	27	,	,	PUNCT
bracis-33605	119	28	 	 	SPACE
bracis-33605	119	29	c	c	NOUN
bracis-33605	119	30	)	)	PUNCT
bracis-33605	119	31	and	and	CCONJ
bracis-33605	119	32	hyperedges	hyperedge	NOUN
bracis-33605	119	33	leading	lead	VERB
bracis-33605	119	34	(	(	PUNCT
bracis-33605	119	35	s	s	PROPN
bracis-33605	119	36	,	,	PUNCT
bracis-33605	119	37	 	 	SPACE
bracis-33605	119	38	c	c	NOUN
bracis-33605	119	39	)	)	PUNCT
bracis-33605	119	40	to	to	ADP
bracis-33605	119	41	its	its	PRON
bracis-33605	119	42	successor	successor	NOUN
bracis-33605	119	43	state	state	NOUN
bracis-33605	119	44	\(s',c'\	\(s',c'\	PUNCT
bracis-33605	119	45	)	)	PUNCT
bracis-33605	119	46	,	,	PUNCT
bracis-33605	119	47	where	where	SCONJ
bracis-33605	119	48	\(s'\	\(s'\	PROPN
bracis-33605	119	49	)	)	PUNCT
bracis-33605	119	50	is	be	AUX
bracis-33605	119	51	the	the	DET
bracis-33605	119	52	successor	successor	NOUN
bracis-33605	119	53	state	state	NOUN
bracis-33605	119	54	in	in	ADP
bracis-33605	119	55	the	the	DET
bracis-33605	119	56	original	original	ADJ
bracis-33605	119	57	mdp	mdp	NOUN
bracis-33605	119	58	and	and	CCONJ
bracis-33605	119	59	\(c	\(c	NOUN
bracis-33605	119	60	'	'	PUNCT
bracis-33605	119	61	=	=	SYM
bracis-33605	119	62	c	c	NOUN
bracis-33605	119	63	+	+	CCONJ
bracis-33605	119	64	c(s	c(	NOUN
bracis-33605	119	65	,	,	PUNCT
bracis-33605	119	66	a)\	a)\	NUM
bracis-33605	119	67	)	)	PUNCT
bracis-33605	119	68	for	for	SCONJ
bracis-33605	119	69	the	the	DET
bracis-33605	119	70	action	action	NOUN
bracis-33605	119	71	performed	perform	VERB
bracis-33605	119	72	a.	a.	NOUN
bracis-33605	119	73	egubs	egub	NOUN
bracis-33605	119	74	-	-	PUNCT
bracis-33605	119	75	ao	ao	PROPN
bracis-33605	119	76	*	*	PROPN
bracis-33605	119	77	solves	solve	NOUN
bracis-33605	119	78	egubs	egub	NOUN
bracis-33605	119	79	by	by	ADP
bracis-33605	119	80	means	mean	NOUN
bracis-33605	119	81	of	of	ADP
bracis-33605	119	82	a	a	DET
bracis-33605	119	83	heuristic	heuristic	ADJ
bracis-33605	119	84	search	search	NOUN
bracis-33605	119	85	with	with	ADP
bracis-33605	119	86	a	a	DET
bracis-33605	119	87	variant	variant	NOUN
bracis-33605	119	88	of	of	ADP
bracis-33605	119	89	the	the	DET
bracis-33605	119	90	ao	ao	PROPN
bracis-33605	119	91	*	*	PROPN
bracis-33605	119	92	algorithm	algorithm	NOUN
bracis-33605	119	93	.	.	PUNCT
bracis-33605	120	1	2.3	2.3	NUM
bracis-33605	120	2	reinforcement	reinforcement	NOUN
bracis-33605	120	3	learning	learning	NOUN
bracis-33605	120	4	in	in	ADP
bracis-33605	120	5	rl	rl	ADP
bracis-33605	120	6	,	,	PUNCT
bracis-33605	120	7	it	it	PRON
bracis-33605	120	8	is	be	AUX
bracis-33605	120	9	common	common	ADJ
bracis-33605	120	10	to	to	PART
bracis-33605	120	11	divide	divide	VERB
bracis-33605	120	12	the	the	DET
bracis-33605	120	13	process	process	NOUN
bracis-33605	120	14	 	 	SPACE
bracis-33605	120	15	into	into	ADP
bracis-33605	120	16	episodes	episode	NOUN
bracis-33605	120	17	(	(	PUNCT
bracis-33605	120	18	represented	represent	VERB
bracis-33605	120	19	by	by	ADP
bracis-33605	120	20	e	e	NOUN
bracis-33605	120	21	)	)	PUNCT
bracis-33605	120	22	.	.	PUNCT
bracis-33605	121	1	an	an	DET
bracis-33605	121	2	episode	episode	NOUN
bracis-33605	121	3	can	can	AUX
bracis-33605	121	4	end	end	VERB
bracis-33605	121	5	in	in	ADP
bracis-33605	121	6	several	several	ADJ
bracis-33605	121	7	ways	way	NOUN
bracis-33605	121	8	,	,	PUNCT
bracis-33605	121	9	for	for	ADP
bracis-33605	121	10	example	example	NOUN
bracis-33605	121	11	,	,	PUNCT
bracis-33605	121	12	when	when	SCONJ
bracis-33605	121	13	a	a	DET
bracis-33605	121	14	goal	goal	NOUN
bracis-33605	121	15	state	state	NOUN
bracis-33605	121	16	,	,	PUNCT
bracis-33605	121	17	a	a	DET
bracis-33605	121	18	dead	dead	ADJ
bracis-33605	121	19	-	-	PUNCT
bracis-33605	121	20	end	end	NOUN
bracis-33605	121	21	or	or	CCONJ
bracis-33605	121	22	a	a	DET
bracis-33605	121	23	pre	pre	ADJ
bracis-33605	121	24	-	-	ADJ
bracis-33605	121	25	defined	define	VERB
bracis-33605	121	26	time	time	NOUN
bracis-33605	121	27	are	be	AUX
bracis-33605	121	28	reached	reach	VERB
bracis-33605	121	29	.	.	PUNCT
bracis-33605	122	1	the	the	DET
bracis-33605	122	2	q	q	NOUN
bracis-33605	122	3	-	-	PUNCT
bracis-33605	122	4	learning	learn	VERB
bracis-33605	122	5	algorithm	algorithm	NOUN
bracis-33605	122	6	[	[	X
bracis-33605	122	7	16	16	NUM
bracis-33605	122	8	]	]	PUNCT
bracis-33605	122	9	is	be	AUX
bracis-33605	122	10	a	a	DET
bracis-33605	122	11	well	well	ADV
bracis-33605	122	12	-	-	PUNCT
bracis-33605	122	13	known	know	VERB
bracis-33605	122	14	algorithm	algorithm	NOUN
bracis-33605	122	15	in	in	ADP
bracis-33605	122	16	rl	rl	PROPN
bracis-33605	122	17	.	.	PUNCT
bracis-33605	123	1	the	the	DET
bracis-33605	123	2	q	q	NOUN
bracis-33605	123	3	-	-	PUNCT
bracis-33605	123	4	value	value	NOUN
bracis-33605	123	5	for	for	ADP
bracis-33605	123	6	all	all	DET
bracis-33605	123	7	\(s_t	\(s_t	PROPN
bracis-33605	123	8	\in	\in	PROPN
bracis-33605	123	9	s\	s\	NOUN
bracis-33605	123	10	)	)	PUNCT
bracis-33605	123	11	in	in	ADP
bracis-33605	123	12	time	time	NOUN
bracis-33605	123	13	t	t	PROPN
bracis-33605	123	14	when	when	SCONJ
bracis-33605	123	15	performing	perform	VERB
bracis-33605	123	16	action	action	NOUN
bracis-33605	123	17	\(a_t	\(a_t	ADJ
bracis-33605	123	18	\in	\in	PROPN
bracis-33605	123	19	a\	a\	PROPN
bracis-33605	123	20	)	)	PUNCT
bracis-33605	123	21	in	in	ADP
bracis-33605	123	22	time	time	NOUN
bracis-33605	123	23	t	t	PROPN
bracis-33605	123	24	and	and	CCONJ
bracis-33605	123	25	following	follow	VERB
bracis-33605	123	26	the	the	DET
bracis-33605	123	27	optimal	optimal	ADJ
bracis-33605	123	28	policy	policy	NOUN
bracis-33605	123	29	\(\pi	\(\pi	PROPN
bracis-33605	123	30	^*\	^*\	X
bracis-33605	123	31	)	)	PUNCT
bracis-33605	123	32	is	be	AUX
bracis-33605	123	33	:	:	PUNCT
bracis-33605	123	34	$	$	SYM
bracis-33605	123	35	$	$	SYM
bracis-33605	123	36	\begin{aligned	\begin{aligne	VERB
bracis-33605	123	37	}	}	PUNCT
bracis-33605	123	38	\begin{aligned	\begin{aligne	VERB
bracis-33605	123	39	}	}	PUNCT
bracis-33605	123	40	q(s_t	q(s_t	PROPN
bracis-33605	123	41	,	,	PUNCT
bracis-33605	123	42	a_t	a_t	PUNCT
bracis-33605	123	43	)	)	PUNCT
bracis-33605	123	44	&	&	CCONJ
bracis-33605	123	45	\leftarrow	\leftarrow	VERB
bracis-33605	123	46	q(s_t	q(s_t	PROPN
bracis-33605	123	47	,	,	PUNCT
bracis-33605	123	48	a_t	a_t	PUNCT
bracis-33605	123	49	)	)	PUNCT
bracis-33605	124	1	+	+	CCONJ
bracis-33605	124	2	\alpha	\alpha	ADP
bracis-33605	124	3	_	_	PRON
bracis-33605	124	4	t\delta	t\delta	NOUN
bracis-33605	124	5	,	,	PUNCT
bracis-33605	124	6	\\	\\	NOUN
bracis-33605	124	7	\delta	\delta	PROPN
bracis-33605	124	8	&	&	CCONJ
bracis-33605	124	9	=	=	PROPN
bracis-33605	124	10	c(s_t	c(s_t	PROPN
bracis-33605	124	11	,	,	PUNCT
bracis-33605	124	12	a_t	a_t	PUNCT
bracis-33605	124	13	)	)	PUNCT
bracis-33605	125	1	+	+	CCONJ
bracis-33605	125	2	\gamma	\gamma	PROPN
bracis-33605	125	3	{	{	PUNCT
bracis-33605	125	4	\min	\min	NOUN
bracis-33605	125	5	}	}	PUNCT
bracis-33605	125	6	_	_	PUNCT
bracis-33605	125	7	{	{	PUNCT
bracis-33605	125	8	a	a	DET
bracis-33605	125	9	'	'	PUNCT
bracis-33605	125	10	\in	\in	ADJ
bracis-33605	125	11	a}q(s_{t+1},a	a}q(s_{t+1},a	NOUN
bracis-33605	125	12	'	'	PART
bracis-33605	125	13	)	)	PUNCT
bracis-33605	125	14	q(s_t	q(s_t	PROPN
bracis-33605	125	15	,	,	PUNCT
bracis-33605	125	16	a_t	a_t	X
bracis-33605	125	17	)	)	PUNCT
bracis-33605	125	18	,	,	PUNCT
bracis-33605	125	19	\end{aligned	\end{aligne	VERB
bracis-33605	125	20	}	}	PUNCT
bracis-33605	125	21	\end{aligned}$$	\end{aligned}$$	X
bracis-33605	125	22	(	(	PUNCT
bracis-33605	125	23	3	3	NUM
bracis-33605	125	24	)	)	PUNCT
bracis-33605	125	25	where	where	SCONJ
bracis-33605	125	26	,	,	PUNCT
bracis-33605	125	27	\(\alpha	\(\alpha	VERB
bracis-33605	125	28	_	_	NOUN
bracis-33605	125	29	t\	t\	VERB
bracis-33605	125	30	)	)	PUNCT
bracis-33605	125	31	is	be	AUX
bracis-33605	125	32	the	the	DET
bracis-33605	125	33	learning	learning	NOUN
bracis-33605	125	34	rate	rate	NOUN
bracis-33605	125	35	(	(	PUNCT
bracis-33605	125	36	\(0\le	\(0\le	PROPN
bracis-33605	125	37	\alpha	\alpha	PROPN
bracis-33605	125	38	_	_	PROPN
bracis-33605	125	39	t	t	PROPN
bracis-33605	125	40	\le	\le	PROPN
bracis-33605	125	41	1\	1\	NUM
bracis-33605	125	42	)	)	PUNCT
bracis-33605	125	43	)	)	PUNCT
bracis-33605	125	44	that	that	PRON
bracis-33605	125	45	determines	determine	VERB
bracis-33605	125	46	the	the	DET
bracis-33605	125	47	learning	learning	NOUN
bracis-33605	125	48	speed	speed	NOUN
bracis-33605	125	49	,	,	PUNCT
bracis-33605	125	50	\(\gamma	\(\gamma	X
bracis-33605	125	51	\	\	NOUN
bracis-33605	125	52	)	)	PUNCT
bracis-33605	125	53	is	be	AUX
bracis-33605	125	54	the	the	DET
bracis-33605	125	55	discount	discount	NOUN
bracis-33605	125	56	factor	factor	NOUN
bracis-33605	125	57	on	on	ADP
bracis-33605	125	58	future	future	ADJ
bracis-33605	125	59	values	value	NOUN
bracis-33605	125	60	and	and	CCONJ
bracis-33605	125	61	\(s_{t+1}\	\(s_{t+1}\	NUM
bracis-33605	125	62	)	)	PUNCT
bracis-33605	125	63	is	be	AUX
bracis-33605	125	64	the	the	DET
bracis-33605	125	65	next	next	ADJ
bracis-33605	125	66	state	state	NOUN
bracis-33605	125	67	after	after	ADP
bracis-33605	125	68	executing	execute	VERB
bracis-33605	125	69	action	action	NOUN
bracis-33605	125	70	\(a_t\	\(a_t\	NUM
bracis-33605	125	71	)	)	PUNCT
bracis-33605	125	72	.	.	PUNCT
bracis-33605	126	1	theorem	theorem	ADJ
bracis-33605	126	2	1	1	NUM
bracis-33605	126	3	(	(	PUNCT
bracis-33605	126	4	convergence	convergence	NOUN
bracis-33605	126	5	of	of	ADP
bracis-33605	126	6	the	the	DET
bracis-33605	126	7	q	q	NOUN
bracis-33605	126	8	-	-	PUNCT
bracis-33605	126	9	learning	learn	VERB
bracis-33605	126	10	[	[	X
bracis-33605	126	11	16	16	NUM
bracis-33605	126	12	]	]	PUNCT
bracis-33605	126	13	)	)	PUNCT
bracis-33605	126	14	let	let	VERB
bracis-33605	126	15	the	the	DET
bracis-33605	126	16	costs	cost	NOUN
bracis-33605	126	17	in	in	ADP
bracis-33605	126	18	each	each	DET
bracis-33605	126	19	time	time	NOUN
bracis-33605	126	20	t	t	NOUN
bracis-33605	126	21	be	be	AUX
bracis-33605	126	22	\(c_t\le	\(c_t\le	PUNCT
bracis-33605	126	23	{	{	PUNCT
bracis-33605	126	24	c}\	c}\	NUM
bracis-33605	126	25	)	)	PUNCT
bracis-33605	126	26	,	,	PUNCT
bracis-33605	126	27	learning	learn	VERB
bracis-33605	126	28	rates	rate	NOUN
bracis-33605	126	29	\(0	\(0	PROPN
bracis-33605	126	30	\le	\le	PROPN
bracis-33605	126	31	\alpha	\alpha	ADP
bracis-33605	126	32	_	_	PROPN
bracis-33605	126	33	t	t	NOUN
bracis-33605	126	34	<	<	X
bracis-33605	126	35	1	1	NUM
bracis-33605	126	36	\	\	NOUN
bracis-33605	126	37	)	)	PUNCT
bracis-33605	126	38	,	,	PUNCT
bracis-33605	126	39	and	and	CCONJ
bracis-33605	126	40	\(\sum	\(\sum	VERB
bracis-33605	127	1	_	_	PRON
bracis-33605	127	2	{	{	PUNCT
bracis-33605	127	3	t=1}^{\infty	t=1}^{\infty	NUM
bracis-33605	127	4	}	}	PUNCT
bracis-33605	127	5	\alpha	\alpha	NOUN
bracis-33605	127	6	_	_	PRON
bracis-33605	127	7	t(s	t(s	PROPN
bracis-33605	127	8	,	,	PUNCT
bracis-33605	127	9	a	a	PRON
bracis-33605	127	10	)	)	PUNCT
bracis-33605	127	11	=	=	SYM
bracis-33605	127	12	\infty	\infty	NOUN
bracis-33605	127	13	,	,	PUNCT
bracis-33605	127	14	\sum	\sum	NOUN
bracis-33605	127	15	_	_	PUNCT
bracis-33605	127	16	{	{	PUNCT
bracis-33605	127	17	t=1}^{\infty	t=1}^{\infty	NUM
bracis-33605	127	18	}	}	PUNCT
bracis-33605	127	19	(	(	PUNCT
bracis-33605	127	20	\alpha	\alpha	ADP
bracis-33605	127	21	_	_	PRON
bracis-33605	127	22	t(s	t(s	PROPN
bracis-33605	127	23	,	,	PUNCT
bracis-33605	127	24	a))^2	a))^2	NOUN
bracis-33605	127	25	<	<	X
bracis-33605	127	26	\infty	\infty	PROPN
bracis-33605	127	27	\	\	PROPN
bracis-33605	127	28	)	)	PUNCT
bracis-33605	127	29	then	then	ADV
bracis-33605	127	30	\(q_t(s	\(q_t(s	PROPN
bracis-33605	127	31	,	,	PUNCT
bracis-33605	127	32	a)\rightarrow	a)\rightarrow	NOUN
bracis-33605	127	33	q^*(s	q^*(s	PROPN
bracis-33605	127	34	,	,	PUNCT
bracis-33605	127	35	a)\	a)\	NUM
bracis-33605	127	36	)	)	PUNCT
bracis-33605	127	37	as	as	ADP
bracis-33605	127	38	\(t\rightarrow	\(t\rightarrow	NOUN
bracis-33605	127	39	\infty	\infty	PROPN
bracis-33605	127	40	\	\	PROPN
bracis-33605	127	41	)	)	PUNCT
bracis-33605	127	42	,	,	PUNCT
bracis-33605	127	43	for	for	ADP
bracis-33605	127	44	every	every	DET
bracis-33605	127	45	state	state	NOUN
bracis-33605	127	46	s	s	NOUN
bracis-33605	127	47	and	and	CCONJ
bracis-33605	127	48	action	action	NOUN
bracis-33605	127	49	a	a	PRON
bracis-33605	127	50	,	,	PUNCT
bracis-33605	127	51	with	with	ADP
bracis-33605	127	52	probability	probability	NOUN
bracis-33605	127	53	1	1	NUM
bracis-33605	127	54	.	.	PUNCT
bracis-33605	128	1	the	the	DET
bracis-33605	128	2	algorithm	algorithm	NOUN
bracis-33605	128	3	proposed	propose	VERB
bracis-33605	128	4	by	by	ADP
bracis-33605	128	5	geibel	geibel	NOUN
bracis-33605	128	6	and	and	CCONJ
bracis-33605	128	7	wysotzki	wysotzki	ADJ
bracis-33605	128	8	[	[	X
bracis-33605	128	9	5	5	NUM
bracis-33605	128	10	]	]	PUNCT
bracis-33605	128	11	is	be	AUX
bracis-33605	128	12	one	one	NUM
bracis-33605	128	13	of	of	ADP
bracis-33605	128	14	the	the	DET
bracis-33605	128	15	few	few	ADJ
bracis-33605	128	16	methods	method	NOUN
bracis-33605	128	17	for	for	ADP
bracis-33605	128	18	rl	rl	PRON
bracis-33605	128	19	designed	design	VERB
bracis-33605	128	20	specifically	specifically	ADV
bracis-33605	128	21	to	to	PART
bracis-33605	128	22	perform	perform	VERB
bracis-33605	128	23	a	a	DET
bracis-33605	128	24	trade	trade	NOUN
bracis-33605	128	25	-	-	PUNCT
bracis-33605	128	26	off	off	NOUN
bracis-33605	128	27	between	between	ADP
bracis-33605	128	28	cost	cost	NOUN
bracis-33605	128	29	and	and	CCONJ
bracis-33605	128	30	probability	probability	NOUN
bracis-33605	128	31	of	of	ADP
bracis-33605	128	32	reaching	reach	VERB
bracis-33605	128	33	an	an	DET
bracis-33605	128	34	error	error	NOUN
bracis-33605	128	35	state	state	NOUN
bracis-33605	128	36	.	.	PUNCT
bracis-33605	129	1	although	although	SCONJ
bracis-33605	129	2	this	this	DET
bracis-33605	129	3	algorithm	algorithm	NOUN
bracis-33605	129	4	uses	use	VERB
bracis-33605	129	5	the	the	DET
bracis-33605	129	6	concept	concept	NOUN
bracis-33605	129	7	of	of	ADP
bracis-33605	129	8	probability	probability	NOUN
bracis-33605	129	9	of	of	ADP
bracis-33605	129	10	reaching	reach	VERB
bracis-33605	129	11	an	an	DET
bracis-33605	129	12	error	error	NOUN
bracis-33605	129	13	state	state	NOUN
bracis-33605	129	14	,	,	PUNCT
bracis-33605	129	15	this	this	DET
bracis-33605	129	16	concept	concept	NOUN
bracis-33605	129	17	is	be	AUX
bracis-33605	129	18	closely	closely	ADV
bracis-33605	129	19	related	relate	VERB
bracis-33605	129	20	to	to	ADP
bracis-33605	129	21	the	the	DET
bracis-33605	129	22	concept	concept	NOUN
bracis-33605	129	23	of	of	ADP
bracis-33605	129	24	probability	probability	NOUN
bracis-33605	129	25	to	to	ADP
bracis-33605	129	26	reaching	reach	VERB
bracis-33605	129	27	the	the	DET
bracis-33605	129	28	goal	goal	NOUN
bracis-33605	129	29	used	use	VERB
bracis-33605	129	30	in	in	ADP
bracis-33605	129	31	this	this	DET
bracis-33605	129	32	work	work	NOUN
bracis-33605	129	33	.	.	PUNCT
bracis-33605	130	1	let	let	AUX
bracis-33605	130	2	\(\varphi	\(\varphi	ADJ
bracis-33605	130	3	\	\	NOUN
bracis-33605	130	4	)	)	PUNCT
bracis-33605	130	5	be	be	AUX
bracis-33605	130	6	the	the	DET
bracis-33605	130	7	set	set	NOUN
bracis-33605	130	8	of	of	ADP
bracis-33605	130	9	error	error	NOUN
bracis-33605	130	10	states	state	NOUN
bracis-33605	130	11	.	.	PUNCT
bracis-33605	131	1	the	the	DET
bracis-33605	131	2	probability	probability	NOUN
bracis-33605	131	3	of	of	ADP
bracis-33605	131	4	reaching	reach	VERB
bracis-33605	131	5	an	an	DET
bracis-33605	131	6	error	error	NOUN
bracis-33605	131	7	state	state	NOUN
bracis-33605	131	8	(	(	PUNCT
bracis-33605	131	9	also	also	ADV
bracis-33605	131	10	called	call	VERB
bracis-33605	131	11	risk	risk	NOUN
bracis-33605	131	12	)	)	PUNCT
bracis-33605	131	13	from	from	ADP
bracis-33605	131	14	an	an	DET
bracis-33605	131	15	initial	initial	ADJ
bracis-33605	131	16	state	state	NOUN
bracis-33605	131	17	s	s	PART
bracis-33605	131	18	is	be	AUX
bracis-33605	131	19	defined	define	VERB
bracis-33605	131	20	as	as	ADP
bracis-33605	131	21	\(\rho	\(\rho	NOUN
bracis-33605	131	22	^{\pi	^{\pi	NUM
bracis-33605	131	23	}	}	PUNCT
bracis-33605	131	24	(	(	PUNCT
bracis-33605	131	25	s	s	X
bracis-33605	131	26	)	)	PUNCT
bracis-33605	131	27	=	=	PUNCT
bracis-33605	131	28	e\left	e\left	VERB
bracis-33605	131	29	[	[	PUNCT
bracis-33605	131	30	\sum	\sum	NOUN
bracis-33605	131	31	^{\infty	^{\infty	ADJ
bracis-33605	131	32	}	}	PUNCT
bracis-33605	131	33	_	_	PUNCT
bracis-33605	131	34	{	{	PUNCT
bracis-33605	131	35	t=0}\bar{\gamma	t=0}\bar{\gamma	PROPN
bracis-33605	131	36	}	}	PUNCT
bracis-33605	131	37	^t\bar{r}_t|s_0	^t\bar{r}_t|s_0	NOUN
bracis-33605	131	38	=	=	NOUN
bracis-33605	131	39	s\right	s\right	NOUN
bracis-33605	131	40	]	]	PUNCT
bracis-33605	131	41	\	\	NOUN
bracis-33605	131	42	)	)	PUNCT
bracis-33605	131	43	,	,	PUNCT
bracis-33605	131	44	where	where	SCONJ
bracis-33605	131	45	\(\bar{\gamma	\(\bar{\gamma	ADP
bracis-33605	131	46	}	}	PUNCT
bracis-33605	131	47	^t\	^t\	X
bracis-33605	131	48	)	)	PUNCT
bracis-33605	131	49	is	be	AUX
bracis-33605	131	50	the	the	DET
bracis-33605	131	51	discount	discount	NOUN
bracis-33605	131	52	factor	factor	NOUN
bracis-33605	131	53	for	for	ADP
bracis-33605	131	54	the	the	DET
bracis-33605	131	55	risk	risk	NOUN
bracis-33605	131	56	at	at	ADP
bracis-33605	131	57	time	time	NOUN
bracis-33605	131	58	t	t	PROPN
bracis-33605	131	59	and	and	CCONJ
bracis-33605	131	60	\(\bar{r}_{s	\(\bar{r}_{s	NOUN
bracis-33605	131	61	,	,	PUNCT
bracis-33605	131	62	a}(s	a}(s	PROPN
bracis-33605	131	63	'	'	PUNCT
bracis-33605	131	64	)	)	PUNCT
bracis-33605	132	1	=	=	PRON
bracis-33605	132	2	{	{	PUNCT
bracis-33605	132	3	\left\	\left\	PROPN
bracis-33605	132	4	{	{	PUNCT
bracis-33605	132	5	\begin{array}{ll	\begin{array}{ll	PROPN
bracis-33605	132	6	}	}	PUNCT
bracis-33605	132	7	1	1	NUM
bracis-33605	132	8	,	,	PUNCT
bracis-33605	132	9	&	&	CCONJ
bracis-33605	132	10	\text	\text	PROPN
bracis-33605	132	11	{	{	PUNCT
bracis-33605	132	12	if	if	SCONJ
bracis-33605	132	13	}	}	PUNCT
bracis-33605	132	14	\,s\in	\,s\in	VERB
bracis-33605	132	15	\varphi	\varphi	PROPN
bracis-33605	132	16	\	\	NOUN
bracis-33605	132	17	,	,	PUNCT
bracis-33605	132	18	\text	\text	SCONJ
bracis-33605	132	19	{	{	PUNCT
bracis-33605	132	20	and}\	and}\	PROPN
bracis-33605	132	21	,	,	PUNCT
bracis-33605	132	22	s	s	PART
bracis-33605	132	23	'	'	PUNCT
bracis-33605	132	24	=	=	SYM
bracis-33605	132	25	\eta	\eta	ADJ
bracis-33605	132	26	\\	\\	NOUN
bracis-33605	132	27	0	0	NUM
bracis-33605	132	28	,	,	PUNCT
bracis-33605	132	29	&	&	CCONJ
bracis-33605	132	30	else	else	ADV
bracis-33605	132	31	\end{array}\right	\end{array}\right	PROPN
bracis-33605	132	32	.	.	PUNCT
bracis-33605	132	33	}	}	PUNCT
bracis-33605	132	34	\	\	NOUN
bracis-33605	132	35	)	)	PUNCT
bracis-33605	132	36	,	,	PUNCT
bracis-33605	132	37	where	where	SCONJ
bracis-33605	132	38	\(\eta	\(\eta	NOUN
bracis-33605	132	39	\	\	NOUN
bracis-33605	132	40	)	)	PUNCT
bracis-33605	132	41	is	be	AUX
bracis-33605	132	42	a	a	DET
bracis-33605	132	43	fictitious	fictitious	ADJ
bracis-33605	132	44	absorbing	absorb	VERB
bracis-33605	132	45	state	state	NOUN
bracis-33605	132	46	where	where	SCONJ
bracis-33605	132	47	the	the	DET
bracis-33605	132	48	agent	agent	NOUN
bracis-33605	132	49	transits	transit	VERB
bracis-33605	132	50	after	after	ADP
bracis-33605	132	51	reaching	reach	VERB
bracis-33605	132	52	an	an	DET
bracis-33605	132	53	error	error	NOUN
bracis-33605	132	54	state	state	NOUN
bracis-33605	132	55	.	.	PUNCT
bracis-33605	133	1	with	with	ADP
bracis-33605	133	2	these	these	DET
bracis-33605	133	3	concepts	concept	NOUN
bracis-33605	133	4	,	,	PUNCT
bracis-33605	133	5	a	a	DET
bracis-33605	133	6	new	new	ADJ
bracis-33605	133	7	value	value	NOUN
bracis-33605	133	8	for	for	ADP
bracis-33605	133	9	the	the	DET
bracis-33605	133	10	risk	risk	NOUN
bracis-33605	133	11	is	be	AUX
bracis-33605	133	12	defined	define	VERB
bracis-33605	133	13	as	as	ADP
bracis-33605	133	14	\(\bar{q}^{\pi	\(\bar{q}^{\pi	NOUN
bracis-33605	133	15	}	}	PUNCT
bracis-33605	133	16	(	(	PUNCT
bracis-33605	133	17	s	s	X
bracis-33605	133	18	,	,	PUNCT
bracis-33605	133	19	a	a	PRON
bracis-33605	133	20	)	)	PUNCT
bracis-33605	133	21	=	=	PUNCT
bracis-33605	133	22	e\left	e\left	VERB
bracis-33605	133	23	[	[	PUNCT
bracis-33605	133	24	\bar{r}_0	\bar{r}_0	ADP
bracis-33605	133	25	+	+	CCONJ
bracis-33605	133	26	\bar{\gamma	\bar{\gamma	PROPN
bracis-33605	133	27	}	}	PUNCT
bracis-33605	133	28	\rho	\rho	NOUN
bracis-33605	133	29	^{\pi	^{\pi	NUM
bracis-33605	133	30	}	}	PUNCT
bracis-33605	133	31	(	(	PUNCT
bracis-33605	133	32	s')\right	s')\right	NOUN
bracis-33605	133	33	]	]	X
bracis-33605	133	34	\	\	NOUN
bracis-33605	133	35	)	)	PUNCT
bracis-33605	133	36	,	,	PUNCT
bracis-33605	133	37	where	where	SCONJ
bracis-33605	133	38	\(s_0	\(s_0	PROPN
bracis-33605	133	39	=	=	SYM
bracis-33605	133	40	s	s	PROPN
bracis-33605	133	41	,	,	PUNCT
bracis-33605	133	42	a_0	a_0	PROPN
bracis-33605	133	43	=	=	SYM
bracis-33605	133	44	a\	a\	NOUN
bracis-33605	133	45	)	)	PUNCT
bracis-33605	133	46	 	 	SPACE
bracis-33605	133	47	and	and	CCONJ
bracis-33605	133	48	\(s'\	\(s'\	PROPN
bracis-33605	133	49	)	)	PUNCT
bracis-33605	133	50	is	be	AUX
bracis-33605	133	51	the	the	DET
bracis-33605	133	52	next	next	ADJ
bracis-33605	133	53	state	state	NOUN
bracis-33605	133	54	.	.	PUNCT
bracis-33605	134	1	then	then	ADV
bracis-33605	134	2	a	a	DET
bracis-33605	134	3	new	new	ADJ
bracis-33605	134	4	general	general	ADJ
bracis-33605	134	5	value	value	NOUN
bracis-33605	134	6	is	be	AUX
bracis-33605	134	7	defined	define	VERB
bracis-33605	134	8	as	as	ADP
bracis-33605	134	9	\(q^{\pi	\(q^{\pi	NOUN
bracis-33605	134	10	}	}	PUNCT
bracis-33605	134	11	_	_	PUNCT
bracis-33605	134	12	{	{	PUNCT
bracis-33605	134	13	\xi	\xi	X
bracis-33605	134	14	}	}	PUNCT
bracis-33605	134	15	(	(	PUNCT
bracis-33605	134	16	s	s	PROPN
bracis-33605	134	17	,	,	PUNCT
bracis-33605	134	18	a	a	PRON
bracis-33605	134	19	)	)	PUNCT
bracis-33605	134	20	=	=	SYM
bracis-33605	134	21	\xi	\xi	PROPN
bracis-33605	134	22	q^{\pi	q^{\pi	PROPN
bracis-33605	134	23	}	}	PUNCT
bracis-33605	134	24	(	(	PUNCT
bracis-33605	134	25	s	s	PROPN
bracis-33605	134	26	,	,	PUNCT
bracis-33605	134	27	a	a	PRON
bracis-33605	134	28	)	)	PUNCT
bracis-33605	134	29	\bar{q}^{\pi	\bar{q}^{\pi	NUM
bracis-33605	134	30	}	}	PUNCT
bracis-33605	134	31	(	(	PUNCT
bracis-33605	134	32	s	s	NOUN
bracis-33605	134	33	,	,	PUNCT
bracis-33605	134	34	a)\	a)\	NUM
bracis-33605	134	35	)	)	PUNCT
bracis-33605	134	36	.	.	PUNCT
bracis-33605	135	1	the	the	DET
bracis-33605	135	2	trade	trade	NOUN
bracis-33605	135	3	-	-	PUNCT
bracis-33605	135	4	off	off	NOUN
bracis-33605	135	5	between	between	ADP
bracis-33605	135	6	the	the	DET
bracis-33605	135	7	probability	probability	NOUN
bracis-33605	135	8	of	of	ADP
bracis-33605	135	9	reaching	reach	VERB
bracis-33605	135	10	an	an	DET
bracis-33605	135	11	error	error	NOUN
bracis-33605	135	12	state	state	NOUN
bracis-33605	135	13	and	and	CCONJ
bracis-33605	135	14	the	the	DET
bracis-33605	135	15	cost	cost	NOUN
bracis-33605	135	16	to	to	ADP
bracis-33605	135	17	goal	goal	NOUN
bracis-33605	135	18	is	be	AUX
bracis-33605	135	19	determined	determine	VERB
bracis-33605	135	20	by	by	ADP
bracis-33605	135	21	varying	vary	VERB
bracis-33605	135	22	the	the	DET
bracis-33605	135	23	weight	weight	NOUN
bracis-33605	135	24	\(\xi	\(\xi	PUNCT
bracis-33605	136	1	\	\	PROPN
bracis-33605	136	2	)	)	PUNCT
bracis-33605	136	3	.	.	PUNCT
bracis-33605	137	1	3	3	NUM
bracis-33605	137	2	rl	rl	NOUN
bracis-33605	137	3	under	under	ADP
bracis-33605	137	4	the	the	DET
bracis-33605	137	5	 	 	SPACE
bracis-33605	137	6	gubs	gubs	ADJ
bracis-33605	137	7	criterion	criterion	NOUN
bracis-33605	137	8	as	as	SCONJ
bracis-33605	137	9	shown	show	VERB
bracis-33605	137	10	in	in	ADP
bracis-33605	137	11	the	the	DET
bracis-33605	137	12	previous	previous	ADJ
bracis-33605	137	13	section	section	NOUN
bracis-33605	137	14	,	,	PUNCT
bracis-33605	137	15	the	the	DET
bracis-33605	137	16	gubs	gubs	ADJ
bracis-33605	137	17	criterion	criterion	NOUN
bracis-33605	137	18	is	be	AUX
bracis-33605	137	19	the	the	DET
bracis-33605	137	20	only	only	ADJ
bracis-33605	137	21	known	know	VERB
bracis-33605	137	22	criterion	criterion	NOUN
bracis-33605	137	23	that	that	PRON
bracis-33605	137	24	guarantees	guarantee	VERB
bracis-33605	137	25	\(\alpha	\(\alpha	NOUN
bracis-33605	137	26	\)-strong	\)-strong	ADP
bracis-33605	137	27	probability	probability	NOUN
bracis-33605	137	28	-	-	PUNCT
bracis-33605	137	29	to	to	ADP
bracis-33605	137	30	-	-	PUNCT
bracis-33605	137	31	goal	goal	NOUN
bracis-33605	137	32	priority	priority	NOUN
bracis-33605	137	33	.	.	PUNCT
bracis-33605	138	1	in	in	ADP
bracis-33605	138	2	this	this	DET
bracis-33605	138	3	work	work	NOUN
bracis-33605	138	4	,	,	PUNCT
bracis-33605	138	5	rl	rl	X
bracis-33605	138	6	algorithms	algorithm	NOUN
bracis-33605	138	7	are	be	AUX
bracis-33605	138	8	introduced	introduce	VERB
bracis-33605	138	9	to	to	PART
bracis-33605	138	10	solve	solve	VERB
bracis-33605	138	11	the	the	DET
bracis-33605	138	12	gubs	gubs	ADJ
bracis-33605	138	13	criterion	criterion	NOUN
bracis-33605	138	14	.	.	PUNCT
bracis-33605	139	1	some	some	DET
bracis-33605	139	2	challenges	challenge	NOUN
bracis-33605	139	3	to	to	PART
bracis-33605	139	4	face	face	VERB
bracis-33605	139	5	in	in	ADP
bracis-33605	139	6	the	the	DET
bracis-33605	139	7	development	development	NOUN
bracis-33605	139	8	of	of	ADP
bracis-33605	139	9	rl	rl	ADP
bracis-33605	139	10	algorithms	algorithm	NOUN
bracis-33605	139	11	that	that	PRON
bracis-33605	139	12	use	use	VERB
bracis-33605	139	13	the	the	DET
bracis-33605	139	14	gubs	gubs	ADJ
bracis-33605	139	15	criterion	criterion	NOUN
bracis-33605	139	16	are	be	AUX
bracis-33605	139	17	:	:	PUNCT
bracis-33605	139	18	(	(	PUNCT
bracis-33605	139	19	1	1	X
bracis-33605	139	20	)	)	PUNCT
bracis-33605	139	21	use	use	VERB
bracis-33605	139	22	extended	extend	VERB
bracis-33605	139	23	states	state	NOUN
bracis-33605	139	24	with	with	ADP
bracis-33605	139	25	cost	cost	NOUN
bracis-33605	139	26	since	since	SCONJ
bracis-33605	139	27	gubs	gub	NOUN
bracis-33605	139	28	finds	find	VERB
bracis-33605	139	29	non	non	ADJ
bracis-33605	139	30	-	-	ADJ
bracis-33605	139	31	stationary	stationary	ADJ
bracis-33605	139	32	policies	policy	NOUN
bracis-33605	139	33	,	,	PUNCT
bracis-33605	139	34	something	something	PRON
bracis-33605	139	35	that	that	PRON
bracis-33605	139	36	is	be	AUX
bracis-33605	139	37	not	not	PART
bracis-33605	139	38	the	the	DET
bracis-33605	139	39	default	default	NOUN
bracis-33605	139	40	behavior	behavior	NOUN
bracis-33605	139	41	of	of	ADP
bracis-33605	139	42	rl	rl	ADP
bracis-33605	139	43	algorithms	algorithm	NOUN
bracis-33605	139	44	;	;	PUNCT
bracis-33605	139	45	(	(	PUNCT
bracis-33605	139	46	2	2	X
bracis-33605	139	47	)	)	PUNCT
bracis-33605	139	48	define	define	VERB
bracis-33605	139	49	the	the	DET
bracis-33605	139	50	type	type	NOUN
bracis-33605	139	51	of	of	ADP
bracis-33605	139	52	the	the	DET
bracis-33605	139	53	initial	initial	ADJ
bracis-33605	139	54	and	and	CCONJ
bracis-33605	139	55	final	final	ADJ
bracis-33605	139	56	policies	policy	NOUN
bracis-33605	139	57	used	use	VERB
bracis-33605	139	58	in	in	ADP
bracis-33605	139	59	the	the	DET
bracis-33605	139	60	algorithms	algorithm	NOUN
bracis-33605	139	61	;	;	PUNCT
bracis-33605	139	62	(	(	PUNCT
bracis-33605	139	63	3	3	X
bracis-33605	139	64	)	)	PUNCT
bracis-33605	139	65	the	the	DET
bracis-33605	139	66	equations	equation	NOUN
bracis-33605	139	67	of	of	ADP
bracis-33605	139	68	the	the	DET
bracis-33605	139	69	gubs	gubs	ADJ
bracis-33605	139	70	-	-	PUNCT
bracis-33605	139	71	vi	vi	NOUN
bracis-33605	139	72	algorithm	algorithm	NOUN
bracis-33605	140	1	[	[	X
bracis-33605	140	2	4	4	X
bracis-33605	140	3	]	]	PUNCT
bracis-33605	140	4	must	must	AUX
bracis-33605	140	5	be	be	AUX
bracis-33605	140	6	adapted	adapt	VERB
bracis-33605	140	7	based	base	VERB
bracis-33605	140	8	on	on	ADP
bracis-33605	140	9	temporal	temporal	ADJ
bracis-33605	140	10	difference	difference	NOUN
bracis-33605	140	11	without	without	ADP
bracis-33605	140	12	using	use	VERB
bracis-33605	140	13	the	the	DET
bracis-33605	140	14	transition	transition	NOUN
bracis-33605	140	15	function	function	VERB
bracis-33605	140	16	a	a	PRON
bracis-33605	140	17	priori	priori	ADV
bracis-33605	140	18	;	;	PUNCT
bracis-33605	140	19	and	and	CCONJ
bracis-33605	140	20	(	(	PUNCT
bracis-33605	140	21	4	4	X
bracis-33605	140	22	)	)	PUNCT
bracis-33605	140	23	define	define	VERB
bracis-33605	140	24	or	or	CCONJ
bracis-33605	140	25	compute	compute	VERB
bracis-33605	140	26	the	the	DET
bracis-33605	140	27	maximum	maximum	ADJ
bracis-33605	140	28	accumulated	accumulate	VERB
bracis-33605	140	29	cost	cost	NOUN
bracis-33605	140	30	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	140	31	)	)	PUNCT
bracis-33605	140	32	that	that	PRON
bracis-33605	140	33	is	be	AUX
bracis-33605	140	34	used	use	VERB
bracis-33605	140	35	by	by	ADP
bracis-33605	140	36	the	the	DET
bracis-33605	140	37	gubs	gubs	ADJ
bracis-33605	140	38	algorithms	algorithm	NOUN
bracis-33605	140	39	.	.	PUNCT
bracis-33605	141	1	the	the	DET
bracis-33605	141	2	algorithms	algorithm	NOUN
bracis-33605	141	3	introduced	introduce	VERB
bracis-33605	141	4	are	be	AUX
bracis-33605	141	5	:	:	PUNCT
bracis-33605	141	6	q	q	ADJ
bracis-33605	141	7	-	-	PUNCT
bracis-33605	141	8	learning	learning	NOUN
bracis-33605	141	9	-	-	PUNCT
bracis-33605	141	10	gubs	gub	NOUN
bracis-33605	141	11	and	and	CCONJ
bracis-33605	141	12	q	q	NOUN
bracis-33605	141	13	-	-	PUNCT
bracis-33605	141	14	learning	learning	NOUN
bracis-33605	141	15	-	-	PUNCT
bracis-33605	141	16	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	141	17	)	)	PUNCT
bracis-33605	141	18	.	.	PUNCT
bracis-33605	142	1	these	these	DET
bracis-33605	142	2	algorithms	algorithm	NOUN
bracis-33605	142	3	do	do	AUX
bracis-33605	142	4	not	not	PART
bracis-33605	142	5	need	need	VERB
bracis-33605	142	6	to	to	PART
bracis-33605	142	7	know	know	VERB
bracis-33605	142	8	transition	transition	NOUN
bracis-33605	142	9	probabilities	probability	NOUN
bracis-33605	142	10	or	or	CCONJ
bracis-33605	142	11	costs	cost	NOUN
bracis-33605	142	12	and	and	CCONJ
bracis-33605	142	13	do	do	AUX
bracis-33605	142	14	not	not	PART
bracis-33605	142	15	use	use	VERB
bracis-33605	142	16	discounts	discount	NOUN
bracis-33605	142	17	or	or	CCONJ
bracis-33605	142	18	penalties	penalty	NOUN
bracis-33605	142	19	to	to	PART
bracis-33605	142	20	find	find	VERB
bracis-33605	142	21	the	the	DET
bracis-33605	142	22	solution	solution	NOUN
bracis-33605	142	23	considering	consider	VERB
bracis-33605	142	24	the	the	DET
bracis-33605	142	25	gubs	gubs	ADJ
bracis-33605	142	26	criterion	criterion	NOUN
bracis-33605	142	27	.	.	PUNCT
bracis-33605	143	1	they	they	PRON
bracis-33605	143	2	are	be	AUX
bracis-33605	143	3	inspired	inspire	VERB
bracis-33605	143	4	by	by	ADP
bracis-33605	143	5	two	two	NUM
bracis-33605	143	6	well	well	ADV
bracis-33605	143	7	-	-	PUNCT
bracis-33605	143	8	known	know	VERB
bracis-33605	143	9	algorithms	algorithm	NOUN
bracis-33605	143	10	based	base	VERB
bracis-33605	143	11	on	on	ADP
bracis-33605	143	12	temporal	temporal	ADJ
bracis-33605	143	13	difference	difference	NOUN
bracis-33605	143	14	,	,	PUNCT
bracis-33605	143	15	that	that	PRON
bracis-33605	143	16	are	be	AUX
bracis-33605	143	17	:	:	PUNCT
bracis-33605	143	18	q	q	ADJ
bracis-33605	143	19	-	-	PUNCT
bracis-33605	143	20	learning	learning	NOUN
bracis-33605	143	21	and	and	CCONJ
bracis-33605	143	22	dyna	dyna	NOUN
bracis-33605	143	23	-	-	PROPN
bracis-33605	143	24	q	q	PROPN
bracis-33605	143	25	(	(	PUNCT
bracis-33605	143	26	a	a	DET
bracis-33605	143	27	model	model	NOUN
bracis-33605	143	28	-	-	PUNCT
bracis-33605	143	29	based	base	VERB
bracis-33605	143	30	algorithm	algorithm	NOUN
bracis-33605	143	31	)	)	PUNCT
bracis-33605	143	32	.	.	PUNCT
bracis-33605	144	1	3.1	3.1	NUM
bracis-33605	144	2	q	q	ADJ
bracis-33605	144	3	-	-	PUNCT
bracis-33605	144	4	learning	learning	NOUN
bracis-33605	144	5	-	-	PUNCT
bracis-33605	144	6	gubs	gub	NOUN
bracis-33605	144	7	the	the	DET
bracis-33605	144	8	q	q	NOUN
bracis-33605	144	9	-	-	PUNCT
bracis-33605	144	10	learning	learning	NOUN
bracis-33605	144	11	-	-	PUNCT
bracis-33605	144	12	gubs	gub	NOUN
bracis-33605	144	13	algorithm	algorithm	NOUN
bracis-33605	144	14	is	be	AUX
bracis-33605	144	15	based	base	VERB
bracis-33605	144	16	on	on	ADP
bracis-33605	144	17	the	the	DET
bracis-33605	144	18	q	q	NOUN
bracis-33605	144	19	-	-	PUNCT
bracis-33605	144	20	learning	learn	VERB
bracis-33605	145	1	[	[	X
bracis-33605	145	2	16	16	NUM
bracis-33605	145	3	]	]	PUNCT
bracis-33605	145	4	and	and	CCONJ
bracis-33605	145	5	gubs	gubs	ADJ
bracis-33605	145	6	-	-	PUNCT
bracis-33605	145	7	vi	vi	NOUN
bracis-33605	145	8	[	[	X
bracis-33605	145	9	4	4	NUM
bracis-33605	145	10	]	]	SYM
bracis-33605	145	11	algorithms	algorithm	NOUN
bracis-33605	145	12	.	.	PUNCT
bracis-33605	146	1	like	like	ADP
bracis-33605	146	2	gubs	gub	NOUN
bracis-33605	146	3	-	-	PUNCT
bracis-33605	146	4	vi	vi	NOUN
bracis-33605	146	5	,	,	PUNCT
bracis-33605	146	6	q	q	ADJ
bracis-33605	146	7	-	-	PUNCT
bracis-33605	146	8	learning	learning	NOUN
bracis-33605	146	9	-	-	PUNCT
bracis-33605	146	10	gubs	gub	NOUN
bracis-33605	146	11	solves	solve	VERB
bracis-33605	146	12	gubs	gub	NOUN
bracis-33605	146	13	for	for	ADP
bracis-33605	146	14	discrete	discrete	ADJ
bracis-33605	146	15	costs	cost	NOUN
bracis-33605	146	16	,	,	PUNCT
bracis-33605	146	17	can	can	AUX
bracis-33605	146	18	use	use	VERB
bracis-33605	146	19	several	several	ADJ
bracis-33605	146	20	utility	utility	NOUN
bracis-33605	146	21	functions	function	NOUN
bracis-33605	146	22	(	(	PUNCT
bracis-33605	146	23	for	for	ADP
bracis-33605	146	24	example	example	NOUN
bracis-33605	146	25	the	the	DET
bracis-33605	146	26	exponential	exponential	ADJ
bracis-33605	146	27	function	function	NOUN
bracis-33605	146	28	and	and	CCONJ
bracis-33605	146	29	others	other	NOUN
bracis-33605	146	30	that	that	PRON
bracis-33605	146	31	hold	hold	VERB
bracis-33605	146	32	the	the	DET
bracis-33605	146	33	properties	property	NOUN
bracis-33605	146	34	defined	define	VERB
bracis-33605	146	35	in	in	ADP
bracis-33605	146	36	[	[	X
bracis-33605	146	37	1	1	NUM
bracis-33605	146	38	]	]	NUM
bracis-33605	146	39	)	)	PUNCT
bracis-33605	146	40	,	,	PUNCT
bracis-33605	146	41	and	and	CCONJ
bracis-33605	146	42	is	be	AUX
bracis-33605	146	43	an	an	DET
bracis-33605	146	44	approximate	approximate	ADJ
bracis-33605	146	45	algorithm	algorithm	NOUN
bracis-33605	146	46	that	that	PRON
bracis-33605	146	47	has	have	VERB
bracis-33605	146	48	as	as	ADP
bracis-33605	146	49	a	a	DET
bracis-33605	146	50	parameter	parameter	NOUN
bracis-33605	146	51	the	the	DET
bracis-33605	146	52	maximum	maximum	ADJ
bracis-33605	146	53	accumulated	accumulate	VERB
bracis-33605	146	54	cost	cost	NOUN
bracis-33605	146	55	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	146	56	)	)	PUNCT
bracis-33605	146	57	for	for	ADP
bracis-33605	146	58	every	every	DET
bracis-33605	146	59	state	state	NOUN
bracis-33605	146	60	.	.	PUNCT
bracis-33605	147	1	based	base	VERB
bracis-33605	147	2	on	on	ADP
bracis-33605	147	3	eqs	eqs	PROPN
bracis-33605	147	4	.	.	PUNCT
bracis-33605	147	5	 	 	SPACE
bracis-33605	147	6	3	3	NUM
bracis-33605	147	7	and	and	CCONJ
bracis-33605	147	8	2	2	NUM
bracis-33605	147	9	,	,	PUNCT
bracis-33605	147	10	the	the	DET
bracis-33605	147	11	following	follow	VERB
bracis-33605	147	12	is	be	AUX
bracis-33605	147	13	the	the	DET
bracis-33605	147	14	modified	modify	VERB
bracis-33605	147	15	bellman	bellman	NOUN
bracis-33605	147	16	equation	equation	NOUN
bracis-33605	147	17	to	to	PART
bracis-33605	147	18	solve	solve	VERB
bracis-33605	147	19	the	the	DET
bracis-33605	147	20	discrete	discrete	ADJ
bracis-33605	147	21	cost	cost	NOUN
bracis-33605	147	22	problem	problem	NOUN
bracis-33605	147	23	(	(	PUNCT
bracis-33605	147	24	definition	definition	NOUN
bracis-33605	147	25	2	2	NUM
bracis-33605	147	26	)	)	PUNCT
bracis-33605	147	27	.	.	PUNCT
bracis-33605	148	1	$	$	SYM
bracis-33605	148	2	$	$	SYM
bracis-33605	148	3	\begin{aligned	\begin{aligne	VERB
bracis-33605	148	4	}	}	PUNCT
bracis-33605	148	5	\begin{aligned	\begin{aligne	VERB
bracis-33605	148	6	}	}	PUNCT
bracis-33605	148	7	q(x_t	q(x_t	NOUN
bracis-33605	148	8	,	,	PUNCT
bracis-33605	148	9	a_t	a_t	X
bracis-33605	148	10	)	)	PUNCT
bracis-33605	148	11	&	&	CCONJ
bracis-33605	148	12	\leftarrow	\leftarrow	PROPN
bracis-33605	148	13	q(x_t	q(x_t	NOUN
bracis-33605	148	14	,	,	PUNCT
bracis-33605	148	15	a_t	a_t	PUNCT
bracis-33605	148	16	)	)	PUNCT
bracis-33605	149	1	+	+	CCONJ
bracis-33605	149	2	\alpha	\alpha	ADP
bracis-33605	149	3	_	_	PROPN
bracis-33605	149	4	t	t	PROPN
bracis-33605	149	5	\delta	\delta	PROPN
bracis-33605	149	6	,	,	PUNCT
bracis-33605	149	7	\\	\\	NOUN
bracis-33605	149	8	\delta	\delta	PROPN
bracis-33605	149	9	&	&	CCONJ
bracis-33605	149	10	\leftarrow	\leftarrow	VERB
bracis-33605	149	11	c'_{t+1}(x_t	c'_{t+1}(x_t	PROPN
bracis-33605	149	12	,	,	PUNCT
bracis-33605	149	13	a_t	a_t	PUNCT
bracis-33605	149	14	)	)	PUNCT
bracis-33605	149	15	+	+	CCONJ
bracis-33605	149	16	\max	\max	PROPN
bracis-33605	149	17	_	_	PUNCT
bracis-33605	149	18	{	{	PUNCT
bracis-33605	149	19	a	a	PRON
bracis-33605	149	20	'	'	PUNCT
bracis-33605	149	21	\in	\in	PROPN
bracis-33605	149	22	\mathcal	\mathcal	PROPN
bracis-33605	149	23	{	{	PUNCT
bracis-33605	149	24	a	a	NOUN
bracis-33605	149	25	}	}	PUNCT
bracis-33605	149	26	}	}	PUNCT
bracis-33605	149	27	q(x_{t+1},a	q(x_{t+1},a	NUM
bracis-33605	149	28	'	'	PUNCT
bracis-33605	149	29	)	)	PUNCT
bracis-33605	149	30	q(x_t	q(x_t	NOUN
bracis-33605	149	31	,	,	PUNCT
bracis-33605	149	32	a_t	a_t	X
bracis-33605	149	33	)	)	PUNCT
bracis-33605	149	34	,	,	PUNCT
bracis-33605	149	35	\end{aligned	\end{aligne	VERB
bracis-33605	149	36	}	}	PUNCT
bracis-33605	149	37	\end{aligned}$$	\end{aligned}$$	X
bracis-33605	149	38	(	(	PUNCT
bracis-33605	149	39	4	4	NUM
bracis-33605	149	40	)	)	PUNCT
bracis-33605	149	41	where	where	SCONJ
bracis-33605	149	42	,	,	PUNCT
bracis-33605	149	43	the	the	DET
bracis-33605	149	44	q	q	NOUN
bracis-33605	149	45	-	-	PUNCT
bracis-33605	149	46	value	value	NOUN
bracis-33605	149	47	is	be	AUX
bracis-33605	149	48	calculated	calculate	VERB
bracis-33605	149	49	for	for	ADP
bracis-33605	149	50	each	each	DET
bracis-33605	149	51	extended	extend	VERB
bracis-33605	149	52	state	state	NOUN
bracis-33605	149	53	\(x_t	\(x_t	NOUN
bracis-33605	149	54	=	=	SYM
bracis-33605	149	55	(	(	PUNCT
bracis-33605	149	56	s_t	s_t	NOUN
bracis-33605	149	57	,	,	PUNCT
bracis-33605	149	58	c_t)\	c_t)\	NOUN
bracis-33605	149	59	)	)	PUNCT
bracis-33605	149	60	considering	consider	VERB
bracis-33605	149	61	the	the	DET
bracis-33605	149	62	utility	utility	NOUN
bracis-33605	149	63	\(c'_{t+1	\(c'_{t+1	X
bracis-33605	149	64	}	}	PUNCT
bracis-33605	149	65	\	\	NOUN
bracis-33605	149	66	)	)	PUNCT
bracis-33605	149	67	of	of	ADP
bracis-33605	149	68	definition	definition	NOUN
bracis-33605	149	69	2	2	NUM
bracis-33605	149	70	.	.	PUNCT
bracis-33605	150	1	notice	notice	VERB
bracis-33605	150	2	that	that	SCONJ
bracis-33605	150	3	eq	eq	ADP
bracis-33605	150	4	.	.	PUNCT
bracis-33605	150	5	 	 	SPACE
bracis-33605	150	6	4	4	NUM
bracis-33605	150	7	does	do	AUX
bracis-33605	150	8	not	not	PART
bracis-33605	150	9	consider	consider	VERB
bracis-33605	150	10	a	a	DET
bracis-33605	150	11	discount	discount	NOUN
bracis-33605	150	12	factor	factor	NOUN
bracis-33605	150	13	.	.	PUNCT
bracis-33605	151	1	furthermore	furthermore	ADV
bracis-33605	151	2	,	,	PUNCT
bracis-33605	151	3	q	q	X
bracis-33605	151	4	-	-	PUNCT
bracis-33605	151	5	learning	learning	NOUN
bracis-33605	151	6	-	-	PUNCT
bracis-33605	151	7	gubs	gub	NOUN
bracis-33605	151	8	uses	use	VERB
bracis-33605	151	9	a	a	DET
bracis-33605	151	10	random	random	ADJ
bracis-33605	151	11	policy	policy	NOUN
bracis-33605	151	12	as	as	ADP
bracis-33605	151	13	an	an	DET
bracis-33605	151	14	initial	initial	ADJ
bracis-33605	151	15	policy	policy	NOUN
bracis-33605	151	16	and	and	CCONJ
bracis-33605	151	17	calculates	calculate	VERB
bracis-33605	151	18	a	a	DET
bracis-33605	151	19	non	non	ADJ
bracis-33605	151	20	-	-	ADJ
bracis-33605	151	21	stationary	stationary	ADJ
bracis-33605	151	22	final	final	ADJ
bracis-33605	151	23	policy	policy	NOUN
bracis-33605	151	24	.	.	PUNCT
bracis-33605	152	1	the	the	DET
bracis-33605	152	2	q	q	ADJ
bracis-33605	152	3	-	-	PUNCT
bracis-33605	152	4	learning	learning	NOUN
bracis-33605	152	5	-	-	PUNCT
bracis-33605	152	6	gubs	gubs	ADJ
bracis-33605	152	7	algorithm	algorithm	NOUN
bracis-33605	152	8	(	(	PUNCT
bracis-33605	152	9	algorithm	algorithm	NOUN
bracis-33605	152	10	 	 	SPACE
bracis-33605	152	11	1	1	NUM
bracis-33605	152	12	)	)	PUNCT
bracis-33605	152	13	first	first	ADV
bracis-33605	152	14	initializes	initialize	VERB
bracis-33605	152	15	the	the	DET
bracis-33605	152	16	q	q	NOUN
bracis-33605	152	17	table	table	NOUN
bracis-33605	152	18	considering	consider	VERB
bracis-33605	152	19	as	as	SCONJ
bracis-33605	152	20	value	value	VERB
bracis-33605	152	21	the	the	DET
bracis-33605	152	22	differences	difference	NOUN
bracis-33605	152	23	between	between	ADP
bracis-33605	152	24	the	the	DET
bracis-33605	152	25	minimum	minimum	ADJ
bracis-33605	152	26	utility	utility	NOUN
bracis-33605	152	27	and	and	CCONJ
bracis-33605	152	28	the	the	DET
bracis-33605	152	29	utility	utility	NOUN
bracis-33605	152	30	for	for	ADP
bracis-33605	152	31	each	each	DET
bracis-33605	152	32	accumulated	accumulate	VERB
bracis-33605	152	33	cost	cost	NOUN
bracis-33605	152	34	c	c	NOUN
bracis-33605	152	35	for	for	ADP
bracis-33605	152	36	states	state	NOUN
bracis-33605	152	37	that	that	PRON
bracis-33605	152	38	are	be	AUX
bracis-33605	152	39	not	not	PART
bracis-33605	152	40	goal	goal	NOUN
bracis-33605	152	41	states	state	NOUN
bracis-33605	152	42	.	.	PUNCT
bracis-33605	153	1	for	for	ADP
bracis-33605	153	2	the	the	DET
bracis-33605	153	3	goal	goal	NOUN
bracis-33605	153	4	states	state	NOUN
bracis-33605	153	5	,	,	PUNCT
bracis-33605	153	6	the	the	DET
bracis-33605	153	7	algorithm	algorithm	NOUN
bracis-33605	153	8	sets	set	VERB
bracis-33605	153	9	the	the	DET
bracis-33605	153	10	q	q	NOUN
bracis-33605	153	11	-	-	PUNCT
bracis-33605	153	12	value	value	NOUN
bracis-33605	153	13	as	as	ADP
bracis-33605	153	14	\(k_g\	\(k_g\	NUM
bracis-33605	153	15	)	)	PUNCT
bracis-33605	153	16	(	(	PUNCT
bracis-33605	153	17	line	line	NOUN
bracis-33605	153	18	25	25	NUM
bracis-33605	153	19	)	)	PUNCT
bracis-33605	153	20	.	.	PUNCT
bracis-33605	154	1	then	then	ADV
bracis-33605	154	2	,	,	PUNCT
bracis-33605	154	3	the	the	DET
bracis-33605	154	4	algorithm	algorithm	NOUN
bracis-33605	154	5	iterates	iterate	VERB
bracis-33605	154	6	over	over	ADP
bracis-33605	154	7	a	a	DET
bracis-33605	154	8	set	set	NOUN
bracis-33605	154	9	of	of	ADP
bracis-33605	154	10	episodes	episode	NOUN
bracis-33605	154	11	in	in	ADP
bracis-33605	154	12	which	which	PRON
bracis-33605	154	13	it	it	PRON
bracis-33605	154	14	computes	compute	VERB
bracis-33605	154	15	\(\alpha	\(\alpha	VERB
bracis-33605	154	16	_	_	NOUN
bracis-33605	154	17	e\	e\	NOUN
bracis-33605	154	18	)	)	PUNCT
bracis-33605	154	19	and	and	CCONJ
bracis-33605	154	20	\(\epsilon	\(\epsilon	NOUN
bracis-33605	154	21	_	_	NOUN
bracis-33605	154	22	e\	e\	NOUN
bracis-33605	154	23	)	)	PUNCT
bracis-33605	154	24	in	in	ADP
bracis-33605	154	25	line	line	NOUN
bracis-33605	154	26	4	4	NUM
bracis-33605	154	27	and	and	CCONJ
bracis-33605	154	28	gets	get	VERB
bracis-33605	154	29	the	the	DET
bracis-33605	154	30	initial	initial	ADJ
bracis-33605	154	31	extended	extended	ADJ
bracis-33605	154	32	state	state	NOUN
bracis-33605	154	33	\(x_0\	\(x_0\	PROPN
bracis-33605	154	34	)	)	PUNCT
bracis-33605	154	35	in	in	ADP
bracis-33605	154	36	line	line	NOUN
bracis-33605	154	37	5	5	NUM
bracis-33605	154	38	.	.	PUNCT
bracis-33605	155	1	at	at	ADP
bracis-33605	155	2	each	each	DET
bracis-33605	155	3	step	step	NOUN
bracis-33605	155	4	,	,	PUNCT
bracis-33605	155	5	an	an	DET
bracis-33605	155	6	action	action	NOUN
bracis-33605	155	7	a	a	PRON
bracis-33605	155	8	is	be	AUX
bracis-33605	155	9	chosen	choose	VERB
bracis-33605	155	10	for	for	ADP
bracis-33605	155	11	the	the	DET
bracis-33605	155	12	current	current	ADJ
bracis-33605	155	13	state	state	NOUN
bracis-33605	155	14	s	s	VERB
bracis-33605	155	15	following	follow	VERB
bracis-33605	155	16	an	an	DET
bracis-33605	155	17	\(\epsilon	\(\epsilon	NOUN
bracis-33605	155	18	_	_	NOUN
bracis-33605	155	19	e\)-greedy	e\)-greedy	NOUN
bracis-33605	155	20	policy	policy	NOUN
bracis-33605	155	21	,	,	PUNCT
bracis-33605	155	22	it	it	PRON
bracis-33605	155	23	acts	act	VERB
bracis-33605	155	24	on	on	ADP
bracis-33605	155	25	the	the	DET
bracis-33605	155	26	environment	environment	NOUN
bracis-33605	155	27	,	,	PUNCT
bracis-33605	155	28	observes	observe	VERB
bracis-33605	155	29	the	the	DET
bracis-33605	155	30	cost	cost	NOUN
bracis-33605	155	31	of	of	ADP
bracis-33605	155	32	the	the	DET
bracis-33605	155	33	action	action	NOUN
bracis-33605	155	34	c(s	c(	VERB
bracis-33605	155	35	,	,	PUNCT
bracis-33605	155	36	 	 	SPACE
bracis-33605	155	37	a	a	NOUN
bracis-33605	155	38	)	)	PUNCT
bracis-33605	155	39	and	and	CCONJ
bracis-33605	155	40	the	the	DET
bracis-33605	155	41	next	next	ADJ
bracis-33605	155	42	state	state	NOUN
bracis-33605	155	43	\(s'\	\(s'\	PROPN
bracis-33605	155	44	)	)	PUNCT
bracis-33605	155	45	(	(	PUNCT
bracis-33605	155	46	lines	line	NOUN
bracis-33605	155	47	11	11	NUM
bracis-33605	155	48	and	and	CCONJ
bracis-33605	155	49	12	12	NUM
bracis-33605	155	50	)	)	PUNCT
bracis-33605	155	51	.	.	PUNCT
bracis-33605	156	1	for	for	ADP
bracis-33605	156	2	goal	goal	NOUN
bracis-33605	156	3	states	state	NOUN
bracis-33605	156	4	\(s	\(s	VERB
bracis-33605	156	5	\in	\in	PROPN
bracis-33605	156	6	\mathcal	\mathcal	PROPN
bracis-33605	156	7	{	{	PUNCT
bracis-33605	156	8	g}\	g}\	PROPN
bracis-33605	156	9	)	)	PUNCT
bracis-33605	156	10	,	,	PUNCT
bracis-33605	156	11	\(c(s	\(c(s	PROPN
bracis-33605	156	12	,	,	PUNCT
bracis-33605	156	13	a	a	DET
bracis-33605	156	14	)	)	PUNCT
bracis-33605	156	15	=	=	SYM
bracis-33605	156	16	0\	0\	PROPN
bracis-33605	156	17	)	)	PUNCT
bracis-33605	156	18	(	(	PUNCT
bracis-33605	156	19	line	line	NOUN
bracis-33605	156	20	14	14	NUM
bracis-33605	156	21	)	)	PUNCT
bracis-33605	156	22	.	.	PUNCT
bracis-33605	157	1	then	then	ADV
bracis-33605	157	2	,	,	PUNCT
bracis-33605	157	3	with	with	ADP
bracis-33605	157	4	this	this	DET
bracis-33605	157	5	data	datum	NOUN
bracis-33605	157	6	,	,	PUNCT
bracis-33605	157	7	the	the	DET
bracis-33605	157	8	q	q	NOUN
bracis-33605	157	9	-	-	PUNCT
bracis-33605	157	10	value	value	NOUN
bracis-33605	157	11	is	be	AUX
bracis-33605	157	12	calculated	calculate	VERB
bracis-33605	157	13	(	(	PUNCT
bracis-33605	157	14	line	line	NOUN
bracis-33605	157	15	18	18	NUM
bracis-33605	157	16	)	)	PUNCT
bracis-33605	157	17	for	for	ADP
bracis-33605	157	18	each	each	DET
bracis-33605	157	19	discrete	discrete	ADJ
bracis-33605	157	20	accumulated	accumulate	VERB
bracis-33605	157	21	cost	cost	NOUN
bracis-33605	157	22	\(c	\(c	NOUN
bracis-33605	157	23	''	''	PUNCT
bracis-33605	157	24	\in	\in	PROPN
bracis-33605	157	25	\{0,	\{0,	PROPN
bracis-33605	157	26	...	...	PUNCT
bracis-33605	157	27	,c_{max}\}\	,c_{max}\}\	PUNCT
bracis-33605	157	28	)	)	PUNCT
bracis-33605	157	29	based	base	VERB
bracis-33605	157	30	on	on	ADP
bracis-33605	157	31	eq	eq	PROPN
bracis-33605	157	32	.	.	PUNCT
bracis-33605	157	33	 	 	SPACE
bracis-33605	157	34	4	4	NUM
bracis-33605	157	35	.	.	PUNCT
bracis-33605	158	1	the	the	DET
bracis-33605	158	2	algorithm	algorithm	NOUN
bracis-33605	158	3	moves	move	VERB
bracis-33605	158	4	to	to	ADP
bracis-33605	158	5	another	another	DET
bracis-33605	158	6	episode	episode	NOUN
bracis-33605	158	7	(	(	PUNCT
bracis-33605	158	8	i	i	NOUN
bracis-33605	158	9	)	)	PUNCT
bracis-33605	158	10	when	when	SCONJ
bracis-33605	158	11	a	a	DET
bracis-33605	158	12	goal	goal	NOUN
bracis-33605	158	13	state	state	NOUN
bracis-33605	158	14	is	be	AUX
bracis-33605	158	15	reached	reach	VERB
bracis-33605	158	16	,	,	PUNCT
bracis-33605	158	17	(	(	PUNCT
bracis-33605	158	18	ii	ii	NOUN
bracis-33605	158	19	)	)	PUNCT
bracis-33605	158	20	when	when	SCONJ
bracis-33605	158	21	the	the	DET
bracis-33605	158	22	accumulated	accumulate	VERB
bracis-33605	158	23	cost	cost	NOUN
bracis-33605	158	24	is	be	AUX
bracis-33605	158	25	greater	great	ADJ
bracis-33605	158	26	than	than	ADP
bracis-33605	158	27	or	or	CCONJ
bracis-33605	158	28	equal	equal	ADJ
bracis-33605	158	29	to	to	ADP
bracis-33605	158	30	the	the	DET
bracis-33605	158	31	value	value	NOUN
bracis-33605	158	32	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	158	33	)	)	PUNCT
bracis-33605	158	34	or	or	CCONJ
bracis-33605	158	35	(	(	PUNCT
bracis-33605	158	36	iii	iii	NOUN
bracis-33605	158	37	)	)	PUNCT
bracis-33605	158	38	when	when	SCONJ
bracis-33605	158	39	a	a	DET
bracis-33605	158	40	dead	dead	ADJ
bracis-33605	158	41	-	-	PUNCT
bracis-33605	158	42	end	end	NOUN
bracis-33605	158	43	is	be	AUX
bracis-33605	158	44	found	find	VERB
bracis-33605	158	45	(	(	PUNCT
bracis-33605	158	46	line	line	NOUN
bracis-33605	158	47	19	19	NUM
bracis-33605	158	48	)	)	PUNCT
bracis-33605	158	49	.	.	PUNCT
bracis-33605	159	1	finally	finally	ADV
bracis-33605	159	2	,	,	PUNCT
bracis-33605	159	3	the	the	DET
bracis-33605	159	4	algorithm	algorithm	NOUN
bracis-33605	159	5	calculates	calculate	VERB
bracis-33605	159	6	the	the	DET
bracis-33605	159	7	gubs	gubs	ADJ
bracis-33605	159	8	policy	policy	NOUN
bracis-33605	159	9	based	base	VERB
bracis-33605	159	10	on	on	ADP
bracis-33605	159	11	the	the	DET
bracis-33605	159	12	q	q	NOUN
bracis-33605	159	13	table	table	NOUN
bracis-33605	159	14	,	,	PUNCT
bracis-33605	159	15	choosing	choose	VERB
bracis-33605	159	16	the	the	DET
bracis-33605	159	17	actions	action	NOUN
bracis-33605	159	18	with	with	ADP
bracis-33605	159	19	the	the	DET
bracis-33605	159	20	highest	high	ADJ
bracis-33605	159	21	value	value	NOUN
bracis-33605	159	22	for	for	ADP
bracis-33605	159	23	each	each	DET
bracis-33605	159	24	extended	extend	VERB
bracis-33605	159	25	state	state	NOUN
bracis-33605	159	26	(	(	PUNCT
bracis-33605	159	27	line	line	NOUN
bracis-33605	159	28	25	25	NUM
bracis-33605	159	29	)	)	PUNCT
bracis-33605	159	30	.	.	PUNCT
bracis-33605	160	1	q	q	X
bracis-33605	160	2	-	-	PUNCT
bracis-33605	160	3	learning	learning	NOUN
bracis-33605	160	4	-	-	PUNCT
bracis-33605	160	5	gubs	gub	NOUN
bracis-33605	160	6	convergence	convergence	NOUN
bracis-33605	160	7	.	.	PUNCT
bracis-33605	161	1	the	the	DET
bracis-33605	161	2	algorithm	algorithm	NOUN
bracis-33605	161	3	finds	find	VERB
bracis-33605	161	4	an	an	DET
bracis-33605	161	5	approximate	approximate	ADJ
bracis-33605	161	6	policy	policy	NOUN
bracis-33605	161	7	for	for	ADP
bracis-33605	161	8	the	the	DET
bracis-33605	161	9	problem	problem	NOUN
bracis-33605	161	10	regardless	regardless	ADV
bracis-33605	161	11	of	of	ADP
bracis-33605	161	12	the	the	DET
bracis-33605	161	13	utility	utility	NOUN
bracis-33605	161	14	function	function	NOUN
bracis-33605	161	15	used	use	VERB
bracis-33605	161	16	.	.	PUNCT
bracis-33605	162	1	the	the	DET
bracis-33605	162	2	proof	proof	NOUN
bracis-33605	162	3	of	of	ADP
bracis-33605	162	4	convergence	convergence	NOUN
bracis-33605	162	5	is	be	AUX
bracis-33605	162	6	based	base	VERB
bracis-33605	162	7	on	on	ADP
bracis-33605	162	8	the	the	DET
bracis-33605	162	9	following	follow	VERB
bracis-33605	162	10	definition	definition	NOUN
bracis-33605	162	11	and	and	CCONJ
bracis-33605	162	12	theorems	theorem	NOUN
bracis-33605	162	13	.	.	PUNCT
bracis-33605	163	1	definition	definition	NOUN
bracis-33605	163	2	4	4	NUM
bracis-33605	163	3	(	(	PUNCT
bracis-33605	163	4	ssps	ssps	VERB
bracis-33605	163	5	with	with	ADP
bracis-33605	163	6	discrete	discrete	ADJ
bracis-33605	163	7	cost	cost	NOUN
bracis-33605	163	8	and	and	CCONJ
bracis-33605	163	9	maximum	maximum	ADJ
bracis-33605	163	10	accumulated	accumulate	VERB
bracis-33605	163	11	cost	cost	NOUN
bracis-33605	163	12	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	163	13	)	)	PUNCT
bracis-33605	163	14	)	)	PUNCT
bracis-33605	163	15	.	.	PUNCT
bracis-33605	164	1	the	the	DET
bracis-33605	164	2	ssp	ssp	NOUN
bracis-33605	164	3	of	of	ADP
bracis-33605	164	4	definition	definition	NOUN
bracis-33605	164	5	2	2	NUM
bracis-33605	164	6	can	can	AUX
bracis-33605	164	7	be	be	AUX
bracis-33605	164	8	restated	restate	VERB
bracis-33605	164	9	into	into	ADP
bracis-33605	164	10	a	a	DET
bracis-33605	164	11	new	new	ADJ
bracis-33605	164	12	ssp	ssp	NOUN
bracis-33605	164	13	with	with	ADP
bracis-33605	164	14	the	the	DET
bracis-33605	164	15	same	same	ADJ
bracis-33605	164	16	elements	element	NOUN
bracis-33605	164	17	but	but	CCONJ
bracis-33605	164	18	with	with	ADP
bracis-33605	164	19	a	a	DET
bracis-33605	164	20	finite	finite	ADJ
bracis-33605	164	21	set	set	NOUN
bracis-33605	164	22	of	of	ADP
bracis-33605	164	23	states	state	NOUN
bracis-33605	164	24	\(x	\(x	PROPN
bracis-33605	164	25	'	'	PUNCT
bracis-33605	164	26	\subseteq	\subseteq	PROPN
bracis-33605	164	27	x\	x\	NOUN
bracis-33605	164	28	)	)	PUNCT
bracis-33605	164	29	where	where	SCONJ
bracis-33605	164	30	,	,	PUNCT
bracis-33605	164	31	\(x'=	\(x'=	PROPN
bracis-33605	164	32	\{x	\{x	X
bracis-33605	164	33	|	|	ADV
bracis-33605	164	34	x	x	ADP
bracis-33605	164	35	=	=	SYM
bracis-33605	164	36	(	(	PUNCT
bracis-33605	164	37	s	s	PROPN
bracis-33605	164	38	,	,	PUNCT
bracis-33605	164	39	c	c	NOUN
bracis-33605	164	40	)	)	PUNCT
bracis-33605	164	41	,	,	PUNCT
bracis-33605	164	42	c	c	PROPN
bracis-33605	164	43	\le	\le	PROPN
bracis-33605	164	44	c_{max	c_{max	NOUN
bracis-33605	164	45	}	}	PUNCT
bracis-33605	164	46	\}\	\}\	NOUN
bracis-33605	164	47	)	)	PUNCT
bracis-33605	164	48	and	and	CCONJ
bracis-33605	164	49	c	c	PROPN
bracis-33605	164	50	is	be	AUX
bracis-33605	164	51	the	the	DET
bracis-33605	164	52	accumulated	accumulate	VERB
bracis-33605	164	53	cost	cost	NOUN
bracis-33605	164	54	.	.	PUNCT
bracis-33605	165	1	theorem	theorem	ADJ
bracis-33605	165	2	2	2	NUM
bracis-33605	165	3	(	(	PUNCT
bracis-33605	165	4	q	q	NOUN
bracis-33605	165	5	-	-	PUNCT
bracis-33605	165	6	learning	learning	NOUN
bracis-33605	165	7	-	-	PUNCT
bracis-33605	165	8	gubs	gub	NOUN
bracis-33605	165	9	convergence	convergence	NOUN
bracis-33605	165	10	)	)	PUNCT
bracis-33605	165	11	.	.	PUNCT
bracis-33605	166	1	let	let	VERB
bracis-33605	166	2	m	m	PRON
bracis-33605	166	3	be	be	AUX
bracis-33605	166	4	the	the	DET
bracis-33605	166	5	ssp	ssp	NOUN
bracis-33605	166	6	of	of	ADP
bracis-33605	166	7	definition	definition	NOUN
bracis-33605	166	8	4	4	NUM
bracis-33605	166	9	where	where	SCONJ
bracis-33605	166	10	,	,	PUNCT
bracis-33605	166	11	in	in	ADP
bracis-33605	166	12	the	the	DET
bracis-33605	166	13	algorithm	algorithm	NOUN
bracis-33605	166	14	,	,	PUNCT
bracis-33605	166	15	\(0	\(0	PROPN
bracis-33605	166	16	\le	\le	PROPN
bracis-33605	166	17	\alpha	\alpha	ADP
bracis-33605	166	18	_	_	PROPN
bracis-33605	166	19	t	t	NOUN
bracis-33605	166	20	<	<	X
bracis-33605	166	21	1	1	NUM
bracis-33605	166	22	\	\	NOUN
bracis-33605	166	23	)	)	PUNCT
bracis-33605	166	24	,	,	PUNCT
bracis-33605	166	25	\(\sum	\(\sum	VERB
bracis-33605	166	26	^{n}_{i=1	^{n}_{i=1	ADP
bracis-33605	166	27	}	}	PUNCT
bracis-33605	166	28	\alpha	\alpha	ADP
bracis-33605	166	29	_	_	NOUN
bracis-33605	166	30	t	t	NOUN
bracis-33605	166	31	=	=	SYM
bracis-33605	166	32	\infty	\infty	PROPN
bracis-33605	166	33	\	\	PROPN
bracis-33605	166	34	)	)	PUNCT
bracis-33605	166	35	and	and	CCONJ
bracis-33605	166	36	\(\sum	\(\sum	VERB
bracis-33605	167	1	^	^	NUM
bracis-33605	167	2	{	{	PUNCT
bracis-33605	167	3	n}_{i=1	n}_{i=1	ADJ
bracis-33605	167	4	}	}	PUNCT
bracis-33605	167	5	\alpha	\alpha	PUNCT
bracis-33605	167	6	_	_	NOUN
bracis-33605	167	7	t^2<\infty	t^2<\infty	X
bracis-33605	167	8	\	\	PROPN
bracis-33605	167	9	)	)	PUNCT
bracis-33605	167	10	,	,	PUNCT
bracis-33605	167	11	and	and	CCONJ
bracis-33605	167	12	as	as	SCONJ
bracis-33605	167	13	the	the	DET
bracis-33605	167	14	number	number	NOUN
bracis-33605	167	15	of	of	ADP
bracis-33605	167	16	episodes	episode	NOUN
bracis-33605	167	17	\(n	\(n	NOUN
bracis-33605	167	18	\rightarrow	\rightarrow	PROPN
bracis-33605	167	19	\infty	\infty	PROPN
bracis-33605	167	20	\	\	PROPN
bracis-33605	167	21	)	)	PUNCT
bracis-33605	167	22	,	,	PUNCT
bracis-33605	167	23	q	q	X
bracis-33605	167	24	-	-	PUNCT
bracis-33605	167	25	learning	learning	NOUN
bracis-33605	167	26	-	-	PUNCT
bracis-33605	167	27	gubs	gub	NOUN
bracis-33605	167	28	converges	converge	NOUN
bracis-33605	167	29	to	to	ADP
bracis-33605	167	30	the	the	DET
bracis-33605	167	31	optimal	optimal	ADJ
bracis-33605	167	32	value	value	NOUN
bracis-33605	167	33	\(q'^*(x	\(q'^*(x	NOUN
bracis-33605	167	34	,	,	PUNCT
bracis-33605	167	35	a)\	a)\	NUM
bracis-33605	167	36	)	)	PUNCT
bracis-33605	167	37	for	for	ADP
bracis-33605	167	38	m.	m.	NOUN
bracis-33605	167	39	proof	proof	NOUN
bracis-33605	167	40	since	since	SCONJ
bracis-33605	167	41	the	the	DET
bracis-33605	167	42	ssp	ssp	NOUN
bracis-33605	167	43	of	of	ADP
bracis-33605	167	44	definition	definition	NOUN
bracis-33605	167	45	4	4	NUM
bracis-33605	167	46	has	have	VERB
bracis-33605	167	47	a	a	DET
bracis-33605	167	48	finite	finite	ADJ
bracis-33605	167	49	number	number	NOUN
bracis-33605	167	50	of	of	ADP
bracis-33605	167	51	states	state	NOUN
bracis-33605	167	52	,	,	PUNCT
bracis-33605	167	53	it	it	PRON
bracis-33605	167	54	is	be	AUX
bracis-33605	167	55	possible	possible	ADJ
bracis-33605	167	56	to	to	PART
bracis-33605	167	57	use	use	VERB
bracis-33605	167	58	the	the	DET
bracis-33605	167	59	same	same	ADJ
bracis-33605	167	60	proof	proof	NOUN
bracis-33605	167	61	of	of	ADP
bracis-33605	167	62	convergence	convergence	NOUN
bracis-33605	167	63	as	as	ADP
bracis-33605	167	64	the	the	DET
bracis-33605	167	65	q	q	NOUN
bracis-33605	167	66	-	-	PUNCT
bracis-33605	167	67	learning	learning	ADJ
bracis-33605	167	68	algorithm	algorithm	NOUN
bracis-33605	167	69	(	(	PUNCT
bracis-33605	167	70	theorem	theorem	NOUN
bracis-33605	167	71	1	1	NUM
bracis-33605	167	72	)	)	PUNCT
bracis-33605	167	73	.	.	PUNCT
bracis-33605	168	1	thus	thus	ADV
bracis-33605	168	2	,	,	PUNCT
bracis-33605	168	3	the	the	DET
bracis-33605	168	4	algorithm	algorithm	NOUN
bracis-33605	168	5	 	 	SPACE
bracis-33605	168	6	1	1	NUM
bracis-33605	168	7	converges	converge	NOUN
bracis-33605	168	8	to	to	ADP
bracis-33605	168	9	the	the	DET
bracis-33605	168	10	optimal	optimal	ADJ
bracis-33605	168	11	value	value	NOUN
bracis-33605	168	12	\(q'^*(x	\(q'^*(x	NOUN
bracis-33605	168	13	,	,	PUNCT
bracis-33605	168	14	a)\	a)\	NUM
bracis-33605	168	15	)	)	PUNCT
bracis-33605	168	16	for	for	ADP
bracis-33605	168	17	the	the	DET
bracis-33605	168	18	ssp	ssp	NOUN
bracis-33605	168	19	of	of	ADP
bracis-33605	168	20	the	the	DET
bracis-33605	168	21	definition	definition	NOUN
bracis-33605	168	22	4	4	NUM
bracis-33605	168	23	.	.	PUNCT
bracis-33605	168	24	theorem	theorem	VERB
bracis-33605	168	25	3	3	NUM
bracis-33605	168	26	(	(	PUNCT
bracis-33605	168	27	the	the	DET
bracis-33605	168	28	q	q	ADJ
bracis-33605	168	29	-	-	PUNCT
bracis-33605	168	30	learning	learning	NOUN
bracis-33605	168	31	-	-	PUNCT
bracis-33605	168	32	gubs	gub	NOUN
bracis-33605	168	33	algorithm	algorithm	NOUN
bracis-33605	168	34	is	be	AUX
bracis-33605	168	35	approximate	approximate	ADJ
bracis-33605	168	36	)	)	PUNCT
bracis-33605	168	37	.	.	PUNCT
bracis-33605	169	1	let	let	VERB
bracis-33605	169	2	m	m	PRON
bracis-33605	169	3	be	be	AUX
bracis-33605	169	4	the	the	DET
bracis-33605	169	5	ssp	ssp	NOUN
bracis-33605	169	6	of	of	ADP
bracis-33605	169	7	definition	definition	NOUN
bracis-33605	169	8	2	2	NUM
bracis-33605	169	9	where	where	SCONJ
bracis-33605	169	10	,	,	PUNCT
bracis-33605	169	11	\(0	\(0	PROPN
bracis-33605	169	12	\le	\le	PROPN
bracis-33605	169	13	\alpha	\alpha	ADP
bracis-33605	169	14	_	_	PROPN
bracis-33605	169	15	t	t	NOUN
bracis-33605	169	16	<	<	X
bracis-33605	169	17	1	1	NUM
bracis-33605	169	18	\	\	NOUN
bracis-33605	169	19	)	)	PUNCT
bracis-33605	169	20	,	,	PUNCT
bracis-33605	169	21	\(\sum	\(\sum	VERB
bracis-33605	169	22	^{n}_{t=1	^{n}_{t=1	ADP
bracis-33605	169	23	}	}	PUNCT
bracis-33605	169	24	\alpha	\alpha	ADP
bracis-33605	169	25	_	_	NOUN
bracis-33605	169	26	t	t	NOUN
bracis-33605	169	27	=	=	SYM
bracis-33605	169	28	\infty	\infty	PROPN
bracis-33605	169	29	\	\	PROPN
bracis-33605	169	30	)	)	PUNCT
bracis-33605	169	31	and	and	CCONJ
bracis-33605	169	32	\(\sum	\(\sum	VERB
bracis-33605	169	33	^{n}_{t=1	^{n}_{t=1	ADP
bracis-33605	169	34	}	}	PUNCT
bracis-33605	169	35	\alpha	\alpha	ADP
bracis-33605	169	36	_	_	NOUN
bracis-33605	169	37	t^2<\infty	t^2<\infty	X
bracis-33605	169	38	\	\	PROPN
bracis-33605	169	39	)	)	PUNCT
bracis-33605	169	40	,	,	PUNCT
bracis-33605	169	41	and	and	CCONJ
bracis-33605	169	42	\(n	\(n	NOUN
bracis-33605	169	43	\rightarrow	\rightarrow	PROPN
bracis-33605	169	44	\infty	\infty	PROPN
bracis-33605	169	45	\	\	PROPN
bracis-33605	169	46	)	)	PUNCT
bracis-33605	169	47	is	be	AUX
bracis-33605	169	48	the	the	DET
bracis-33605	169	49	number	number	NOUN
bracis-33605	169	50	of	of	ADP
bracis-33605	169	51	episodes	episode	NOUN
bracis-33605	169	52	,	,	PUNCT
bracis-33605	169	53	so	so	ADV
bracis-33605	169	54	:	:	PUNCT
bracis-33605	169	55	1	1	X
bracis-33605	169	56	.	.	X
bracis-33605	169	57	q	q	X
bracis-33605	169	58	-	-	PUNCT
bracis-33605	169	59	learning	learning	NOUN
bracis-33605	169	60	-	-	PUNCT
bracis-33605	169	61	gubs	gub	NOUN
bracis-33605	169	62	converges	converge	NOUN
bracis-33605	169	63	to	to	ADP
bracis-33605	169	64	the	the	DET
bracis-33605	169	65	value	value	NOUN
bracis-33605	169	66	\(q'^*(x	\(q'^*(x	NOUN
bracis-33605	169	67	,	,	PUNCT
bracis-33605	169	68	a)\	a)\	NUM
bracis-33605	169	69	)	)	PUNCT
bracis-33605	170	1	where	where	SCONJ
bracis-33605	170	2	,	,	PUNCT
bracis-33605	170	3	\(q^*(x	\(q^*(x	PROPN
bracis-33605	170	4	,	,	PUNCT
bracis-33605	170	5	a	a	PRON
bracis-33605	170	6	)	)	PUNCT
bracis-33605	170	7	q'^*(x	q'^*(x	ADJ
bracis-33605	170	8	,	,	PUNCT
bracis-33605	170	9	a	a	PRON
bracis-33605	170	10	)	)	PUNCT
bracis-33605	170	11	\ge	\ge	PROPN
bracis-33605	170	12	0\	0\	PROPN
bracis-33605	170	13	)	)	PUNCT
bracis-33605	170	14	for	for	ADP
bracis-33605	170	15	every	every	DET
bracis-33605	170	16	extended	extended	ADJ
bracis-33605	170	17	state	state	NOUN
bracis-33605	170	18	x	x	NOUN
bracis-33605	170	19	and	and	CCONJ
bracis-33605	170	20	action	action	NOUN
bracis-33605	170	21	a	a	DET
bracis-33605	170	22	,	,	PUNCT
bracis-33605	170	23	and	and	CCONJ
bracis-33605	170	24	\(q^*(x	\(q^*(x	NOUN
bracis-33605	170	25	,	,	PUNCT
bracis-33605	170	26	a)\	a)\	NUM
bracis-33605	170	27	)	)	PUNCT
bracis-33605	170	28	is	be	AUX
bracis-33605	170	29	the	the	DET
bracis-33605	170	30	optimal	optimal	ADJ
bracis-33605	170	31	value	value	NOUN
bracis-33605	170	32	for	for	ADP
bracis-33605	170	33	the	the	DET
bracis-33605	170	34	problem	problem	NOUN
bracis-33605	170	35	in	in	ADP
bracis-33605	170	36	definition	definition	NOUN
bracis-33605	170	37	2	2	NUM
bracis-33605	170	38	;	;	PUNCT
bracis-33605	170	39	2	2	NUM
bracis-33605	170	40	.	.	PUNCT
bracis-33605	171	1	\(\lim	\(\lim	NOUN
bracis-33605	172	1	_	_	PRON
bracis-33605	172	2	{	{	PUNCT
bracis-33605	172	3	c_{max	c_{max	VERB
bracis-33605	172	4	}	}	PUNCT
bracis-33605	172	5	\rightarrow	\rightarrow	PROPN
bracis-33605	172	6	\infty	\infty	PROPN
bracis-33605	172	7	}	}	PUNCT
bracis-33605	172	8	q'^*(x	q'^*(x	PROPN
bracis-33605	172	9	,	,	PUNCT
bracis-33605	172	10	a	a	PRON
bracis-33605	172	11	)	)	PUNCT
bracis-33605	172	12	=	=	SYM
bracis-33605	172	13	q^*(x	q^*(x	NOUN
bracis-33605	172	14	,	,	PUNCT
bracis-33605	172	15	a)\	a)\	NUM
bracis-33605	172	16	)	)	PUNCT
bracis-33605	172	17	.	.	PUNCT
bracis-33605	173	1	proof	proof	NOUN
bracis-33605	173	2	the	the	DET
bracis-33605	173	3	q	q	NOUN
bracis-33605	173	4	-	-	PUNCT
bracis-33605	173	5	learning	learning	NOUN
bracis-33605	173	6	-	-	PUNCT
bracis-33605	173	7	gubs	gub	NOUN
bracis-33605	173	8	algorithm	algorithm	NOUN
bracis-33605	173	9	restates	restate	VERB
bracis-33605	173	10	the	the	DET
bracis-33605	173	11	original	original	ADJ
bracis-33605	173	12	ssp	ssp	NOUN
bracis-33605	173	13	m	m	VERB
bracis-33605	173	14	(	(	PUNCT
bracis-33605	173	15	definition	definition	NOUN
bracis-33605	173	16	2	2	NUM
bracis-33605	173	17	)	)	PUNCT
bracis-33605	173	18	into	into	ADP
bracis-33605	173	19	an	an	DET
bracis-33605	173	20	ssp	ssp	NOUN
bracis-33605	173	21	\(m'\	\(m'\	NOUN
bracis-33605	173	22	)	)	PUNCT
bracis-33605	173	23	with	with	ADP
bracis-33605	173	24	maximum	maximum	ADJ
bracis-33605	173	25	cost	cost	NOUN
bracis-33605	173	26	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	173	27	)	)	PUNCT
bracis-33605	173	28	(	(	PUNCT
bracis-33605	173	29	definition	definition	NOUN
bracis-33605	173	30	4	4	NUM
bracis-33605	173	31	)	)	PUNCT
bracis-33605	173	32	.	.	PUNCT
bracis-33605	174	1	since	since	SCONJ
bracis-33605	174	2	\(m'\	\(m'\	NOUN
bracis-33605	174	3	)	)	PUNCT
bracis-33605	174	4	has	have	VERB
bracis-33605	174	5	a	a	DET
bracis-33605	174	6	set	set	NOUN
bracis-33605	174	7	of	of	ADP
bracis-33605	174	8	states	state	NOUN
bracis-33605	174	9	\(x	\(x	PROPN
bracis-33605	174	10	'	'	PUNCT
bracis-33605	174	11	\subseteq	\subseteq	PROPN
bracis-33605	174	12	x\	x\	NOUN
bracis-33605	174	13	)	)	PUNCT
bracis-33605	174	14	,	,	PUNCT
bracis-33605	174	15	then	then	ADV
bracis-33605	174	16	\(q^*(x	\(q^*(x	PROPN
bracis-33605	174	17	,	,	PUNCT
bracis-33605	174	18	a	a	PRON
bracis-33605	174	19	)	)	PUNCT
bracis-33605	174	20	q'^*(x	q'^*(x	ADJ
bracis-33605	174	21	,	,	PUNCT
bracis-33605	174	22	a	a	PRON
bracis-33605	174	23	)	)	PUNCT
bracis-33605	174	24	\ge	\ge	PROPN
bracis-33605	174	25	0\	0\	PROPN
bracis-33605	174	26	)	)	PUNCT
bracis-33605	174	27	.	.	PUNCT
bracis-33605	175	1	let	let	VERB
bracis-33605	175	2	\(q^{\pi	\(q^{\pi	NOUN
bracis-33605	175	3	*	*	NOUN
bracis-33605	175	4	}	}	PUNCT
bracis-33605	175	5	\	\	NOUN
bracis-33605	175	6	)	)	PUNCT
bracis-33605	175	7	be	be	AUX
bracis-33605	175	8	the	the	DET
bracis-33605	175	9	optimal	optimal	ADJ
bracis-33605	175	10	policy	policy	NOUN
bracis-33605	175	11	value	value	NOUN
bracis-33605	175	12	of	of	ADP
bracis-33605	175	13	the	the	DET
bracis-33605	175	14	original	original	ADJ
bracis-33605	175	15	ssp	ssp	NOUN
bracis-33605	175	16	in	in	ADP
bracis-33605	175	17	the	the	DET
bracis-33605	175	18	ssp	ssp	NOUN
bracis-33605	175	19	of	of	ADP
bracis-33605	175	20	definition	definition	NOUN
bracis-33605	175	21	4	4	NUM
bracis-33605	175	22	,	,	PUNCT
bracis-33605	175	23	then	then	ADV
bracis-33605	175	24	\(q^*(x	\(q^*(x	PROPN
bracis-33605	175	25	,	,	PUNCT
bracis-33605	175	26	a	a	PRON
bracis-33605	175	27	)	)	PUNCT
bracis-33605	176	1	\ge	\ge	PROPN
bracis-33605	176	2	q'^	q'^	NOUN
bracis-33605	176	3	*	*	SYM
bracis-33605	176	4	(	(	PUNCT
bracis-33605	176	5	x	x	X
bracis-33605	176	6	,	,	PUNCT
bracis-33605	176	7	a	a	PRON
bracis-33605	176	8	)	)	PUNCT
bracis-33605	176	9	\ge	\ge	PROPN
bracis-33605	176	10	q^{\pi	q^{\pi	NOUN
bracis-33605	176	11	*	*	PUNCT
bracis-33605	176	12	}	}	PUNCT
bracis-33605	176	13	(	(	PUNCT
bracis-33605	176	14	x	x	NOUN
bracis-33605	176	15	,	,	PUNCT
bracis-33605	176	16	a)\	a)\	NUM
bracis-33605	176	17	)	)	PUNCT
bracis-33605	176	18	.	.	PUNCT
bracis-33605	177	1	the	the	DET
bracis-33605	177	2	difference	difference	NOUN
bracis-33605	177	3	between	between	ADP
bracis-33605	177	4	these	these	DET
bracis-33605	177	5	values	value	NOUN
bracis-33605	177	6	is	be	AUX
bracis-33605	177	7	given	give	VERB
bracis-33605	177	8	by	by	ADP
bracis-33605	177	9	the	the	DET
bracis-33605	177	10	traces	trace	NOUN
bracis-33605	177	11	that	that	PRON
bracis-33605	177	12	do	do	AUX
bracis-33605	177	13	not	not	PART
bracis-33605	177	14	reach	reach	VERB
bracis-33605	177	15	the	the	DET
bracis-33605	177	16	goal	goal	NOUN
bracis-33605	177	17	before	before	ADP
bracis-33605	177	18	the	the	DET
bracis-33605	177	19	cost	cost	NOUN
bracis-33605	177	20	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	177	21	)	)	PUNCT
bracis-33605	177	22	.	.	PUNCT
bracis-33605	178	1	the	the	PRON
bracis-33605	178	2	greater	great	ADJ
bracis-33605	178	3	the	the	DET
bracis-33605	178	4	value	value	NOUN
bracis-33605	178	5	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	178	6	)	)	PUNCT
bracis-33605	178	7	,	,	PUNCT
bracis-33605	178	8	the	the	PRON
bracis-33605	178	9	lower	low	ADJ
bracis-33605	178	10	the	the	DET
bracis-33605	178	11	probability	probability	NOUN
bracis-33605	178	12	of	of	ADP
bracis-33605	178	13	having	have	VERB
bracis-33605	178	14	these	these	DET
bracis-33605	178	15	traces	trace	NOUN
bracis-33605	178	16	,	,	PUNCT
bracis-33605	178	17	therefore	therefore	ADV
bracis-33605	178	18	\(q^*(x	\(q^*(x	PROPN
bracis-33605	178	19	,	,	PUNCT
bracis-33605	178	20	a	a	PRON
bracis-33605	178	21	)	)	PUNCT
bracis-33605	178	22	\rightarrow	\rightarrow	PROPN
bracis-33605	178	23	q^{\pi	q^{\pi	PROPN
bracis-33605	178	24	*	*	PUNCT
bracis-33605	178	25	}	}	PUNCT
bracis-33605	178	26	(	(	PUNCT
bracis-33605	178	27	x	x	NOUN
bracis-33605	178	28	,	,	PUNCT
bracis-33605	178	29	a)\	a)\	NUM
bracis-33605	178	30	)	)	PUNCT
bracis-33605	178	31	and	and	CCONJ
bracis-33605	178	32	\	\	PROPN
bracis-33605	178	33	(	(	PUNCT
bracis-33605	178	34	q^{\pi	q^{\pi	PROPN
bracis-33605	178	35	*	*	PUNCT
bracis-33605	178	36	}	}	PUNCT
bracis-33605	178	37	(	(	PUNCT
bracis-33605	178	38	x	x	X
bracis-33605	178	39	,	,	PUNCT
bracis-33605	178	40	a	a	PRON
bracis-33605	178	41	)	)	PUNCT
bracis-33605	179	1	\rightarrow	\rightarrow	PROPN
bracis-33605	179	2	q'^	q'^	PROPN
bracis-33605	179	3	*	*	PUNCT
bracis-33605	179	4	(	(	PUNCT
bracis-33605	179	5	x	x	NOUN
bracis-33605	179	6	,	,	PUNCT
bracis-33605	179	7	a)\	a)\	NUM
bracis-33605	179	8	)	)	PUNCT
bracis-33605	179	9	.	.	PUNCT
bracis-33605	180	1	thus	thus	ADV
bracis-33605	180	2	\(\lim	\(\lim	NOUN
bracis-33605	180	3	_	_	PRON
bracis-33605	180	4	{	{	PUNCT
bracis-33605	180	5	c_{max	c_{max	VERB
bracis-33605	180	6	}	}	PUNCT
bracis-33605	180	7	\rightarrow	\rightarrow	PROPN
bracis-33605	180	8	\infty	\infty	PROPN
bracis-33605	180	9	}	}	PUNCT
bracis-33605	180	10	q'^*(x	q'^*(x	PROPN
bracis-33605	180	11	,	,	PUNCT
bracis-33605	180	12	a	a	PRON
bracis-33605	180	13	)	)	PUNCT
bracis-33605	180	14	=	=	SYM
bracis-33605	180	15	q^*(x	q^*(x	NOUN
bracis-33605	180	16	,	,	PUNCT
bracis-33605	180	17	a)\	a)\	NUM
bracis-33605	180	18	)	)	PUNCT
bracis-33605	180	19	.	.	PUNCT
bracis-33605	181	1	3.2	3.2	NUM
bracis-33605	181	2	q	q	X
bracis-33605	181	3	-	-	PUNCT
bracis-33605	181	4	learning	learning	NOUN
bracis-33605	181	5	-	-	PUNCT
bracis-33605	181	6	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	181	7	)	)	PUNCT
bracis-33605	181	8	q	q	ADJ
bracis-33605	181	9	-	-	PUNCT
bracis-33605	181	10	learning	learning	NOUN
bracis-33605	181	11	-	-	PUNCT
bracis-33605	181	12	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	181	13	)	)	PUNCT
bracis-33605	181	14	is	be	AUX
bracis-33605	181	15	based	base	VERB
bracis-33605	181	16	on	on	ADP
bracis-33605	181	17	gubs	gub	NOUN
bracis-33605	181	18	-	-	PUNCT
bracis-33605	181	19	vi	vi	NOUN
bracis-33605	181	20	,	,	PUNCT
bracis-33605	181	21	egubs	egub	NOUN
bracis-33605	181	22	-	-	PUNCT
bracis-33605	181	23	vi	vi	NOUN
bracis-33605	181	24	,	,	PUNCT
bracis-33605	181	25	q	q	NOUN
bracis-33605	181	26	-	-	PUNCT
bracis-33605	181	27	learning	learning	NOUN
bracis-33605	181	28	and	and	CCONJ
bracis-33605	181	29	dyna	dyna	NOUN
bracis-33605	181	30	-	-	PROPN
bracis-33605	181	31	q	q	NOUN
bracis-33605	182	1	[	[	X
bracis-33605	182	2	11	11	NUM
bracis-33605	182	3	]	]	PUNCT
bracis-33605	182	4	algorithms	algorithm	NOUN
bracis-33605	182	5	.	.	PUNCT
bracis-33605	183	1	as	as	ADP
bracis-33605	183	2	egubs	egub	NOUN
bracis-33605	183	3	-	-	PUNCT
bracis-33605	183	4	vi	vi	NOUN
bracis-33605	183	5	,	,	PUNCT
bracis-33605	183	6	q	q	ADJ
bracis-33605	183	7	-	-	PUNCT
bracis-33605	183	8	learning	learning	NOUN
bracis-33605	183	9	-	-	PUNCT
bracis-33605	183	10	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	183	11	)	)	PUNCT
bracis-33605	183	12	computes	compute	VERB
bracis-33605	183	13	a	a	DET
bracis-33605	183	14	lexicographic	lexicographic	ADJ
bracis-33605	183	15	policy	policy	NOUN
bracis-33605	183	16	as	as	ADP
bracis-33605	183	17	an	an	DET
bracis-33605	183	18	initial	initial	ADJ
bracis-33605	183	19	policy	policy	NOUN
bracis-33605	183	20	and	and	CCONJ
bracis-33605	183	21	calculates	calculate	VERB
bracis-33605	183	22	a	a	DET
bracis-33605	183	23	non	non	ADJ
bracis-33605	183	24	-	-	ADJ
bracis-33605	183	25	stationary	stationary	ADJ
bracis-33605	183	26	final	final	ADJ
bracis-33605	183	27	policy	policy	NOUN
bracis-33605	183	28	.	.	PUNCT
bracis-33605	184	1	although	although	SCONJ
bracis-33605	184	2	q	q	NOUN
bracis-33605	184	3	-	-	PUNCT
bracis-33605	184	4	learning	learning	NOUN
bracis-33605	184	5	-	-	PUNCT
bracis-33605	184	6	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	184	7	)	)	PUNCT
bracis-33605	184	8	is	be	AUX
bracis-33605	184	9	based	base	VERB
bracis-33605	184	10	on	on	ADP
bracis-33605	184	11	the	the	DET
bracis-33605	184	12	gubs	gubs	ADJ
bracis-33605	184	13	-	-	PUNCT
bracis-33605	184	14	vi	vi	NOUN
bracis-33605	184	15	algorithm	algorithm	NOUN
bracis-33605	184	16	that	that	PRON
bracis-33605	184	17	solves	solve	VERB
bracis-33605	184	18	the	the	DET
bracis-33605	184	19	gubs	gubs	ADJ
bracis-33605	184	20	criterion	criterion	NOUN
bracis-33605	184	21	,	,	PUNCT
bracis-33605	184	22	 	 	SPACE
bracis-33605	184	23	it	it	PRON
bracis-33605	184	24	calculates	calculate	VERB
bracis-33605	184	25	the	the	DET
bracis-33605	184	26	lexicographic	lexicographic	ADJ
bracis-33605	184	27	policy	policy	NOUN
bracis-33605	184	28	with	with	ADP
bracis-33605	184	29	the	the	DET
bracis-33605	184	30	risk	risk	NOUN
bracis-33605	184	31	-	-	PUNCT
bracis-33605	184	32	sensitive	sensitive	ADJ
bracis-33605	184	33	lexicographic	lexicographic	ADJ
bracis-33605	184	34	-	-	PUNCT
bracis-33605	184	35	vi	vi	NOUN
bracis-33605	184	36	algorithm	algorithm	NOUN
bracis-33605	184	37	(	(	PUNCT
bracis-33605	184	38	[	[	X
bracis-33605	184	39	3	3	NUM
bracis-33605	184	40	]	]	PUNCT
bracis-33605	184	41	)	)	PUNCT
bracis-33605	184	42	which	which	PRON
bracis-33605	184	43	is	be	AUX
bracis-33605	184	44	based	base	VERB
bracis-33605	184	45	on	on	ADP
bracis-33605	184	46	the	the	DET
bracis-33605	184	47	exponential	exponential	ADJ
bracis-33605	184	48	utility	utility	NOUN
bracis-33605	184	49	function	function	NOUN
bracis-33605	184	50	.	.	PUNCT
bracis-33605	185	1	thus	thus	ADV
bracis-33605	185	2	,	,	PUNCT
bracis-33605	185	3	q	q	X
bracis-33605	185	4	-	-	PUNCT
bracis-33605	185	5	learning	learning	NOUN
bracis-33605	185	6	-	-	PUNCT
bracis-33605	185	7	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	185	8	)	)	PUNCT
bracis-33605	185	9	solves	solve	VERB
bracis-33605	185	10	the	the	DET
bracis-33605	185	11	egubs	egub	NOUN
bracis-33605	185	12	criterion	criterion	NOUN
bracis-33605	185	13	.	.	PUNCT
bracis-33605	186	1	different	different	ADJ
bracis-33605	186	2	 	 	SPACE
bracis-33605	186	3	from	from	ADP
bracis-33605	186	4	q	q	ADJ
bracis-33605	186	5	-	-	PUNCT
bracis-33605	186	6	learning	learning	NOUN
bracis-33605	186	7	-	-	PUNCT
bracis-33605	186	8	gubs	gub	NOUN
bracis-33605	186	9	,	,	PUNCT
bracis-33605	186	10	q	q	ADJ
bracis-33605	186	11	-	-	PUNCT
bracis-33605	186	12	learning	learning	NOUN
bracis-33605	186	13	-	-	PUNCT
bracis-33605	186	14	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	186	15	)	)	PUNCT
bracis-33605	186	16	allows	allow	VERB
bracis-33605	186	17	to	to	PART
bracis-33605	186	18	calculate	calculate	VERB
bracis-33605	186	19	the	the	DET
bracis-33605	186	20	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	186	21	)	)	PUNCT
bracis-33605	186	22	value	value	NOUN
bracis-33605	186	23	automatically	automatically	ADV
bracis-33605	186	24	.	.	PUNCT
bracis-33605	187	1	in	in	ADP
bracis-33605	187	2	the	the	DET
bracis-33605	187	3	q	q	NOUN
bracis-33605	187	4	-	-	PUNCT
bracis-33605	187	5	learning	learning	NOUN
bracis-33605	187	6	-	-	PUNCT
bracis-33605	187	7	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	187	8	)	)	PUNCT
bracis-33605	187	9	algorithm	algorithm	NOUN
bracis-33605	187	10	,	,	PUNCT
bracis-33605	187	11	a	a	DET
bracis-33605	187	12	lexicographic	lexicographic	ADJ
bracis-33605	187	13	policy	policy	NOUN
bracis-33605	187	14	\(\pi	\(\pi	ADP
bracis-33605	187	15	_	_	PUNCT
bracis-33605	187	16	l\	l\	NOUN
bracis-33605	187	17	)	)	PUNCT
bracis-33605	187	18	is	be	AUX
bracis-33605	187	19	found	find	VERB
bracis-33605	187	20	based	base	VERB
bracis-33605	187	21	on	on	ADP
bracis-33605	187	22	a	a	DET
bracis-33605	187	23	model	model	NOUN
bracis-33605	187	24	of	of	ADP
bracis-33605	187	25	the	the	DET
bracis-33605	187	26	transition	transition	NOUN
bracis-33605	187	27	probability	probability	NOUN
bracis-33605	187	28	\(\hat{p}\	\(\hat{p}\	PROPN
bracis-33605	187	29	)	)	PUNCT
bracis-33605	187	30	and	and	CCONJ
bracis-33605	187	31	the	the	DET
bracis-33605	187	32	costs	cost	NOUN
bracis-33605	187	33	of	of	ADP
bracis-33605	187	34	actions	action	NOUN
bracis-33605	187	35	\(\hat{c}\	\(\hat{c}\	PROPN
bracis-33605	187	36	)	)	PUNCT
bracis-33605	187	37	in	in	ADP
bracis-33605	187	38	the	the	DET
bracis-33605	187	39	ssp	ssp	NOUN
bracis-33605	187	40	,	,	PUNCT
bracis-33605	187	41	which	which	PRON
bracis-33605	187	42	is	be	AUX
bracis-33605	187	43	calculated	calculate	VERB
bracis-33605	187	44	at	at	ADP
bracis-33605	187	45	each	each	DET
bracis-33605	187	46	f	f	PROPN
bracis-33605	187	47	episodes	episode	NOUN
bracis-33605	187	48	with	with	ADP
bracis-33605	187	49	the	the	DET
bracis-33605	187	50	risk	risk	NOUN
bracis-33605	187	51	-	-	PUNCT
bracis-33605	187	52	sensitive	sensitive	ADJ
bracis-33605	187	53	lexicographic	lexicographic	ADJ
bracis-33605	187	54	-	-	PUNCT
bracis-33605	187	55	vi	vi	NOUN
bracis-33605	187	56	algorithm	algorithm	NOUN
bracis-33605	187	57	[	[	X
bracis-33605	187	58	3	3	NUM
bracis-33605	187	59	]	]	PUNCT
bracis-33605	187	60	.	.	PUNCT
bracis-33605	188	1	as	as	ADP
bracis-33605	188	2	in	in	ADP
bracis-33605	188	3	[	[	X
bracis-33605	188	4	11	11	NUM
bracis-33605	188	5	]	]	PUNCT
bracis-33605	188	6	and	and	CCONJ
bracis-33605	188	7	[	[	X
bracis-33605	188	8	2	2	X
bracis-33605	188	9	]	]	PUNCT
bracis-33605	188	10	the	the	DET
bracis-33605	188	11	model	model	NOUN
bracis-33605	188	12	of	of	ADP
bracis-33605	188	13	probabilities	probability	NOUN
bracis-33605	188	14	is	be	AUX
bracis-33605	188	15	calculated	calculate	VERB
bracis-33605	188	16	by	by	ADP
bracis-33605	188	17	\(\hat{p}(s	\(\hat{p}(s	PROPN
bracis-33605	188	18	,	,	PUNCT
bracis-33605	188	19	a	a	DET
bracis-33605	188	20	,	,	PUNCT
bracis-33605	188	21	s	s	NOUN
bracis-33605	188	22	'	'	PUNCT
bracis-33605	188	23	)	)	PUNCT
bracis-33605	189	1	=	=	VERB
bracis-33605	189	2	\frac{\sum	\frac{\sum	PROPN
bracis-33605	189	3	n(s	n(s	PROPN
bracis-33605	189	4	,	,	PUNCT
bracis-33605	189	5	a	a	PRON
bracis-33605	189	6	,	,	PUNCT
bracis-33605	189	7	s')}{\sum	s')}{\sum	NOUN
bracis-33605	189	8	_	_	PUNCT
bracis-33605	189	9	{	{	PUNCT
bracis-33605	189	10	s''}n(s	s''}n(s	NOUN
bracis-33605	189	11	,	,	PUNCT
bracis-33605	189	12	a	a	PRON
bracis-33605	189	13	,	,	PUNCT
bracis-33605	189	14	s'')}\	s'')}\	NUM
bracis-33605	189	15	)	)	PUNCT
bracis-33605	189	16	,	,	PUNCT
bracis-33605	189	17	and	and	CCONJ
bracis-33605	189	18	the	the	DET
bracis-33605	189	19	model	model	NOUN
bracis-33605	189	20	of	of	ADP
bracis-33605	189	21	cost	cost	NOUN
bracis-33605	189	22	is	be	AUX
bracis-33605	189	23	calculated	calculate	VERB
bracis-33605	189	24	by	by	ADP
bracis-33605	189	25	\	\	PROPN
bracis-33605	189	26	(	(	PUNCT
bracis-33605	189	27	\hat{c}(s	\hat{c}(s	PROPN
bracis-33605	189	28	,	,	PUNCT
bracis-33605	189	29	a	a	PRON
bracis-33605	189	30	)	)	PUNCT
bracis-33605	189	31	=	=	SYM
bracis-33605	189	32	\frac{\sum	\frac{\sum	NOUN
bracis-33605	189	33	c(s	c(	NOUN
bracis-33605	189	34	,	,	PUNCT
bracis-33605	189	35	a)}{\sum	a)}{\sum	VERB
bracis-33605	190	1	_	_	PRON
bracis-33605	190	2	{	{	PUNCT
bracis-33605	190	3	s''}n(s	s''}n(s	NOUN
bracis-33605	190	4	,	,	PUNCT
bracis-33605	190	5	a	a	PRON
bracis-33605	190	6	,	,	PUNCT
bracis-33605	190	7	s'')}\	s'')}\	NOUN
bracis-33605	190	8	)	)	PUNCT
bracis-33605	190	9	where	where	SCONJ
bracis-33605	190	10	,	,	PUNCT
bracis-33605	190	11	\(n(s	\(n(s	PROPN
bracis-33605	190	12	,	,	PUNCT
bracis-33605	190	13	a	a	DET
bracis-33605	190	14	,	,	PUNCT
bracis-33605	190	15	s')\	s')\	NOUN
bracis-33605	190	16	)	)	PUNCT
bracis-33605	190	17	is	be	AUX
bracis-33605	190	18	the	the	DET
bracis-33605	190	19	number	number	NOUN
bracis-33605	190	20	of	of	ADP
bracis-33605	190	21	times	time	NOUN
bracis-33605	190	22	a	a	DET
bracis-33605	190	23	state	state	NOUN
bracis-33605	190	24	s	s	PART
bracis-33605	190	25	is	be	AUX
bracis-33605	190	26	visited	visit	VERB
bracis-33605	190	27	,	,	PUNCT
bracis-33605	190	28	the	the	DET
bracis-33605	190	29	action	action	NOUN
bracis-33605	190	30	a	a	PRON
bracis-33605	190	31	is	be	AUX
bracis-33605	190	32	performed	perform	VERB
bracis-33605	190	33	and	and	CCONJ
bracis-33605	190	34	passed	pass	VERB
bracis-33605	190	35	to	to	ADP
bracis-33605	190	36	the	the	DET
bracis-33605	190	37	state	state	NOUN
bracis-33605	190	38	\(s'\	\(s'\	PROPN
bracis-33605	190	39	)	)	PUNCT
bracis-33605	190	40	.	.	PUNCT
bracis-33605	191	1	to	to	PART
bracis-33605	191	2	compute	compute	VERB
bracis-33605	191	3	the	the	DET
bracis-33605	191	4	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	191	5	)	)	PUNCT
bracis-33605	191	6	value	value	NOUN
bracis-33605	191	7	,	,	PUNCT
bracis-33605	191	8	after	after	ADP
bracis-33605	191	9	calculating	calculate	VERB
bracis-33605	191	10	the	the	DET
bracis-33605	191	11	lexicographic	lexicographic	ADJ
bracis-33605	191	12	policy	policy	NOUN
bracis-33605	191	13	\(\pi	\(\pi	ADP
bracis-33605	192	1	_	_	PUNCT
bracis-33605	193	1	l\	l\	NOUN
bracis-33605	193	2	)	)	PUNCT
bracis-33605	193	3	,	,	PUNCT
bracis-33605	193	4	the	the	DET
bracis-33605	193	5	algorithm	algorithm	NOUN
bracis-33605	193	6	q	q	NOUN
bracis-33605	193	7	-	-	PUNCT
bracis-33605	193	8	learning	learning	NOUN
bracis-33605	193	9	-	-	PUNCT
bracis-33605	193	10	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	193	11	)	)	PUNCT
bracis-33605	193	12	evaluates	evaluate	VERB
bracis-33605	193	13	the	the	DET
bracis-33605	193	14	expected	expect	VERB
bracis-33605	193	15	cost	cost	NOUN
bracis-33605	193	16	to	to	PART
bracis-33605	193	17	goal	goal	VERB
bracis-33605	193	18	\(\overline{c}^{\pi	\(\overline{c}^{\pi	NOUN
bracis-33605	193	19	_	_	PRON
bracis-33605	193	20	l}_g(s)\	l}_g(s)\	NOUN
bracis-33605	193	21	)	)	PUNCT
bracis-33605	193	22	of	of	ADP
bracis-33605	193	23	the	the	DET
bracis-33605	193	24	policy	policy	NOUN
bracis-33605	193	25	\(\pi	\(\pi	ADP
bracis-33605	193	26	_	_	PUNCT
bracis-33605	194	1	l\	l\	NOUN
bracis-33605	194	2	)	)	PUNCT
bracis-33605	194	3	iteratively	iteratively	ADV
bracis-33605	194	4	for	for	ADP
bracis-33605	194	5	each	each	DET
bracis-33605	194	6	state	state	NOUN
bracis-33605	194	7	s	s	VERB
bracis-33605	194	8	based	base	VERB
bracis-33605	194	9	on	on	ADP
bracis-33605	194	10	eq	eq	PROPN
bracis-33605	194	11	.	.	PUNCT
bracis-33605	194	12	 	 	SPACE
bracis-33605	194	13	5	5	NUM
bracis-33605	194	14	,	,	PUNCT
bracis-33605	194	15	without	without	ADP
bracis-33605	194	16	considering	consider	VERB
bracis-33605	194	17	the	the	DET
bracis-33605	194	18	infinite	infinite	ADJ
bracis-33605	194	19	costs	cost	NOUN
bracis-33605	194	20	of	of	ADP
bracis-33605	194	21	the	the	DET
bracis-33605	194	22	dead	dead	ADJ
bracis-33605	194	23	-	-	PUNCT
bracis-33605	194	24	end	end	NOUN
bracis-33605	194	25	states	state	NOUN
bracis-33605	194	26	.	.	PUNCT
bracis-33605	195	1	$	$	SYM
bracis-33605	195	2	$	$	SYM
bracis-33605	195	3	\begin{aligned	\begin{aligne	VERB
bracis-33605	195	4	}	}	PUNCT
bracis-33605	195	5	\overline{c}^{\pi	\overline{c}^{\pi	PROPN
bracis-33605	195	6	_	_	PUNCT
bracis-33605	195	7	l}_g(s	l}_g(s	PROPN
bracis-33605	195	8	)	)	PUNCT
bracis-33605	195	9	=	=	PRON
bracis-33605	195	10	{	{	PUNCT
bracis-33605	195	11	\left\	\left\	PROPN
bracis-33605	195	12	{	{	PUNCT
bracis-33605	195	13	\begin{array}{ll	\begin{array}{ll	PROPN
bracis-33605	195	14	}	}	PUNCT
bracis-33605	195	15	0	0	PUNCT
bracis-33605	195	16	&	&	CCONJ
bracis-33605	195	17	,	,	PUNCT
bracis-33605	195	18	s\in	s\in	VERB
bracis-33605	195	19	\mathcal	\mathcal	PROPN
bracis-33605	195	20	{	{	PUNCT
bracis-33605	195	21	g	g	NOUN
bracis-33605	195	22	}	}	PUNCT
bracis-33605	195	23	\cup	\cup	VERB
bracis-33605	195	24	\mathcal	\mathcal	PROPN
bracis-33605	195	25	{	{	PUNCT
bracis-33605	195	26	d	d	NOUN
bracis-33605	195	27	}	}	PUNCT
bracis-33605	195	28	\\	\\	NOUN
bracis-33605	195	29	\hat{c}(s,\pi	\hat{c}(s,\pi	PROPN
bracis-33605	195	30	_	_	PUNCT
bracis-33605	195	31	l(s	l(s	PROPN
bracis-33605	195	32	)	)	PUNCT
bracis-33605	195	33	)	)	PUNCT
bracis-33605	196	1	+	+	NUM
bracis-33605	196	2	\displaystyle	\displaystyle	PROPN
bracis-33605	196	3	\sum	\sum	NOUN
bracis-33605	196	4	_	_	PUNCT
bracis-33605	196	5	{	{	PUNCT
bracis-33605	196	6	s	s	X
bracis-33605	196	7	'	'	CCONJ
bracis-33605	196	8	\in	\in	ADJ
bracis-33605	196	9	s}\hat{p}(s,\pi	s}\hat{p}(s,\pi	NOUN
bracis-33605	196	10	_	_	NOUN
bracis-33605	196	11	l(s),s')\overline{c}^{\pi	l(s),s')\overline{c}^{\pi	NOUN
bracis-33605	196	12	_	_	PUNCT
bracis-33605	196	13	l}_g(s	l}_g(s	PROPN
bracis-33605	196	14	'	'	PUNCT
bracis-33605	196	15	)	)	PUNCT
bracis-33605	196	16	&	&	CCONJ
bracis-33605	196	17	,	,	PUNCT
bracis-33605	196	18	otherwise	otherwise	ADV
bracis-33605	196	19	\end{array}\right	\end{array}\right	PROPN
bracis-33605	196	20	.	.	PUNCT
bracis-33605	196	21	}	}	PUNCT
bracis-33605	196	22	\end{aligned}$$	\end{aligned}$$	X
bracis-33605	196	23	(	(	PUNCT
bracis-33605	196	24	5	5	NUM
bracis-33605	196	25	)	)	PUNCT
bracis-33605	196	26	the	the	DET
bracis-33605	196	27	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	196	28	)	)	PUNCT
bracis-33605	196	29	value	value	NOUN
bracis-33605	196	30	is	be	AUX
bracis-33605	196	31	equal	equal	ADJ
bracis-33605	196	32	to	to	ADP
bracis-33605	196	33	the	the	DET
bracis-33605	196	34	cost	cost	NOUN
bracis-33605	196	35	of	of	ADP
bracis-33605	196	36	the	the	DET
bracis-33605	196	37	lexicographic	lexicographic	ADJ
bracis-33605	196	38	policy	policy	NOUN
bracis-33605	196	39	from	from	ADP
bracis-33605	196	40	the	the	DET
bracis-33605	196	41	initial	initial	ADJ
bracis-33605	196	42	state	state	NOUN
bracis-33605	196	43	\(s_0\	\(s_0\	PROPN
bracis-33605	196	44	)	)	PUNCT
bracis-33605	196	45	to	to	ADP
bracis-33605	196	46	the	the	DET
bracis-33605	196	47	goal	goal	NOUN
bracis-33605	196	48	:	:	PUNCT
bracis-33605	196	49	$	$	SYM
bracis-33605	196	50	$	$	SYM
bracis-33605	196	51	\begin{aligned	\begin{aligne	VERB
bracis-33605	196	52	}	}	PUNCT
bracis-33605	196	53	c_{max	c_{max	VERB
bracis-33605	196	54	}	}	PUNCT
bracis-33605	196	55	=	=	PUNCT
bracis-33605	196	56	\overline{c}^{\pi	\overline{c}^{\pi	PROPN
bracis-33605	196	57	_	_	PROPN
bracis-33605	196	58	l}_g(s_0	l}_g(s_0	PROPN
bracis-33605	196	59	)	)	PUNCT
bracis-33605	196	60	.	.	PUNCT
bracis-33605	197	1	\end{aligned}$$	\end{aligned}$$	PROPN
bracis-33605	197	2	(	(	PUNCT
bracis-33605	197	3	6	6	NUM
bracis-33605	197	4	)	)	PUNCT
bracis-33605	197	5	since	since	SCONJ
bracis-33605	197	6	the	the	DET
bracis-33605	197	7	steps	step	NOUN
bracis-33605	197	8	for	for	ADP
bracis-33605	197	9	computing	compute	VERB
bracis-33605	197	10	the	the	DET
bracis-33605	197	11	lexicographic	lexicographic	ADJ
bracis-33605	197	12	policy	policy	NOUN
bracis-33605	197	13	and	and	CCONJ
bracis-33605	197	14	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	197	15	)	)	PUNCT
bracis-33605	197	16	are	be	AUX
bracis-33605	197	17	expensive	expensive	ADJ
bracis-33605	197	18	in	in	ADP
bracis-33605	197	19	terms	term	NOUN
bracis-33605	197	20	of	of	ADP
bracis-33605	197	21	processing	processing	NOUN
bracis-33605	197	22	time	time	NOUN
bracis-33605	197	23	cost	cost	NOUN
bracis-33605	197	24	,	,	PUNCT
bracis-33605	197	25	and	and	CCONJ
bracis-33605	197	26	this	this	DET
bracis-33605	197	27	cost	cost	NOUN
bracis-33605	197	28	depends	depend	VERB
bracis-33605	197	29	on	on	ADP
bracis-33605	197	30	the	the	DET
bracis-33605	197	31	number	number	NOUN
bracis-33605	197	32	of	of	ADP
bracis-33605	197	33	states	state	NOUN
bracis-33605	197	34	and	and	CCONJ
bracis-33605	197	35	the	the	DET
bracis-33605	197	36	type	type	NOUN
bracis-33605	197	37	of	of	ADP
bracis-33605	197	38	problem	problem	NOUN
bracis-33605	197	39	,	,	PUNCT
bracis-33605	197	40	these	these	DET
bracis-33605	197	41	steps	step	NOUN
bracis-33605	197	42	are	be	AUX
bracis-33605	197	43	only	only	ADV
bracis-33605	197	44	performed	perform	VERB
bracis-33605	197	45	in	in	ADP
bracis-33605	197	46	some	some	DET
bracis-33605	197	47	episodes	episode	NOUN
bracis-33605	197	48	.	.	PUNCT
bracis-33605	198	1	q	q	X
bracis-33605	198	2	-	-	PUNCT
bracis-33605	198	3	learning	learning	NOUN
bracis-33605	198	4	-	-	PUNCT
bracis-33605	198	5	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	198	6	)	)	PUNCT
bracis-33605	198	7	(	(	PUNCT
bracis-33605	198	8	algorithm	algorithm	NOUN
bracis-33605	198	9	 	 	SPACE
bracis-33605	198	10	2	2	NUM
bracis-33605	198	11	)	)	PUNCT
bracis-33605	198	12	,	,	PUNCT
bracis-33605	198	13	in	in	ADP
bracis-33605	198	14	addition	addition	NOUN
bracis-33605	198	15	to	to	ADP
bracis-33605	198	16	the	the	DET
bracis-33605	198	17	inputs	input	NOUN
bracis-33605	198	18	of	of	ADP
bracis-33605	198	19	the	the	DET
bracis-33605	198	20	q	q	NOUN
bracis-33605	198	21	-	-	PUNCT
bracis-33605	198	22	learning	learning	NOUN
bracis-33605	198	23	-	-	PUNCT
bracis-33605	198	24	gubs	gub	NOUN
bracis-33605	198	25	algorithm	algorithm	NOUN
bracis-33605	198	26	,	,	PUNCT
bracis-33605	198	27	receives	receive	VERB
bracis-33605	198	28	as	as	ADP
bracis-33605	198	29	input	input	NOUN
bracis-33605	198	30	the	the	DET
bracis-33605	198	31	minimum	minimum	NOUN
bracis-33605	198	32	error	error	NOUN
bracis-33605	198	33	\(\varepsilon	\(\varepsilon	NOUN
bracis-33605	198	34	_	_	PUNCT
bracis-33605	199	1	l\	l\	NOUN
bracis-33605	199	2	)	)	PUNCT
bracis-33605	199	3	used	use	VERB
bracis-33605	199	4	in	in	ADP
bracis-33605	199	5	the	the	DET
bracis-33605	199	6	risk	risk	NOUN
bracis-33605	199	7	-	-	PUNCT
bracis-33605	199	8	sensitive	sensitive	ADJ
bracis-33605	199	9	lexicographic	lexicographic	ADJ
bracis-33605	199	10	-	-	PUNCT
bracis-33605	199	11	vi	vi	NOUN
bracis-33605	199	12	algorithm	algorithm	NOUN
bracis-33605	199	13	and	and	CCONJ
bracis-33605	199	14	the	the	DET
bracis-33605	199	15	amount	amount	NOUN
bracis-33605	199	16	f	f	PROPN
bracis-33605	199	17	of	of	ADP
bracis-33605	199	18	episodes	episode	NOUN
bracis-33605	199	19	needed	need	VERB
bracis-33605	199	20	before	before	ADP
bracis-33605	199	21	recalculating	recalculate	VERB
bracis-33605	199	22	the	the	DET
bracis-33605	199	23	lexicographic	lexicographic	ADJ
bracis-33605	199	24	policy	policy	NOUN
bracis-33605	199	25	.	.	PUNCT
bracis-33605	200	1	after	after	ADP
bracis-33605	200	2	initializing	initialize	VERB
bracis-33605	200	3	the	the	DET
bracis-33605	200	4	values	value	NOUN
bracis-33605	200	5	of	of	ADP
bracis-33605	200	6	q	q	NOUN
bracis-33605	200	7	,	,	PUNCT
bracis-33605	200	8	n(s	n(s	PROPN
bracis-33605	200	9	,	,	PUNCT
bracis-33605	200	10	 	 	SPACE
bracis-33605	200	11	a	a	PRON
bracis-33605	200	12	)	)	PUNCT
bracis-33605	200	13	,	,	PUNCT
bracis-33605	200	14	\(n(s	\(n(s	PROPN
bracis-33605	200	15	,	,	PUNCT
bracis-33605	200	16	a	a	DET
bracis-33605	200	17	,	,	PUNCT
bracis-33605	200	18	s')\	s')\	NOUN
bracis-33605	200	19	)	)	PUNCT
bracis-33605	200	20	,	,	PUNCT
bracis-33605	200	21	\(\hat{c}(s	\(\hat{c}(s	PROPN
bracis-33605	200	22	,	,	PUNCT
bracis-33605	200	23	a)\	a)\	NUM
bracis-33605	200	24	)	)	PUNCT
bracis-33605	200	25	,	,	PUNCT
bracis-33605	200	26	\(\hat{c	\(\hat{c	NOUN
bracis-33605	200	27	}	}	PUNCT
bracis-33605	200	28	(	(	PUNCT
bracis-33605	200	29	s	s	PROPN
bracis-33605	200	30	,	,	PUNCT
bracis-33605	200	31	a)\	a)\	NUM
bracis-33605	200	32	)	)	PUNCT
bracis-33605	200	33	and	and	CCONJ
bracis-33605	200	34	\(\hat{p}(s	\(\hat{p}(s	PROPN
bracis-33605	200	35	,	,	PUNCT
bracis-33605	200	36	a	a	DET
bracis-33605	200	37	,	,	PUNCT
bracis-33605	200	38	s')\	s')\	NOUN
bracis-33605	200	39	)	)	PUNCT
bracis-33605	200	40	;	;	PUNCT
bracis-33605	200	41	algorithm	algorithm	NOUN
bracis-33605	200	42	 	 	SPACE
bracis-33605	200	43	2	2	NUM
bracis-33605	200	44	performs	perform	VERB
bracis-33605	200	45	a	a	DET
bracis-33605	200	46	set	set	NOUN
bracis-33605	200	47	of	of	ADP
bracis-33605	200	48	episodes	episode	NOUN
bracis-33605	200	49	.	.	PUNCT
bracis-33605	201	1	in	in	ADP
bracis-33605	201	2	each	each	DET
bracis-33605	201	3	episode	episode	NOUN
bracis-33605	201	4	,	,	PUNCT
bracis-33605	201	5	it	it	PRON
bracis-33605	201	6	performs	perform	VERB
bracis-33605	201	7	the	the	DET
bracis-33605	201	8	same	same	ADJ
bracis-33605	201	9	operations	operation	NOUN
bracis-33605	201	10	to	to	PART
bracis-33605	201	11	interact	interact	VERB
bracis-33605	201	12	with	with	ADP
bracis-33605	201	13	the	the	DET
bracis-33605	201	14	environment	environment	NOUN
bracis-33605	201	15	and	and	CCONJ
bracis-33605	201	16	to	to	PART
bracis-33605	201	17	calculate	calculate	VERB
bracis-33605	201	18	the	the	DET
bracis-33605	201	19	value	value	NOUN
bracis-33605	201	20	q	q	PUNCT
bracis-33605	201	21	as	as	ADP
bracis-33605	201	22	in	in	ADP
bracis-33605	201	23	the	the	DET
bracis-33605	201	24	q	q	NOUN
bracis-33605	201	25	-	-	PUNCT
bracis-33605	201	26	learning	learning	NOUN
bracis-33605	201	27	-	-	PUNCT
bracis-33605	201	28	gubs	gub	NOUN
bracis-33605	201	29	algorithm	algorithm	NOUN
bracis-33605	201	30	.	.	PUNCT
bracis-33605	202	1	furthermore	furthermore	ADV
bracis-33605	202	2	,	,	PUNCT
bracis-33605	202	3	each	each	DET
bracis-33605	202	4	f	f	PROPN
bracis-33605	202	5	episodes	episode	NOUN
bracis-33605	202	6	,	,	PUNCT
bracis-33605	202	7	algorithm	algorithm	NOUN
bracis-33605	202	8	 	 	SPACE
bracis-33605	202	9	2	2	NUM
bracis-33605	202	10	recalculates	recalculate	VERB
bracis-33605	202	11	the	the	DET
bracis-33605	202	12	model	model	NOUN
bracis-33605	202	13	of	of	ADP
bracis-33605	202	14	probabilities	probability	NOUN
bracis-33605	202	15	\(\hat{p}\	\(\hat{p}\	PROPN
bracis-33605	202	16	)	)	PUNCT
bracis-33605	202	17	and	and	CCONJ
bracis-33605	202	18	costs	cost	VERB
bracis-33605	202	19	\(\hat{c}\	\(\hat{c}\	PROPN
bracis-33605	202	20	)	)	PUNCT
bracis-33605	202	21	based	base	VERB
bracis-33605	202	22	on	on	ADP
bracis-33605	202	23	the	the	DET
bracis-33605	202	24	counts	count	NOUN
bracis-33605	202	25	of	of	ADP
bracis-33605	202	26	the	the	DET
bracis-33605	202	27	visits	visit	NOUN
bracis-33605	202	28	to	to	ADP
bracis-33605	202	29	states	state	NOUN
bracis-33605	202	30	n(s	n(s	PROPN
bracis-33605	202	31	,	,	PUNCT
bracis-33605	202	32	 	 	SPACE
bracis-33605	202	33	a	a	NOUN
bracis-33605	202	34	)	)	PUNCT
bracis-33605	202	35	and	and	CCONJ
bracis-33605	202	36	\(n(s	\(n(s	PROPN
bracis-33605	202	37	,	,	PUNCT
bracis-33605	202	38	a	a	DET
bracis-33605	202	39	,	,	PUNCT
bracis-33605	202	40	s')\	s')\	NOUN
bracis-33605	202	41	)	)	PUNCT
bracis-33605	202	42	and	and	CCONJ
bracis-33605	202	43	the	the	DET
bracis-33605	202	44	accumulated	accumulate	VERB
bracis-33605	202	45	cost	cost	NOUN
bracis-33605	202	46	of	of	ADP
bracis-33605	202	47	the	the	DET
bracis-33605	202	48	visits	visit	NOUN
bracis-33605	202	49	to	to	PART
bracis-33605	202	50	states	state	VERB
bracis-33605	202	51	\(\hat{c}(s	\(\hat{c}(s	PROPN
bracis-33605	202	52	,	,	PUNCT
bracis-33605	202	53	a)\	a)\	NUM
bracis-33605	202	54	)	)	PUNCT
bracis-33605	202	55	(	(	PUNCT
bracis-33605	202	56	lines	line	NOUN
bracis-33605	202	57	8	8	NUM
bracis-33605	202	58	and	and	CCONJ
bracis-33605	202	59	9	9	NUM
bracis-33605	202	60	)	)	PUNCT
bracis-33605	202	61	.	.	PUNCT
bracis-33605	203	1	it	it	PRON
bracis-33605	203	2	uses	use	VERB
bracis-33605	203	3	these	these	DET
bracis-33605	203	4	models	model	NOUN
bracis-33605	203	5	to	to	PART
bracis-33605	203	6	calculate	calculate	VERB
bracis-33605	203	7	the	the	DET
bracis-33605	203	8	lexicographic	lexicographic	ADJ
bracis-33605	203	9	policy	policy	NOUN
bracis-33605	203	10	(	(	PUNCT
bracis-33605	203	11	line	line	NOUN
bracis-33605	203	12	11	11	NUM
bracis-33605	203	13	)	)	PUNCT
bracis-33605	203	14	using	use	VERB
bracis-33605	203	15	the	the	DET
bracis-33605	203	16	risk	risk	NOUN
bracis-33605	203	17	-	-	PUNCT
bracis-33605	203	18	sensitive	sensitive	ADJ
bracis-33605	203	19	lexicographic	lexicographic	ADJ
bracis-33605	203	20	-	-	PUNCT
bracis-33605	203	21	vi	vi	NOUN
bracis-33605	203	22	algorithm	algorithm	NOUN
bracis-33605	203	23	[	[	X
bracis-33605	203	24	3	3	NUM
bracis-33605	203	25	]	]	PUNCT
bracis-33605	203	26	.	.	PUNCT
bracis-33605	204	1	the	the	DET
bracis-33605	204	2	value	value	NOUN
bracis-33605	204	3	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	204	4	)	)	PUNCT
bracis-33605	204	5	is	be	AUX
bracis-33605	204	6	calculated	calculate	VERB
bracis-33605	204	7	in	in	ADP
bracis-33605	204	8	line	line	NOUN
bracis-33605	204	9	12	12	NUM
bracis-33605	204	10	based	base	VERB
bracis-33605	204	11	on	on	ADP
bracis-33605	204	12	eq	eq	PROPN
bracis-33605	204	13	.	.	PROPN
bracis-33605	204	14	 	 	SPACE
bracis-33605	204	15	6	6	NUM
bracis-33605	204	16	.	.	PUNCT
bracis-33605	205	1	algorithm	algorithm	NOUN
bracis-33605	205	2	 	 	SPACE
bracis-33605	205	3	2	2	NUM
bracis-33605	205	4	calculates	calculate	VERB
bracis-33605	205	5	the	the	DET
bracis-33605	205	6	gubs	gubs	ADJ
bracis-33605	205	7	value	value	NOUN
bracis-33605	205	8	in	in	ADP
bracis-33605	205	9	line	line	NOUN
bracis-33605	205	10	25	25	NUM
bracis-33605	205	11	based	base	VERB
bracis-33605	205	12	on	on	ADP
bracis-33605	205	13	eq	eq	PROPN
bracis-33605	205	14	.	.	PUNCT
bracis-33605	205	15	 	 	SPACE
bracis-33605	205	16	4	4	X
bracis-33605	205	17	.	.	PUNCT
bracis-33605	206	1	finally	finally	ADV
bracis-33605	206	2	,	,	PUNCT
bracis-33605	206	3	based	base	VERB
bracis-33605	206	4	on	on	ADP
bracis-33605	206	5	the	the	DET
bracis-33605	206	6	gubs	gub	NOUN
bracis-33605	206	7	value	value	VERB
bracis-33605	206	8	the	the	DET
bracis-33605	206	9	optimal	optimal	ADJ
bracis-33605	206	10	gubs	gub	NOUN
bracis-33605	206	11	policy	policy	NOUN
bracis-33605	206	12	\(\pi	\(\pi	PROPN
bracis-33605	206	13	^*\	^*\	X
bracis-33605	206	14	)	)	PUNCT
bracis-33605	206	15	is	be	AUX
bracis-33605	206	16	calculated	calculate	VERB
bracis-33605	206	17	in	in	ADP
bracis-33605	206	18	line	line	NOUN
bracis-33605	206	19	26	26	NUM
bracis-33605	206	20	.	.	PUNCT
bracis-33605	207	1	the	the	DET
bracis-33605	207	2	counts	count	NOUN
bracis-33605	207	3	n(s	n(s	PROPN
bracis-33605	207	4	,	,	PUNCT
bracis-33605	207	5	 	 	SPACE
bracis-33605	207	6	a	a	NOUN
bracis-33605	207	7	)	)	PUNCT
bracis-33605	207	8	and	and	CCONJ
bracis-33605	207	9	\(n(s	\(n(s	PROPN
bracis-33605	207	10	,	,	PUNCT
bracis-33605	207	11	a	a	DET
bracis-33605	207	12	,	,	PUNCT
bracis-33605	207	13	s')\	s')\	NOUN
bracis-33605	207	14	)	)	PUNCT
bracis-33605	207	15	and	and	CCONJ
bracis-33605	207	16	the	the	DET
bracis-33605	207	17	accumulated	accumulate	VERB
bracis-33605	207	18	cost	cost	NOUN
bracis-33605	207	19	\(\hat{c}(s	\(\hat{c}(s	PROPN
bracis-33605	207	20	,	,	PUNCT
bracis-33605	207	21	a)\	a)\	NUM
bracis-33605	207	22	)	)	PUNCT
bracis-33605	207	23	used	use	VERB
bracis-33605	207	24	in	in	ADP
bracis-33605	207	25	the	the	DET
bracis-33605	207	26	model	model	NOUN
bracis-33605	207	27	calculation	calculation	NOUN
bracis-33605	207	28	,	,	PUNCT
bracis-33605	207	29	are	be	AUX
bracis-33605	207	30	updated	update	VERB
bracis-33605	207	31	in	in	ADP
bracis-33605	207	32	each	each	DET
bracis-33605	207	33	episode	episode	NOUN
bracis-33605	207	34	(	(	PUNCT
bracis-33605	207	35	line	line	NOUN
bracis-33605	207	36	27	27	NUM
bracis-33605	207	37	)	)	PUNCT
bracis-33605	207	38	.	.	PUNCT
bracis-33605	208	1	4	4	NUM
bracis-33605	208	2	experiments	experiment	NOUN
bracis-33605	208	3	the	the	DET
bracis-33605	208	4	experiments	experiment	NOUN
bracis-33605	208	5	were	be	AUX
bracis-33605	208	6	carried	carry	VERB
bracis-33605	208	7	out	out	ADP
bracis-33605	208	8	in	in	ADP
bracis-33605	208	9	two	two	NUM
bracis-33605	208	10	domains	domain	NOUN
bracis-33605	208	11	:	:	PUNCT
bracis-33605	208	12	the	the	DET
bracis-33605	208	13	river	river	NOUN
bracis-33605	208	14	[	[	X
bracis-33605	208	15	4	4	NUM
bracis-33605	208	16	]	]	PUNCT
bracis-33605	208	17	(	(	PUNCT
bracis-33605	208	18	see	see	VERB
bracis-33605	208	19	sect	sect	NOUN
bracis-33605	208	20	.	.	PUNCT
bracis-33605	208	21	 	 	SPACE
bracis-33605	208	22	1	1	X
bracis-33605	208	23	)	)	PUNCT
bracis-33605	208	24	and	and	CCONJ
bracis-33605	208	25	the	the	DET
bracis-33605	208	26	navigation	navigation	NOUN
bracis-33605	208	27	domains	domain	NOUN
bracis-33605	208	28	.	.	PUNCT
bracis-33605	209	1	in	in	ADP
bracis-33605	209	2	the	the	DET
bracis-33605	209	3	navigation	navigation	NOUN
bracis-33605	209	4	domain	domain	NOUN
bracis-33605	209	5	[	[	X
bracis-33605	209	6	10	10	NUM
bracis-33605	209	7	]	]	PUNCT
bracis-33605	209	8	,	,	PUNCT
bracis-33605	209	9	a	a	DET
bracis-33605	209	10	robot	robot	NOUN
bracis-33605	209	11	is	be	AUX
bracis-33605	209	12	walking	walk	VERB
bracis-33605	209	13	in	in	ADP
bracis-33605	209	14	a	a	DET
bracis-33605	209	15	grid	grid	NOUN
bracis-33605	209	16	with	with	ADP
bracis-33605	209	17	dimensions	dimension	NOUN
bracis-33605	209	18	\(n_x	\(n_x	NOUN
bracis-33605	209	19	\times	\times	ADP
bracis-33605	209	20	n_y\	n_y\	NOUN
bracis-33605	209	21	)	)	PUNCT
bracis-33605	209	22	.	.	PUNCT
bracis-33605	210	1	in	in	ADP
bracis-33605	210	2	this	this	DET
bracis-33605	210	3	domain	domain	NOUN
bracis-33605	210	4	,	,	PUNCT
bracis-33605	210	5	the	the	DET
bracis-33605	210	6	 	 	SPACE
bracis-33605	210	7	robot	robot	NOUN
bracis-33605	210	8	starts	start	VERB
bracis-33605	210	9	from	from	ADP
bracis-33605	210	10	an	an	DET
bracis-33605	210	11	initial	initial	ADJ
bracis-33605	210	12	position	position	NOUN
bracis-33605	210	13	\(s_0\	\(s_0\	NOUN
bracis-33605	210	14	)	)	PUNCT
bracis-33605	210	15	and	and	CCONJ
bracis-33605	210	16	tries	try	VERB
bracis-33605	210	17	to	to	PART
bracis-33605	210	18	 	 	SPACE
bracis-33605	210	19	go	go	VERB
bracis-33605	210	20	to	to	ADP
bracis-33605	210	21	a	a	DET
bracis-33605	210	22	destination	destination	NOUN
bracis-33605	210	23	position	position	NOUN
bracis-33605	210	24	.	.	PUNCT
bracis-33605	211	1	in	in	ADP
bracis-33605	211	2	the	the	DET
bracis-33605	211	3	first	first	ADJ
bracis-33605	211	4	and	and	CCONJ
bracis-33605	211	5	last	last	ADJ
bracis-33605	211	6	lines	line	NOUN
bracis-33605	211	7	(	(	PUNCT
bracis-33605	211	8	\(y	\(y	NOUN
bracis-33605	211	9	=	=	SYM
bracis-33605	211	10	1\	1\	NUM
bracis-33605	211	11	)	)	PUNCT
bracis-33605	211	12	and	and	CCONJ
bracis-33605	211	13	\(y	\(y	PRON
bracis-33605	211	14	=	=	SYM
bracis-33605	211	15	n_y\	n_y\	NOUN
bracis-33605	211	16	)	)	PUNCT
bracis-33605	211	17	)	)	PUNCT
bracis-33605	211	18	the	the	DET
bracis-33605	211	19	actions	action	NOUN
bracis-33605	211	20	are	be	AUX
bracis-33605	211	21	deterministic	deterministic	ADJ
bracis-33605	211	22	and	and	CCONJ
bracis-33605	211	23	in	in	ADP
bracis-33605	211	24	the	the	DET
bracis-33605	211	25	intermediate	intermediate	ADJ
bracis-33605	211	26	lines	line	NOUN
bracis-33605	211	27	,	,	PUNCT
bracis-33605	211	28	the	the	DET
bracis-33605	211	29	agent	agent	NOUN
bracis-33605	211	30	has	have	VERB
bracis-33605	211	31	a	a	DET
bracis-33605	211	32	probability	probability	NOUN
bracis-33605	211	33	p(x	p(x	NOUN
bracis-33605	211	34	,	,	PUNCT
bracis-33605	211	35	 	 	SPACE
bracis-33605	211	36	y	y	NOUN
bracis-33605	211	37	)	)	PUNCT
bracis-33605	211	38	of	of	ADP
bracis-33605	211	39	disappearing	disappear	VERB
bracis-33605	211	40	,	,	PUNCT
bracis-33605	211	41	going	go	VERB
bracis-33605	211	42	to	to	ADP
bracis-33605	211	43	a	a	DET
bracis-33605	211	44	dead	dead	ADJ
bracis-33605	211	45	-	-	PUNCT
bracis-33605	211	46	end	end	NOUN
bracis-33605	211	47	state	state	NOUN
bracis-33605	211	48	.	.	PUNCT
bracis-33605	212	1	the	the	DET
bracis-33605	212	2	probability	probability	NOUN
bracis-33605	212	3	of	of	ADP
bracis-33605	212	4	disappearing	disappear	VERB
bracis-33605	212	5	is	be	AUX
bracis-33605	212	6	greater	great	ADJ
bracis-33605	212	7	as	as	SCONJ
bracis-33605	212	8	the	the	DET
bracis-33605	212	9	agent	agent	NOUN
bracis-33605	212	10	is	be	AUX
bracis-33605	212	11	closer	close	ADJ
bracis-33605	212	12	to	to	ADP
bracis-33605	212	13	the	the	DET
bracis-33605	212	14	goal	goal	NOUN
bracis-33605	212	15	.	.	PUNCT
bracis-33605	213	1	although	although	SCONJ
bracis-33605	213	2	these	these	DET
bracis-33605	213	3	domains	domain	NOUN
bracis-33605	213	4	are	be	AUX
bracis-33605	213	5	simple	simple	ADJ
bracis-33605	213	6	,	,	PUNCT
bracis-33605	213	7	they	they	PRON
bracis-33605	213	8	have	have	VERB
bracis-33605	213	9	characteristics	characteristic	NOUN
bracis-33605	213	10	that	that	PRON
bracis-33605	213	11	make	make	VERB
bracis-33605	213	12	them	they	PRON
bracis-33605	213	13	difficult	difficult	ADJ
bracis-33605	213	14	.	.	PUNCT
bracis-33605	214	1	these	these	DET
bracis-33605	214	2	problems	problem	NOUN
bracis-33605	214	3	have	have	VERB
bracis-33605	214	4	uncertainty	uncertainty	NOUN
bracis-33605	214	5	,	,	PUNCT
bracis-33605	214	6	risk	risk	NOUN
bracis-33605	214	7	and	and	CCONJ
bracis-33605	214	8	complex	complex	ADJ
bracis-33605	214	9	decision	decision	NOUN
bracis-33605	214	10	making	making	NOUN
bracis-33605	214	11	.	.	PUNCT
bracis-33605	215	1	they	they	PRON
bracis-33605	215	2	were	be	AUX
bracis-33605	215	3	modeled	model	VERB
bracis-33605	215	4	as	as	ADP
bracis-33605	215	5	ssps	ssps	NOUN
bracis-33605	215	6	with	with	ADP
bracis-33605	215	7	extended	extended	ADJ
bracis-33605	215	8	states	state	NOUN
bracis-33605	215	9	,	,	PUNCT
bracis-33605	215	10	which	which	PRON
bracis-33605	215	11	include	include	VERB
bracis-33605	215	12	the	the	DET
bracis-33605	215	13	position	position	NOUN
bracis-33605	215	14	and	and	CCONJ
bracis-33605	215	15	accumulated	accumulate	VERB
bracis-33605	215	16	cost	cost	NOUN
bracis-33605	215	17	.	.	PUNCT
bracis-33605	216	1	actions	action	NOUN
bracis-33605	216	2	correspond	correspond	VERB
bracis-33605	216	3	to	to	ADP
bracis-33605	216	4	decisions	decision	NOUN
bracis-33605	216	5	to	to	PART
bracis-33605	216	6	move	move	VERB
bracis-33605	216	7	and	and	CCONJ
bracis-33605	216	8	have	have	AUX
bracis-33605	216	9	cost	cost	VERB
bracis-33605	216	10	1	1	NUM
bracis-33605	216	11	.	.	PUNCT
bracis-33605	217	1	in	in	ADP
bracis-33605	217	2	the	the	DET
bracis-33605	217	3	river	river	NOUN
bracis-33605	217	4	domain	domain	NOUN
bracis-33605	217	5	the	the	DET
bracis-33605	217	6	experiments	experiment	NOUN
bracis-33605	217	7	were	be	AUX
bracis-33605	217	8	carried	carry	VERB
bracis-33605	217	9	out	out	ADP
bracis-33605	217	10	in	in	ADP
bracis-33605	217	11	an	an	DET
bracis-33605	217	12	environment	environment	NOUN
bracis-33605	217	13	of	of	ADP
bracis-33605	217	14	size	size	NOUN
bracis-33605	217	15	\(n_y	\(n_y	NOUN
bracis-33605	217	16	=	=	SYM
bracis-33605	217	17	20\	20\	NUM
bracis-33605	217	18	)	)	PUNCT
bracis-33605	217	19	and	and	CCONJ
bracis-33605	217	20	\(n_x=5\	\(n_x=5\	PROPN
bracis-33605	217	21	)	)	PUNCT
bracis-33605	217	22	,	,	PUNCT
bracis-33605	217	23	the	the	DET
bracis-33605	217	24	initial	initial	ADJ
bracis-33605	217	25	state	state	NOUN
bracis-33605	217	26	is	be	AUX
bracis-33605	217	27	on	on	ADP
bracis-33605	217	28	a	a	DET
bracis-33605	217	29	bank	bank	NOUN
bracis-33605	217	30	of	of	ADP
bracis-33605	217	31	the	the	DET
bracis-33605	217	32	river	river	NOUN
bracis-33605	217	33	\(x_0	\(x_0	PROPN
bracis-33605	217	34	=	=	SYM
bracis-33605	217	35	1	1	NUM
bracis-33605	217	36	,	,	PUNCT
bracis-33605	217	37	y_0	y_0	PROPN
bracis-33605	217	38	=	=	SYM
bracis-33605	217	39	2\	2\	PROPN
bracis-33605	217	40	)	)	PUNCT
bracis-33605	217	41	,	,	PUNCT
bracis-33605	217	42	the	the	DET
bracis-33605	217	43	goal	goal	NOUN
bracis-33605	217	44	on	on	ADP
bracis-33605	217	45	the	the	DET
bracis-33605	217	46	other	other	ADJ
bracis-33605	217	47	bank	bank	NOUN
bracis-33605	217	48	at	at	ADP
bracis-33605	217	49	position	position	NOUN
bracis-33605	217	50	\(x_g	\(x_g	ADJ
bracis-33605	217	51	=	=	SYM
bracis-33605	217	52	n_x	n_x	NUM
bracis-33605	217	53	,	,	PUNCT
bracis-33605	217	54	y_g	y_g	PUNCT
bracis-33605	217	55	=	=	SYM
bracis-33605	217	56	1\	1\	NUM
bracis-33605	217	57	)	)	PUNCT
bracis-33605	217	58	and	and	CCONJ
bracis-33605	217	59	two	two	NUM
bracis-33605	217	60	probability	probability	NOUN
bracis-33605	217	61	values	value	NOUN
bracis-33605	217	62	were	be	AUX
bracis-33605	217	63	used	use	VERB
bracis-33605	217	64	\(p=0.4\	\(p=0.4\	PROPN
bracis-33605	217	65	)	)	PUNCT
bracis-33605	217	66	and	and	CCONJ
bracis-33605	217	67	\(p=0.6\	\(p=0.6\	NUM
bracis-33605	217	68	)	)	PUNCT
bracis-33605	217	69	.	.	PUNCT
bracis-33605	218	1	these	these	PRON
bracis-33605	218	2	are	be	AUX
bracis-33605	218	3	the	the	DET
bracis-33605	218	4	same	same	ADJ
bracis-33605	218	5	values	value	NOUN
bracis-33605	218	6	of	of	ADP
bracis-33605	218	7	p	p	NOUN
bracis-33605	218	8	used	use	VERB
bracis-33605	218	9	by	by	ADP
bracis-33605	218	10	[	[	X
bracis-33605	218	11	4	4	X
bracis-33605	218	12	]	]	PUNCT
bracis-33605	218	13	and	and	CCONJ
bracis-33605	218	14	they	they	PRON
bracis-33605	218	15	show	show	VERB
bracis-33605	218	16	different	different	ADJ
bracis-33605	218	17	preferences	preference	NOUN
bracis-33605	218	18	for	for	ADP
bracis-33605	218	19	proper	proper	ADJ
bracis-33605	218	20	policies	policy	NOUN
bracis-33605	218	21	.	.	PUNCT
bracis-33605	219	1	in	in	ADP
bracis-33605	219	2	the	the	DET
bracis-33605	219	3	navigation	navigation	NOUN
bracis-33605	219	4	domain	domain	NOUN
bracis-33605	219	5	,	,	PUNCT
bracis-33605	219	6	the	the	DET
bracis-33605	219	7	initial	initial	ADJ
bracis-33605	219	8	state	state	NOUN
bracis-33605	219	9	is	be	AUX
bracis-33605	219	10	at	at	ADP
bracis-33605	219	11	\(x_0	\(x_0	PROPN
bracis-33605	219	12	=	=	SYM
bracis-33605	219	13	n_x	n_x	NUM
bracis-33605	219	14	,	,	PUNCT
bracis-33605	219	15	y_0	y_0	PROPN
bracis-33605	219	16	=	=	SYM
bracis-33605	219	17	n_y\	n_y\	NOUN
bracis-33605	219	18	)	)	PUNCT
bracis-33605	219	19	,	,	PUNCT
bracis-33605	219	20	the	the	DET
bracis-33605	219	21	goal	goal	NOUN
bracis-33605	219	22	is	be	AUX
bracis-33605	219	23	at	at	ADP
bracis-33605	219	24	position	position	NOUN
bracis-33605	219	25	\(x_g	\(x_g	ADJ
bracis-33605	219	26	=	=	SYM
bracis-33605	219	27	n_x	n_x	NUM
bracis-33605	219	28	,	,	PUNCT
bracis-33605	219	29	y_g	y_g	PUNCT
bracis-33605	219	30	=	=	SYM
bracis-33605	219	31	1\	1\	NUM
bracis-33605	219	32	)	)	PUNCT
bracis-33605	219	33	,	,	PUNCT
bracis-33605	219	34	and	and	CCONJ
bracis-33605	219	35	two	two	NUM
bracis-33605	219	36	instances	instance	NOUN
bracis-33605	219	37	were	be	AUX
bracis-33605	219	38	used	use	VERB
bracis-33605	219	39	.	.	PUNCT
bracis-33605	220	1	an	an	DET
bracis-33605	220	2	instance	instance	NOUN
bracis-33605	220	3	of	of	ADP
bracis-33605	220	4	size	size	NOUN
bracis-33605	220	5	\(n_x	\(n_x	PROPN
bracis-33605	220	6	=	=	SYM
bracis-33605	220	7	15	15	NUM
bracis-33605	220	8	\times	\time	NOUN
bracis-33605	220	9	n_y	n_y	NOUN
bracis-33605	220	10	=	=	SYM
bracis-33605	220	11	3\	3\	NUM
bracis-33605	220	12	)	)	PUNCT
bracis-33605	220	13	,	,	PUNCT
bracis-33605	220	14	with	with	ADP
bracis-33605	220	15	a	a	DET
bracis-33605	220	16	minimum	minimum	ADJ
bracis-33605	220	17	probability	probability	NOUN
bracis-33605	220	18	of	of	ADP
bracis-33605	220	19	disappearing	disappear	VERB
bracis-33605	220	20	of	of	ADP
bracis-33605	220	21	0.2	0.2	NUM
bracis-33605	220	22	and	and	CCONJ
bracis-33605	220	23	another	another	DET
bracis-33605	220	24	instance	instance	NOUN
bracis-33605	220	25	of	of	ADP
bracis-33605	220	26	size	size	NOUN
bracis-33605	220	27	\(n_x	\(n_x	PROPN
bracis-33605	220	28	=	=	SYM
bracis-33605	220	29	10\	10\	NUM
bracis-33605	220	30	)	)	PUNCT
bracis-33605	220	31	\(\times	\(\time	NOUN
bracis-33605	220	32	\	\	NOUN
bracis-33605	220	33	)	)	PUNCT
bracis-33605	220	34	\(n_y	\(n_y	NOUN
bracis-33605	220	35	=	=	SYM
bracis-33605	220	36	4\	4\	PROPN
bracis-33605	220	37	)	)	PUNCT
bracis-33605	220	38	with	with	ADP
bracis-33605	220	39	a	a	DET
bracis-33605	220	40	minimum	minimum	ADJ
bracis-33605	220	41	probability	probability	NOUN
bracis-33605	220	42	of	of	ADP
bracis-33605	220	43	disappearing	disappear	VERB
bracis-33605	220	44	of	of	ADP
bracis-33605	220	45	0.1	0.1	NUM
bracis-33605	220	46	.	.	PUNCT
bracis-33605	221	1	table	table	NOUN
bracis-33605	221	2	 	 	SPACE
bracis-33605	221	3	1	1	NUM
bracis-33605	221	4	shows	show	VERB
bracis-33605	221	5	the	the	DET
bracis-33605	221	6	parameters	parameter	NOUN
bracis-33605	221	7	used	use	VERB
bracis-33605	221	8	by	by	ADP
bracis-33605	221	9	the	the	DET
bracis-33605	221	10	algorithms	algorithm	NOUN
bracis-33605	221	11	.	.	PUNCT
bracis-33605	222	1	a	a	DET
bracis-33605	222	2	function	function	NOUN
bracis-33605	222	3	with	with	ADP
bracis-33605	222	4	linear	linear	ADJ
bracis-33605	222	5	decay	decay	NOUN
bracis-33605	222	6	was	be	AUX
bracis-33605	222	7	used	use	VERB
bracis-33605	222	8	to	to	PART
bracis-33605	222	9	vary	vary	VERB
bracis-33605	222	10	\(\alpha	\(\alpha	X
bracis-33605	222	11	\	\	NOUN
bracis-33605	222	12	)	)	PUNCT
bracis-33605	222	13	and	and	CCONJ
bracis-33605	222	14	\(\epsilon	\(\epsilon	NOUN
bracis-33605	222	15	\	\	NOUN
bracis-33605	222	16	)	)	PUNCT
bracis-33605	222	17	from	from	ADP
bracis-33605	222	18	their	their	PRON
bracis-33605	222	19	respective	respective	ADJ
bracis-33605	222	20	maximum	maximum	ADJ
bracis-33605	222	21	values	value	NOUN
bracis-33605	222	22	to	to	ADP
bracis-33605	222	23	the	the	DET
bracis-33605	222	24	minimum	minimum	ADJ
bracis-33605	222	25	values	value	NOUN
bracis-33605	222	26	of	of	ADP
bracis-33605	222	27	both	both	DET
bracis-33605	222	28	parameters	parameter	NOUN
bracis-33605	222	29	.	.	PUNCT
bracis-33605	223	1	the	the	DET
bracis-33605	223	2	initial	initial	ADJ
bracis-33605	223	3	extended	extended	ADJ
bracis-33605	223	4	state	state	NOUN
bracis-33605	223	5	was	be	AUX
bracis-33605	223	6	chosen	choose	VERB
bracis-33605	223	7	randomly	randomly	ADV
bracis-33605	223	8	for	for	ADP
bracis-33605	223	9	each	each	DET
bracis-33605	223	10	episode	episode	NOUN
bracis-33605	223	11	.	.	PUNCT
bracis-33605	224	1	table	table	NOUN
bracis-33605	224	2	 	 	SPACE
bracis-33605	224	3	2	2	NUM
bracis-33605	224	4	shows	show	VERB
bracis-33605	224	5	the	the	DET
bracis-33605	224	6	parameters	parameter	NOUN
bracis-33605	224	7	used	use	VERB
bracis-33605	224	8	for	for	ADP
bracis-33605	224	9	the	the	DET
bracis-33605	224	10	river	river	NOUN
bracis-33605	224	11	and	and	CCONJ
bracis-33605	224	12	navigation	navigation	NOUN
bracis-33605	224	13	problems	problem	NOUN
bracis-33605	224	14	.	.	PUNCT
bracis-33605	225	1	table	table	NOUN
bracis-33605	225	2	1	1	NUM
bracis-33605	225	3	.	.	PUNCT
bracis-33605	226	1	parameters	parameter	NOUN
bracis-33605	226	2	for	for	ADP
bracis-33605	226	3	the	the	DET
bracis-33605	226	4	algorithms.full	algorithms.full	NOUN
bracis-33605	226	5	size	size	NOUN
bracis-33605	226	6	table	table	NOUN
bracis-33605	226	7	table	table	NOUN
bracis-33605	226	8	2	2	NUM
bracis-33605	226	9	.	.	PUNCT
bracis-33605	226	10	parameters	parameter	NOUN
bracis-33605	226	11	for	for	ADP
bracis-33605	226	12	the	the	DET
bracis-33605	226	13	river	river	NOUN
bracis-33605	226	14	and	and	CCONJ
bracis-33605	226	15	navigation	navigation	NOUN
bracis-33605	226	16	problems.full	problems.full	ADJ
bracis-33605	226	17	size	size	NOUN
bracis-33605	226	18	table	table	NOUN
bracis-33605	226	19	in	in	ADP
bracis-33605	226	20	sects	sect	NOUN
bracis-33605	226	21	.	.	PUNCT
bracis-33605	226	22	 	 	SPACE
bracis-33605	226	23	4.1	4.1	NUM
bracis-33605	226	24	and	and	CCONJ
bracis-33605	226	25	4.2	4.2	NUM
bracis-33605	226	26	,	,	PUNCT
bracis-33605	226	27	the	the	DET
bracis-33605	226	28	proposed	propose	VERB
bracis-33605	226	29	algorithms	algorithm	NOUN
bracis-33605	226	30	are	be	AUX
bracis-33605	226	31	compared	compare	VERB
bracis-33605	226	32	with	with	ADP
bracis-33605	226	33	the	the	DET
bracis-33605	226	34	gubs	gubs	ADJ
bracis-33605	226	35	-	-	PUNCT
bracis-33605	226	36	vi	vi	NOUN
bracis-33605	226	37	algorithm	algorithm	NOUN
bracis-33605	226	38	.	.	PUNCT
bracis-33605	227	1	this	this	DET
bracis-33605	227	2	choice	choice	NOUN
bracis-33605	227	3	is	be	AUX
bracis-33605	227	4	because	because	SCONJ
bracis-33605	227	5	the	the	DET
bracis-33605	227	6	proposed	propose	VERB
bracis-33605	227	7	algorithms	algorithm	NOUN
bracis-33605	227	8	are	be	AUX
bracis-33605	227	9	based	base	VERB
bracis-33605	227	10	on	on	ADP
bracis-33605	227	11	gubs	gub	NOUN
bracis-33605	227	12	-	-	PUNCT
bracis-33605	227	13	vi	vi	NOUN
bracis-33605	227	14	.	.	PUNCT
bracis-33605	228	1	the	the	DET
bracis-33605	228	2	purpose	purpose	NOUN
bracis-33605	228	3	of	of	ADP
bracis-33605	228	4	this	this	DET
bracis-33605	228	5	comparison	comparison	NOUN
bracis-33605	228	6	is	be	AUX
bracis-33605	228	7	to	to	PART
bracis-33605	228	8	verify	verify	VERB
bracis-33605	228	9	if	if	SCONJ
bracis-33605	228	10	the	the	DET
bracis-33605	228	11	proposed	propose	VERB
bracis-33605	228	12	algorithms	algorithm	NOUN
bracis-33605	228	13	found	find	VERB
bracis-33605	228	14	the	the	DET
bracis-33605	228	15	same	same	ADJ
bracis-33605	228	16	results	result	NOUN
bracis-33605	228	17	obtained	obtain	VERB
bracis-33605	228	18	by	by	ADP
bracis-33605	228	19	gubs	gub	NOUN
bracis-33605	228	20	-	-	PUNCT
bracis-33605	228	21	vi	vi	NOUN
bracis-33605	228	22	.	.	PUNCT
bracis-33605	228	23	theoretically	theoretically	ADV
bracis-33605	228	24	,	,	PUNCT
bracis-33605	228	25	the	the	DET
bracis-33605	228	26	proposed	propose	VERB
bracis-33605	228	27	algorithms	algorithm	NOUN
bracis-33605	228	28	should	should	AUX
bracis-33605	228	29	obtain	obtain	VERB
bracis-33605	228	30	the	the	DET
bracis-33605	228	31	same	same	ADJ
bracis-33605	228	32	result	result	NOUN
bracis-33605	228	33	,	,	PUNCT
bracis-33605	228	34	however	however	ADV
bracis-33605	228	35	,	,	PUNCT
bracis-33605	228	36	since	since	SCONJ
bracis-33605	228	37	they	they	PRON
bracis-33605	228	38	use	use	VERB
bracis-33605	228	39	sampling	sample	VERB
bracis-33605	228	40	,	,	PUNCT
bracis-33605	228	41	they	they	PRON
bracis-33605	228	42	could	could	AUX
bracis-33605	228	43	obtain	obtain	VERB
bracis-33605	228	44	a	a	DET
bracis-33605	228	45	very	very	ADV
bracis-33605	228	46	close	close	ADJ
bracis-33605	228	47	solution	solution	NOUN
bracis-33605	228	48	within	within	ADP
bracis-33605	228	49	the	the	DET
bracis-33605	228	50	number	number	NOUN
bracis-33605	228	51	of	of	ADP
bracis-33605	228	52	episodes	episode	NOUN
bracis-33605	228	53	used	use	VERB
bracis-33605	228	54	.	.	PUNCT
bracis-33605	229	1	in	in	ADP
bracis-33605	229	2	the	the	DET
bracis-33605	229	3	figures	figure	NOUN
bracis-33605	229	4	shown	show	VERB
bracis-33605	229	5	in	in	ADP
bracis-33605	229	6	sects	sect	NOUN
bracis-33605	229	7	.	.	PUNCT
bracis-33605	230	1	4.1	4.1	NUM
bracis-33605	230	2	and	and	CCONJ
bracis-33605	230	3	4.2	4.2	NUM
bracis-33605	230	4	,	,	PUNCT
bracis-33605	230	5	the	the	DET
bracis-33605	230	6	x	x	VERB
bracis-33605	230	7	axis	axis	NOUN
bracis-33605	230	8	has	have	VERB
bracis-33605	230	9	the	the	DET
bracis-33605	230	10	values	value	NOUN
bracis-33605	230	11	of	of	ADP
bracis-33605	230	12	\(k_g\	\(k_g\	NOUN
bracis-33605	230	13	)	)	PUNCT
bracis-33605	230	14	expressed	express	VERB
bracis-33605	230	15	in	in	ADP
bracis-33605	230	16	logarithm	logarithm	NOUN
bracis-33605	230	17	base	base	NOUN
bracis-33605	230	18	2	2	NUM
bracis-33605	230	19	and	and	CCONJ
bracis-33605	230	20	the	the	DET
bracis-33605	230	21	y	y	PROPN
bracis-33605	230	22	axis	axis	NOUN
bracis-33605	230	23	shows	show	VERB
bracis-33605	230	24	the	the	DET
bracis-33605	230	25	probability	probability	NOUN
bracis-33605	230	26	and	and	CCONJ
bracis-33605	230	27	cost	cost	NOUN
bracis-33605	230	28	of	of	ADP
bracis-33605	230	29	reaching	reach	VERB
bracis-33605	230	30	the	the	DET
bracis-33605	230	31	goal	goal	NOUN
bracis-33605	230	32	from	from	ADP
bracis-33605	230	33	the	the	DET
bracis-33605	230	34	initial	initial	ADJ
bracis-33605	230	35	state	state	NOUN
bracis-33605	230	36	.	.	PUNCT
bracis-33605	231	1	each	each	DET
bracis-33605	231	2	point	point	NOUN
bracis-33605	231	3	in	in	ADP
bracis-33605	231	4	the	the	DET
bracis-33605	231	5	figures	figure	NOUN
bracis-33605	231	6	is	be	AUX
bracis-33605	231	7	the	the	DET
bracis-33605	231	8	average	average	NOUN
bracis-33605	231	9	of	of	ADP
bracis-33605	231	10	5	5	NUM
bracis-33605	231	11	executions	execution	NOUN
bracis-33605	231	12	of	of	ADP
bracis-33605	231	13	algorithms	algorithm	NOUN
bracis-33605	231	14	.	.	PUNCT
bracis-33605	232	1	additionally	additionally	ADV
bracis-33605	232	2	,	,	PUNCT
bracis-33605	232	3	the	the	DET
bracis-33605	232	4	standard	standard	ADJ
bracis-33605	232	5	deviation	deviation	NOUN
bracis-33605	232	6	is	be	AUX
bracis-33605	232	7	shown	show	VERB
bracis-33605	232	8	.	.	PUNCT
bracis-33605	233	1	we	we	PRON
bracis-33605	233	2	also	also	ADV
bracis-33605	233	3	record	record	VERB
bracis-33605	233	4	the	the	DET
bracis-33605	233	5	learning	learning	NOUN
bracis-33605	233	6	curves	curve	NOUN
bracis-33605	233	7	of	of	ADP
bracis-33605	233	8	these	these	DET
bracis-33605	233	9	experiments	experiment	NOUN
bracis-33605	233	10	.	.	PUNCT
bracis-33605	234	1	these	these	DET
bracis-33605	234	2	results	result	NOUN
bracis-33605	234	3	show	show	VERB
bracis-33605	234	4	that	that	SCONJ
bracis-33605	234	5	the	the	DET
bracis-33605	234	6	algorithm	algorithm	NOUN
bracis-33605	234	7	that	that	PRON
bracis-33605	234	8	converges	converge	VERB
bracis-33605	234	9	the	the	DET
bracis-33605	234	10	fastest	fast	ADJ
bracis-33605	234	11	is	be	AUX
bracis-33605	234	12	q	q	ADJ
bracis-33605	234	13	-	-	PUNCT
bracis-33605	234	14	learning	learning	NOUN
bracis-33605	234	15	-	-	PUNCT
bracis-33605	234	16	gubs	gub	NOUN
bracis-33605	234	17	algorithm	algorithm	NOUN
bracis-33605	234	18	.	.	PUNCT
bracis-33605	235	1	we	we	PRON
bracis-33605	235	2	do	do	AUX
bracis-33605	235	3	not	not	PART
bracis-33605	235	4	include	include	VERB
bracis-33605	235	5	these	these	DET
bracis-33605	235	6	curves	curve	NOUN
bracis-33605	235	7	for	for	ADP
bracis-33605	235	8	space	space	NOUN
bracis-33605	235	9	reasons	reason	NOUN
bracis-33605	235	10	.	.	PUNCT
bracis-33605	236	1	in	in	ADP
bracis-33605	236	2	sect	sect	NOUN
bracis-33605	236	3	.	.	PUNCT
bracis-33605	236	4	 	 	SPACE
bracis-33605	236	5	4.3	4.3	NUM
bracis-33605	236	6	,	,	PUNCT
bracis-33605	236	7	we	we	PRON
bracis-33605	236	8	compare	compare	VERB
bracis-33605	236	9	the	the	DET
bracis-33605	236	10	proposed	propose	VERB
bracis-33605	236	11	algorithms	algorithm	NOUN
bracis-33605	236	12	with	with	ADP
bracis-33605	236	13	the	the	DET
bracis-33605	236	14	geibel	geibel	NOUN
bracis-33605	236	15	and	and	CCONJ
bracis-33605	236	16	wysotzki	wysotzki	ADJ
bracis-33605	236	17	criterion	criterion	NOUN
bracis-33605	236	18	[	[	X
bracis-33605	236	19	5	5	NUM
bracis-33605	236	20	]	]	PUNCT
bracis-33605	236	21	.	.	PUNCT
bracis-33605	237	1	notice	notice	VERB
bracis-33605	237	2	that	that	SCONJ
bracis-33605	237	3	,	,	PUNCT
bracis-33605	237	4	the	the	DET
bracis-33605	237	5	gubs	gubs	ADJ
bracis-33605	237	6	criterion	criterion	NOUN
bracis-33605	237	7	and	and	CCONJ
bracis-33605	237	8	other	other	ADJ
bracis-33605	237	9	criteria	criterion	NOUN
bracis-33605	237	10	(	(	PUNCT
bracis-33605	237	11	fsspude	fsspude	NOUN
bracis-33605	237	12	and	and	CCONJ
bracis-33605	237	13	discount	discount	NOUN
bracis-33605	237	14	cost	cost	NOUN
bracis-33605	237	15	)	)	PUNCT
bracis-33605	237	16	were	be	AUX
bracis-33605	237	17	compared	compare	VERB
bracis-33605	237	18	by	by	ADP
bracis-33605	237	19	[	[	X
bracis-33605	237	20	1	1	NUM
bracis-33605	237	21	]	]	PUNCT
bracis-33605	237	22	.	.	PUNCT
bracis-33605	238	1	4.1	4.1	NUM
bracis-33605	238	2	results	result	NOUN
bracis-33605	238	3	of	of	ADP
bracis-33605	238	4	 	 	SPACE
bracis-33605	238	5	the	the	DET
bracis-33605	238	6	 	 	SPACE
bracis-33605	238	7	river	river	NOUN
bracis-33605	238	8	domain	domain	NOUN
bracis-33605	238	9	figures	figure	NOUN
bracis-33605	238	10	 	 	SPACE
bracis-33605	238	11	1a	1a	NOUN
bracis-33605	238	12	)	)	PUNCT
bracis-33605	238	13	and	and	CCONJ
bracis-33605	238	14	1b	1b	NUM
bracis-33605	238	15	)	)	PUNCT
bracis-33605	238	16	show	show	VERB
bracis-33605	238	17	the	the	DET
bracis-33605	238	18	probability	probability	NOUN
bracis-33605	238	19	to	to	ADP
bracis-33605	238	20	goal	goal	NOUN
bracis-33605	238	21	obtained	obtain	VERB
bracis-33605	238	22	by	by	ADP
bracis-33605	238	23	the	the	DET
bracis-33605	238	24	gubs	gubs	ADJ
bracis-33605	238	25	-	-	PUNCT
bracis-33605	238	26	vi	vi	NOUN
bracis-33605	238	27	algorithm	algorithm	NOUN
bracis-33605	238	28	and	and	CCONJ
bracis-33605	238	29	the	the	DET
bracis-33605	238	30	proposed	propose	VERB
bracis-33605	238	31	algorithms	algorithm	NOUN
bracis-33605	238	32	q	q	NOUN
bracis-33605	238	33	-	-	PUNCT
bracis-33605	238	34	learning	learning	NOUN
bracis-33605	238	35	-	-	PUNCT
bracis-33605	238	36	gubs	gub	NOUN
bracis-33605	238	37	and	and	CCONJ
bracis-33605	238	38	q	q	NOUN
bracis-33605	238	39	-	-	PUNCT
bracis-33605	238	40	learning	learning	NOUN
bracis-33605	238	41	-	-	PUNCT
bracis-33605	238	42	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	238	43	)	)	PUNCT
bracis-33605	238	44	in	in	ADP
bracis-33605	238	45	the	the	DET
bracis-33605	238	46	river	river	NOUN
bracis-33605	238	47	problems	problem	NOUN
bracis-33605	238	48	with	with	ADP
bracis-33605	238	49	\(p=0.4\	\(p=0.4\	PROPN
bracis-33605	238	50	)	)	PUNCT
bracis-33605	238	51	and	and	CCONJ
bracis-33605	238	52	\(p=0.6\	\(p=0.6\	NUM
bracis-33605	238	53	)	)	PUNCT
bracis-33605	238	54	,	,	PUNCT
bracis-33605	238	55	respectively	respectively	ADV
bracis-33605	238	56	.	.	PUNCT
bracis-33605	239	1	the	the	DET
bracis-33605	239	2	probabilities	probability	NOUN
bracis-33605	239	3	obtained	obtain	VERB
bracis-33605	239	4	by	by	ADP
bracis-33605	239	5	the	the	DET
bracis-33605	239	6	algorithms	algorithm	NOUN
bracis-33605	239	7	are	be	AUX
bracis-33605	239	8	very	very	ADV
bracis-33605	239	9	close	close	ADJ
bracis-33605	239	10	to	to	ADP
bracis-33605	239	11	the	the	DET
bracis-33605	239	12	results	result	NOUN
bracis-33605	239	13	obtained	obtain	VERB
bracis-33605	239	14	by	by	ADP
bracis-33605	239	15	the	the	DET
bracis-33605	239	16	algorithm	algorithm	NOUN
bracis-33605	239	17	gubs	gub	NOUN
bracis-33605	239	18	-	-	PUNCT
bracis-33605	239	19	vi	vi	NOUN
bracis-33605	239	20	for	for	ADP
bracis-33605	239	21	both	both	DET
bracis-33605	239	22	instances	instance	NOUN
bracis-33605	239	23	of	of	ADP
bracis-33605	239	24	the	the	DET
bracis-33605	239	25	domain	domain	NOUN
bracis-33605	239	26	.	.	PUNCT
bracis-33605	240	1	the	the	DET
bracis-33605	240	2	proper	proper	ADJ
bracis-33605	240	3	policy	policy	NOUN
bracis-33605	240	4	for	for	ADP
bracis-33605	240	5	the	the	DET
bracis-33605	240	6	\(p=0.4\	\(p=0.4\	PROPN
bracis-33605	240	7	)	)	PUNCT
bracis-33605	240	8	instance	instance	NOUN
bracis-33605	240	9	is	be	AUX
bracis-33605	240	10	reached	reach	VERB
bracis-33605	240	11	when	when	SCONJ
bracis-33605	240	12	\(k_g	\(k_g	PROPN
bracis-33605	240	13	>	>	X
bracis-33605	240	14	16\	16\	NUM
bracis-33605	240	15	)	)	PUNCT
bracis-33605	240	16	(	(	PUNCT
bracis-33605	240	17	\(\log	\(\log	NOUN
bracis-33605	240	18	(	(	PUNCT
bracis-33605	240	19	k_g)=4\	k_g)=4\	PROPN
bracis-33605	240	20	)	)	PUNCT
bracis-33605	240	21	)	)	PUNCT
bracis-33605	240	22	,	,	PUNCT
bracis-33605	240	23	and	and	CCONJ
bracis-33605	240	24	the	the	DET
bracis-33605	240	25	proposed	propose	VERB
bracis-33605	240	26	algorithms	algorithm	NOUN
bracis-33605	240	27	find	find	VERB
bracis-33605	240	28	policies	policy	NOUN
bracis-33605	240	29	with	with	ADP
bracis-33605	240	30	probabilities	probability	NOUN
bracis-33605	240	31	increasingly	increasingly	ADV
bracis-33605	240	32	closer	close	ADJ
bracis-33605	240	33	to	to	ADP
bracis-33605	240	34	1	1	NUM
bracis-33605	240	35	.	.	PUNCT
bracis-33605	241	1	for	for	ADP
bracis-33605	241	2	the	the	DET
bracis-33605	241	3	instance	instance	NOUN
bracis-33605	241	4	with	with	ADP
bracis-33605	241	5	\(p=0.6\	\(p=0.6\	NUM
bracis-33605	241	6	)	)	PUNCT
bracis-33605	241	7	,	,	PUNCT
bracis-33605	241	8	the	the	DET
bracis-33605	241	9	optimal	optimal	ADJ
bracis-33605	241	10	policy	policy	NOUN
bracis-33605	241	11	when	when	SCONJ
bracis-33605	241	12	\(k_g	\(k_g	PROPN
bracis-33605	241	13	\ge	\ge	PROPN
bracis-33605	241	14	0.5\	0.5\	NUM
bracis-33605	241	15	)	)	PUNCT
bracis-33605	241	16	(	(	PUNCT
bracis-33605	241	17	\(\log	\(\log	X
bracis-33605	241	18	(	(	PUNCT
bracis-33605	241	19	k_g)=-1\	k_g)=-1\	NOUN
bracis-33605	241	20	)	)	PUNCT
bracis-33605	241	21	)	)	PUNCT
bracis-33605	241	22	is	be	AUX
bracis-33605	241	23	a	a	DET
bracis-33605	241	24	proper	proper	ADJ
bracis-33605	241	25	policy	policy	NOUN
bracis-33605	241	26	(	(	PUNCT
bracis-33605	241	27	i.e.	i.e.	X
bracis-33605	241	28	crossing	cross	VERB
bracis-33605	241	29	the	the	DET
bracis-33605	241	30	bridge	bridge	NOUN
bracis-33605	241	31	)	)	PUNCT
bracis-33605	241	32	.	.	PUNCT
bracis-33605	242	1	in	in	ADP
bracis-33605	242	2	this	this	DET
bracis-33605	242	3	case	case	NOUN
bracis-33605	242	4	,	,	PUNCT
bracis-33605	242	5	the	the	DET
bracis-33605	242	6	proposed	propose	VERB
bracis-33605	242	7	algorithms	algorithm	NOUN
bracis-33605	242	8	find	find	VERB
bracis-33605	242	9	the	the	DET
bracis-33605	242	10	proper	proper	ADJ
bracis-33605	242	11	policy	policy	NOUN
bracis-33605	242	12	.	.	PUNCT
bracis-33605	243	1	figures	figure	NOUN
bracis-33605	243	2	 	 	SPACE
bracis-33605	243	3	1c	1c	NOUN
bracis-33605	243	4	)	)	PUNCT
bracis-33605	243	5	and	and	CCONJ
bracis-33605	243	6	1d	1d	NUM
bracis-33605	243	7	)	)	PUNCT
bracis-33605	243	8	compare	compare	VERB
bracis-33605	243	9	the	the	DET
bracis-33605	243	10	costs	cost	NOUN
bracis-33605	243	11	reached	reach	VERB
bracis-33605	243	12	from	from	ADP
bracis-33605	243	13	the	the	DET
bracis-33605	243	14	initial	initial	ADJ
bracis-33605	243	15	state	state	NOUN
bracis-33605	243	16	by	by	ADP
bracis-33605	243	17	the	the	DET
bracis-33605	243	18	three	three	NUM
bracis-33605	243	19	algorithms	algorithm	NOUN
bracis-33605	243	20	for	for	ADP
bracis-33605	243	21	the	the	DET
bracis-33605	243	22	same	same	ADJ
bracis-33605	243	23	instances	instance	NOUN
bracis-33605	243	24	of	of	ADP
bracis-33605	243	25	the	the	DET
bracis-33605	243	26	river	river	NOUN
bracis-33605	243	27	domain	domain	NOUN
bracis-33605	243	28	.	.	PUNCT
bracis-33605	244	1	the	the	DET
bracis-33605	244	2	results	result	NOUN
bracis-33605	244	3	are	be	AUX
bracis-33605	244	4	very	very	ADV
bracis-33605	244	5	close	close	ADJ
bracis-33605	244	6	between	between	ADP
bracis-33605	244	7	the	the	DET
bracis-33605	244	8	algorithms	algorithm	NOUN
bracis-33605	244	9	for	for	ADP
bracis-33605	244	10	both	both	DET
bracis-33605	244	11	instances	instance	NOUN
bracis-33605	244	12	.	.	PUNCT
bracis-33605	245	1	fig	fig	NOUN
bracis-33605	245	2	.	.	PUNCT
bracis-33605	246	1	1	1	NUM
bracis-33605	246	2	.	.	X
bracis-33605	246	3	probability	probability	NOUN
bracis-33605	246	4	to	to	ADP
bracis-33605	246	5	goal	goal	NOUN
bracis-33605	246	6	and	and	CCONJ
bracis-33605	246	7	cost	cost	NOUN
bracis-33605	246	8	to	to	ADP
bracis-33605	246	9	goal	goal	NOUN
bracis-33605	246	10	found	find	VERB
bracis-33605	246	11	by	by	ADP
bracis-33605	246	12	q	q	NOUN
bracis-33605	246	13	-	-	PUNCT
bracis-33605	246	14	learning	learning	NOUN
bracis-33605	246	15	-	-	PUNCT
bracis-33605	246	16	gubs	gub	NOUN
bracis-33605	246	17	,	,	PUNCT
bracis-33605	246	18	q	q	ADJ
bracis-33605	246	19	-	-	PUNCT
bracis-33605	246	20	learning	learning	NOUN
bracis-33605	246	21	-	-	PUNCT
bracis-33605	246	22	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	246	23	)	)	PUNCT
bracis-33605	246	24	and	and	CCONJ
bracis-33605	246	25	gubs	gubs	ADJ
bracis-33605	246	26	-	-	PUNCT
bracis-33605	246	27	vi	vi	NOUN
bracis-33605	246	28	algorithms	algorithms	NOUN
bracis-33605	246	29	for	for	ADP
bracis-33605	246	30	different	different	ADJ
bracis-33605	246	31	values	value	NOUN
bracis-33605	246	32	of	of	ADP
bracis-33605	246	33	\(k_g\	\(k_g\	NOUN
bracis-33605	246	34	)	)	PUNCT
bracis-33605	246	35	for	for	ADP
bracis-33605	246	36	instance	instance	NOUN
bracis-33605	246	37	\(n_y	\(n_y	NOUN
bracis-33605	246	38	=	=	NOUN
bracis-33605	246	39	20	20	NUM
bracis-33605	246	40	\times	\time	NOUN
bracis-33605	246	41	n_x=5\	n_x=5\	NOUN
bracis-33605	246	42	)	)	PUNCT
bracis-33605	246	43	of	of	ADP
bracis-33605	246	44	the	the	DET
bracis-33605	246	45	river	river	NOUN
bracis-33605	246	46	domain	domain	NOUN
bracis-33605	246	47	.	.	PUNCT
bracis-33605	247	1	full	full	ADJ
bracis-33605	247	2	size	size	NOUN
bracis-33605	247	3	image	image	NOUN
bracis-33605	247	4	4.2	4.2	NUM
bracis-33605	247	5	results	result	NOUN
bracis-33605	247	6	of	of	ADP
bracis-33605	247	7	 	 	SPACE
bracis-33605	247	8	the	the	DET
bracis-33605	247	9	 	 	SPACE
bracis-33605	247	10	navigation	navigation	NOUN
bracis-33605	247	11	domain	domain	NOUN
bracis-33605	247	12	figures	figure	NOUN
bracis-33605	247	13	 	 	SPACE
bracis-33605	247	14	2a	2a	NUM
bracis-33605	247	15	)	)	PUNCT
bracis-33605	247	16	and	and	CCONJ
bracis-33605	247	17	2b	2b	NOUN
bracis-33605	247	18	)	)	PUNCT
bracis-33605	247	19	show	show	VERB
bracis-33605	247	20	the	the	DET
bracis-33605	247	21	probabilities	probability	NOUN
bracis-33605	247	22	to	to	ADP
bracis-33605	247	23	goal	goal	NOUN
bracis-33605	247	24	obtained	obtain	VERB
bracis-33605	247	25	by	by	ADP
bracis-33605	247	26	the	the	DET
bracis-33605	247	27	three	three	NUM
bracis-33605	247	28	algorithms	algorithm	NOUN
bracis-33605	247	29	in	in	ADP
bracis-33605	247	30	the	the	DET
bracis-33605	247	31	two	two	NUM
bracis-33605	247	32	described	describe	VERB
bracis-33605	247	33	instances	instance	NOUN
bracis-33605	247	34	of	of	ADP
bracis-33605	247	35	the	the	DET
bracis-33605	247	36	navigation	navigation	NOUN
bracis-33605	247	37	domain	domain	NOUN
bracis-33605	247	38	.	.	PUNCT
bracis-33605	248	1	the	the	DET
bracis-33605	248	2	probabilities	probability	NOUN
bracis-33605	248	3	found	find	VERB
bracis-33605	248	4	by	by	ADP
bracis-33605	248	5	the	the	DET
bracis-33605	248	6	proposed	propose	VERB
bracis-33605	248	7	algorithms	algorithm	NOUN
bracis-33605	248	8	are	be	AUX
bracis-33605	248	9	very	very	ADV
bracis-33605	248	10	close	close	ADJ
bracis-33605	248	11	to	to	ADP
bracis-33605	248	12	the	the	DET
bracis-33605	248	13	probabilities	probability	NOUN
bracis-33605	248	14	found	find	VERB
bracis-33605	248	15	by	by	ADP
bracis-33605	248	16	the	the	DET
bracis-33605	248	17	gubs	gubs	ADJ
bracis-33605	248	18	-	-	PUNCT
bracis-33605	248	19	vi	vi	NOUN
bracis-33605	248	20	algorithm	algorithm	NOUN
bracis-33605	248	21	for	for	ADP
bracis-33605	248	22	both	both	DET
bracis-33605	248	23	instances	instance	NOUN
bracis-33605	248	24	of	of	ADP
bracis-33605	248	25	the	the	DET
bracis-33605	248	26	problem	problem	NOUN
bracis-33605	248	27	.	.	PUNCT
bracis-33605	249	1	however	however	ADV
bracis-33605	249	2	,	,	PUNCT
bracis-33605	249	3	there	there	PRON
bracis-33605	249	4	is	be	VERB
bracis-33605	249	5	a	a	DET
bracis-33605	249	6	larger	large	ADJ
bracis-33605	249	7	standard	standard	ADJ
bracis-33605	249	8	deviation	deviation	NOUN
bracis-33605	249	9	for	for	ADP
bracis-33605	249	10	values	value	NOUN
bracis-33605	249	11	of	of	ADP
bracis-33605	249	12	\(k_g\	\(k_g\	NOUN
bracis-33605	249	13	)	)	PUNCT
bracis-33605	249	14	that	that	PRON
bracis-33605	249	15	have	have	VERB
bracis-33605	249	16	policies	policy	NOUN
bracis-33605	249	17	close	close	ADJ
bracis-33605	249	18	to	to	ADP
bracis-33605	249	19	maxprob	maxprob	NOUN
bracis-33605	249	20	and	and	CCONJ
bracis-33605	249	21	correspond	correspond	VERB
bracis-33605	249	22	to	to	ADP
bracis-33605	249	23	the	the	DET
bracis-33605	249	24	furthest	furth	ADJ
bracis-33605	249	25	route	route	NOUN
bracis-33605	249	26	from	from	ADP
bracis-33605	249	27	the	the	DET
bracis-33605	249	28	goal	goal	NOUN
bracis-33605	249	29	(	(	PUNCT
bracis-33605	249	30	the	the	DET
bracis-33605	249	31	probability	probability	NOUN
bracis-33605	249	32	for	for	ADP
bracis-33605	249	33	the	the	DET
bracis-33605	249	34	goal	goal	NOUN
bracis-33605	249	35	is	be	AUX
bracis-33605	249	36	0.8	0.8	NUM
bracis-33605	249	37	for	for	ADP
bracis-33605	249	38	the	the	DET
bracis-33605	249	39	instance	instance	NOUN
bracis-33605	249	40	\(n_x	\(n_x	PROPN
bracis-33605	249	41	=	=	SYM
bracis-33605	249	42	15	15	NUM
bracis-33605	249	43	\times	\time	NOUN
bracis-33605	249	44	n_y	n_y	NOUN
bracis-33605	249	45	=	=	SYM
bracis-33605	249	46	3\	3\	NUM
bracis-33605	249	47	)	)	PUNCT
bracis-33605	249	48	,	,	PUNCT
bracis-33605	249	49	and	and	CCONJ
bracis-33605	249	50	0.81	0.81	NUM
bracis-33605	249	51	for	for	ADP
bracis-33605	249	52	the	the	DET
bracis-33605	249	53	instance	instance	NOUN
bracis-33605	249	54	\(n_x	\(n_x	PROPN
bracis-33605	249	55	=	=	NUM
bracis-33605	249	56	10	10	NUM
bracis-33605	249	57	\times	\time	NOUN
bracis-33605	249	58	n_y	n_y	ADP
bracis-33605	249	59	=	=	SYM
bracis-33605	249	60	4\	4\	PROPN
bracis-33605	249	61	)	)	PUNCT
bracis-33605	249	62	)	)	PUNCT
bracis-33605	249	63	.	.	PUNCT
bracis-33605	250	1	the	the	DET
bracis-33605	250	2	algorithms	algorithm	NOUN
bracis-33605	250	3	find	find	VERB
bracis-33605	250	4	a	a	DET
bracis-33605	250	5	policy	policy	NOUN
bracis-33605	250	6	according	accord	VERB
bracis-33605	250	7	to	to	ADP
bracis-33605	250	8	\(k_g\	\(k_g\	NUM
bracis-33605	250	9	)	)	PUNCT
bracis-33605	250	10	.	.	PUNCT
bracis-33605	251	1	for	for	ADP
bracis-33605	251	2	values	value	NOUN
bracis-33605	251	3	\(k_g	\(k_g	PROPN
bracis-33605	251	4	\ge	\ge	PROPN
bracis-33605	251	5	0.3\	0.3\	NOUN
bracis-33605	251	6	)	)	PUNCT
bracis-33605	251	7	(	(	PUNCT
bracis-33605	251	8	\(\log	\(\log	X
bracis-33605	251	9	(	(	PUNCT
bracis-33605	251	10	k_g	k_g	PROPN
bracis-33605	251	11	)	)	PUNCT
bracis-33605	251	12	=	=	PUNCT
bracis-33605	251	13	-1.7\	-1.7\	ADJ
bracis-33605	251	14	)	)	PUNCT
bracis-33605	251	15	)	)	PUNCT
bracis-33605	252	1	in	in	ADP
bracis-33605	252	2	instance	instance	NOUN
bracis-33605	252	3	\(n_x	\(n_x	PROPN
bracis-33605	252	4	=	=	SYM
bracis-33605	252	5	15	15	NUM
bracis-33605	252	6	\times	\time	NOUN
bracis-33605	252	7	n_y	n_y	NOUN
bracis-33605	252	8	=	=	SYM
bracis-33605	252	9	3\	3\	NUM
bracis-33605	252	10	)	)	PUNCT
bracis-33605	252	11	and	and	CCONJ
bracis-33605	252	12	\(k_g	\(k_g	PROPN
bracis-33605	252	13	\ge	\ge	PROPN
bracis-33605	252	14	0.05\	0.05\	PROPN
bracis-33605	252	15	)	)	PUNCT
bracis-33605	252	16	(	(	PUNCT
bracis-33605	252	17	\(\log	\(\log	X
bracis-33605	252	18	(	(	PUNCT
bracis-33605	252	19	k_g	k_g	PROPN
bracis-33605	252	20	)	)	PUNCT
bracis-33605	252	21	=	=	PUNCT
bracis-33605	252	22	-4.3\	-4.3\	PUNCT
bracis-33605	252	23	)	)	PUNCT
bracis-33605	252	24	)	)	PUNCT
bracis-33605	252	25	in	in	ADP
bracis-33605	252	26	the	the	DET
bracis-33605	252	27	instance	instance	NOUN
bracis-33605	252	28	\(n_x	\(n_x	PROPN
bracis-33605	252	29	=	=	NUM
bracis-33605	252	30	10	10	NUM
bracis-33605	252	31	\times	\time	NOUN
bracis-33605	252	32	n_y	n_y	ADP
bracis-33605	252	33	=	=	SYM
bracis-33605	252	34	4\	4\	PROPN
bracis-33605	252	35	)	)	PUNCT
bracis-33605	252	36	,	,	PUNCT
bracis-33605	252	37	the	the	DET
bracis-33605	252	38	algorithm	algorithm	NOUN
bracis-33605	252	39	q	q	NOUN
bracis-33605	252	40	-	-	PUNCT
bracis-33605	252	41	learning	learning	NOUN
bracis-33605	252	42	-	-	PUNCT
bracis-33605	252	43	gubs	gub	NOUN
bracis-33605	252	44	can	can	AUX
bracis-33605	252	45	find	find	VERB
bracis-33605	252	46	the	the	DET
bracis-33605	252	47	maxprob	maxprob	NOUN
bracis-33605	252	48	.	.	PUNCT
bracis-33605	253	1	however	however	ADV
bracis-33605	253	2	,	,	PUNCT
bracis-33605	253	3	due	due	ADP
bracis-33605	253	4	to	to	ADP
bracis-33605	253	5	the	the	DET
bracis-33605	253	6	parameters	parameter	NOUN
bracis-33605	253	7	used	use	VERB
bracis-33605	253	8	,	,	PUNCT
bracis-33605	253	9	the	the	DET
bracis-33605	253	10	algorithm	algorithm	NOUN
bracis-33605	253	11	q	q	NOUN
bracis-33605	253	12	-	-	PUNCT
bracis-33605	253	13	learning	learning	NOUN
bracis-33605	253	14	-	-	PUNCT
bracis-33605	253	15	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	253	16	)	)	PUNCT
bracis-33605	253	17	can	can	AUX
bracis-33605	253	18	not	not	PART
bracis-33605	253	19	find	find	VERB
bracis-33605	253	20	the	the	DET
bracis-33605	253	21	maxprob	maxprob	ADJ
bracis-33605	253	22	policy	policy	NOUN
bracis-33605	253	23	in	in	ADP
bracis-33605	253	24	the	the	DET
bracis-33605	253	25	instance	instance	NOUN
bracis-33605	253	26	\(n_x	\(n_x	PROPN
bracis-33605	253	27	=	=	SYM
bracis-33605	253	28	15	15	NUM
bracis-33605	253	29	\times	\time	NOUN
bracis-33605	253	30	n_y	n_y	NOUN
bracis-33605	253	31	=	=	SYM
bracis-33605	253	32	3\	3\	NUM
bracis-33605	253	33	)	)	PUNCT
bracis-33605	253	34	in	in	ADP
bracis-33605	253	35	any	any	PRON
bracis-33605	253	36	of	of	ADP
bracis-33605	253	37	the	the	DET
bracis-33605	253	38	5	5	NUM
bracis-33605	253	39	executions	execution	NOUN
bracis-33605	253	40	.	.	PUNCT
bracis-33605	254	1	figures	figure	NOUN
bracis-33605	254	2	 	 	SPACE
bracis-33605	254	3	2c	2c	NUM
bracis-33605	254	4	)	)	PUNCT
bracis-33605	254	5	and	and	CCONJ
bracis-33605	254	6	2d	2d	NOUN
bracis-33605	254	7	)	)	PUNCT
bracis-33605	254	8	compare	compare	VERB
bracis-33605	254	9	the	the	DET
bracis-33605	254	10	costs	cost	NOUN
bracis-33605	254	11	reached	reach	VERB
bracis-33605	254	12	from	from	ADP
bracis-33605	254	13	the	the	DET
bracis-33605	254	14	initial	initial	ADJ
bracis-33605	254	15	state	state	NOUN
bracis-33605	254	16	by	by	ADP
bracis-33605	254	17	the	the	DET
bracis-33605	254	18	three	three	NUM
bracis-33605	254	19	algorithms	algorithm	NOUN
bracis-33605	254	20	for	for	ADP
bracis-33605	254	21	the	the	DET
bracis-33605	254	22	same	same	ADJ
bracis-33605	254	23	two	two	NUM
bracis-33605	254	24	instances	instance	NOUN
bracis-33605	254	25	of	of	ADP
bracis-33605	254	26	the	the	DET
bracis-33605	254	27	navigation	navigation	NOUN
bracis-33605	254	28	domain	domain	NOUN
bracis-33605	254	29	.	.	PUNCT
bracis-33605	255	1	the	the	DET
bracis-33605	255	2	policy	policy	NOUN
bracis-33605	255	3	costs	cost	NOUN
bracis-33605	255	4	calculated	calculate	VERB
bracis-33605	255	5	by	by	ADP
bracis-33605	255	6	the	the	DET
bracis-33605	255	7	q	q	NOUN
bracis-33605	255	8	-	-	PUNCT
bracis-33605	255	9	learning	learning	NOUN
bracis-33605	255	10	-	-	PUNCT
bracis-33605	255	11	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	255	12	)	)	PUNCT
bracis-33605	255	13	algorithm	algorithm	NOUN
bracis-33605	255	14	are	be	AUX
bracis-33605	255	15	very	very	ADV
bracis-33605	255	16	close	close	ADJ
bracis-33605	255	17	to	to	ADP
bracis-33605	255	18	the	the	DET
bracis-33605	255	19	policy	policy	NOUN
bracis-33605	255	20	costs	cost	NOUN
bracis-33605	255	21	calculated	calculate	VERB
bracis-33605	255	22	by	by	ADP
bracis-33605	255	23	the	the	DET
bracis-33605	255	24	algorithm	algorithm	NOUN
bracis-33605	255	25	gubs	gub	NOUN
bracis-33605	255	26	-	-	PUNCT
bracis-33605	255	27	vi	vi	NOUN
bracis-33605	255	28	.	.	NOUN
bracis-33605	255	29	fig	fig	NOUN
bracis-33605	255	30	.	.	PUNCT
bracis-33605	256	1	2	2	X
bracis-33605	256	2	.	.	X
bracis-33605	256	3	probability	probability	NOUN
bracis-33605	256	4	to	to	ADP
bracis-33605	256	5	goal	goal	NOUN
bracis-33605	256	6	and	and	CCONJ
bracis-33605	256	7	cost	cost	NOUN
bracis-33605	256	8	to	to	ADP
bracis-33605	256	9	goal	goal	NOUN
bracis-33605	256	10	found	find	VERB
bracis-33605	256	11	by	by	ADP
bracis-33605	256	12	q	q	NOUN
bracis-33605	256	13	-	-	PUNCT
bracis-33605	256	14	learning	learning	NOUN
bracis-33605	256	15	-	-	PUNCT
bracis-33605	256	16	gubs	gub	NOUN
bracis-33605	256	17	,	,	PUNCT
bracis-33605	256	18	q	q	ADJ
bracis-33605	256	19	-	-	PUNCT
bracis-33605	256	20	learning	learning	NOUN
bracis-33605	256	21	-	-	PUNCT
bracis-33605	256	22	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	256	23	)	)	PUNCT
bracis-33605	256	24	and	and	CCONJ
bracis-33605	256	25	gubs	gubs	ADJ
bracis-33605	256	26	-	-	PUNCT
bracis-33605	256	27	vi	vi	NOUN
bracis-33605	256	28	algorithms	algorithms	NOUN
bracis-33605	256	29	for	for	ADP
bracis-33605	256	30	different	different	ADJ
bracis-33605	256	31	values	value	NOUN
bracis-33605	256	32	of	of	ADP
bracis-33605	256	33	\(k_g\	\(k_g\	NOUN
bracis-33605	256	34	)	)	PUNCT
bracis-33605	256	35	for	for	ADP
bracis-33605	256	36	the	the	DET
bracis-33605	256	37	navigation	navigation	NOUN
bracis-33605	256	38	domain	domain	NOUN
bracis-33605	256	39	.	.	PUNCT
bracis-33605	257	1	full	full	ADJ
bracis-33605	257	2	size	size	NOUN
bracis-33605	257	3	image	image	NOUN
bracis-33605	257	4	the	the	DET
bracis-33605	257	5	q	q	NOUN
bracis-33605	257	6	-	-	PUNCT
bracis-33605	257	7	learning	learning	NOUN
bracis-33605	257	8	-	-	PUNCT
bracis-33605	257	9	gubs	gub	NOUN
bracis-33605	257	10	algorithm	algorithm	NOUN
bracis-33605	257	11	finds	find	VERB
bracis-33605	257	12	policies	policy	NOUN
bracis-33605	257	13	with	with	ADP
bracis-33605	257	14	higher	high	ADJ
bracis-33605	257	15	costs	cost	NOUN
bracis-33605	257	16	than	than	ADP
bracis-33605	257	17	the	the	DET
bracis-33605	257	18	policies	policy	NOUN
bracis-33605	257	19	found	find	VERB
bracis-33605	257	20	by	by	ADP
bracis-33605	257	21	the	the	DET
bracis-33605	257	22	gubs	gubs	ADJ
bracis-33605	257	23	-	-	PUNCT
bracis-33605	257	24	vi	vi	NOUN
bracis-33605	257	25	algorithm	algorithm	NOUN
bracis-33605	257	26	,	,	PUNCT
bracis-33605	257	27	especially	especially	ADV
bracis-33605	257	28	for	for	ADP
bracis-33605	257	29	large	large	ADJ
bracis-33605	257	30	values	value	NOUN
bracis-33605	257	31	of	of	ADP
bracis-33605	257	32	\(k_g\	\(k_g\	NUM
bracis-33605	257	33	)	)	PUNCT
bracis-33605	257	34	.	.	PUNCT
bracis-33605	258	1	next	next	ADV
bracis-33605	258	2	,	,	PUNCT
bracis-33605	258	3	we	we	PRON
bracis-33605	258	4	analyze	analyze	VERB
bracis-33605	258	5	this	this	DET
bracis-33605	258	6	behavior	behavior	NOUN
bracis-33605	258	7	.	.	PUNCT
bracis-33605	259	1	when	when	SCONJ
bracis-33605	259	2	\(k_g\	\(k_g\	NOUN
bracis-33605	259	3	)	)	PUNCT
bracis-33605	259	4	is	be	AUX
bracis-33605	259	5	large	large	ADJ
bracis-33605	259	6	,	,	PUNCT
bracis-33605	259	7	the	the	DET
bracis-33605	259	8	q	q	NOUN
bracis-33605	259	9	-	-	PUNCT
bracis-33605	259	10	learning	learning	NOUN
bracis-33605	259	11	-	-	PUNCT
bracis-33605	259	12	gubs	gub	NOUN
bracis-33605	259	13	algorithm	algorithm	NOUN
bracis-33605	259	14	oscillates	oscillate	VERB
bracis-33605	259	15	more	more	ADV
bracis-33605	259	16	especially	especially	ADV
bracis-33605	259	17	during	during	ADP
bracis-33605	259	18	the	the	DET
bracis-33605	259	19	learning	learning	NOUN
bracis-33605	259	20	of	of	ADP
bracis-33605	259	21	probabilities	probability	NOUN
bracis-33605	259	22	in	in	ADP
bracis-33605	259	23	states	state	NOUN
bracis-33605	259	24	with	with	ADP
bracis-33605	259	25	uncertainty	uncertainty	NOUN
bracis-33605	259	26	,	,	PUNCT
bracis-33605	259	27	it	it	PRON
bracis-33605	259	28	is	be	AUX
bracis-33605	259	29	difficult	difficult	ADJ
bracis-33605	259	30	to	to	PART
bracis-33605	259	31	reach	reach	VERB
bracis-33605	259	32	the	the	DET
bracis-33605	259	33	exact	exact	ADJ
bracis-33605	259	34	value	value	NOUN
bracis-33605	259	35	of	of	ADP
bracis-33605	259	36	the	the	DET
bracis-33605	259	37	actions	action	NOUN
bracis-33605	259	38	since	since	ADV
bracis-33605	259	39	,	,	PUNCT
bracis-33605	259	40	in	in	ADP
bracis-33605	259	41	addition	addition	NOUN
bracis-33605	259	42	,	,	PUNCT
bracis-33605	259	43	the	the	DET
bracis-33605	259	44	difference	difference	NOUN
bracis-33605	259	45	between	between	ADP
bracis-33605	259	46	action	action	NOUN
bracis-33605	259	47	values	value	NOUN
bracis-33605	259	48	are	be	AUX
bracis-33605	259	49	very	very	ADV
bracis-33605	259	50	small	small	ADJ
bracis-33605	259	51	in	in	ADP
bracis-33605	259	52	this	this	DET
bracis-33605	259	53	problem	problem	NOUN
bracis-33605	259	54	.	.	PUNCT
bracis-33605	260	1	then	then	ADV
bracis-33605	260	2	,	,	PUNCT
bracis-33605	260	3	the	the	DET
bracis-33605	260	4	found	find	VERB
bracis-33605	260	5	policy	policy	NOUN
bracis-33605	260	6	uses	use	VERB
bracis-33605	260	7	actions	action	NOUN
bracis-33605	260	8	with	with	ADP
bracis-33605	260	9	values	value	NOUN
bracis-33605	260	10	close	close	ADJ
bracis-33605	260	11	to	to	ADP
bracis-33605	260	12	the	the	DET
bracis-33605	260	13	optimal	optimal	ADJ
bracis-33605	260	14	values	value	NOUN
bracis-33605	260	15	for	for	ADP
bracis-33605	260	16	some	some	DET
bracis-33605	260	17	extended	extended	ADJ
bracis-33605	260	18	states	state	NOUN
bracis-33605	260	19	,	,	PUNCT
bracis-33605	260	20	but	but	CCONJ
bracis-33605	260	21	not	not	PART
bracis-33605	260	22	the	the	DET
bracis-33605	260	23	optimal	optimal	ADJ
bracis-33605	260	24	one	one	NOUN
bracis-33605	260	25	.	.	PUNCT
bracis-33605	261	1	it	it	PRON
bracis-33605	261	2	causes	cause	VERB
bracis-33605	261	3	the	the	DET
bracis-33605	261	4	formation	formation	NOUN
bracis-33605	261	5	of	of	ADP
bracis-33605	261	6	cycles	cycle	NOUN
bracis-33605	261	7	,	,	PUNCT
bracis-33605	261	8	in	in	ADP
bracis-33605	261	9	which	which	PRON
bracis-33605	261	10	the	the	DET
bracis-33605	261	11	policy	policy	NOUN
bracis-33605	261	12	can	can	AUX
bracis-33605	261	13	return	return	VERB
bracis-33605	261	14	to	to	ADP
bracis-33605	261	15	the	the	DET
bracis-33605	261	16	same	same	ADJ
bracis-33605	261	17	real	real	ADJ
bracis-33605	261	18	state	state	NOUN
bracis-33605	261	19	but	but	CCONJ
bracis-33605	261	20	with	with	ADP
bracis-33605	261	21	a	a	DET
bracis-33605	261	22	higher	high	ADJ
bracis-33605	261	23	cost	cost	NOUN
bracis-33605	261	24	.	.	PUNCT
bracis-33605	262	1	thus	thus	ADV
bracis-33605	262	2	,	,	PUNCT
bracis-33605	262	3	the	the	DET
bracis-33605	262	4	total	total	ADJ
bracis-33605	262	5	cost	cost	NOUN
bracis-33605	262	6	to	to	ADP
bracis-33605	262	7	the	the	DET
bracis-33605	262	8	goal	goal	NOUN
bracis-33605	262	9	is	be	AUX
bracis-33605	262	10	increased	increase	VERB
bracis-33605	262	11	.	.	PUNCT
bracis-33605	263	1	as	as	ADV
bracis-33605	263	2	far	far	ADV
bracis-33605	263	3	as	as	SCONJ
bracis-33605	263	4	we	we	PRON
bracis-33605	263	5	have	have	AUX
bracis-33605	263	6	analyzed	analyze	VERB
bracis-33605	263	7	,	,	PUNCT
bracis-33605	263	8	by	by	ADP
bracis-33605	263	9	changing	change	VERB
bracis-33605	263	10	the	the	DET
bracis-33605	263	11	parameters	parameter	NOUN
bracis-33605	263	12	,	,	PUNCT
bracis-33605	263	13	it	it	PRON
bracis-33605	263	14	is	be	AUX
bracis-33605	263	15	possible	possible	ADJ
bracis-33605	263	16	to	to	PART
bracis-33605	263	17	reach	reach	VERB
bracis-33605	263	18	the	the	DET
bracis-33605	263	19	right	right	ADJ
bracis-33605	263	20	cost	cost	NOUN
bracis-33605	263	21	,	,	PUNCT
bracis-33605	263	22	for	for	ADP
bracis-33605	263	23	example	example	NOUN
bracis-33605	263	24	,	,	PUNCT
bracis-33605	263	25	by	by	ADP
bracis-33605	263	26	increasing	increase	VERB
bracis-33605	263	27	the	the	DET
bracis-33605	263	28	number	number	NOUN
bracis-33605	263	29	of	of	ADP
bracis-33605	263	30	episodes	episode	NOUN
bracis-33605	263	31	.	.	PUNCT
bracis-33605	264	1	4.3	4.3	NUM
bracis-33605	264	2	comparison	comparison	NOUN
bracis-33605	264	3	with	with	ADP
bracis-33605	264	4	 	 	SPACE
bracis-33605	264	5	the	the	DET
bracis-33605	264	6	 	 	SPACE
bracis-33605	264	7	geibel	geibel	NOUN
bracis-33605	264	8	and	and	CCONJ
bracis-33605	264	9	 	 	SPACE
bracis-33605	264	10	wysotzki	wysotzki	ADJ
bracis-33605	264	11	criterion	criterion	NOUN
bracis-33605	264	12	we	we	PRON
bracis-33605	264	13	compare	compare	VERB
bracis-33605	264	14	the	the	DET
bracis-33605	264	15	proposed	propose	VERB
bracis-33605	264	16	algorithms	algorithm	NOUN
bracis-33605	264	17	with	with	ADP
bracis-33605	264	18	the	the	DET
bracis-33605	264	19	geibel	geibel	NOUN
bracis-33605	264	20	and	and	CCONJ
bracis-33605	264	21	wysotzki	wysotzki	ADJ
bracis-33605	264	22	criterion	criterion	NOUN
bracis-33605	264	23	[	[	X
bracis-33605	264	24	5	5	NUM
bracis-33605	264	25	]	]	PUNCT
bracis-33605	264	26	.	.	PUNCT
bracis-33605	265	1	for	for	ADP
bracis-33605	265	2	this	this	DET
bracis-33605	265	3	comparison	comparison	NOUN
bracis-33605	265	4	,	,	PUNCT
bracis-33605	265	5	we	we	PRON
bracis-33605	265	6	varied	vary	VERB
bracis-33605	265	7	the	the	DET
bracis-33605	265	8	parameter	parameter	NOUN
bracis-33605	265	9	\(\xi	\(\xi	PUNCT
bracis-33605	265	10	\	\	PROPN
bracis-33605	265	11	)	)	PUNCT
bracis-33605	265	12	between	between	ADP
bracis-33605	265	13	the	the	DET
bracis-33605	265	14	values	value	NOUN
bracis-33605	265	15	\([0.1	\([0.1	NOUN
bracis-33605	265	16	,	,	PUNCT
bracis-33605	265	17	1	1	NUM
bracis-33605	265	18	\times	\time	NOUN
bracis-33605	265	19	10	10	NUM
bracis-33605	265	20	^	^	SYM
bracis-33605	265	21	7]\	7]\	NUM
bracis-33605	265	22	)	)	PUNCT
bracis-33605	265	23	to	to	PART
bracis-33605	265	24	evaluate	evaluate	VERB
bracis-33605	265	25	if	if	SCONJ
bracis-33605	265	26	geibel	geibel	NOUN
bracis-33605	265	27	and	and	CCONJ
bracis-33605	265	28	wysotzki	wysotzki	ADJ
bracis-33605	265	29	criterion	criterion	NOUN
bracis-33605	265	30	can	can	AUX
bracis-33605	265	31	compute	compute	VERB
bracis-33605	265	32	the	the	DET
bracis-33605	265	33	optimal	optimal	ADJ
bracis-33605	265	34	policies	policy	NOUN
bracis-33605	265	35	for	for	ADP
bracis-33605	265	36	egubs	egub	NOUN
bracis-33605	265	37	only	only	ADV
bracis-33605	265	38	by	by	ADP
bracis-33605	265	39	choosing	choose	VERB
bracis-33605	265	40	the	the	DET
bracis-33605	265	41	parameters	parameter	NOUN
bracis-33605	265	42	appropriately	appropriately	ADV
bracis-33605	265	43	and	and	CCONJ
bracis-33605	265	44	,	,	PUNCT
bracis-33605	265	45	if	if	SCONJ
bracis-33605	265	46	not	not	PART
bracis-33605	265	47	,	,	PUNCT
bracis-33605	265	48	how	how	SCONJ
bracis-33605	265	49	close	close	ADJ
bracis-33605	265	50	it	it	PRON
bracis-33605	265	51	can	can	AUX
bracis-33605	265	52	get	get	VERB
bracis-33605	265	53	to	to	ADP
bracis-33605	265	54	the	the	DET
bracis-33605	265	55	optimal	optimal	ADJ
bracis-33605	265	56	value	value	NOUN
bracis-33605	265	57	.	.	PUNCT
bracis-33605	266	1	the	the	DET
bracis-33605	266	2	discounts	discount	NOUN
bracis-33605	266	3	\(\gamma	\(\gamma	X
bracis-33605	266	4	=	=	SYM
bracis-33605	266	5	0.9\	0.9\	NOUN
bracis-33605	266	6	)	)	PUNCT
bracis-33605	266	7	 	 	SPACE
bracis-33605	266	8	and	and	CCONJ
bracis-33605	266	9	\(\bar{\gamma	\(\bar{\gamma	VERB
bracis-33605	266	10	}	}	PUNCT
bracis-33605	266	11	=	=	SYM
bracis-33605	266	12	1\	1\	NUM
bracis-33605	266	13	)	)	PUNCT
bracis-33605	266	14	were	be	AUX
bracis-33605	266	15	used	use	VERB
bracis-33605	266	16	.	.	PUNCT
bracis-33605	267	1	figure	figure	VERB
bracis-33605	267	2	 	 	SPACE
bracis-33605	267	3	3	3	NUM
bracis-33605	267	4	shows	show	VERB
bracis-33605	267	5	the	the	DET
bracis-33605	267	6	\(log_2(\xi	\(log_2(\xi	NOUN
bracis-33605	267	7	)	)	PUNCT
bracis-33605	267	8	\	\	NOUN
bracis-33605	267	9	)	)	PUNCT
bracis-33605	267	10	on	on	ADP
bracis-33605	267	11	the	the	DET
bracis-33605	267	12	x	x	NOUN
bracis-33605	267	13	axis	axis	NOUN
bracis-33605	267	14	and	and	CCONJ
bracis-33605	267	15	,	,	PUNCT
bracis-33605	267	16	on	on	ADP
bracis-33605	267	17	the	the	DET
bracis-33605	267	18	y	y	NOUN
bracis-33605	267	19	-	-	PUNCT
bracis-33605	267	20	axis	axis	NOUN
bracis-33605	267	21	,	,	PUNCT
bracis-33605	267	22	the	the	DET
bracis-33605	267	23	values	value	NOUN
bracis-33605	267	24	corresponding	correspond	VERB
bracis-33605	267	25	to	to	ADP
bracis-33605	267	26	the	the	DET
bracis-33605	267	27	optimal	optimal	ADJ
bracis-33605	267	28	policy	policy	NOUN
bracis-33605	267	29	found	find	VERB
bracis-33605	267	30	by	by	ADP
bracis-33605	267	31	the	the	DET
bracis-33605	267	32	gubs	gub	NOUN
bracis-33605	267	33	-	-	PUNCT
bracis-33605	267	34	vi	vi	NOUN
bracis-33605	267	35	,	,	PUNCT
bracis-33605	267	36	q	q	ADJ
bracis-33605	267	37	-	-	PUNCT
bracis-33605	267	38	learning	learning	NOUN
bracis-33605	267	39	-	-	PUNCT
bracis-33605	267	40	gubs	gub	NOUN
bracis-33605	267	41	and	and	CCONJ
bracis-33605	267	42	q	q	NOUN
bracis-33605	267	43	-	-	PUNCT
bracis-33605	267	44	learning	learning	ADJ
bracis-33605	267	45	-	-	PUNCT
bracis-33605	267	46	egubs+cmax	egubs+cmax	PROPN
bracis-33605	267	47	algorithms	algorithm	NOUN
bracis-33605	267	48	with	with	ADP
bracis-33605	267	49	\(k_g	\(k_g	PROPN
bracis-33605	267	50	=	=	SYM
bracis-33605	267	51	5\	5\	PROPN
bracis-33605	267	52	)	)	PUNCT
bracis-33605	267	53	and	and	CCONJ
bracis-33605	267	54	\(c_{max	\(c_{max	ADJ
bracis-33605	267	55	}	}	PUNCT
bracis-33605	267	56	=	=	SYM
bracis-33605	267	57	100\	100\	NUM
bracis-33605	267	58	)	)	PUNCT
bracis-33605	267	59	,	,	PUNCT
bracis-33605	267	60	and	and	CCONJ
bracis-33605	267	61	the	the	DET
bracis-33605	267	62	values	value	NOUN
bracis-33605	267	63	of	of	ADP
bracis-33605	267	64	the	the	DET
bracis-33605	267	65	optimal	optimal	ADJ
bracis-33605	267	66	policies	policy	NOUN
bracis-33605	267	67	found	find	VERB
bracis-33605	267	68	by	by	ADP
bracis-33605	267	69	the	the	DET
bracis-33605	267	70	geibel	geibel	NOUN
bracis-33605	267	71	and	and	CCONJ
bracis-33605	267	72	wysotzki	wysotzki	ADJ
bracis-33605	267	73	criterion	criterion	NOUN
bracis-33605	267	74	evaluated	evaluate	VERB
bracis-33605	267	75	with	with	ADP
bracis-33605	267	76	the	the	DET
bracis-33605	267	77	gubs	gubs	ADJ
bracis-33605	267	78	criterion	criterion	NOUN
bracis-33605	267	79	for	for	ADP
bracis-33605	267	80	the	the	DET
bracis-33605	267	81	river	river	NOUN
bracis-33605	267	82	domain	domain	NOUN
bracis-33605	267	83	of	of	ADP
bracis-33605	267	84	size	size	NOUN
bracis-33605	267	85	\(n_y	\(n_y	NOUN
bracis-33605	267	86	=	=	SYM
bracis-33605	267	87	20\	20\	NUM
bracis-33605	267	88	)	)	PUNCT
bracis-33605	267	89	and	and	CCONJ
bracis-33605	267	90	\(n_x=5\	\(n_x=5\	PROPN
bracis-33605	267	91	)	)	PUNCT
bracis-33605	267	92	with	with	ADP
bracis-33605	267	93	\(p=0.4\	\(p=0.4\	PROPN
bracis-33605	267	94	)	)	PUNCT
bracis-33605	267	95	.	.	PUNCT
bracis-33605	268	1	each	each	DET
bracis-33605	268	2	point	point	NOUN
bracis-33605	268	3	in	in	ADP
bracis-33605	268	4	the	the	DET
bracis-33605	268	5	figure	figure	NOUN
bracis-33605	268	6	is	be	AUX
bracis-33605	268	7	the	the	DET
bracis-33605	268	8	average	average	NOUN
bracis-33605	268	9	of	of	ADP
bracis-33605	268	10	5	5	NUM
bracis-33605	268	11	executions	execution	NOUN
bracis-33605	268	12	of	of	ADP
bracis-33605	268	13	the	the	DET
bracis-33605	268	14	algorithms	algorithm	NOUN
bracis-33605	268	15	.	.	PUNCT
bracis-33605	269	1	the	the	DET
bracis-33605	269	2	values	value	NOUN
bracis-33605	269	3	found	find	VERB
bracis-33605	269	4	by	by	ADP
bracis-33605	269	5	the	the	DET
bracis-33605	269	6	algorithms	algorithm	NOUN
bracis-33605	269	7	q	q	NOUN
bracis-33605	269	8	-	-	PUNCT
bracis-33605	269	9	learning	learning	NOUN
bracis-33605	269	10	-	-	PUNCT
bracis-33605	269	11	egubs+cmax	egubs+cmax	PROPN
bracis-33605	269	12	,	,	PUNCT
bracis-33605	269	13	q	q	ADJ
bracis-33605	269	14	-	-	PUNCT
bracis-33605	269	15	learning	learning	NOUN
bracis-33605	269	16	-	-	PUNCT
bracis-33605	269	17	gubs	gub	NOUN
bracis-33605	269	18	and	and	CCONJ
bracis-33605	269	19	gubs	gub	NOUN
bracis-33605	269	20	-	-	PUNCT
bracis-33605	269	21	vi	vi	NOUN
bracis-33605	269	22	are	be	AUX
bracis-33605	269	23	close	close	ADJ
bracis-33605	269	24	.	.	PUNCT
bracis-33605	270	1	when	when	SCONJ
bracis-33605	270	2	\(log_2(\xi	\(log_2(\xi	X
bracis-33605	270	3	)	)	PUNCT
bracis-33605	270	4	=	=	SYM
bracis-33605	270	5	1\	1\	NUM
bracis-33605	270	6	)	)	PUNCT
bracis-33605	270	7	,	,	PUNCT
bracis-33605	270	8	the	the	DET
bracis-33605	270	9	values	value	NOUN
bracis-33605	270	10	of	of	ADP
bracis-33605	270	11	the	the	DET
bracis-33605	270	12	policies	policy	NOUN
bracis-33605	270	13	found	find	VERB
bracis-33605	270	14	by	by	ADP
bracis-33605	270	15	the	the	DET
bracis-33605	270	16	geibel	geibel	NOUN
bracis-33605	270	17	and	and	CCONJ
bracis-33605	270	18	wysotzki	wysotzki	ADJ
bracis-33605	270	19	criterion	criterion	NOUN
bracis-33605	270	20	are	be	AUX
bracis-33605	270	21	close	close	ADJ
bracis-33605	270	22	to	to	ADP
bracis-33605	270	23	the	the	DET
bracis-33605	270	24	ones	one	NOUN
bracis-33605	270	25	found	find	VERB
bracis-33605	270	26	by	by	ADP
bracis-33605	270	27	gubs	gub	NOUN
bracis-33605	270	28	-	-	PUNCT
bracis-33605	270	29	vi	vi	NOUN
bracis-33605	270	30	,	,	PUNCT
bracis-33605	270	31	q	q	ADJ
bracis-33605	270	32	-	-	PUNCT
bracis-33605	270	33	learning	learning	NOUN
bracis-33605	270	34	-	-	PUNCT
bracis-33605	270	35	egubs+cmax	egubs+cmax	PROPN
bracis-33605	270	36	and	and	CCONJ
bracis-33605	270	37	q	q	ADJ
bracis-33605	270	38	-	-	PUNCT
bracis-33605	270	39	learning	learning	NOUN
bracis-33605	270	40	-	-	PUNCT
bracis-33605	270	41	gubs	gub	NOUN
bracis-33605	270	42	.	.	PUNCT
bracis-33605	271	1	notice	notice	VERB
bracis-33605	271	2	that	that	SCONJ
bracis-33605	271	3	,	,	PUNCT
bracis-33605	271	4	to	to	PART
bracis-33605	271	5	carry	carry	VERB
bracis-33605	271	6	out	out	ADP
bracis-33605	271	7	this	this	DET
bracis-33605	271	8	experiment	experiment	NOUN
bracis-33605	271	9	,	,	PUNCT
bracis-33605	271	10	an	an	DET
bracis-33605	271	11	indefinite	indefinite	ADJ
bracis-33605	271	12	set	set	NOUN
bracis-33605	271	13	of	of	ADP
bracis-33605	271	14	values	value	NOUN
bracis-33605	271	15	for	for	ADP
bracis-33605	271	16	\(\xi	\(\xi	PROPN
bracis-33605	271	17	\	\	NOUN
bracis-33605	271	18	)	)	PUNCT
bracis-33605	271	19	had	have	VERB
bracis-33605	271	20	to	to	PART
bracis-33605	271	21	be	be	AUX
bracis-33605	271	22	varied	varied	ADJ
bracis-33605	271	23	,	,	PUNCT
bracis-33605	271	24	this	this	PRON
bracis-33605	271	25	is	be	AUX
bracis-33605	271	26	because	because	SCONJ
bracis-33605	271	27	there	there	PRON
bracis-33605	271	28	is	be	VERB
bracis-33605	271	29	no	no	DET
bracis-33605	271	30	clear	clear	ADJ
bracis-33605	271	31	procedure	procedure	NOUN
bracis-33605	271	32	to	to	PART
bracis-33605	271	33	choose	choose	VERB
bracis-33605	271	34	a	a	DET
bracis-33605	271	35	certain	certain	ADJ
bracis-33605	271	36	value	value	NOUN
bracis-33605	271	37	of	of	ADP
bracis-33605	271	38	\(\xi	\(\xi	PROPN
bracis-33605	271	39	\	\	PROPN
bracis-33605	271	40	)	)	PUNCT
bracis-33605	271	41	for	for	ADP
bracis-33605	271	42	this	this	DET
bracis-33605	271	43	criterion	criterion	NOUN
bracis-33605	271	44	,	,	PUNCT
bracis-33605	271	45	to	to	PART
bracis-33605	271	46	have	have	VERB
bracis-33605	271	47	a	a	DET
bracis-33605	271	48	guarantee	guarantee	NOUN
bracis-33605	271	49	of	of	ADP
bracis-33605	271	50	finding	find	VERB
bracis-33605	271	51	the	the	DET
bracis-33605	271	52	resulting	result	VERB
bracis-33605	271	53	policy	policy	NOUN
bracis-33605	271	54	under	under	ADP
bracis-33605	271	55	gubs	gub	NOUN
bracis-33605	271	56	.	.	PUNCT
bracis-33605	272	1	in	in	ADP
bracis-33605	272	2	certain	certain	ADJ
bracis-33605	272	3	situations	situation	NOUN
bracis-33605	272	4	,	,	PUNCT
bracis-33605	272	5	it	it	PRON
bracis-33605	272	6	may	may	AUX
bracis-33605	272	7	not	not	PART
bracis-33605	272	8	be	be	AUX
bracis-33605	272	9	feasible	feasible	ADJ
bracis-33605	272	10	to	to	PART
bracis-33605	272	11	carry	carry	VERB
bracis-33605	272	12	out	out	ADP
bracis-33605	272	13	this	this	DET
bracis-33605	272	14	type	type	NOUN
bracis-33605	272	15	of	of	ADP
bracis-33605	272	16	procedure	procedure	NOUN
bracis-33605	272	17	to	to	PART
bracis-33605	272	18	guarantee	guarantee	VERB
bracis-33605	272	19	finding	find	VERB
bracis-33605	272	20	certain	certain	ADJ
bracis-33605	272	21	policies	policy	NOUN
bracis-33605	272	22	.	.	PUNCT
bracis-33605	273	1	fig	fig	NOUN
bracis-33605	273	2	.	.	PUNCT
bracis-33605	274	1	3	3	X
bracis-33605	274	2	.	.	X
bracis-33605	274	3	comparison	comparison	NOUN
bracis-33605	274	4	of	of	ADP
bracis-33605	274	5	the	the	DET
bracis-33605	274	6	value	value	NOUN
bracis-33605	274	7	of	of	ADP
bracis-33605	274	8	the	the	DET
bracis-33605	274	9	optimal	optimal	ADJ
bracis-33605	274	10	policy	policy	NOUN
bracis-33605	274	11	for	for	ADP
bracis-33605	274	12	gubs	gub	NOUN
bracis-33605	274	13	-	-	PUNCT
bracis-33605	274	14	vi	vi	NOUN
bracis-33605	274	15	,	,	PUNCT
bracis-33605	274	16	q	q	ADJ
bracis-33605	274	17	-	-	PUNCT
bracis-33605	274	18	learning	learning	NOUN
bracis-33605	274	19	-	-	PUNCT
bracis-33605	274	20	gubs	gub	NOUN
bracis-33605	274	21	and	and	CCONJ
bracis-33605	274	22	q	q	NOUN
bracis-33605	274	23	-	-	PUNCT
bracis-33605	274	24	learning	learning	NOUN
bracis-33605	274	25	-	-	PUNCT
bracis-33605	274	26	egubs+cmax	egubs+cmax	PROPN
bracis-33605	274	27	with	with	ADP
bracis-33605	274	28	the	the	DET
bracis-33605	274	29	value	value	NOUN
bracis-33605	274	30	of	of	ADP
bracis-33605	274	31	optimal	optimal	ADJ
bracis-33605	274	32	policies	policy	NOUN
bracis-33605	274	33	for	for	ADP
bracis-33605	274	34	the	the	DET
bracis-33605	274	35	geibel	geibel	NOUN
bracis-33605	274	36	and	and	CCONJ
bracis-33605	274	37	wysotzki	wysotzki	ADJ
bracis-33605	274	38	criterion	criterion	NOUN
bracis-33605	274	39	for	for	ADP
bracis-33605	274	40	different	different	ADJ
bracis-33605	274	41	values	value	NOUN
bracis-33605	274	42	of	of	ADP
bracis-33605	274	43	\(\xi	\(\xi	PROPN
bracis-33605	274	44	\	\	PROPN
bracis-33605	274	45	)	)	PUNCT
bracis-33605	274	46	for	for	ADP
bracis-33605	274	47	the	the	DET
bracis-33605	274	48	river	river	NOUN
bracis-33605	274	49	domain	domain	NOUN
bracis-33605	274	50	\(n_y	\(n_y	NOUN
bracis-33605	274	51	=	=	SYM
bracis-33605	274	52	20\	20\	NUM
bracis-33605	274	53	)	)	PUNCT
bracis-33605	274	54	and	and	CCONJ
bracis-33605	274	55	\(n_x=5\	\(n_x=5\	PROPN
bracis-33605	274	56	)	)	PUNCT
bracis-33605	274	57	with	with	ADP
bracis-33605	274	58	\(p=0.4\	\(p=0.4\	PROPN
bracis-33605	274	59	)	)	PUNCT
bracis-33605	274	60	.	.	PUNCT
bracis-33605	275	1	full	full	ADJ
bracis-33605	275	2	size	size	NOUN
bracis-33605	275	3	image	image	NOUN
bracis-33605	275	4	5	5	NUM
bracis-33605	275	5	conclusion	conclusion	NOUN
bracis-33605	275	6	in	in	ADP
bracis-33605	275	7	this	this	DET
bracis-33605	275	8	work	work	NOUN
bracis-33605	275	9	,	,	PUNCT
bracis-33605	275	10	we	we	PRON
bracis-33605	275	11	propose	propose	VERB
bracis-33605	275	12	the	the	DET
bracis-33605	275	13	first	first	ADJ
bracis-33605	275	14	two	two	NUM
bracis-33605	275	15	algorithms	algorithm	NOUN
bracis-33605	275	16	for	for	ADP
bracis-33605	275	17	rl	rl	PRON
bracis-33605	275	18	that	that	PRON
bracis-33605	275	19	use	use	VERB
bracis-33605	275	20	the	the	DET
bracis-33605	275	21	gubs	gubs	ADJ
bracis-33605	275	22	criterion	criterion	NOUN
bracis-33605	275	23	:	:	PUNCT
bracis-33605	275	24	q	q	ADJ
bracis-33605	275	25	-	-	PUNCT
bracis-33605	275	26	learning	learning	NOUN
bracis-33605	275	27	-	-	PUNCT
bracis-33605	275	28	gubs	gub	NOUN
bracis-33605	275	29	and	and	CCONJ
bracis-33605	275	30	q	q	NOUN
bracis-33605	275	31	-	-	PUNCT
bracis-33605	275	32	learning	learning	NOUN
bracis-33605	275	33	-	-	PUNCT
bracis-33605	275	34	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	275	35	)	)	PUNCT
bracis-33605	275	36	to	to	PART
bracis-33605	275	37	make	make	VERB
bracis-33605	275	38	a	a	DET
bracis-33605	275	39	trade	trade	NOUN
bracis-33605	275	40	-	-	PUNCT
bracis-33605	275	41	off	off	NOUN
bracis-33605	275	42	between	between	ADP
bracis-33605	275	43	probability	probability	NOUN
bracis-33605	275	44	to	to	ADP
bracis-33605	275	45	goal	goal	NOUN
bracis-33605	275	46	and	and	CCONJ
bracis-33605	275	47	cost	cost	VERB
bracis-33605	275	48	to	to	ADP
bracis-33605	275	49	goal	goal	NOUN
bracis-33605	275	50	.	.	PUNCT
bracis-33605	276	1	q	q	X
bracis-33605	276	2	-	-	PUNCT
bracis-33605	276	3	learning	learning	NOUN
bracis-33605	276	4	-	-	PUNCT
bracis-33605	276	5	gubs	gub	NOUN
bracis-33605	276	6	is	be	AUX
bracis-33605	276	7	an	an	DET
bracis-33605	276	8	approximate	approximate	ADJ
bracis-33605	276	9	algorithm	algorithm	NOUN
bracis-33605	276	10	and	and	CCONJ
bracis-33605	276	11	can	can	AUX
bracis-33605	276	12	use	use	VERB
bracis-33605	276	13	several	several	ADJ
bracis-33605	276	14	utility	utility	NOUN
bracis-33605	276	15	functions	function	NOUN
bracis-33605	276	16	.	.	PUNCT
bracis-33605	277	1	q	q	X
bracis-33605	277	2	-	-	PUNCT
bracis-33605	277	3	learning	learning	NOUN
bracis-33605	277	4	-	-	PUNCT
bracis-33605	277	5	egubs+\(c_{max}\	egubs+\(c_{max}\	PROPN
bracis-33605	277	6	)	)	PUNCT
bracis-33605	277	7	is	be	AUX
bracis-33605	277	8	a	a	DET
bracis-33605	277	9	model	model	NOUN
bracis-33605	277	10	-	-	PUNCT
bracis-33605	277	11	based	base	VERB
bracis-33605	277	12	algorithm	algorithm	NOUN
bracis-33605	277	13	that	that	PRON
bracis-33605	277	14	calculates	calculate	VERB
bracis-33605	277	15	the	the	DET
bracis-33605	277	16	\(c_{max}\	\(c_{max}\	NOUN
bracis-33605	277	17	)	)	PUNCT
bracis-33605	277	18	value	value	NOUN
bracis-33605	277	19	for	for	ADP
bracis-33605	277	20	the	the	DET
bracis-33605	277	21	particular	particular	ADJ
bracis-33605	277	22	case	case	NOUN
bracis-33605	277	23	of	of	ADP
bracis-33605	277	24	the	the	DET
bracis-33605	277	25	exponential	exponential	ADJ
bracis-33605	277	26	utility	utility	NOUN
bracis-33605	277	27	function	function	NOUN
bracis-33605	277	28	.	.	PUNCT
bracis-33605	278	1	experiments	experiment	NOUN
bracis-33605	278	2	were	be	AUX
bracis-33605	278	3	carried	carry	VERB
bracis-33605	278	4	out	out	ADP
bracis-33605	278	5	with	with	ADP
bracis-33605	278	6	the	the	DET
bracis-33605	278	7	proposed	propose	VERB
bracis-33605	278	8	algorithms	algorithm	NOUN
bracis-33605	278	9	with	with	ADP
bracis-33605	278	10	several	several	ADJ
bracis-33605	278	11	instances	instance	NOUN
bracis-33605	278	12	corresponding	correspond	VERB
bracis-33605	278	13	to	to	ADP
bracis-33605	278	14	two	two	NUM
bracis-33605	278	15	domains	domain	NOUN
bracis-33605	278	16	.	.	PUNCT
bracis-33605	279	1	the	the	DET
bracis-33605	279	2	results	result	NOUN
bracis-33605	279	3	show	show	VERB
bracis-33605	279	4	that	that	SCONJ
bracis-33605	279	5	these	these	DET
bracis-33605	279	6	algorithms	algorithm	NOUN
bracis-33605	279	7	can	can	AUX
bracis-33605	279	8	find	find	VERB
bracis-33605	279	9	policies	policy	NOUN
bracis-33605	279	10	that	that	PRON
bracis-33605	279	11	make	make	VERB
bracis-33605	279	12	a	a	DET
bracis-33605	279	13	trade	trade	NOUN
bracis-33605	279	14	-	-	PUNCT
bracis-33605	279	15	off	off	NOUN
bracis-33605	279	16	between	between	ADP
bracis-33605	279	17	probability	probability	NOUN
bracis-33605	279	18	and	and	CCONJ
bracis-33605	279	19	cost	cost	NOUN
bracis-33605	279	20	according	accord	VERB
bracis-33605	279	21	to	to	ADP
bracis-33605	279	22	the	the	DET
bracis-33605	279	23	parameter	parameter	NOUN
bracis-33605	279	24	\(k_g\	\(k_g\	PROPN
bracis-33605	279	25	)	)	PUNCT
bracis-33605	279	26	and	and	CCONJ
bracis-33605	279	27	according	accord	VERB
bracis-33605	279	28	to	to	ADP
bracis-33605	279	29	the	the	DET
bracis-33605	279	30	utility	utility	NOUN
bracis-33605	279	31	function	function	NOUN
bracis-33605	279	32	used	use	VERB
bracis-33605	279	33	.	.	PUNCT
bracis-33605	280	1	we	we	PRON
bracis-33605	280	2	also	also	ADV
bracis-33605	280	3	compare	compare	VERB
bracis-33605	280	4	the	the	DET
bracis-33605	280	5	proposed	propose	VERB
bracis-33605	280	6	algorithms	algorithm	NOUN
bracis-33605	280	7	with	with	ADP
bracis-33605	280	8	geibel	geibel	NOUN
bracis-33605	280	9	and	and	CCONJ
bracis-33605	280	10	wysotzki	wysotzki	ADJ
bracis-33605	280	11	criterion	criterion	NOUN
bracis-33605	280	12	[	[	X
bracis-33605	280	13	5	5	NUM
bracis-33605	280	14	]	]	PUNCT
bracis-33605	280	15	.	.	PUNCT
bracis-33605	281	1	the	the	DET
bracis-33605	281	2	values	value	NOUN
bracis-33605	281	3	of	of	ADP
bracis-33605	281	4	the	the	DET
bracis-33605	281	5	policies	policy	NOUN
bracis-33605	281	6	found	find	VERB
bracis-33605	281	7	by	by	ADP
bracis-33605	281	8	geibel	geibel	NOUN
bracis-33605	281	9	and	and	CCONJ
bracis-33605	281	10	wysotzki	wysotzki	ADJ
bracis-33605	281	11	criterion	criterion	NOUN
bracis-33605	281	12	for	for	ADP
bracis-33605	281	13	some	some	DET
bracis-33605	281	14	values	value	NOUN
bracis-33605	281	15	of	of	ADP
bracis-33605	281	16	parameters	parameter	NOUN
bracis-33605	281	17	are	be	AUX
bracis-33605	281	18	close	close	ADJ
bracis-33605	281	19	to	to	ADP
bracis-33605	281	20	those	those	PRON
bracis-33605	281	21	found	find	VERB
bracis-33605	281	22	by	by	ADP
bracis-33605	281	23	the	the	DET
bracis-33605	281	24	proposed	propose	VERB
bracis-33605	281	25	algorithms	algorithm	NOUN
bracis-33605	281	26	.	.	PUNCT
bracis-33605	282	1	different	different	ADJ
bracis-33605	282	2	from	from	ADP
bracis-33605	282	3	geibel	geibel	NOUN
bracis-33605	282	4	and	and	CCONJ
bracis-33605	282	5	wysotzki	wysotzki	ADJ
bracis-33605	282	6	criterion	criterion	NOUN
bracis-33605	282	7	that	that	PRON
bracis-33605	282	8	is	be	AUX
bracis-33605	282	9	based	base	VERB
bracis-33605	282	10	on	on	ADP
bracis-33605	282	11	\(\xi	\(\xi	PROPN
bracis-33605	282	12	\	\	PROPN
bracis-33605	282	13	)	)	PUNCT
bracis-33605	282	14	and	and	CCONJ
bracis-33605	282	15	discount	discount	NOUN
bracis-33605	282	16	factors	factor	NOUN
bracis-33605	282	17	,	,	PUNCT
bracis-33605	282	18	where	where	SCONJ
bracis-33605	282	19	the	the	DET
bracis-33605	282	20	choice	choice	NOUN
bracis-33605	282	21	of	of	ADP
bracis-33605	282	22	parameters	parameter	NOUN
bracis-33605	282	23	depends	depend	VERB
bracis-33605	282	24	on	on	ADP
bracis-33605	282	25	the	the	DET
bracis-33605	282	26	problem	problem	NOUN
bracis-33605	282	27	;	;	PUNCT
bracis-33605	282	28	the	the	DET
bracis-33605	282	29	parameters	parameter	NOUN
bracis-33605	282	30	of	of	ADP
bracis-33605	282	31	gubs	gub	NOUN
bracis-33605	282	32	(	(	PUNCT
bracis-33605	282	33	\(k_g\	\(k_g\	NOUN
bracis-33605	282	34	)	)	PUNCT
bracis-33605	282	35	,	,	PUNCT
bracis-33605	282	36	\(u_{max}\	\(u_{max}\	NOUN
bracis-33605	282	37	)	)	PUNCT
bracis-33605	282	38	and	and	CCONJ
bracis-33605	282	39	\(u_{min}\	\(u_{min}\	PROPN
bracis-33605	282	40	)	)	PUNCT
bracis-33605	282	41	)	)	PUNCT
bracis-33605	282	42	do	do	AUX
bracis-33605	282	43	not	not	PART
bracis-33605	282	44	depend	depend	VERB
bracis-33605	282	45	on	on	ADP
bracis-33605	282	46	the	the	DET
bracis-33605	282	47	problem	problem	NOUN
bracis-33605	282	48	,	,	PUNCT
bracis-33605	282	49	so	so	ADV
bracis-33605	282	50	,	,	PUNCT
bracis-33605	282	51	the	the	DET
bracis-33605	282	52	gubs	gubs	ADJ
bracis-33605	282	53	parameters	parameter	NOUN
bracis-33605	282	54	of	of	ADP
bracis-33605	282	55	the	the	DET
bracis-33605	282	56	proposed	propose	VERB
bracis-33605	282	57	algorithms	algorithm	NOUN
bracis-33605	282	58	can	can	AUX
bracis-33605	282	59	be	be	AUX
bracis-33605	282	60	rationally	rationally	ADV
bracis-33605	282	61	set	set	VERB
bracis-33605	282	62	.	.	PUNCT
bracis-33605	283	1	the	the	DET
bracis-33605	283	2	proposed	propose	VERB
bracis-33605	283	3	algorithms	algorithm	NOUN
bracis-33605	283	4	could	could	AUX
bracis-33605	283	5	be	be	AUX
bracis-33605	283	6	the	the	DET
bracis-33605	283	7	basis	basis	NOUN
bracis-33605	283	8	for	for	ADP
bracis-33605	283	9	designing	design	VERB
bracis-33605	283	10	more	more	ADV
bracis-33605	283	11	sophisticated	sophisticated	ADJ
bracis-33605	283	12	rl	rl	NOUN
bracis-33605	283	13	algorithms	algorithm	NOUN
bracis-33605	283	14	that	that	PRON
bracis-33605	283	15	can	can	AUX
bracis-33605	283	16	work	work	VERB
bracis-33605	283	17	in	in	ADP
bracis-33605	283	18	more	more	ADJ
bracis-33605	283	19	complex	complex	ADJ
bracis-33605	283	20	domains	domain	NOUN
bracis-33605	283	21	.	.	PUNCT
bracis-33605	284	1	notes	note	VERB
bracis-33605	284	2	1.notice	1.notice	NUM
bracis-33605	284	3	that	that	DET
bracis-33605	284	4	\(c'\	\(c'\	NOUN
bracis-33605	284	5	)	)	PUNCT
bracis-33605	284	6	is	be	AUX
bracis-33605	284	7	an	an	DET
bracis-33605	284	8	utility	utility	NOUN
bracis-33605	284	9	over	over	ADP
bracis-33605	284	10	costs	cost	NOUN
bracis-33605	284	11	and	and	CCONJ
bracis-33605	284	12	the	the	DET
bracis-33605	284	13	optimal	optimal	ADJ
bracis-33605	284	14	policy	policy	NOUN
bracis-33605	284	15	must	must	AUX
bracis-33605	284	16	maximize	maximize	VERB
bracis-33605	284	17	the	the	DET
bracis-33605	284	18	utility	utility	NOUN
bracis-33605	284	19	.	.	PUNCT
bracis-33605	285	1	references	reference	NOUN
bracis-33605	285	2	crispino	crispino	PROPN
bracis-33605	285	3	,	,	PUNCT
bracis-33605	285	4	g.n	g.n	PROPN
bracis-33605	285	5	.	.	PROPN
bracis-33605	285	6	,	,	PUNCT
bracis-33605	285	7	freire	freire	PROPN
bracis-33605	285	8	,	,	PUNCT
bracis-33605	285	9	v.	v.	PROPN
bracis-33605	285	10	,	,	PUNCT
bracis-33605	285	11	delgado	delgado	PROPN
bracis-33605	285	12	,	,	PUNCT
bracis-33605	285	13	k.v	k.v	PROPN
bracis-33605	285	14	.	.	PUNCT
bracis-33605	285	15	:	:	PUNCT
bracis-33605	286	1	gubs	gubs	ADJ
bracis-33605	286	2	criterion	criterion	NOUN
bracis-33605	286	3	:	:	PUNCT
bracis-33605	286	4	arbitrary	arbitrary	ADJ
bracis-33605	286	5	trade	trade	NOUN
bracis-33605	286	6	-	-	PUNCT
bracis-33605	286	7	offs	off	NOUN
bracis-33605	286	8	between	between	ADP
bracis-33605	286	9	cost	cost	NOUN
bracis-33605	286	10	and	and	CCONJ
bracis-33605	286	11	probability	probability	NOUN
bracis-33605	286	12	-	-	PUNCT
bracis-33605	286	13	to	to	ADP
bracis-33605	286	14	-	-	PUNCT
bracis-33605	286	15	goal	goal	NOUN
bracis-33605	286	16	in	in	ADP
bracis-33605	286	17	stochastic	stochastic	ADJ
bracis-33605	286	18	planning	planning	NOUN
bracis-33605	286	19	based	base	VERB
bracis-33605	286	20	on	on	ADP
bracis-33605	286	21	expected	expect	VERB
bracis-33605	286	22	utility	utility	NOUN
bracis-33605	286	23	theory	theory	NOUN
bracis-33605	286	24	.	.	PUNCT
bracis-33605	287	1	artif	artif	INTJ
bracis-33605	287	2	.	.	PUNCT
bracis-33605	288	1	intell	intell	PROPN
bracis-33605	288	2	.	.	PUNCT
bracis-33605	289	1	316(c	316(c	NUM
bracis-33605	289	2	)	)	PUNCT
bracis-33605	289	3	(	(	PUNCT
bracis-33605	289	4	2023	2023	NUM
bracis-33605	289	5	)	)	PUNCT
bracis-33605	289	6	google	google	NOUN
bracis-33605	289	7	scholar	scholar	NOUN
bracis-33605	289	8	  	  	SPACE
bracis-33605	289	9	faycal	faycal	ADJ
bracis-33605	289	10	,	,	PUNCT
bracis-33605	289	11	t.	t.	PROPN
bracis-33605	289	12	,	,	PUNCT
bracis-33605	289	13	zito	zito	PROPN
bracis-33605	289	14	,	,	PUNCT
bracis-33605	289	15	c.	c.	PROPN
bracis-33605	289	16	:	:	PUNCT
bracis-33605	289	17	dyna	dyna	PROPN
bracis-33605	289	18	-	-	PUNCT
bracis-33605	289	19	t	t	PROPN
bracis-33605	289	20	:	:	PUNCT
bracis-33605	289	21	dyna	dyna	NOUN
bracis-33605	289	22	-	-	PUNCT
bracis-33605	289	23	q	q	ADJ
bracis-33605	289	24	and	and	CCONJ
bracis-33605	289	25	upper	upper	ADJ
bracis-33605	289	26	confidence	confidence	NOUN
bracis-33605	289	27	bounds	bound	NOUN
bracis-33605	289	28	applied	apply	VERB
bracis-33605	289	29	to	to	ADP
bracis-33605	289	30	trees	tree	NOUN
bracis-33605	289	31	.	.	PUNCT
bracis-33605	290	1	arxiv	arxiv	PROPN
bracis-33605	290	2	abs/2201.04502	abs/2201.04502	PROPN
bracis-33605	290	3	(	(	PUNCT
bracis-33605	290	4	2022	2022	NUM
bracis-33605	290	5	)	)	PUNCT
bracis-33605	290	6	google	google	PROPN
bracis-33605	290	7	scholar	scholar	NOUN
bracis-33605	290	8	  	  	SPACE
bracis-33605	290	9	freire	freire	NOUN
bracis-33605	290	10	,	,	PUNCT
bracis-33605	290	11	v.	v.	PROPN
bracis-33605	290	12	,	,	PUNCT
bracis-33605	290	13	delgado	delgado	PROPN
bracis-33605	290	14	,	,	PUNCT
bracis-33605	290	15	k.v	k.v	PROPN
bracis-33605	290	16	.	.	PROPN
bracis-33605	290	17	,	,	PUNCT
bracis-33605	290	18	reis	reis	NOUN
bracis-33605	290	19	,	,	PUNCT
bracis-33605	290	20	w.a.s	w.a.s	PROPN
bracis-33605	290	21	.	.	PROPN
bracis-33605	290	22	:	:	PUNCT
bracis-33605	291	1	an	an	DET
bracis-33605	291	2	exact	exact	ADJ
bracis-33605	291	3	algorithm	algorithm	NOUN
bracis-33605	291	4	to	to	PART
bracis-33605	291	5	make	make	VERB
bracis-33605	291	6	a	a	DET
bracis-33605	291	7	trade	trade	NOUN
bracis-33605	291	8	-	-	PUNCT
bracis-33605	291	9	off	off	NOUN
bracis-33605	291	10	between	between	ADP
bracis-33605	291	11	cost	cost	NOUN
bracis-33605	291	12	and	and	CCONJ
bracis-33605	291	13	probability	probability	NOUN
bracis-33605	291	14	in	in	ADP
bracis-33605	291	15	ssps	ssps	NOUN
bracis-33605	291	16	.	.	PUNCT
bracis-33605	292	1	in	in	ADP
bracis-33605	292	2	:	:	PUNCT
bracis-33605	292	3	international	international	ADJ
bracis-33605	292	4	conference	conference	NOUN
bracis-33605	292	5	on	on	ADP
bracis-33605	292	6	automated	automate	VERB
bracis-33605	292	7	planning	planning	NOUN
bracis-33605	292	8	and	and	CCONJ
bracis-33605	292	9	scheduling	scheduling	NOUN
bracis-33605	292	10	2019	2019	NUM
bracis-33605	292	11	,	,	PUNCT
bracis-33605	292	12	vol	vol	NOUN
bracis-33605	292	13	.	.	PUNCT
bracis-33605	292	14	 	 	SPACE
bracis-33605	292	15	29	29	NUM
bracis-33605	292	16	,	,	PUNCT
bracis-33605	292	17	pp	pp	ADJ
bracis-33605	292	18	.	.	PUNCT
bracis-33605	293	1	146–154	146–154	NUM
bracis-33605	293	2	(	(	PUNCT
bracis-33605	293	3	2019	2019	NUM
bracis-33605	293	4	)	)	PUNCT
bracis-33605	293	5	google	google	PROPN
bracis-33605	293	6	scholar	scholar	NOUN
bracis-33605	293	7	  	  	SPACE
bracis-33605	293	8	freire	freire	NOUN
bracis-33605	293	9	,	,	PUNCT
bracis-33605	293	10	v.	v.	PROPN
bracis-33605	293	11	,	,	PUNCT
bracis-33605	293	12	delgado	delgado	PROPN
bracis-33605	293	13	,	,	PUNCT
bracis-33605	293	14	k.v	k.v	PROPN
bracis-33605	293	15	.	.	PUNCT
bracis-33605	293	16	:	:	PUNCT
bracis-33605	294	1	gubs	gub	NOUN
bracis-33605	294	2	:	:	PUNCT
bracis-33605	294	3	a	a	DET
bracis-33605	294	4	utility	utility	NOUN
bracis-33605	294	5	-	-	PUNCT
bracis-33605	294	6	based	base	VERB
bracis-33605	294	7	semantic	semantic	NOUN
bracis-33605	294	8	for	for	ADP
bracis-33605	294	9	goal	goal	NOUN
bracis-33605	294	10	-	-	PUNCT
bracis-33605	294	11	directed	direct	VERB
bracis-33605	294	12	markov	markov	NOUN
bracis-33605	294	13	decision	decision	NOUN
bracis-33605	294	14	processes	process	NOUN
bracis-33605	294	15	.	.	PUNCT
bracis-33605	295	1	in	in	ADP
bracis-33605	295	2	:	:	PUNCT
bracis-33605	295	3	international	international	ADJ
bracis-33605	295	4	conference	conference	NOUN
bracis-33605	295	5	on	on	ADP
bracis-33605	295	6	autonomous	autonomous	ADJ
bracis-33605	295	7	agents	agent	NOUN
bracis-33605	295	8	and	and	CCONJ
bracis-33605	295	9	multiagent	multiagent	NOUN
bracis-33605	295	10	systems	system	NOUN
bracis-33605	295	11	,	,	PUNCT
bracis-33605	295	12	pp	pp	ADV
bracis-33605	295	13	.	.	PUNCT
bracis-33605	296	1	741–749	741–749	NUM
bracis-33605	296	2	(	(	PUNCT
bracis-33605	296	3	2017	2017	NUM
bracis-33605	296	4	)	)	PUNCT
bracis-33605	296	5	google	google	NOUN
bracis-33605	296	6	scholar	scholar	NOUN
bracis-33605	296	7	  	  	SPACE
bracis-33605	296	8	geibel	geibel	NOUN
bracis-33605	296	9	,	,	PUNCT
bracis-33605	296	10	p.	p.	NOUN
bracis-33605	296	11	,	,	PUNCT
bracis-33605	296	12	wysotzki	wysotzki	ADV
bracis-33605	296	13	,	,	PUNCT
bracis-33605	296	14	f.	f.	NOUN
bracis-33605	296	15	:	:	PUNCT
bracis-33605	296	16	risk	risk	NOUN
bracis-33605	296	17	-	-	PUNCT
bracis-33605	296	18	sensitive	sensitive	ADJ
bracis-33605	296	19	reinforcement	reinforcement	NOUN
bracis-33605	296	20	learning	learning	NOUN
bracis-33605	296	21	applied	apply	VERB
bracis-33605	296	22	to	to	PART
bracis-33605	296	23	control	control	VERB
bracis-33605	296	24	under	under	ADP
bracis-33605	296	25	constraints	constraint	NOUN
bracis-33605	296	26	.	.	PUNCT
bracis-33605	297	1	j.	j.	PROPN
bracis-33605	297	2	artif	artif	PROPN
bracis-33605	297	3	.	.	PUNCT
bracis-33605	298	1	intell	intell	PROPN
bracis-33605	298	2	.	.	PUNCT
bracis-33605	299	1	res	re	NOUN
bracis-33605	299	2	.	.	PROPN
bracis-33605	300	1	24	24	NUM
bracis-33605	300	2	,	,	PUNCT
bracis-33605	300	3	81–108	81–108	NUM
bracis-33605	300	4	(	(	PUNCT
bracis-33605	300	5	2005	2005	NUM
bracis-33605	300	6	)	)	PUNCT
bracis-33605	300	7	article	article	NOUN
bracis-33605	300	8	  	  	SPACE
bracis-33605	300	9	math	math	NOUN
bracis-33605	300	10	  	  	SPACE
bracis-33605	300	11	google	google	PROPN
bracis-33605	300	12	scholar	scholar	NOUN
bracis-33605	300	13	  	  	SPACE
bracis-33605	300	14	kolobov	kolobov	PROPN
bracis-33605	300	15	,	,	PUNCT
bracis-33605	300	16	a.	a.	NOUN
bracis-33605	300	17	,	,	PUNCT
bracis-33605	300	18	mausam	mausam	PROPN
bracis-33605	300	19	,	,	PUNCT
bracis-33605	300	20	weld	weld	NOUN
bracis-33605	300	21	,	,	PUNCT
bracis-33605	300	22	d.s	d.s	PROPN
bracis-33605	300	23	.	.	PROPN
bracis-33605	300	24	:	:	PUNCT
bracis-33605	300	25	a	a	DET
bracis-33605	300	26	theory	theory	NOUN
bracis-33605	300	27	of	of	ADP
bracis-33605	300	28	goal	goal	NOUN
bracis-33605	300	29	-	-	PUNCT
bracis-33605	300	30	oriented	orient	VERB
bracis-33605	300	31	mdps	mdps	NOUN
bracis-33605	300	32	with	with	ADP
bracis-33605	300	33	dead	dead	ADJ
bracis-33605	300	34	ends	end	NOUN
bracis-33605	300	35	.	.	PUNCT
bracis-33605	301	1	in	in	ADP
bracis-33605	301	2	:	:	PUNCT
bracis-33605	301	3	conference	conference	NOUN
bracis-33605	301	4	on	on	ADP
bracis-33605	301	5	uncertainty	uncertainty	NOUN
bracis-33605	301	6	in	in	ADP
bracis-33605	301	7	artificial	artificial	ADJ
bracis-33605	301	8	intelligence	intelligence	NOUN
bracis-33605	301	9	,	,	PUNCT
bracis-33605	301	10	pp	pp	ADP
bracis-33605	301	11	.	.	PUNCT
bracis-33605	302	1	438–447	438–447	NUM
bracis-33605	302	2	.	.	PUNCT
bracis-33605	303	1	auai	auai	PROPN
bracis-33605	303	2	press	press	PROPN
bracis-33605	303	3	,	,	PUNCT
bracis-33605	303	4	arlington	arlington	PROPN
bracis-33605	303	5	,	,	PUNCT
bracis-33605	303	6	virginia	virginia	PROPN
bracis-33605	303	7	,	,	PUNCT
bracis-33605	303	8	usa	usa	PROPN
bracis-33605	303	9	(	(	PUNCT
bracis-33605	303	10	2012	2012	NUM
bracis-33605	303	11	)	)	PUNCT
bracis-33605	303	12	google	google	PROPN
bracis-33605	303	13	scholar	scholar	NOUN
bracis-33605	303	14	  	  	SPACE
bracis-33605	303	15	kolobov	kolobov	PROPN
bracis-33605	303	16	,	,	PUNCT
bracis-33605	303	17	a.	a.	NOUN
bracis-33605	303	18	,	,	PUNCT
bracis-33605	303	19	mausam	mausam	PROPN
bracis-33605	303	20	,	,	PUNCT
bracis-33605	303	21	m.	m.	NOUN
bracis-33605	303	22	,	,	PUNCT
bracis-33605	303	23	weld	weld	NOUN
bracis-33605	303	24	,	,	PUNCT
bracis-33605	303	25	d.	d.	PROPN
bracis-33605	303	26	,	,	PUNCT
bracis-33605	303	27	geffner	geffner	PROPN
bracis-33605	303	28	,	,	PUNCT
bracis-33605	303	29	h.	h.	PROPN
bracis-33605	303	30	:	:	PUNCT
bracis-33605	303	31	heuristic	heuristic	ADJ
bracis-33605	303	32	search	search	NOUN
bracis-33605	303	33	for	for	ADP
bracis-33605	303	34	generalized	generalized	ADJ
bracis-33605	303	35	stochastic	stochastic	ADJ
bracis-33605	303	36	shortest	short	ADJ
bracis-33605	303	37	path	path	NOUN
bracis-33605	303	38	mdps	mdps	NOUN
bracis-33605	303	39	.	.	PUNCT
bracis-33605	304	1	in	in	ADP
bracis-33605	304	2	:	:	PUNCT
bracis-33605	304	3	international	international	ADJ
bracis-33605	304	4	conference	conference	NOUN
bracis-33605	304	5	on	on	ADP
bracis-33605	304	6	automated	automate	VERB
bracis-33605	304	7	planning	planning	NOUN
bracis-33605	304	8	and	and	CCONJ
bracis-33605	304	9	scheduling	scheduling	NOUN
bracis-33605	304	10	2011	2011	NUM
bracis-33605	304	11	,	,	PUNCT
bracis-33605	304	12	vol	vol	NOUN
bracis-33605	304	13	.	.	PUNCT
bracis-33605	304	14	 	 	SPACE
bracis-33605	304	15	21	21	NUM
bracis-33605	304	16	,	,	PUNCT
bracis-33605	304	17	pp	pp	ADJ
bracis-33605	304	18	.	.	PUNCT
bracis-33605	305	1	130–137	130–137	NUM
bracis-33605	305	2	(	(	PUNCT
bracis-33605	305	3	2011	2011	NUM
bracis-33605	305	4	)	)	PUNCT
bracis-33605	305	5	google	google	PROPN
bracis-33605	305	6	scholar	scholar	NOUN
bracis-33605	305	7	  	  	SPACE
bracis-33605	305	8	park	park	NOUN
bracis-33605	305	9	,	,	PUNCT
bracis-33605	305	10	i.w	i.w	PROPN
bracis-33605	305	11	.	.	PROPN
bracis-33605	305	12	,	,	PUNCT
bracis-33605	305	13	kim	kim	PROPN
bracis-33605	305	14	,	,	PUNCT
bracis-33605	305	15	j.h	j.h	PROPN
bracis-33605	305	16	.	.	PROPN
bracis-33605	305	17	,	,	PUNCT
bracis-33605	305	18	park	park	PROPN
bracis-33605	305	19	,	,	PUNCT
bracis-33605	305	20	k.h	k.h	PROPN
bracis-33605	305	21	.	.	PROPN
bracis-33605	305	22	:	:	PUNCT
bracis-33605	305	23	accelerated	accelerate	VERB
bracis-33605	305	24	q	q	ADV
bracis-33605	305	25	-	-	PUNCT
bracis-33605	305	26	learning	learning	NOUN
bracis-33605	305	27	for	for	ADP
bracis-33605	305	28	fail	fail	ADJ
bracis-33605	305	29	state	state	NOUN
bracis-33605	305	30	and	and	CCONJ
bracis-33605	305	31	action	action	NOUN
bracis-33605	305	32	spaces	space	NOUN
bracis-33605	305	33	.	.	PUNCT
bracis-33605	306	1	in	in	ADP
bracis-33605	306	2	:	:	PUNCT
bracis-33605	306	3	2008	2008	NUM
bracis-33605	306	4	ieee	ieee	PROPN
bracis-33605	306	5	international	international	ADJ
bracis-33605	306	6	conference	conference	NOUN
bracis-33605	306	7	on	on	ADP
bracis-33605	306	8	systems	system	NOUN
bracis-33605	306	9	,	,	PUNCT
bracis-33605	306	10	man	man	NOUN
bracis-33605	306	11	and	and	CCONJ
bracis-33605	306	12	cybernetics	cybernetic	NOUN
bracis-33605	306	13	,	,	PUNCT
bracis-33605	306	14	pp	pp	X
bracis-33605	306	15	.	.	PUNCT
bracis-33605	307	1	763–767	763–767	NUM
bracis-33605	307	2	(	(	PUNCT
bracis-33605	307	3	2008	2008	NUM
bracis-33605	307	4	)	)	PUNCT
bracis-33605	307	5	google	google	PROPN
bracis-33605	307	6	scholar	scholar	NOUN
bracis-33605	307	7	  	  	SPACE
bracis-33605	307	8	patek	patek	NOUN
bracis-33605	307	9	,	,	PUNCT
bracis-33605	307	10	s.d	s.d	PROPN
bracis-33605	307	11	.	.	PUNCT
bracis-33605	307	12	:	:	PUNCT
bracis-33605	307	13	on	on	ADP
bracis-33605	307	14	terminating	terminate	VERB
bracis-33605	307	15	markov	markov	NOUN
bracis-33605	307	16	decision	decision	NOUN
bracis-33605	307	17	processes	process	NOUN
bracis-33605	307	18	with	with	ADP
bracis-33605	307	19	a	a	DET
bracis-33605	307	20	risk	risk	NOUN
bracis-33605	307	21	-	-	PUNCT
bracis-33605	307	22	averse	averse	ADJ
bracis-33605	307	23	objective	objective	ADJ
bracis-33605	307	24	function	function	NOUN
bracis-33605	307	25	.	.	PUNCT
bracis-33605	308	1	automatica	automatica	PROPN
bracis-33605	308	2	37(9	37(9	NUM
bracis-33605	308	3	)	)	PUNCT
bracis-33605	308	4	,	,	PUNCT
bracis-33605	308	5	1379–1386	1379–1386	NUM
bracis-33605	308	6	(	(	PUNCT
bracis-33605	308	7	2001	2001	NUM
bracis-33605	308	8	)	)	PUNCT
bracis-33605	308	9	article	article	NOUN
bracis-33605	308	10	  	  	SPACE
bracis-33605	308	11	math	math	NOUN
bracis-33605	308	12	  	  	SPACE
bracis-33605	308	13	google	google	PROPN
bracis-33605	308	14	scholar	scholar	NOUN
bracis-33605	308	15	  	  	SPACE
bracis-33605	308	16	sanner	sanner	NOUN
bracis-33605	308	17	,	,	PUNCT
bracis-33605	308	18	s.	s.	PROPN
bracis-33605	308	19	,	,	PUNCT
bracis-33605	308	20	yoon	yoon	PROPN
bracis-33605	308	21	,	,	PUNCT
bracis-33605	308	22	s.	s.	PROPN
bracis-33605	308	23	:	:	PUNCT
bracis-33605	308	24	ippc	ippc	VERB
bracis-33605	308	25	results	result	VERB
bracis-33605	308	26	presentation	presentation	NOUN
bracis-33605	308	27	in	in	ADP
bracis-33605	308	28	:	:	PUNCT
bracis-33605	308	29	international	international	ADJ
bracis-33605	308	30	conference	conference	NOUN
bracis-33605	308	31	on	on	ADP
bracis-33605	308	32	automated	automate	VERB
bracis-33605	308	33	planning	planning	NOUN
bracis-33605	308	34	and	and	CCONJ
bracis-33605	308	35	scheduling	scheduling	NOUN
bracis-33605	308	36	(	(	PUNCT
bracis-33605	308	37	2011	2011	NUM
bracis-33605	308	38	)	)	PUNCT
bracis-33605	308	39	.	.	PUNCT
bracis-33605	309	1	http://users.cecs.anu.edu.au/ssanner/ippc_2011/ippc_2011_presentation.pdf	http://users.cecs.anu.edu.au/ssanner/ippc_2011/ippc_2011_presentation.pdf	PROPN
bracis-33605	309	2	.	.	PUNCT
bracis-33605	310	1	accessed	access	VERB
bracis-33605	310	2	aug	aug	PROPN
bracis-33605	310	3	2024	2024	NUM
bracis-33605	310	4	sutton	sutton	PROPN
bracis-33605	310	5	,	,	PUNCT
bracis-33605	310	6	r.s	r.s	PROPN
bracis-33605	310	7	.	.	PROPN
bracis-33605	310	8	:	:	PUNCT
bracis-33605	311	1	integrated	integrate	VERB
bracis-33605	311	2	architectures	architecture	NOUN
bracis-33605	311	3	for	for	ADP
bracis-33605	311	4	learning	learning	NOUN
bracis-33605	311	5	,	,	PUNCT
bracis-33605	311	6	planning	planning	NOUN
bracis-33605	311	7	,	,	PUNCT
bracis-33605	311	8	and	and	CCONJ
bracis-33605	311	9	reacting	react	VERB
bracis-33605	311	10	based	base	VERB
bracis-33605	311	11	on	on	ADP
bracis-33605	311	12	approximating	approximate	VERB
bracis-33605	311	13	dynamic	dynamic	ADJ
bracis-33605	311	14	programming	programming	NOUN
bracis-33605	311	15	.	.	PUNCT
bracis-33605	312	1	in	in	ADP
bracis-33605	312	2	:	:	PUNCT
bracis-33605	312	3	international	international	ADJ
bracis-33605	312	4	conference	conference	NOUN
bracis-33605	312	5	on	on	ADP
bracis-33605	312	6	machine	machine	NOUN
bracis-33605	312	7	learning	learning	NOUN
bracis-33605	312	8	,	,	PUNCT
bracis-33605	312	9	pp	pp	ADJ
bracis-33605	312	10	.	.	PUNCT
bracis-33605	313	1	216–224	216–224	NUM
bracis-33605	313	2	.	.	PUNCT
bracis-33605	314	1	morgan	morgan	PROPN
bracis-33605	314	2	kaufmann	kaufmann	PROPN
bracis-33605	314	3	(	(	PUNCT
bracis-33605	314	4	1990	1990	NUM
bracis-33605	314	5	)	)	PUNCT
bracis-33605	314	6	google	google	NOUN
bracis-33605	314	7	scholar	scholar	NOUN
bracis-33605	314	8	  	  	SPACE
bracis-33605	314	9	tarbouriech	tarbouriech	NOUN
bracis-33605	314	10	,	,	PUNCT
bracis-33605	314	11	j.	j.	PROPN
bracis-33605	314	12	:	:	PUNCT
bracis-33605	314	13	goal	goal	NOUN
bracis-33605	314	14	-	-	PUNCT
bracis-33605	314	15	oriented	orient	VERB
bracis-33605	314	16	exploration	exploration	NOUN
bracis-33605	314	17	for	for	ADP
bracis-33605	314	18	reinforcement	reinforcement	NOUN
bracis-33605	314	19	learning	learning	NOUN
bracis-33605	314	20	.	.	PUNCT
bracis-33605	315	1	ph.d	ph.d	PROPN
bracis-33605	315	2	.	.	PUNCT
bracis-33605	316	1	thesis	thesis	PROPN
bracis-33605	316	2	,	,	PUNCT
bracis-33605	316	3	université	université	PROPN
bracis-33605	316	4	de	de	PROPN
bracis-33605	316	5	lille	lille	X
bracis-33605	316	6	(	(	PUNCT
bracis-33605	316	7	2022	2022	NUM
bracis-33605	316	8	)	)	PUNCT
bracis-33605	316	9	google	google	NOUN
bracis-33605	316	10	scholar	scholar	NOUN
bracis-33605	316	11	  	  	SPACE
bracis-33605	316	12	teichteil	teichteil	NOUN
bracis-33605	316	13	-	-	PUNCT
bracis-33605	316	14	konigsbuch	konigsbuch	NOUN
bracis-33605	316	15	,	,	PUNCT
bracis-33605	316	16	f.	f.	PROPN
bracis-33605	316	17	,	,	PUNCT
bracis-33605	316	18	vidal	vidal	NOUN
bracis-33605	316	19	,	,	PUNCT
bracis-33605	316	20	v.	v.	PROPN
bracis-33605	316	21	,	,	PUNCT
bracis-33605	316	22	infantes	infante	NOUN
bracis-33605	316	23	,	,	PUNCT
bracis-33605	316	24	g.	g.	NOUN
bracis-33605	316	25	:	:	PUNCT
bracis-33605	316	26	extending	extend	VERB
bracis-33605	316	27	classical	classical	ADJ
bracis-33605	316	28	planning	planning	NOUN
bracis-33605	316	29	heuristics	heuristic	NOUN
bracis-33605	316	30	to	to	PART
bracis-33605	316	31	probabilistic	probabilistic	ADJ
bracis-33605	316	32	planning	planning	NOUN
bracis-33605	316	33	with	with	ADP
bracis-33605	316	34	dead	dead	ADJ
bracis-33605	316	35	-	-	PUNCT
bracis-33605	316	36	ends	end	NOUN
bracis-33605	316	37	.	.	PUNCT
bracis-33605	317	1	in	in	ADP
bracis-33605	317	2	:	:	PUNCT
bracis-33605	317	3	aaai	aaai	PROPN
bracis-33605	317	4	conference	conference	NOUN
bracis-33605	317	5	on	on	ADP
bracis-33605	317	6	artificial	artificial	ADJ
bracis-33605	317	7	intelligence	intelligence	NOUN
bracis-33605	317	8	,	,	PUNCT
bracis-33605	317	9	pp	pp	ADJ
bracis-33605	317	10	.	.	PUNCT
bracis-33605	317	11	1017–1022	1017–1022	NUM
bracis-33605	317	12	.	.	PUNCT
bracis-33605	318	1	aaai	aaai	PROPN
bracis-33605	318	2	press	press	PROPN
bracis-33605	318	3	(	(	PUNCT
bracis-33605	318	4	2011	2011	NUM
bracis-33605	318	5	)	)	PUNCT
bracis-33605	318	6	google	google	PROPN
bracis-33605	318	7	scholar	scholar	NOUN
bracis-33605	318	8	  	  	SPACE
bracis-33605	318	9	teichteil	teichteil	NOUN
bracis-33605	318	10	-	-	PUNCT
bracis-33605	318	11	königsbuch	königsbuch	NOUN
bracis-33605	318	12	,	,	PUNCT
bracis-33605	318	13	f.	f.	NOUN
bracis-33605	318	14	:	:	PUNCT
bracis-33605	318	15	stochastic	stochastic	ADJ
bracis-33605	318	16	safest	safe	ADJ
bracis-33605	318	17	and	and	CCONJ
bracis-33605	318	18	shortest	short	ADJ
bracis-33605	318	19	path	path	NOUN
bracis-33605	318	20	problems	problem	NOUN
bracis-33605	318	21	.	.	PUNCT
bracis-33605	319	1	in	in	ADP
bracis-33605	319	2	:	:	PUNCT
bracis-33605	319	3	aaai	aaai	PROPN
bracis-33605	319	4	conference	conference	NOUN
bracis-33605	319	5	on	on	ADP
bracis-33605	319	6	artificial	artificial	ADJ
bracis-33605	319	7	intelligence	intelligence	NOUN
bracis-33605	319	8	,	,	PUNCT
bracis-33605	319	9	vol	vol	NOUN
bracis-33605	319	10	.	.	PUNCT
bracis-33605	319	11	 	 	SPACE
bracis-33605	319	12	26	26	NUM
bracis-33605	319	13	,	,	PUNCT
bracis-33605	319	14	pp	pp	ADJ
bracis-33605	319	15	.	.	PUNCT
bracis-33605	320	1	1825–1831	1825–1831	NUM
bracis-33605	320	2	(	(	PUNCT
bracis-33605	320	3	2012	2012	NUM
bracis-33605	320	4	)	)	PUNCT
bracis-33605	320	5	google	google	PROPN
bracis-33605	320	6	scholar	scholar	NOUN
bracis-33605	320	7	  	  	SPACE
bracis-33605	320	8	trevizan	trevizan	NOUN
bracis-33605	320	9	,	,	PUNCT
bracis-33605	320	10	f.w	f.w	PROPN
bracis-33605	320	11	.	.	PROPN
bracis-33605	320	12	,	,	PUNCT
bracis-33605	320	13	teichteil	teichteil	NOUN
bracis-33605	320	14	-	-	PUNCT
bracis-33605	320	15	königsbuch	königsbuch	PROPN
bracis-33605	320	16	,	,	PUNCT
bracis-33605	320	17	f.	f.	PROPN
bracis-33605	320	18	,	,	PUNCT
bracis-33605	320	19	thiébaux	thiébaux	PROPN
bracis-33605	320	20	,	,	PUNCT
bracis-33605	320	21	s.	s.	PROPN
bracis-33605	320	22	:	:	PUNCT
bracis-33605	320	23	efficient	efficient	ADJ
bracis-33605	320	24	solutions	solution	NOUN
bracis-33605	320	25	for	for	ADP
bracis-33605	320	26	stochastic	stochastic	ADJ
bracis-33605	320	27	shortest	short	ADJ
bracis-33605	320	28	path	path	NOUN
bracis-33605	320	29	problems	problem	NOUN
bracis-33605	320	30	with	with	ADP
bracis-33605	320	31	dead	dead	ADJ
bracis-33605	320	32	ends	end	NOUN
bracis-33605	320	33	.	.	PUNCT
bracis-33605	321	1	in	in	ADP
bracis-33605	321	2	:	:	PUNCT
bracis-33605	321	3	conference	conference	NOUN
bracis-33605	321	4	on	on	ADP
bracis-33605	321	5	uncertainty	uncertainty	NOUN
bracis-33605	321	6	in	in	ADP
bracis-33605	321	7	artificial	artificial	ADJ
bracis-33605	321	8	intelligence	intelligence	NOUN
bracis-33605	321	9	(	(	PUNCT
bracis-33605	321	10	2017	2017	NUM
bracis-33605	321	11	)	)	PUNCT
bracis-33605	321	12	google	google	PROPN
bracis-33605	321	13	scholar	scholar	NOUN
bracis-33605	321	14	  	  	SPACE
bracis-33605	321	15	watkins	watkin	NOUN
bracis-33605	321	16	,	,	PUNCT
bracis-33605	321	17	c.j.c.h	c.j.c.h	PROPN
bracis-33605	321	18	.	.	PROPN
bracis-33605	321	19	,	,	PUNCT
bracis-33605	321	20	dayan	dayan	PROPN
bracis-33605	321	21	,	,	PUNCT
bracis-33605	321	22	p.	p.	NOUN
bracis-33605	321	23	:	:	PUNCT
bracis-33605	321	24	technical	technical	ADJ
bracis-33605	321	25	note	note	NOUN
bracis-33605	321	26	.	.	PUNCT
bracis-33605	322	1	q	q	X
bracis-33605	322	2	-	-	PUNCT
bracis-33605	322	3	learning	learning	NOUN
bracis-33605	322	4	.	.	PUNCT
bracis-33605	323	1	mach	mach	NOUN
bracis-33605	323	2	.	.	PUNCT
bracis-33605	324	1	learn	learn	VERB
bracis-33605	324	2	.	.	PUNCT
bracis-33605	325	1	8(7	8(7	NUM
bracis-33605	325	2	)	)	PUNCT
bracis-33605	325	3	,	,	PUNCT
bracis-33605	325	4	279–292	279–292	NUM
bracis-33605	325	5	(	(	PUNCT
bracis-33605	325	6	1992	1992	NUM
bracis-33605	325	7	)	)	PUNCT
bracis-33605	325	8	article	article	NOUN
bracis-33605	325	9	  	  	SPACE
bracis-33605	325	10	math	math	NOUN
bracis-33605	325	11	  	  	SPACE
bracis-33605	325	12	google	google	PROPN
bracis-33605	325	13	scholar	scholar	NOUN
bracis-33605	325	14	  	  	SPACE
bracis-33605	325	15	download	download	NOUN
bracis-33605	325	16	references	reference	NOUN
bracis-33605	325	17	acknowledgments	acknowledgment	NOUN
bracis-33605	325	18	this	this	DET
bracis-33605	325	19	study	study	NOUN
bracis-33605	325	20	was	be	AUX
bracis-33605	325	21	supported	support	VERB
bracis-33605	325	22	in	in	ADP
bracis-33605	325	23	part	part	NOUN
bracis-33605	325	24	by	by	ADP
bracis-33605	325	25	the	the	DET
bracis-33605	325	26	conselho	conselho	PROPN
bracis-33605	325	27	nacional	nacional	PROPN
bracis-33605	325	28	de	de	X
bracis-33605	325	29	desenvolvimento	desenvolvimento	PROPN
bracis-33605	325	30	científico	científico	PROPN
bracis-33605	325	31	e	e	NOUN
bracis-33605	325	32	tecnológico	tecnológico	PROPN
bracis-33605	325	33	(	(	PUNCT
bracis-33605	325	34	cnpq	cnpq	NOUN
bracis-33605	325	35	)	)	PUNCT
bracis-33605	325	36	–	–	PUNCT
bracis-33605	325	37	finance	finance	NOUN
bracis-33605	325	38	code	code	NOUN
bracis-33605	325	39	001	001	NOUN
bracis-33605	325	40	,	,	PUNCT
bracis-33605	325	41	by	by	ADP
bracis-33605	325	42	the	the	DET
bracis-33605	325	43	são	são	PROPN
bracis-33605	325	44	paulo	paulo	PROPN
bracis-33605	325	45	research	research	PROPN
bracis-33605	325	46	foundation	foundation	PROPN
bracis-33605	325	47	(	(	PUNCT
bracis-33605	325	48	fapesp	fapesp	NOUN
bracis-33605	325	49	)	)	PUNCT
bracis-33605	325	50	grant	grant	NOUN
bracis-33605	325	51	\(\#\)2018/11236	\(\#\)2018/11236	NOUN
bracis-33605	325	52	-	-	SYM
bracis-33605	325	53	9	9	NUM
bracis-33605	325	54	and	and	CCONJ
bracis-33605	325	55	the	the	DET
bracis-33605	325	56	center	center	NOUN
bracis-33605	325	57	for	for	ADP
bracis-33605	325	58	artificial	artificial	ADJ
bracis-33605	325	59	intelligence	intelligence	NOUN
bracis-33605	325	60	(	(	PUNCT
bracis-33605	325	61	c4ai	c4ai	NOUN
bracis-33605	325	62	-	-	PUNCT
bracis-33605	325	63	usp	usp	NOUN
bracis-33605	325	64	)	)	PUNCT
bracis-33605	325	65	,	,	PUNCT
bracis-33605	325	66	with	with	ADP
bracis-33605	325	67	support	support	NOUN
bracis-33605	325	68	 	 	SPACE
bracis-33605	325	69	by	by	ADP
bracis-33605	325	70	fapesp	fapesp	NOUN
bracis-33605	325	71	(	(	PUNCT
bracis-33605	325	72	grant	grant	VERB
bracis-33605	325	73	#	#	SYM
bracis-33605	325	74	2019/07665	2019/07665	NUM
bracis-33605	325	75	-	-	SYM
bracis-33605	325	76	4	4	NUM
bracis-33605	325	77	)	)	PUNCT
bracis-33605	325	78	and	and	CCONJ
bracis-33605	325	79	by	by	ADP
bracis-33605	325	80	the	the	DET
bracis-33605	325	81	ibm	ibm	PROPN
bracis-33605	325	82	corporation	corporation	NOUN
bracis-33605	325	83	.	.	PUNCT
bracis-33605	326	1	author	author	NOUN
bracis-33605	326	2	information	information	NOUN
bracis-33605	326	3	authors	author	NOUN
bracis-33605	326	4	and	and	CCONJ
bracis-33605	326	5	affiliations	affiliation	NOUN
bracis-33605	326	6	university	university	PROPN
bracis-33605	326	7	of	of	ADP
bracis-33605	326	8	são	são	PROPN
bracis-33605	326	9	paulo	paulo	PROPN
bracis-33605	326	10	,	,	PUNCT
bracis-33605	326	11	cidade	cidade	NOUN
bracis-33605	326	12	universitária	universitária	PROPN
bracis-33605	326	13	,	,	PUNCT
bracis-33605	326	14	são	são	PROPN
bracis-33605	326	15	paulo	paulo	PROPN
bracis-33605	326	16	,	,	PUNCT
bracis-33605	326	17	sp	sp	PROPN
bracis-33605	326	18	,	,	PUNCT
bracis-33605	326	19	brazil	brazil	PROPN
bracis-33605	326	20	christian	christian	PROPN
bracis-33605	326	21	delgado	delgado	VERB
bracis-33605	326	22	polar	polar	ADJ
bracis-33605	326	23	,	,	PUNCT
bracis-33605	326	24	 	 	SPACE
bracis-33605	326	25	karina	karina	PROPN
bracis-33605	326	26	valdivia	valdivia	PROPN
bracis-33605	326	27	delgado	delgado	PROPN
bracis-33605	326	28	 	 	SPACE
bracis-33605	326	29	&	&	CCONJ
bracis-33605	326	30	 	 	SPACE
bracis-33605	326	31	valdinei	valdinei	PROPN
bracis-33605	326	32	freire	freire	PROPN
bracis-33605	326	33	school	school	NOUN
bracis-33605	326	34	of	of	ADP
bracis-33605	326	35	computer	computer	NOUN
bracis-33605	326	36	science	science	NOUN
bracis-33605	326	37	,	,	PUNCT
bracis-33605	326	38	san	san	PROPN
bracis-33605	326	39	pablo	pablo	PROPN
bracis-33605	326	40	catholic	catholic	PROPN
bracis-33605	326	41	university	university	PROPN
bracis-33605	326	42	,	,	PUNCT
bracis-33605	326	43	arequipa	arequipa	NOUN
bracis-33605	326	44	,	,	PUNCT
bracis-33605	326	45	peru	peru	PROPN
bracis-33605	326	46	christian	christian	PROPN
bracis-33605	326	47	delgado	delgado	VERB
bracis-33605	326	48	polar	polar	ADJ
bracis-33605	326	49	authors	author	NOUN
bracis-33605	326	50	christian	christian	PROPN
bracis-33605	326	51	delgado	delgado	AUX
bracis-33605	326	52	polarview	polarview	VERB
bracis-33605	326	53	author	author	NOUN
bracis-33605	326	54	publications	publication	NOUN
bracis-33605	326	55	search	search	NOUN
bracis-33605	326	56	author	author	NOUN
bracis-33605	326	57	on	on	ADP
bracis-33605	326	58	:	:	PUNCT
bracis-33605	326	59	pubmed	pubmed	PROPN
bracis-33605	326	60	 	 	SPACE
bracis-33605	326	61	google	google	PROPN
bracis-33605	326	62	scholar	scholar	PROPN
bracis-33605	326	63	karina	karina	PROPN
bracis-33605	326	64	valdivia	valdivia	PROPN
bracis-33605	326	65	delgadoview	delgadoview	PROPN
bracis-33605	326	66	author	author	NOUN
bracis-33605	326	67	publications	publication	NOUN
bracis-33605	326	68	search	search	NOUN
bracis-33605	326	69	author	author	NOUN
bracis-33605	326	70	on	on	ADP
bracis-33605	326	71	:	:	PUNCT
bracis-33605	326	72	pubmed	pubmed	PROPN
bracis-33605	326	73	 	 	SPACE
bracis-33605	326	74	google	google	PROPN
bracis-33605	326	75	scholar	scholar	NOUN
bracis-33605	326	76	valdinei	valdinei	NOUN
bracis-33605	326	77	freireview	freireview	NOUN
bracis-33605	326	78	author	author	NOUN
bracis-33605	326	79	publications	publication	NOUN
bracis-33605	326	80	search	search	NOUN
bracis-33605	326	81	author	author	NOUN
bracis-33605	326	82	on	on	ADP
bracis-33605	326	83	:	:	PUNCT
bracis-33605	326	84	pubmed	pubmed	PROPN
bracis-33605	326	85	 	 	SPACE
bracis-33605	326	86	google	google	PROPN
bracis-33605	326	87	scholar	scholar	NOUN
bracis-33605	326	88	corresponding	correspond	VERB
bracis-33605	326	89	author	author	NOUN
bracis-33605	326	90	correspondence	correspondence	NOUN
bracis-33605	326	91	to	to	ADP
bracis-33605	326	92	christian	christian	PROPN
bracis-33605	326	93	delgado	delgado	VERB
bracis-33605	326	94	polar	polar	ADJ
bracis-33605	326	95	.	.	PUNCT
bracis-33605	327	1	editor	editor	NOUN
bracis-33605	327	2	information	information	NOUN
bracis-33605	327	3	editors	editor	NOUN
bracis-33605	327	4	and	and	CCONJ
bracis-33605	327	5	affiliations	affiliation	NOUN
bracis-33605	327	6	universidade	universidade	VERB
bracis-33605	327	7	federal	federal	PROPN
bracis-33605	327	8	fluminense	fluminense	PROPN
bracis-33605	327	9	,	,	PUNCT
bracis-33605	327	10	niterói	niterói	PROPN
bracis-33605	327	11	,	,	PUNCT
bracis-33605	327	12	brazil	brazil	PROPN
bracis-33605	327	13	aline	aline	PROPN
bracis-33605	327	14	paes	paes	PROPN
bracis-33605	327	15	instituto	instituto	PROPN
bracis-33605	327	16	tecnológico	tecnológico	PROPN
bracis-33605	327	17	de	de	PROPN
bracis-33605	327	18	aeronáutica	aeronáutica	PROPN
bracis-33605	327	19	,	,	PUNCT
bracis-33605	327	20	são	são	PROPN
bracis-33605	327	21	josé	josé	PROPN
bracis-33605	327	22	dos	dos	PROPN
bracis-33605	327	23	campos	campos	PROPN
bracis-33605	327	24	,	,	PUNCT
bracis-33605	327	25	brazil	brazil	PROPN
bracis-33605	327	26	filipe	filipe	PROPN
bracis-33605	327	27	a.	a.	PROPN
bracis-33605	327	28	n.	n.	PROPN
bracis-33605	327	29	verri	verri	PROPN
bracis-33605	327	30	rights	right	NOUN
bracis-33605	327	31	and	and	CCONJ
bracis-33605	327	32	permissions	permission	NOUN
bracis-33605	327	33	reprints	reprint	NOUN
bracis-33605	327	34	and	and	CCONJ
bracis-33605	327	35	permissions	permission	VERB
bracis-33605	327	36	copyright	copyright	NOUN
bracis-33605	327	37	information	information	NOUN
bracis-33605	327	38	©	©	ADP
bracis-33605	327	39	2025	2025	NUM
bracis-33605	327	40	the	the	DET
bracis-33605	327	41	author(s	author(s	NOUN
bracis-33605	327	42	)	)	PUNCT
bracis-33605	327	43	,	,	PUNCT
bracis-33605	327	44	under	under	ADP
bracis-33605	327	45	exclusive	exclusive	ADJ
bracis-33605	327	46	license	license	NOUN
bracis-33605	327	47	to	to	ADP
bracis-33605	327	48	springer	springer	NOUN
bracis-33605	327	49	nature	nature	PROPN
bracis-33605	327	50	switzerland	switzerland	PROPN
bracis-33605	327	51	ag	ag	PROPN
bracis-33605	327	52	about	about	ADP
bracis-33605	327	53	this	this	DET
bracis-33605	327	54	paper	paper	NOUN
bracis-33605	327	55	cite	cite	VERB
bracis-33605	327	56	this	this	DET
bracis-33605	327	57	paper	paper	NOUN
bracis-33605	327	58	polar	polar	NOUN
bracis-33605	327	59	,	,	PUNCT
bracis-33605	327	60	c.d	c.d	PROPN
bracis-33605	327	61	.	.	PROPN
bracis-33605	327	62	,	,	PUNCT
bracis-33605	327	63	delgado	delgado	PROPN
bracis-33605	327	64	,	,	PUNCT
bracis-33605	327	65	k.v	k.v	PROPN
bracis-33605	327	66	.	.	PROPN
bracis-33605	327	67	,	,	PUNCT
bracis-33605	327	68	freire	freire	PROPN
bracis-33605	327	69	,	,	PUNCT
bracis-33605	327	70	v.	v.	PROPN
bracis-33605	327	71	(	(	PUNCT
bracis-33605	327	72	2025	2025	NUM
bracis-33605	327	73	)	)	PUNCT
bracis-33605	327	74	.	.	PUNCT
bracis-33605	328	1	reinforcement	reinforcement	NOUN
bracis-33605	328	2	learning	learn	VERB
bracis-33605	328	3	with	with	ADP
bracis-33605	328	4	 	 	SPACE
bracis-33605	328	5	utility	utility	NOUN
bracis-33605	328	6	-	-	PUNCT
bracis-33605	328	7	based	base	VERB
bracis-33605	328	8	semantic	semantic	NOUN
bracis-33605	328	9	for	for	ADP
bracis-33605	328	10	 	 	SPACE
bracis-33605	328	11	goals	goal	NOUN
bracis-33605	328	12	.	.	PUNCT
bracis-33605	329	1	in	in	ADP
bracis-33605	329	2	:	:	PUNCT
bracis-33605	329	3	paes	paes	PROPN
bracis-33605	329	4	,	,	PUNCT
bracis-33605	329	5	a.	a.	PROPN
bracis-33605	329	6	,	,	PUNCT
bracis-33605	329	7	verri	verri	NOUN
bracis-33605	329	8	,	,	PUNCT
bracis-33605	329	9	f.a.n	f.a.n	ADJ
bracis-33605	329	10	.	.	PUNCT
bracis-33605	329	11	(	(	PUNCT
bracis-33605	329	12	eds	ed	NOUN
bracis-33605	329	13	)	)	PUNCT
bracis-33605	329	14	intelligent	intelligent	ADJ
bracis-33605	329	15	systems	system	NOUN
bracis-33605	329	16	.	.	PUNCT
bracis-33605	329	17	bracis	bracis	PROPN
bracis-33605	329	18	2024	2024	NUM
bracis-33605	329	19	.	.	PUNCT
bracis-33605	330	1	lecture	lecture	NOUN
bracis-33605	330	2	notes	note	NOUN
bracis-33605	330	3	in	in	ADP
bracis-33605	330	4	computer	computer	NOUN
bracis-33605	330	5	science	science	NOUN
bracis-33605	330	6	(	(	PUNCT
bracis-33605	330	7	)	)	PUNCT
bracis-33605	330	8	,	,	PUNCT
bracis-33605	330	9	vol	vol	NOUN
bracis-33605	330	10	15413	15413	NUM
bracis-33605	330	11	.	.	PUNCT
bracis-33605	331	1	springer	springer	NOUN
bracis-33605	331	2	,	,	PUNCT
bracis-33605	331	3	cham	cham	PROPN
bracis-33605	331	4	.	.	PUNCT
bracis-33605	332	1	https://doi.org/10.1007/978-3-031-79032-4_25	https://doi.org/10.1007/978-3-031-79032-4_25	PROPN
bracis-33605	332	2	download	download	NOUN
bracis-33605	332	3	citation	citation	NOUN
bracis-33605	332	4	.ris	.ris	PUNCT
bracis-33605	333	1	.enw	.enw	PROPN
bracis-33605	333	2	.bib	.bib	PUNCT
bracis-33605	334	1	doi	doi	PROPN
bracis-33605	334	2	:	:	PUNCT
bracis-33605	334	3	https://doi.org/10.1007/978-3-031-79032-4_25	https://doi.org/10.1007/978-3-031-79032-4_25	PUNCT
bracis-33605	334	4	published	publish	VERB
bracis-33605	334	5	:	:	PUNCT
bracis-33605	334	6	30	30	NUM
bracis-33605	334	7	january	january	PROPN
bracis-33605	334	8	2025	2025	NUM
bracis-33605	334	9	publisher	publisher	NOUN
bracis-33605	334	10	name	name	NOUN
bracis-33605	334	11	:	:	PUNCT
bracis-33605	334	12	springer	springer	NOUN
bracis-33605	334	13	,	,	PUNCT
bracis-33605	334	14	cham	cham	PROPN
bracis-33605	334	15	print	print	PROPN
bracis-33605	334	16	isbn	isbn	PROPN
bracis-33605	334	17	:	:	PUNCT
bracis-33605	334	18	978	978	NUM
bracis-33605	334	19	-	-	SYM
bracis-33605	334	20	3	3	NUM
bracis-33605	334	21	-	-	PUNCT
bracis-33605	334	22	031	031	NUM
bracis-33605	334	23	-	-	PUNCT
bracis-33605	334	24	79031	79031	NUM
bracis-33605	334	25	-	-	SYM
bracis-33605	334	26	7	7	NUM
bracis-33605	334	27	online	online	ADJ
bracis-33605	334	28	isbn	isbn	NOUN
bracis-33605	334	29	:	:	PUNCT
bracis-33605	334	30	978	978	NUM
bracis-33605	334	31	-	-	SYM
bracis-33605	334	32	3	3	NUM
bracis-33605	334	33	-	-	PUNCT
bracis-33605	334	34	031	031	NUM
bracis-33605	334	35	-	-	PUNCT
bracis-33605	334	36	79032	79032	NUM
bracis-33605	334	37	-	-	SYM
bracis-33605	334	38	4	4	NUM
bracis-33605	334	39	ebook	ebook	NOUN
bracis-33605	334	40	packages	package	NOUN
bracis-33605	334	41	:	:	PUNCT
bracis-33605	334	42	computer	computer	NOUN
bracis-33605	334	43	sciencecomputer	sciencecomputer	NOUN
bracis-33605	334	44	science	science	NOUN
bracis-33605	334	45	(	(	PUNCT
bracis-33605	334	46	r0	r0	NOUN
bracis-33605	334	47	)	)	PUNCT
bracis-33605	334	48	share	share	VERB
bracis-33605	334	49	this	this	DET
bracis-33605	334	50	paper	paper	NOUN
bracis-33605	334	51	anyone	anyone	PRON
bracis-33605	334	52	you	you	PRON
bracis-33605	334	53	share	share	VERB
bracis-33605	334	54	the	the	DET
bracis-33605	334	55	following	follow	VERB
bracis-33605	334	56	link	link	NOUN
bracis-33605	334	57	with	with	ADP
bracis-33605	334	58	will	will	AUX
bracis-33605	334	59	be	be	AUX
bracis-33605	334	60	able	able	ADJ
bracis-33605	334	61	to	to	PART
bracis-33605	334	62	read	read	VERB
bracis-33605	334	63	this	this	DET
bracis-33605	334	64	content	content	NOUN
bracis-33605	334	65	:	:	PUNCT
bracis-33605	334	66	get	get	VERB
bracis-33605	334	67	shareable	shareable	ADJ
bracis-33605	334	68	linksorry	linksorry	NOUN
bracis-33605	334	69	,	,	PUNCT
bracis-33605	334	70	a	a	DET
bracis-33605	334	71	shareable	shareable	ADJ
bracis-33605	334	72	link	link	NOUN
bracis-33605	334	73	is	be	AUX
bracis-33605	334	74	not	not	PART
bracis-33605	334	75	currently	currently	ADV
bracis-33605	334	76	available	available	ADJ
bracis-33605	334	77	for	for	ADP
bracis-33605	334	78	this	this	DET
bracis-33605	334	79	article	article	NOUN
bracis-33605	334	80	.	.	PUNCT
bracis-33605	335	1	copy	copy	VERB
bracis-33605	335	2	shareable	shareable	ADJ
bracis-33605	335	3	link	link	NOUN
bracis-33605	335	4	to	to	PART
bracis-33605	335	5	clipboard	clipboard	NOUN
bracis-33605	335	6	provided	provide	VERB
bracis-33605	335	7	by	by	ADP
bracis-33605	335	8	the	the	DET
bracis-33605	335	9	springer	springer	NOUN
bracis-33605	335	10	nature	nature	PROPN
bracis-33605	335	11	sharedit	sharedit	PROPN
bracis-33605	335	12	content	content	NOUN
bracis-33605	335	13	-	-	PUNCT
bracis-33605	335	14	sharing	share	VERB
bracis-33605	335	15	initiative	initiative	NOUN
bracis-33605	335	16	keywords	keyword	NOUN
bracis-33605	335	17	reinforcement	reinforcement	NOUN
bracis-33605	335	18	learning	learn	VERB
bracis-33605	335	19	dead	dead	ADJ
bracis-33605	335	20	-	-	PUNCT
bracis-33605	335	21	end	end	NOUN
bracis-33605	335	22	q	q	NOUN
bracis-33605	335	23	-	-	PUNCT
bracis-33605	335	24	learning	learn	VERB
bracis-33605	335	25	gubs	gub	NOUN
bracis-33605	335	26	publish	publish	VERB
bracis-33605	335	27	with	with	ADP
bracis-33605	335	28	us	us	PROPN
bracis-33605	335	29	policies	policy	NOUN
bracis-33605	335	30	and	and	CCONJ
bracis-33605	335	31	ethics	ethic	NOUN
bracis-33605	335	32	search	search	NOUN
bracis-33605	335	33	search	search	NOUN
bracis-33605	335	34	by	by	ADP
bracis-33605	335	35	keyword	keyword	NOUN
bracis-33605	335	36	or	or	CCONJ
bracis-33605	335	37	author	author	NOUN
bracis-33605	335	38	search	search	NOUN
bracis-33605	335	39	navigation	navigation	NOUN
bracis-33605	335	40	find	find	VERB
bracis-33605	335	41	a	a	DET
bracis-33605	335	42	journal	journal	NOUN
bracis-33605	335	43	publish	publish	VERB
bracis-33605	335	44	with	with	ADP
bracis-33605	335	45	us	we	PRON
bracis-33605	335	46	track	track	VERB
bracis-33605	335	47	your	your	PRON
bracis-33605	335	48	research	research	NOUN
bracis-33605	335	49	discover	discover	VERB
bracis-33605	335	50	content	content	NOUN
bracis-33605	335	51	journals	journal	NOUN
bracis-33605	335	52	a	a	DET
bracis-33605	335	53	-	-	PUNCT
bracis-33605	335	54	z	z	NOUN
bracis-33605	335	55	books	book	NOUN
bracis-33605	335	56	a	a	DET
bracis-33605	335	57	-	-	PUNCT
bracis-33605	335	58	z	z	NOUN
bracis-33605	335	59	publish	publish	NOUN
bracis-33605	335	60	with	with	ADP
bracis-33605	335	61	us	us	PROPN
bracis-33605	335	62	journal	journal	PROPN
bracis-33605	335	63	finder	finder	PROPN
bracis-33605	335	64	publish	publish	VERB
bracis-33605	335	65	your	your	PRON
bracis-33605	335	66	research	research	NOUN
bracis-33605	335	67	language	language	NOUN
bracis-33605	335	68	editing	edit	VERB
bracis-33605	335	69	open	open	ADJ
bracis-33605	335	70	access	access	NOUN
bracis-33605	335	71	publishing	publishing	NOUN
bracis-33605	335	72	products	product	NOUN
bracis-33605	335	73	and	and	CCONJ
bracis-33605	335	74	services	service	NOUN
bracis-33605	335	75	our	our	PRON
bracis-33605	335	76	products	product	NOUN
bracis-33605	335	77	librarians	librarian	VERB
bracis-33605	335	78	societies	society	NOUN
bracis-33605	335	79	partners	partner	NOUN
bracis-33605	335	80	and	and	CCONJ
bracis-33605	335	81	advertisers	advertiser	NOUN
bracis-33605	335	82	our	our	PRON
bracis-33605	335	83	brands	brand	NOUN
bracis-33605	335	84	springer	springer	NOUN
bracis-33605	335	85	nature	nature	PROPN
bracis-33605	335	86	portfolio	portfolio	PROPN
bracis-33605	335	87	bmc	bmc	PROPN
bracis-33605	335	88	palgrave	palgrave	PROPN
bracis-33605	335	89	macmillan	macmillan	PROPN
bracis-33605	335	90	apress	apress	PROPN
bracis-33605	335	91	discover	discover	VERB
bracis-33605	335	92	your	your	PRON
bracis-33605	335	93	privacy	privacy	NOUN
bracis-33605	335	94	choices	choice	NOUN
bracis-33605	335	95	/	/	SYM
bracis-33605	335	96	manage	manage	NOUN
bracis-33605	335	97	cookies	cookie	NOUN
bracis-33605	335	98	your	your	PRON
bracis-33605	335	99	us	us	PROPN
bracis-33605	336	1	state	state	NOUN
bracis-33605	336	2	privacy	privacy	NOUN
bracis-33605	336	3	rights	right	NOUN
bracis-33605	336	4	accessibility	accessibility	NOUN
bracis-33605	336	5	statement	statement	NOUN
bracis-33605	336	6	terms	term	NOUN
bracis-33605	336	7	and	and	CCONJ
bracis-33605	336	8	conditions	condition	NOUN
bracis-33605	336	9	privacy	privacy	NOUN
bracis-33605	336	10	policy	policy	NOUN
bracis-33605	336	11	help	help	NOUN
bracis-33605	336	12	and	and	CCONJ
bracis-33605	336	13	support	support	VERB
bracis-33605	336	14	legal	legal	ADJ
bracis-33605	336	15	notice	notice	NOUN
bracis-33605	336	16	cancel	cancel	VERB
bracis-33605	336	17	contracts	contract	NOUN
bracis-33605	336	18	here	here	ADV
bracis-33605	336	19	129.74.145.123	129.74.145.123	NUM
bracis-33605	336	20	hesburgh	hesburgh	PROPN
bracis-33605	336	21	library	library	PROPN
bracis-33605	336	22	er	er	INTJ
bracis-33605	336	23	unit	unit	NOUN
bracis-33605	336	24	(	(	PUNCT
bracis-33605	336	25	3005732405	3005732405	NUM
bracis-33605	336	26	)	)	PUNCT
bracis-33605	336	27	northeast	northeast	ADJ
bracis-33605	336	28	research	research	NOUN
bracis-33605	336	29	libraries	library	NOUN
bracis-33605	336	30	(	(	PUNCT
bracis-33605	336	31	nerl	nerl	PROPN
bracis-33605	336	32	)	)	PUNCT
bracis-33605	336	33	(	(	PUNCT
bracis-33605	336	34	8200828607	8200828607	NUM
bracis-33605	336	35	)	)	PUNCT
bracis-33605	336	36	nerl	nerl	VERB
bracis-33605	336	37	ta	ta	X
bracis-33605	336	38	account	account	NOUN
bracis-33605	336	39	(	(	PUNCT
bracis-33605	336	40	3006206169	3006206169	NUM
bracis-33605	336	41	)	)	PUNCT
bracis-33605	336	42	university	university	NOUN
bracis-33605	336	43	of	of	ADP
bracis-33605	336	44	notre	notre	PROPN
bracis-33605	336	45	dame	dame	PROPN
bracis-33605	336	46	hesburgh	hesburgh	PROPN
bracis-33605	336	47	library	library	NOUN
bracis-33605	336	48	(	(	PUNCT
bracis-33605	336	49	3000184373	3000184373	NUM
bracis-33605	336	50	)	)	PUNCT
bracis-33605	337	1	©	©	ADP
bracis-33605	337	2	2025	2025	NUM
bracis-33605	337	3	springer	springer	NOUN
bracis-33605	337	4	nature	nature	NOUN