id	sid	tid	token	lemma	pos
ap-5030	1	1	acta	acta	PROPN
ap-5030	1	2	polytechnica	polytechnica	PROPN
ap-5030	1	3	doi:10.14311	doi:10.14311	PROPN
ap-5030	1	4	/	/	SYM
ap-5030	1	5	ap.2019.59.0518	ap.2019.59.0518	ADJ
ap-5030	1	6	acta	acta	PROPN
ap-5030	1	7	polytechnica	polytechnica	PROPN
ap-5030	1	8	59(5):518–526	59(5):518–526	PROPN
ap-5030	1	9	,	,	PUNCT
ap-5030	1	10	2019	2019	NUM
ap-5030	1	11	©	©	PROPN
ap-5030	1	12	czech	czech	PROPN
ap-5030	1	13	technical	technical	PROPN
ap-5030	1	14	university	university	PROPN
ap-5030	1	15	in	in	ADP
ap-5030	1	16	prague	prague	PROPN
ap-5030	1	17	,	,	PUNCT
ap-5030	1	18	2019	2019	NUM
ap-5030	1	19	available	available	ADJ
ap-5030	1	20	online	online	ADV
ap-5030	1	21	at	at	ADP
ap-5030	1	22	https://ojs.cvut.cz/ojs/index.php/ap	https://ojs.cvut.cz/ojs/index.php/ap	PROPN
ap-5030	1	23	model	model	NOUN
ap-5030	1	24	-	-	PUNCT
ap-5030	1	25	based	base	VERB
ap-5030	1	26	security	security	NOUN
ap-5030	1	27	analysis	analysis	NOUN
ap-5030	1	28	of	of	ADP
ap-5030	1	29	fpga	fpga	PROPN
ap-5030	1	30	designs	design	NOUN
ap-5030	1	31	through	through	ADP
ap-5030	1	32	reinforcement	reinforcement	NOUN
ap-5030	1	33	learning	learn	VERB
ap-5030	1	34	michael	michael	PROPN
ap-5030	1	35	vetter	vetter	PROPN
ap-5030	1	36	university	university	PROPN
ap-5030	1	37	of	of	ADP
ap-5030	1	38	west	west	PROPN
ap-5030	1	39	bohemia	bohemia	PROPN
ap-5030	1	40	,	,	PUNCT
ap-5030	1	41	faculty	faculty	NOUN
ap-5030	1	42	of	of	ADP
ap-5030	1	43	applied	apply	VERB
ap-5030	1	44	sciences	science	NOUN
ap-5030	1	45	,	,	PUNCT
ap-5030	1	46	department	department	NOUN
ap-5030	1	47	of	of	ADP
ap-5030	1	48	computer	computer	NOUN
ap-5030	1	49	science	science	NOUN
ap-5030	1	50	and	and	CCONJ
ap-5030	1	51	engineering	engineering	NOUN
ap-5030	1	52	,	,	PUNCT
ap-5030	1	53	technická	technická	NOUN
ap-5030	1	54	8	8	NUM
ap-5030	1	55	,	,	PUNCT
ap-5030	1	56	301	301	NUM
ap-5030	1	57	00	00	NUM
ap-5030	1	58	plzeň	plzeň	ADJ
ap-5030	1	59	,	,	PUNCT
ap-5030	1	60	czech	czech	PROPN
ap-5030	1	61	republic	republic	NOUN
ap-5030	1	62	correspondence	correspondence	NOUN
ap-5030	1	63	:	:	PUNCT
ap-5030	2	1	michaelvetter@protonmail.com	michaelvetter@protonmail.com	X
ap-5030	2	2	abstract	abstract	ADJ
ap-5030	2	3	.	.	PUNCT
ap-5030	3	1	finding	find	VERB
ap-5030	3	2	potential	potential	ADJ
ap-5030	3	3	security	security	NOUN
ap-5030	3	4	weaknesses	weakness	NOUN
ap-5030	3	5	in	in	ADP
ap-5030	3	6	any	any	DET
ap-5030	3	7	complex	complex	NOUN
ap-5030	3	8	it	it	PRON
ap-5030	3	9	system	system	NOUN
ap-5030	3	10	is	be	AUX
ap-5030	3	11	an	an	DET
ap-5030	3	12	important	important	ADJ
ap-5030	3	13	and	and	CCONJ
ap-5030	3	14	often	often	ADV
ap-5030	3	15	challenging	challenge	VERB
ap-5030	3	16	task	task	NOUN
ap-5030	3	17	best	well	ADV
ap-5030	3	18	started	start	VERB
ap-5030	3	19	in	in	ADP
ap-5030	3	20	the	the	DET
ap-5030	3	21	early	early	ADJ
ap-5030	3	22	stages	stage	NOUN
ap-5030	3	23	of	of	ADP
ap-5030	3	24	the	the	DET
ap-5030	3	25	development	development	NOUN
ap-5030	3	26	process	process	NOUN
ap-5030	3	27	.	.	PUNCT
ap-5030	4	1	we	we	PRON
ap-5030	4	2	present	present	VERB
ap-5030	4	3	a	a	DET
ap-5030	4	4	method	method	NOUN
ap-5030	4	5	that	that	PRON
ap-5030	4	6	transforms	transform	VERB
ap-5030	4	7	this	this	DET
ap-5030	4	8	task	task	NOUN
ap-5030	4	9	for	for	ADP
ap-5030	4	10	fpga	fpga	PROPN
ap-5030	4	11	designs	design	NOUN
ap-5030	4	12	into	into	ADP
ap-5030	4	13	a	a	DET
ap-5030	4	14	reinforcement	reinforcement	NOUN
ap-5030	4	15	learning	learning	NOUN
ap-5030	4	16	(	(	PUNCT
ap-5030	4	17	rl	rl	NOUN
ap-5030	4	18	)	)	PUNCT
ap-5030	4	19	problem	problem	NOUN
ap-5030	4	20	.	.	PUNCT
ap-5030	5	1	this	this	DET
ap-5030	5	2	paper	paper	NOUN
ap-5030	5	3	introduces	introduce	VERB
ap-5030	5	4	a	a	DET
ap-5030	5	5	method	method	NOUN
ap-5030	5	6	to	to	PART
ap-5030	5	7	generate	generate	VERB
ap-5030	5	8	a	a	DET
ap-5030	5	9	markov	markov	NOUN
ap-5030	5	10	decision	decision	NOUN
ap-5030	5	11	process	process	NOUN
ap-5030	5	12	based	base	VERB
ap-5030	5	13	rl	rl	PROPN
ap-5030	5	14	model	model	NOUN
ap-5030	5	15	from	from	ADP
ap-5030	5	16	a	a	DET
ap-5030	5	17	formal	formal	ADJ
ap-5030	5	18	,	,	PUNCT
ap-5030	5	19	high	high	ADJ
ap-5030	5	20	-	-	PUNCT
ap-5030	5	21	level	level	NOUN
ap-5030	5	22	system	system	NOUN
ap-5030	5	23	description	description	NOUN
ap-5030	5	24	(	(	PUNCT
ap-5030	5	25	formulated	formulate	VERB
ap-5030	5	26	in	in	ADP
ap-5030	5	27	the	the	DET
ap-5030	5	28	domain	domain	NOUN
ap-5030	5	29	-	-	PUNCT
ap-5030	5	30	specific	specific	ADJ
ap-5030	5	31	language	language	NOUN
ap-5030	5	32	)	)	PUNCT
ap-5030	5	33	of	of	ADP
ap-5030	5	34	the	the	DET
ap-5030	5	35	system	system	NOUN
ap-5030	5	36	under	under	ADP
ap-5030	5	37	review	review	NOUN
ap-5030	5	38	and	and	CCONJ
ap-5030	5	39	different	different	ADJ
ap-5030	5	40	,	,	PUNCT
ap-5030	5	41	quantified	quantified	ADJ
ap-5030	5	42	assumptions	assumption	NOUN
ap-5030	5	43	about	about	ADP
ap-5030	5	44	the	the	DET
ap-5030	5	45	system	system	NOUN
ap-5030	5	46	’s	’s	PART
ap-5030	5	47	security	security	NOUN
ap-5030	5	48	.	.	PUNCT
ap-5030	6	1	probabilistic	probabilistic	ADJ
ap-5030	6	2	transitions	transition	NOUN
ap-5030	6	3	and	and	CCONJ
ap-5030	6	4	the	the	DET
ap-5030	6	5	reward	reward	NOUN
ap-5030	6	6	function	function	NOUN
ap-5030	6	7	can	can	AUX
ap-5030	6	8	be	be	AUX
ap-5030	6	9	used	use	VERB
ap-5030	6	10	to	to	PART
ap-5030	6	11	model	model	VERB
ap-5030	6	12	the	the	DET
ap-5030	6	13	varying	vary	VERB
ap-5030	6	14	resilience	resilience	NOUN
ap-5030	6	15	of	of	ADP
ap-5030	6	16	different	different	ADJ
ap-5030	6	17	elements	element	NOUN
ap-5030	6	18	against	against	ADP
ap-5030	6	19	attacks	attack	NOUN
ap-5030	6	20	and	and	CCONJ
ap-5030	6	21	the	the	DET
ap-5030	6	22	capabilities	capability	NOUN
ap-5030	6	23	of	of	ADP
ap-5030	6	24	an	an	DET
ap-5030	6	25	attacker	attacker	NOUN
ap-5030	6	26	.	.	PUNCT
ap-5030	7	1	this	this	DET
ap-5030	7	2	information	information	NOUN
ap-5030	7	3	is	be	AUX
ap-5030	7	4	then	then	ADV
ap-5030	7	5	used	use	VERB
ap-5030	7	6	to	to	PART
ap-5030	7	7	determine	determine	VERB
ap-5030	7	8	a	a	DET
ap-5030	7	9	plausible	plausible	ADJ
ap-5030	7	10	data	data	NOUN
ap-5030	7	11	exfiltration	exfiltration	NOUN
ap-5030	7	12	strategy	strategy	NOUN
ap-5030	7	13	.	.	PUNCT
ap-5030	8	1	an	an	DET
ap-5030	8	2	example	example	NOUN
ap-5030	8	3	with	with	ADP
ap-5030	8	4	multiple	multiple	ADJ
ap-5030	8	5	scenarios	scenario	NOUN
ap-5030	8	6	illustrates	illustrate	VERB
ap-5030	8	7	the	the	DET
ap-5030	8	8	workflow	workflow	NOUN
ap-5030	8	9	.	.	PUNCT
ap-5030	9	1	a	a	DET
ap-5030	9	2	discussion	discussion	NOUN
ap-5030	9	3	of	of	ADP
ap-5030	9	4	supplementary	supplementary	ADJ
ap-5030	9	5	techniques	technique	NOUN
ap-5030	9	6	like	like	ADP
ap-5030	9	7	hierarchical	hierarchical	ADJ
ap-5030	9	8	learning	learning	NOUN
ap-5030	9	9	and	and	CCONJ
ap-5030	9	10	deep	deep	ADJ
ap-5030	9	11	neural	neural	ADJ
ap-5030	9	12	networks	network	NOUN
ap-5030	9	13	concludes	conclude	VERB
ap-5030	9	14	this	this	DET
ap-5030	9	15	paper	paper	NOUN
ap-5030	9	16	.	.	PUNCT
ap-5030	10	1	keywords	keyword	NOUN
ap-5030	10	2	:	:	PUNCT
ap-5030	10	3	fpga	fpga	PROPN
ap-5030	10	4	,	,	PUNCT
ap-5030	10	5	it	it	PRON
ap-5030	10	6	security	security	NOUN
ap-5030	10	7	,	,	PUNCT
ap-5030	10	8	model	model	NOUN
ap-5030	10	9	-	-	PUNCT
ap-5030	10	10	driven	drive	VERB
ap-5030	10	11	design	design	NOUN
ap-5030	10	12	,	,	PUNCT
ap-5030	10	13	reinforcement	reinforcement	NOUN
ap-5030	10	14	learning	learning	NOUN
ap-5030	10	15	,	,	PUNCT
ap-5030	10	16	machine	machine	NOUN
ap-5030	10	17	learning	learning	NOUN
ap-5030	10	18	.	.	PUNCT
ap-5030	11	1	1	1	X
ap-5030	11	2	.	.	X
ap-5030	11	3	introduction	introduction	NOUN
ap-5030	11	4	securing	secure	VERB
ap-5030	11	5	any	any	DET
ap-5030	11	6	(	(	PUNCT
ap-5030	11	7	non	non	ADJ
ap-5030	11	8	-	-	ADJ
ap-5030	11	9	trivial	trivial	ADJ
ap-5030	11	10	)	)	PUNCT
ap-5030	11	11	computer	computer	NOUN
ap-5030	11	12	system	system	NOUN
ap-5030	11	13	against	against	ADP
ap-5030	11	14	malicious	malicious	ADJ
ap-5030	11	15	actors	actor	NOUN
ap-5030	11	16	is	be	AUX
ap-5030	11	17	a	a	DET
ap-5030	11	18	challenging	challenging	ADJ
ap-5030	11	19	task	task	NOUN
ap-5030	11	20	.	.	PUNCT
ap-5030	12	1	the	the	DET
ap-5030	12	2	defender	defender	NOUN
ap-5030	12	3	has	have	VERB
ap-5030	12	4	to	to	PART
ap-5030	12	5	identify	identify	VERB
ap-5030	12	6	and	and	CCONJ
ap-5030	12	7	protect	protect	VERB
ap-5030	12	8	all	all	DET
ap-5030	12	9	possible	possible	ADJ
ap-5030	12	10	paths	path	NOUN
ap-5030	12	11	of	of	ADP
ap-5030	12	12	entry	entry	NOUN
ap-5030	12	13	while	while	SCONJ
ap-5030	12	14	an	an	DET
ap-5030	12	15	attacker	attacker	NOUN
ap-5030	12	16	has	have	VERB
ap-5030	12	17	to	to	PART
ap-5030	12	18	find	find	VERB
ap-5030	12	19	only	only	ADV
ap-5030	12	20	one	one	NUM
ap-5030	12	21	feasible	feasible	ADJ
ap-5030	12	22	way	way	NOUN
ap-5030	12	23	to	to	PART
ap-5030	12	24	breach	breach	VERB
ap-5030	12	25	the	the	DET
ap-5030	12	26	system	system	NOUN
ap-5030	12	27	’s	’s	PART
ap-5030	12	28	security	security	NOUN
ap-5030	12	29	properties	property	NOUN
ap-5030	12	30	.	.	PUNCT
ap-5030	13	1	as	as	ADP
ap-5030	13	2	a	a	DET
ap-5030	13	3	system	system	NOUN
ap-5030	13	4	matures	mature	NOUN
ap-5030	13	5	,	,	PUNCT
ap-5030	13	6	new	new	ADJ
ap-5030	13	7	threats	threat	NOUN
ap-5030	13	8	can	can	AUX
ap-5030	13	9	arise	arise	VERB
ap-5030	13	10	and	and	CCONJ
ap-5030	13	11	the	the	DET
ap-5030	13	12	strength	strength	NOUN
ap-5030	13	13	of	of	ADP
ap-5030	13	14	defence	defence	NOUN
ap-5030	13	15	mechanisms	mechanism	NOUN
ap-5030	13	16	may	may	AUX
ap-5030	13	17	erode	erode	VERB
ap-5030	13	18	.	.	PUNCT
ap-5030	14	1	current	current	ADJ
ap-5030	14	2	threat	threat	NOUN
ap-5030	14	3	modelling	modelling	NOUN
ap-5030	14	4	approaches	approach	NOUN
ap-5030	14	5	are	be	AUX
ap-5030	14	6	often	often	ADV
ap-5030	14	7	informal	informal	ADJ
ap-5030	14	8	,	,	PUNCT
ap-5030	14	9	provide	provide	VERB
ap-5030	14	10	little	little	ADJ
ap-5030	14	11	automation	automation	NOUN
ap-5030	14	12	and	and	CCONJ
ap-5030	14	13	have	have	VERB
ap-5030	14	14	a	a	DET
ap-5030	14	15	tendency	tendency	NOUN
ap-5030	14	16	to	to	PART
ap-5030	14	17	neglect	neglect	VERB
ap-5030	14	18	the	the	DET
ap-5030	14	19	defence	defence	NOUN
ap-5030	14	20	in	in	ADP
ap-5030	14	21	-	-	PUNCT
ap-5030	14	22	depth	depth	NOUN
ap-5030	14	23	aspect	aspect	NOUN
ap-5030	14	24	of	of	ADP
ap-5030	14	25	a	a	DET
ap-5030	14	26	design	design	NOUN
ap-5030	14	27	by	by	ADP
ap-5030	14	28	focusing	focus	VERB
ap-5030	14	29	on	on	ADP
ap-5030	14	30	the	the	DET
ap-5030	14	31	attack	attack	NOUN
ap-5030	14	32	surface	surface	NOUN
ap-5030	14	33	.	.	PUNCT
ap-5030	15	1	attack	attack	NOUN
ap-5030	15	2	trees	tree	NOUN
ap-5030	16	1	[	[	X
ap-5030	16	2	1	1	NUM
ap-5030	16	3	]	]	PUNCT
ap-5030	16	4	are	be	AUX
ap-5030	16	5	a	a	DET
ap-5030	16	6	common	common	ADJ
ap-5030	16	7	method	method	NOUN
ap-5030	16	8	to	to	PART
ap-5030	16	9	model	model	VERB
ap-5030	16	10	an	an	DET
ap-5030	16	11	attacker	attacker	NOUN
ap-5030	16	12	’s	’s	PART
ap-5030	16	13	options	option	NOUN
ap-5030	16	14	but	but	CCONJ
ap-5030	16	15	their	their	PRON
ap-5030	16	16	creation	creation	NOUN
ap-5030	16	17	is	be	AUX
ap-5030	16	18	often	often	ADV
ap-5030	16	19	tedious	tedious	ADJ
ap-5030	16	20	and	and	CCONJ
ap-5030	16	21	they	they	PRON
ap-5030	16	22	provide	provide	VERB
ap-5030	16	23	little	little	ADJ
ap-5030	16	24	information	information	NOUN
ap-5030	16	25	about	about	ADP
ap-5030	16	26	the	the	DET
ap-5030	16	27	most	most	ADV
ap-5030	16	28	efficient	efficient	ADJ
ap-5030	16	29	way	way	NOUN
ap-5030	16	30	to	to	PART
ap-5030	16	31	breach	breach	VERB
ap-5030	16	32	a	a	DET
ap-5030	16	33	system	system	NOUN
ap-5030	16	34	.	.	PUNCT
ap-5030	17	1	we	we	PRON
ap-5030	17	2	address	address	VERB
ap-5030	17	3	this	this	DET
ap-5030	17	4	problem	problem	NOUN
ap-5030	17	5	for	for	ADP
ap-5030	17	6	fpga	fpga	PROPN
ap-5030	17	7	[	[	X
ap-5030	17	8	2	2	NUM
ap-5030	17	9	]	]	PUNCT
ap-5030	17	10	based	base	VERB
ap-5030	17	11	designs	design	NOUN
ap-5030	17	12	through	through	ADP
ap-5030	17	13	a	a	DET
ap-5030	17	14	combination	combination	NOUN
ap-5030	17	15	of	of	ADP
ap-5030	17	16	a	a	DET
ap-5030	17	17	text	text	NOUN
ap-5030	17	18	-	-	PUNCT
ap-5030	17	19	based	base	VERB
ap-5030	17	20	domain	domain	NOUN
ap-5030	17	21	specific	specific	ADJ
ap-5030	17	22	language	language	NOUN
ap-5030	17	23	(	(	PUNCT
ap-5030	17	24	dsl	dsl	PROPN
ap-5030	17	25	)	)	PUNCT
ap-5030	17	26	for	for	ADP
ap-5030	17	27	the	the	DET
ap-5030	17	28	design	design	NOUN
ap-5030	17	29	description	description	NOUN
ap-5030	17	30	and	and	CCONJ
ap-5030	17	31	a	a	DET
ap-5030	17	32	quantified	quantified	ADJ
ap-5030	17	33	assessment	assessment	NOUN
ap-5030	17	34	of	of	ADP
ap-5030	17	35	the	the	DET
ap-5030	17	36	security	security	NOUN
ap-5030	17	37	properties	property	NOUN
ap-5030	17	38	.	.	PUNCT
ap-5030	18	1	this	this	DET
ap-5030	18	2	description	description	NOUN
ap-5030	18	3	is	be	AUX
ap-5030	18	4	automatically	automatically	ADV
ap-5030	18	5	transformed	transform	VERB
ap-5030	18	6	into	into	ADP
ap-5030	18	7	a	a	DET
ap-5030	18	8	processable	processable	ADJ
ap-5030	18	9	markov	markov	NOUN
ap-5030	18	10	decision	decision	NOUN
ap-5030	18	11	process	process	NOUN
ap-5030	18	12	(	(	PUNCT
ap-5030	18	13	mdp	mdp	PROPN
ap-5030	18	14	)	)	PUNCT
ap-5030	18	15	.	.	PUNCT
ap-5030	19	1	an	an	DET
ap-5030	19	2	agent	agent	NOUN
ap-5030	19	3	is	be	AUX
ap-5030	19	4	trained	train	VERB
ap-5030	19	5	on	on	ADP
ap-5030	19	6	this	this	DET
ap-5030	19	7	mdp	mdp	NOUN
ap-5030	19	8	to	to	PART
ap-5030	19	9	exfiltrate	exfiltrate	VERB
ap-5030	19	10	all	all	DET
ap-5030	19	11	data	datum	NOUN
ap-5030	19	12	stored	store	VERB
ap-5030	19	13	within	within	ADP
ap-5030	19	14	an	an	DET
ap-5030	19	15	fpga	fpga	NOUN
ap-5030	19	16	design	design	NOUN
ap-5030	19	17	using	use	VERB
ap-5030	19	18	the	the	DET
ap-5030	19	19	most	most	ADV
ap-5030	19	20	efficient	efficient	ADJ
ap-5030	19	21	sequence	sequence	NOUN
ap-5030	19	22	of	of	ADP
ap-5030	19	23	predefined	predefine	VERB
ap-5030	19	24	actions	action	NOUN
ap-5030	19	25	.	.	PUNCT
ap-5030	20	1	this	this	DET
ap-5030	20	2	task	task	NOUN
ap-5030	20	3	is	be	AUX
ap-5030	20	4	performed	perform	VERB
ap-5030	20	5	by	by	ADP
ap-5030	20	6	well	well	ADV
ap-5030	20	7	-	-	PUNCT
ap-5030	20	8	established	establish	VERB
ap-5030	20	9	reinforcement	reinforcement	NOUN
ap-5030	20	10	learning	learning	NOUN
ap-5030	20	11	(	(	PUNCT
ap-5030	20	12	rl	rl	NOUN
ap-5030	20	13	)	)	PUNCT
ap-5030	21	1	[	[	X
ap-5030	21	2	3	3	X
ap-5030	21	3	]	]	X
ap-5030	21	4	[	[	X
ap-5030	21	5	4	4	NUM
ap-5030	21	6	]	]	SYM
ap-5030	21	7	algorithms	algorithm	NOUN
ap-5030	21	8	.	.	PUNCT
ap-5030	22	1	the	the	DET
ap-5030	22	2	result	result	NOUN
ap-5030	22	3	of	of	ADP
ap-5030	22	4	this	this	DET
ap-5030	22	5	analysis	analysis	NOUN
ap-5030	22	6	is	be	AUX
ap-5030	22	7	one	one	NUM
ap-5030	22	8	or	or	CCONJ
ap-5030	22	9	more	more	ADJ
ap-5030	22	10	attack	attack	NOUN
ap-5030	22	11	sequences	sequence	NOUN
ap-5030	22	12	that	that	PRON
ap-5030	22	13	provide	provide	VERB
ap-5030	22	14	the	the	DET
ap-5030	22	15	user	user	NOUN
ap-5030	22	16	with	with	ADP
ap-5030	22	17	insights	insight	NOUN
ap-5030	22	18	about	about	ADP
ap-5030	22	19	an	an	DET
ap-5030	22	20	attacker	attacker	NOUN
ap-5030	22	21	’s	’s	PART
ap-5030	22	22	strategy	strategy	NOUN
ap-5030	22	23	under	under	ADP
ap-5030	22	24	the	the	DET
ap-5030	22	25	described	describe	VERB
ap-5030	22	26	circumstances	circumstance	NOUN
ap-5030	22	27	.	.	PUNCT
ap-5030	23	1	we	we	PRON
ap-5030	23	2	illustrate	illustrate	VERB
ap-5030	23	3	the	the	DET
ap-5030	23	4	feasibility	feasibility	NOUN
ap-5030	23	5	of	of	ADP
ap-5030	23	6	this	this	DET
ap-5030	23	7	approach	approach	NOUN
ap-5030	23	8	through	through	ADP
ap-5030	23	9	a	a	DET
ap-5030	23	10	series	series	NOUN
ap-5030	23	11	of	of	ADP
ap-5030	23	12	experiments	experiment	NOUN
ap-5030	23	13	,	,	PUNCT
ap-5030	23	14	assess	assess	VERB
ap-5030	23	15	its	its	PRON
ap-5030	23	16	constraints	constraint	NOUN
ap-5030	23	17	and	and	CCONJ
ap-5030	23	18	conclude	conclude	VERB
ap-5030	23	19	with	with	ADP
ap-5030	23	20	a	a	DET
ap-5030	23	21	discussion	discussion	NOUN
ap-5030	23	22	about	about	ADP
ap-5030	23	23	our	our	PRON
ap-5030	23	24	model	model	NOUN
ap-5030	23	25	’s	’s	PART
ap-5030	23	26	limits	limit	NOUN
ap-5030	23	27	and	and	CCONJ
ap-5030	23	28	possible	possible	ADJ
ap-5030	23	29	methods	method	NOUN
ap-5030	23	30	to	to	PART
ap-5030	23	31	lift	lift	VERB
ap-5030	23	32	them	they	PRON
ap-5030	23	33	.	.	PUNCT
ap-5030	24	1	2	2	X
ap-5030	24	2	.	.	X
ap-5030	24	3	related	relate	VERB
ap-5030	24	4	work	work	NOUN
ap-5030	24	5	the	the	DET
ap-5030	24	6	approach	approach	NOUN
ap-5030	24	7	described	describe	VERB
ap-5030	24	8	in	in	ADP
ap-5030	24	9	this	this	DET
ap-5030	24	10	work	work	NOUN
ap-5030	24	11	operates	operate	VERB
ap-5030	24	12	on	on	ADP
ap-5030	24	13	the	the	DET
ap-5030	24	14	intersection	intersection	NOUN
ap-5030	24	15	between	between	ADP
ap-5030	24	16	it	it	PRON
ap-5030	24	17	security	security	NOUN
ap-5030	24	18	,	,	PUNCT
ap-5030	24	19	model	model	NOUN
ap-5030	24	20	-	-	PUNCT
ap-5030	24	21	driven	drive	VERB
ap-5030	24	22	development	development	NOUN
ap-5030	24	23	(	(	PUNCT
ap-5030	24	24	mdd	mdd	PROPN
ap-5030	24	25	)	)	PUNCT
ap-5030	24	26	and	and	CCONJ
ap-5030	24	27	machine	machine	NOUN
ap-5030	24	28	learning	learning	NOUN
ap-5030	24	29	.	.	PUNCT
ap-5030	25	1	model	model	NOUN
ap-5030	25	2	-	-	PUNCT
ap-5030	25	3	driven	drive	VERB
ap-5030	25	4	development	development	NOUN
ap-5030	25	5	[	[	X
ap-5030	25	6	5	5	NUM
ap-5030	25	7	]	]	PUNCT
ap-5030	25	8	is	be	AUX
ap-5030	25	9	a	a	DET
ap-5030	25	10	common	common	ADJ
ap-5030	25	11	technique	technique	NOUN
ap-5030	25	12	to	to	PART
ap-5030	25	13	master	master	VERB
ap-5030	25	14	the	the	DET
ap-5030	25	15	complexity	complexity	NOUN
ap-5030	25	16	of	of	ADP
ap-5030	25	17	modern	modern	ADJ
ap-5030	25	18	systems	system	NOUN
ap-5030	25	19	.	.	PUNCT
ap-5030	26	1	general	general	ADJ
ap-5030	26	2	purpose	purpose	NOUN
ap-5030	26	3	languages	language	NOUN
ap-5030	26	4	like	like	ADP
ap-5030	26	5	sysml	sysml	NOUN
ap-5030	26	6	[	[	X
ap-5030	26	7	6	6	NUM
ap-5030	26	8	]	]	PUNCT
ap-5030	26	9	and	and	CCONJ
ap-5030	26	10	aadl	aadl	NOUN
ap-5030	26	11	[	[	X
ap-5030	26	12	7	7	NUM
ap-5030	26	13	]	]	PUNCT
ap-5030	26	14	flatten	flatten	VERB
ap-5030	26	15	the	the	DET
ap-5030	26	16	learning	learning	NOUN
ap-5030	26	17	curve	curve	NOUN
ap-5030	26	18	,	,	PUNCT
ap-5030	26	19	advance	advance	VERB
ap-5030	26	20	best	good	ADJ
ap-5030	26	21	practices	practice	NOUN
ap-5030	26	22	and	and	CCONJ
ap-5030	26	23	prevent	prevent	VERB
ap-5030	26	24	vendor	vendor	NOUN
ap-5030	26	25	lock	lock	NOUN
ap-5030	26	26	-	-	PUNCT
ap-5030	26	27	in	in	NOUN
ap-5030	26	28	,	,	PUNCT
ap-5030	26	29	but	but	CCONJ
ap-5030	26	30	,	,	PUNCT
ap-5030	26	31	to	to	ADP
ap-5030	26	32	the	the	DET
ap-5030	26	33	authors	author	NOUN
ap-5030	26	34	knowledge	knowledge	NOUN
ap-5030	26	35	,	,	PUNCT
ap-5030	26	36	no	no	DET
ap-5030	26	37	standardized	standardized	ADJ
ap-5030	26	38	support	support	NOUN
ap-5030	26	39	for	for	ADP
ap-5030	26	40	a	a	DET
ap-5030	26	41	security	security	NOUN
ap-5030	26	42	analysis	analysis	NOUN
ap-5030	26	43	exists	exist	VERB
ap-5030	26	44	in	in	ADP
ap-5030	26	45	either	either	PRON
ap-5030	26	46	of	of	ADP
ap-5030	26	47	these	these	DET
ap-5030	26	48	languages	language	NOUN
ap-5030	26	49	.	.	PUNCT
ap-5030	27	1	proposals	proposal	NOUN
ap-5030	27	2	like	like	ADP
ap-5030	27	3	[	[	X
ap-5030	27	4	8	8	NUM
ap-5030	27	5	]	]	PUNCT
ap-5030	27	6	for	for	ADP
ap-5030	27	7	aadl	aadl	NOUN
ap-5030	27	8	have	have	AUX
ap-5030	27	9	been	be	AUX
ap-5030	27	10	made	make	VERB
ap-5030	27	11	but	but	CCONJ
ap-5030	27	12	not	not	PART
ap-5030	27	13	widely	widely	ADV
ap-5030	27	14	adopted	adopt	VERB
ap-5030	27	15	.	.	PUNCT
ap-5030	28	1	threat	threat	NOUN
ap-5030	28	2	modelling	model	VERB
ap-5030	28	3	[	[	X
ap-5030	28	4	9	9	NUM
ap-5030	28	5	]	]	PUNCT
ap-5030	28	6	is	be	AUX
ap-5030	28	7	often	often	ADV
ap-5030	28	8	performed	perform	VERB
ap-5030	28	9	informally	informally	ADV
ap-5030	28	10	using	use	VERB
ap-5030	28	11	tools	tool	NOUN
ap-5030	28	12	as	as	ADV
ap-5030	28	13	simple	simple	ADJ
ap-5030	28	14	as	as	ADP
ap-5030	28	15	paper	paper	NOUN
ap-5030	28	16	,	,	PUNCT
ap-5030	28	17	whiteboards	whiteboard	NOUN
ap-5030	28	18	or	or	CCONJ
ap-5030	28	19	general	general	ADJ
ap-5030	28	20	purpose	purpose	NOUN
ap-5030	28	21	diagram	diagram	NOUN
ap-5030	28	22	software	software	NOUN
ap-5030	28	23	.	.	PUNCT
ap-5030	29	1	dedicated	dedicated	ADJ
ap-5030	29	2	threat	threat	NOUN
ap-5030	29	3	modelling	model	VERB
ap-5030	29	4	tools	tool	NOUN
ap-5030	29	5	,	,	PUNCT
ap-5030	29	6	including	include	VERB
ap-5030	29	7	microsoft	microsoft	PROPN
ap-5030	29	8	threat	threat	NOUN
ap-5030	29	9	modelling	model	VERB
ap-5030	29	10	tool	tool	NOUN
ap-5030	29	11	[	[	X
ap-5030	29	12	10	10	NUM
ap-5030	29	13	]	]	PUNCT
ap-5030	29	14	and	and	CCONJ
ap-5030	29	15	owasp	owasp	PROPN
ap-5030	29	16	threat	threat	NOUN
ap-5030	29	17	dragon	dragon	NOUN
ap-5030	30	1	[	[	PUNCT
ap-5030	30	2	11	11	NUM
ap-5030	30	3	]	]	PUNCT
ap-5030	30	4	,	,	PUNCT
ap-5030	30	5	provide	provide	VERB
ap-5030	30	6	a	a	DET
ap-5030	30	7	suitable	suitable	ADJ
ap-5030	30	8	structure	structure	NOUN
ap-5030	30	9	and	and	CCONJ
ap-5030	30	10	user	user	NOUN
ap-5030	30	11	interface	interface	NOUN
ap-5030	30	12	but	but	CCONJ
ap-5030	30	13	little	little	ADJ
ap-5030	30	14	automation	automation	NOUN
ap-5030	30	15	.	.	PUNCT
ap-5030	31	1	machine	machine	NOUN
ap-5030	31	2	learning	learning	NOUN
ap-5030	31	3	has	have	AUX
ap-5030	31	4	seen	see	VERB
ap-5030	31	5	a	a	DET
ap-5030	31	6	boom	boom	NOUN
ap-5030	31	7	in	in	ADP
ap-5030	31	8	recent	recent	ADJ
ap-5030	31	9	years	year	NOUN
ap-5030	31	10	,	,	PUNCT
ap-5030	31	11	mainly	mainly	ADV
ap-5030	31	12	powered	power	VERB
ap-5030	31	13	by	by	ADP
ap-5030	31	14	a	a	DET
ap-5030	31	15	trifecta	trifecta	NOUN
ap-5030	31	16	of	of	ADP
ap-5030	31	17	big	big	ADJ
ap-5030	31	18	data	datum	NOUN
ap-5030	31	19	,	,	PUNCT
ap-5030	31	20	parallel	parallel	ADJ
ap-5030	31	21	data	datum	NOUN
ap-5030	31	22	processing	processing	NOUN
ap-5030	31	23	power	power	NOUN
ap-5030	31	24	provided	provide	VERB
ap-5030	31	25	by	by	ADP
ap-5030	31	26	gpus	gpus	PROPN
ap-5030	31	27	(	(	PUNCT
ap-5030	31	28	graphics	graphic	NOUN
ap-5030	31	29	processing	processing	NOUN
ap-5030	31	30	unit	unit	NOUN
ap-5030	31	31	)	)	PUNCT
ap-5030	31	32	and	and	CCONJ
ap-5030	31	33	new	new	ADJ
ap-5030	31	34	ml	ml	NOUN
ap-5030	31	35	architectures	architecture	NOUN
ap-5030	31	36	that	that	PRON
ap-5030	31	37	utilize	utilize	VERB
ap-5030	31	38	both	both	PRON
ap-5030	31	39	.	.	PUNCT
ap-5030	32	1	[	[	X
ap-5030	32	2	12	12	NUM
ap-5030	32	3	]	]	PUNCT
ap-5030	32	4	discusses	discuss	VERB
ap-5030	32	5	the	the	DET
ap-5030	32	6	application	application	NOUN
ap-5030	32	7	of	of	ADP
ap-5030	32	8	machine	machine	NOUN
ap-5030	32	9	learning	learn	VERB
ap-5030	32	10	for	for	ADP
ap-5030	32	11	a	a	DET
ap-5030	32	12	malware	malware	NOUN
ap-5030	32	13	detection	detection	NOUN
ap-5030	32	14	from	from	ADP
ap-5030	32	15	a	a	DET
ap-5030	32	16	practitioner	practitioner	NOUN
ap-5030	32	17	’s	’s	PART
ap-5030	32	18	perspective.[13	perspective.[13	NOUN
ap-5030	32	19	]	]	PUNCT
ap-5030	32	20	presents	present	VERB
ap-5030	32	21	both	both	CCONJ
ap-5030	32	22	a	a	DET
ap-5030	32	23	comprehensive	comprehensive	ADJ
ap-5030	32	24	review	review	NOUN
ap-5030	32	25	of	of	ADP
ap-5030	32	26	the	the	DET
ap-5030	32	27	scientific	scientific	ADJ
ap-5030	32	28	literature	literature	NOUN
ap-5030	32	29	and	and	CCONJ
ap-5030	32	30	a	a	DET
ap-5030	32	31	number	number	NOUN
ap-5030	32	32	of	of	ADP
ap-5030	32	33	research	research	NOUN
ap-5030	32	34	papers	paper	NOUN
ap-5030	32	35	with	with	ADP
ap-5030	32	36	applications	application	NOUN
ap-5030	32	37	ranging	range	VERB
ap-5030	32	38	from	from	ADP
ap-5030	32	39	the	the	DET
ap-5030	32	40	detection	detection	NOUN
ap-5030	32	41	of	of	ADP
ap-5030	32	42	malware	malware	NOUN
ap-5030	32	43	to	to	PART
ap-5030	32	44	automatically	automatically	ADV
ap-5030	32	45	generated	generate	VERB
ap-5030	32	46	penetration	penetration	NOUN
ap-5030	32	47	test	test	NOUN
ap-5030	32	48	plans	plan	NOUN
ap-5030	32	49	.	.	PUNCT
ap-5030	33	1	machine	machine	NOUN
ap-5030	33	2	learning	learning	NOUN
ap-5030	33	3	has	have	AUX
ap-5030	33	4	also	also	ADV
ap-5030	33	5	been	be	AUX
ap-5030	33	6	used	use	VERB
ap-5030	33	7	to	to	PART
ap-5030	33	8	generate	generate	VERB
ap-5030	33	9	malicious	malicious	ADJ
ap-5030	33	10	inputs	input	NOUN
ap-5030	33	11	with	with	ADP
ap-5030	33	12	complexity	complexity	NOUN
ap-5030	33	13	beyond	beyond	ADP
ap-5030	33	14	the	the	DET
ap-5030	33	15	conventional	conventional	ADJ
ap-5030	33	16	fuzzing	fuzzing	NOUN
ap-5030	33	17	[	[	X
ap-5030	33	18	14	14	NUM
ap-5030	33	19	]	]	PUNCT
ap-5030	33	20	methods	method	NOUN
ap-5030	33	21	.	.	PUNCT
ap-5030	34	1	[	[	X
ap-5030	34	2	15	15	NUM
ap-5030	34	3	]	]	PUNCT
ap-5030	34	4	presents	present	VERB
ap-5030	34	5	a	a	DET
ap-5030	34	6	generative	generative	ADJ
ap-5030	34	7	adversarial	adversarial	ADJ
ap-5030	34	8	networks	network	NOUN
ap-5030	34	9	(	(	PUNCT
ap-5030	34	10	gan	gan	PROPN
ap-5030	34	11	)	)	PUNCT
ap-5030	35	1	[	[	X
ap-5030	35	2	16	16	NUM
ap-5030	35	3	]	]	PUNCT
ap-5030	35	4	based	base	VERB
ap-5030	35	5	attack	attack	NOUN
ap-5030	35	6	against	against	ADP
ap-5030	35	7	a	a	DET
ap-5030	35	8	fingerprint	fingerprint	NOUN
ap-5030	35	9	scanner	scanner	NOUN
ap-5030	35	10	.	.	PUNCT
ap-5030	36	1	in	in	ADP
ap-5030	36	2	[	[	X
ap-5030	36	3	17	17	NUM
ap-5030	36	4	]	]	X
ap-5030	36	5	,	,	PUNCT
ap-5030	36	6	a	a	DET
ap-5030	36	7	similar	similar	ADJ
ap-5030	36	8	approach	approach	NOUN
ap-5030	36	9	is	be	AUX
ap-5030	36	10	used	use	VERB
ap-5030	36	11	against	against	ADP
ap-5030	36	12	the	the	DET
ap-5030	36	13	computer	computer	NOUN
ap-5030	36	14	vision	vision	NOUN
ap-5030	36	15	system	system	NOUN
ap-5030	36	16	of	of	ADP
ap-5030	36	17	an	an	DET
ap-5030	36	18	electric	electric	ADJ
ap-5030	36	19	vehicle	vehicle	NOUN
ap-5030	36	20	.	.	PUNCT
ap-5030	37	1	in	in	ADP
ap-5030	37	2	[	[	X
ap-5030	37	3	18	18	NUM
ap-5030	37	4	]	]	PUNCT
ap-5030	37	5	,	,	PUNCT
ap-5030	37	6	the	the	DET
ap-5030	37	7	authors	author	NOUN
ap-5030	37	8	use	use	VERB
ap-5030	37	9	markov	markov	NOUN
ap-5030	37	10	518	518	NUM
ap-5030	37	11	https://doi.org/10.14311/ap.2019.59.0518	https://doi.org/10.14311/ap.2019.59.0518	PRON
ap-5030	37	12	https://ojs.cvut.cz/ojs/index.php/ap	https://ojs.cvut.cz/ojs/index.php/ap	PROPN
ap-5030	37	13	vol	vol	NOUN
ap-5030	37	14	.	.	PUNCT
ap-5030	38	1	59	59	NUM
ap-5030	38	2	no	no	NOUN
ap-5030	38	3	.	.	PUNCT
ap-5030	39	1	5/2019	5/2019	NUM
ap-5030	39	2	security	security	NOUN
ap-5030	39	3	analysis	analysis	NOUN
ap-5030	39	4	of	of	ADP
ap-5030	39	5	fpgas	fpgas	NOUN
ap-5030	39	6	designs	design	NOUN
ap-5030	39	7	through	through	ADP
ap-5030	39	8	reinforcement	reinforcement	NOUN
ap-5030	39	9	learning	learning	NOUN
ap-5030	39	10	decision	decision	NOUN
ap-5030	39	11	processes	process	NOUN
ap-5030	39	12	to	to	PART
ap-5030	39	13	craft	craft	VERB
ap-5030	39	14	stealthy	stealthy	ADJ
ap-5030	39	15	attack	attack	NOUN
ap-5030	39	16	sequences	sequence	NOUN
ap-5030	39	17	against	against	ADP
ap-5030	39	18	a	a	DET
ap-5030	39	19	cyber	cyber	NOUN
ap-5030	39	20	-	-	PUNCT
ap-5030	39	21	physical	physical	ADJ
ap-5030	39	22	system	system	NOUN
ap-5030	39	23	.	.	PUNCT
ap-5030	40	1	mdps	mdps	NOUN
ap-5030	40	2	can	can	AUX
ap-5030	40	3	be	be	AUX
ap-5030	40	4	used	use	VERB
ap-5030	40	5	for	for	ADP
ap-5030	40	6	defensive	defensive	ADJ
ap-5030	40	7	purposes	purpose	NOUN
ap-5030	40	8	e.g.	e.g.	ADV
ap-5030	40	9	to	to	PART
ap-5030	40	10	detect	detect	VERB
ap-5030	40	11	attacks	attack	NOUN
ap-5030	40	12	[	[	X
ap-5030	40	13	19	19	NUM
ap-5030	40	14	]	]	PUNCT
ap-5030	40	15	.	.	PUNCT
ap-5030	41	1	the	the	DET
ap-5030	41	2	cyber	cyber	ADJ
ap-5030	41	3	grand	grand	ADJ
ap-5030	41	4	challenge	challenge	NOUN
ap-5030	41	5	2016	2016	NUM
ap-5030	42	1	[	[	X
ap-5030	42	2	20	20	NUM
ap-5030	42	3	]	]	PUNCT
ap-5030	42	4	demonstrated	demonstrate	VERB
ap-5030	42	5	that	that	SCONJ
ap-5030	42	6	modern	modern	ADJ
ap-5030	42	7	computer	computer	NOUN
ap-5030	42	8	programs	program	NOUN
ap-5030	42	9	are	be	AUX
ap-5030	42	10	,	,	PUNCT
ap-5030	42	11	in	in	ADP
ap-5030	42	12	principle	principle	NOUN
ap-5030	42	13	,	,	PUNCT
ap-5030	42	14	able	able	ADJ
ap-5030	42	15	to	to	PART
ap-5030	42	16	detect	detect	VERB
ap-5030	42	17	,	,	PUNCT
ap-5030	42	18	exploit	exploit	VERB
ap-5030	42	19	and	and	CCONJ
ap-5030	42	20	patch	patch	VERB
ap-5030	42	21	previously	previously	ADV
ap-5030	42	22	unknown	unknown	ADJ
ap-5030	42	23	vulnerabilities	vulnerability	NOUN
ap-5030	42	24	in	in	ADP
ap-5030	42	25	other	other	ADJ
ap-5030	42	26	cyber	cyber	NOUN
ap-5030	42	27	systems	system	NOUN
ap-5030	42	28	without	without	ADP
ap-5030	42	29	a	a	DET
ap-5030	42	30	human	human	ADJ
ap-5030	42	31	intervention	intervention	NOUN
ap-5030	42	32	.	.	PUNCT
ap-5030	43	1	it	it	PRON
ap-5030	43	2	is	be	AUX
ap-5030	43	3	generally	generally	ADV
ap-5030	43	4	assumed	assume	VERB
ap-5030	43	5	that	that	SCONJ
ap-5030	43	6	most	most	ADJ
ap-5030	43	7	research	research	NOUN
ap-5030	43	8	in	in	ADP
ap-5030	43	9	this	this	DET
ap-5030	43	10	field	field	NOUN
ap-5030	43	11	is	be	AUX
ap-5030	43	12	done	do	VERB
ap-5030	43	13	by	by	ADP
ap-5030	43	14	military	military	ADJ
ap-5030	43	15	contractors	contractor	NOUN
ap-5030	43	16	and	and	CCONJ
ap-5030	43	17	intelligence	intelligence	NOUN
ap-5030	43	18	agencies	agency	NOUN
ap-5030	43	19	and	and	CCONJ
ap-5030	43	20	,	,	PUNCT
ap-5030	43	21	therefore	therefore	ADV
ap-5030	43	22	,	,	PUNCT
ap-5030	43	23	a	a	DET
ap-5030	43	24	secret	secret	NOUN
ap-5030	44	1	[	[	X
ap-5030	44	2	21	21	NUM
ap-5030	44	3	]	]	X
ap-5030	44	4	advances	advance	NOUN
ap-5030	44	5	in	in	ADP
ap-5030	44	6	other	other	ADJ
ap-5030	44	7	,	,	PUNCT
ap-5030	44	8	more	more	ADV
ap-5030	44	9	visible	visible	ADJ
ap-5030	44	10	,	,	PUNCT
ap-5030	44	11	domains	domain	NOUN
ap-5030	44	12	like	like	ADP
ap-5030	44	13	games	game	NOUN
ap-5030	44	14	can	can	AUX
ap-5030	44	15	provide	provide	VERB
ap-5030	44	16	us	we	PRON
ap-5030	44	17	with	with	ADP
ap-5030	44	18	valuable	valuable	ADJ
ap-5030	44	19	insights	insight	NOUN
ap-5030	44	20	about	about	ADP
ap-5030	44	21	the	the	DET
ap-5030	44	22	capabilities	capability	NOUN
ap-5030	44	23	and	and	CCONJ
ap-5030	44	24	limits	limit	NOUN
ap-5030	44	25	of	of	ADP
ap-5030	44	26	current	current	ADJ
ap-5030	44	27	machine	machine	NOUN
ap-5030	44	28	learning	learning	NOUN
ap-5030	44	29	systems	system	NOUN
ap-5030	44	30	.	.	PUNCT
ap-5030	45	1	milestones	milestone	NOUN
ap-5030	45	2	for	for	ADP
ap-5030	45	3	this	this	DET
ap-5030	45	4	progress	progress	NOUN
ap-5030	45	5	include	include	VERB
ap-5030	45	6	chess	chess	NOUN
ap-5030	45	7	with	with	ADP
ap-5030	45	8	deep	deep	ADJ
ap-5030	45	9	blue	blue	ADJ
ap-5030	45	10	(	(	PUNCT
ap-5030	45	11	chess	chess	NOUN
ap-5030	45	12	)	)	PUNCT
ap-5030	46	1	[	[	X
ap-5030	46	2	22	22	NUM
ap-5030	46	3	]	]	PUNCT
ap-5030	46	4	,	,	PUNCT
ap-5030	46	5	jeopardy	jeopardy	VERB
ap-5030	46	6	with	with	ADP
ap-5030	46	7	watson	watson	PROPN
ap-5030	47	1	[	[	X
ap-5030	47	2	23	23	NUM
ap-5030	47	3	]	]	PUNCT
ap-5030	47	4	,	,	PUNCT
ap-5030	47	5	pacman[24	pacman[24	PROPN
ap-5030	47	6	]	]	PUNCT
ap-5030	47	7	and	and	CCONJ
ap-5030	47	8	diverse	diverse	ADJ
ap-5030	47	9	atari	atari	ADJ
ap-5030	47	10	2600	2600	NUM
ap-5030	47	11	arcade	arcade	NOUN
ap-5030	47	12	games	game	NOUN
ap-5030	47	13	[	[	X
ap-5030	47	14	25	25	NUM
ap-5030	47	15	]	]	PUNCT
ap-5030	47	16	,	,	PUNCT
ap-5030	47	17	the	the	DET
ap-5030	47	18	board	board	NOUN
ap-5030	47	19	game	game	NOUN
ap-5030	47	20	go	go	VERB
ap-5030	47	21	[	[	X
ap-5030	47	22	26	26	NUM
ap-5030	47	23	]	]	PUNCT
ap-5030	47	24	,	,	PUNCT
ap-5030	47	25	as	as	ADV
ap-5030	47	26	well	well	ADV
ap-5030	47	27	as	as	ADP
ap-5030	47	28	the	the	DET
ap-5030	47	29	real	real	ADJ
ap-5030	47	30	-	-	PUNCT
ap-5030	47	31	time	time	NOUN
ap-5030	47	32	strategy	strategy	NOUN
ap-5030	47	33	game	game	NOUN
ap-5030	47	34	starcraft	starcraft	PROPN
ap-5030	47	35	ii	ii	PROPN
ap-5030	48	1	[	[	X
ap-5030	48	2	27	27	NUM
ap-5030	48	3	]	]	PUNCT
ap-5030	48	4	and	and	CCONJ
ap-5030	48	5	quake	quake	NOUN
ap-5030	48	6	iii	iii	NUM
ap-5030	49	1	[	[	X
ap-5030	49	2	28	28	NUM
ap-5030	49	3	]	]	PUNCT
ap-5030	49	4	.	.	PUNCT
ap-5030	50	1	unresolved	unresolved	ADJ
ap-5030	50	2	,	,	PUNCT
ap-5030	50	3	but	but	CCONJ
ap-5030	50	4	intensely	intensely	ADV
ap-5030	50	5	researched	research	VERB
ap-5030	50	6	problems	problem	NOUN
ap-5030	50	7	,	,	PUNCT
ap-5030	50	8	like	like	ADP
ap-5030	50	9	autonomous	autonomous	ADJ
ap-5030	50	10	driving	driving	NOUN
ap-5030	50	11	,	,	PUNCT
ap-5030	50	12	indicate	indicate	VERB
ap-5030	50	13	the	the	DET
ap-5030	50	14	limits	limit	NOUN
ap-5030	50	15	of	of	ADP
ap-5030	50	16	the	the	DET
ap-5030	50	17	current	current	ADJ
ap-5030	50	18	technology	technology	NOUN
ap-5030	50	19	when	when	SCONJ
ap-5030	50	20	the	the	DET
ap-5030	50	21	problem	problem	NOUN
ap-5030	50	22	space	space	NOUN
ap-5030	50	23	becomes	become	VERB
ap-5030	50	24	too	too	ADV
ap-5030	50	25	large	large	ADJ
ap-5030	50	26	.	.	PUNCT
ap-5030	51	1	advances	advance	NOUN
ap-5030	51	2	in	in	ADP
ap-5030	51	3	other	other	ADJ
ap-5030	51	4	fields	field	NOUN
ap-5030	51	5	of	of	ADP
ap-5030	51	6	statistics	statistic	NOUN
ap-5030	51	7	have	have	AUX
ap-5030	51	8	provided	provide	VERB
ap-5030	51	9	us	we	PRON
ap-5030	51	10	with	with	ADP
ap-5030	51	11	methods	method	NOUN
ap-5030	51	12	to	to	PART
ap-5030	51	13	determinate	determinate	VERB
ap-5030	51	14	casual	casual	ADJ
ap-5030	51	15	effects	effect	NOUN
ap-5030	51	16	[	[	X
ap-5030	51	17	29][30	29][30	X
ap-5030	51	18	]	]	X
ap-5030	51	19	in	in	ADP
ap-5030	51	20	the	the	DET
ap-5030	51	21	absence	absence	NOUN
ap-5030	51	22	of	of	ADP
ap-5030	51	23	randomized	randomized	ADJ
ap-5030	51	24	experiments	experiment	NOUN
ap-5030	51	25	and	and	CCONJ
ap-5030	51	26	the	the	DET
ap-5030	51	27	exploration	exploration	NOUN
ap-5030	51	28	of	of	ADP
ap-5030	51	29	counterfactual	counterfactual	ADJ
ap-5030	51	30	scenarios	scenario	NOUN
ap-5030	51	31	.	.	PUNCT
ap-5030	52	1	progress	progress	NOUN
ap-5030	52	2	in	in	ADP
ap-5030	52	3	probabilistic	probabilistic	ADJ
ap-5030	52	4	programming	programming	NOUN
ap-5030	53	1	[	[	X
ap-5030	53	2	31	31	NUM
ap-5030	53	3	]	]	PUNCT
ap-5030	54	1	[	[	X
ap-5030	54	2	32	32	NUM
ap-5030	54	3	]	]	PUNCT
ap-5030	54	4	have	have	AUX
ap-5030	54	5	eased	ease	VERB
ap-5030	54	6	the	the	DET
ap-5030	54	7	usage	usage	NOUN
ap-5030	54	8	of	of	ADP
ap-5030	54	9	bayesian	bayesian	NOUN
ap-5030	54	10	methods	method	NOUN
ap-5030	54	11	to	to	PART
ap-5030	54	12	e.g.	e.g.	ADV
ap-5030	54	13	incorporate	incorporate	VERB
ap-5030	54	14	knowledge	knowledge	NOUN
ap-5030	54	15	about	about	ADP
ap-5030	54	16	a	a	DET
ap-5030	54	17	system	system	NOUN
ap-5030	54	18	as	as	ADP
ap-5030	54	19	informed	informed	ADJ
ap-5030	54	20	priors	prior	NOUN
ap-5030	54	21	.	.	PUNCT
ap-5030	55	1	we	we	PRON
ap-5030	55	2	can	can	AUX
ap-5030	55	3	conclude	conclude	VERB
ap-5030	55	4	that	that	SCONJ
ap-5030	55	5	the	the	DET
ap-5030	55	6	advancement	advancement	NOUN
ap-5030	55	7	of	of	ADP
ap-5030	55	8	machine	machine	NOUN
ap-5030	55	9	learning	learn	VERB
ap-5030	55	10	in	in	ADP
ap-5030	55	11	recent	recent	ADJ
ap-5030	55	12	years	year	NOUN
ap-5030	55	13	can	can	AUX
ap-5030	55	14	and	and	CCONJ
ap-5030	55	15	has	have	AUX
ap-5030	55	16	been	be	AUX
ap-5030	55	17	used	use	VERB
ap-5030	55	18	to	to	PART
ap-5030	55	19	improve	improve	VERB
ap-5030	55	20	both	both	DET
ap-5030	55	21	attacks	attack	NOUN
ap-5030	55	22	and	and	CCONJ
ap-5030	55	23	defences	defence	NOUN
ap-5030	55	24	of	of	ADP
ap-5030	55	25	existing	exist	VERB
ap-5030	55	26	cyber	cyber	ADJ
ap-5030	55	27	physical	physical	ADJ
ap-5030	55	28	systems	system	NOUN
ap-5030	55	29	.	.	PUNCT
ap-5030	56	1	the	the	DET
ap-5030	56	2	machine	machine	NOUN
ap-5030	56	3	learning	learn	VERB
ap-5030	56	4	tools	tool	NOUN
ap-5030	56	5	available	available	ADJ
ap-5030	56	6	today	today	NOUN
ap-5030	56	7	must	must	AUX
ap-5030	56	8	be	be	AUX
ap-5030	56	9	tailored	tailor	VERB
ap-5030	56	10	to	to	ADP
ap-5030	56	11	the	the	DET
ap-5030	56	12	application	application	NOUN
ap-5030	56	13	at	at	ADP
ap-5030	56	14	hand	hand	NOUN
ap-5030	56	15	and	and	CCONJ
ap-5030	56	16	used	use	VERB
ap-5030	56	17	with	with	ADP
ap-5030	56	18	a	a	DET
ap-5030	56	19	careful	careful	ADJ
ap-5030	56	20	knowledge	knowledge	NOUN
ap-5030	56	21	of	of	ADP
ap-5030	56	22	their	their	PRON
ap-5030	56	23	limits	limit	NOUN
ap-5030	56	24	.	.	PUNCT
ap-5030	57	1	to	to	ADP
ap-5030	57	2	the	the	DET
ap-5030	57	3	author	author	NOUN
ap-5030	57	4	’s	’s	PART
ap-5030	57	5	knowledge	knowledge	NOUN
ap-5030	57	6	,	,	PUNCT
ap-5030	57	7	no	no	DET
ap-5030	57	8	such	such	ADJ
ap-5030	57	9	system	system	NOUN
ap-5030	57	10	exists	exist	VERB
ap-5030	57	11	that	that	PRON
ap-5030	57	12	uses	use	VERB
ap-5030	57	13	reinforcement	reinforcement	NOUN
ap-5030	57	14	learning	learning	NOUN
ap-5030	57	15	to	to	PART
ap-5030	57	16	determine	determine	VERB
ap-5030	57	17	a	a	DET
ap-5030	57	18	systems	system	NOUN
ap-5030	57	19	vulnerabilities	vulnerability	NOUN
ap-5030	57	20	in	in	ADP
ap-5030	57	21	the	the	DET
ap-5030	57	22	analysis	analysis	NOUN
ap-5030	57	23	and	and	CCONJ
ap-5030	57	24	design	design	NOUN
ap-5030	57	25	phase	phase	NOUN
ap-5030	57	26	.	.	PUNCT
ap-5030	58	1	3	3	X
ap-5030	58	2	.	.	X
ap-5030	58	3	the	the	DET
ap-5030	58	4	descriptive	descriptive	ADJ
ap-5030	58	5	model	model	NOUN
ap-5030	58	6	of	of	ADP
ap-5030	58	7	the	the	DET
ap-5030	58	8	fpga	fpga	PROPN
ap-5030	58	9	design	design	NOUN
ap-5030	58	10	it	it	PRON
ap-5030	58	11	is	be	AUX
ap-5030	58	12	common	common	ADJ
ap-5030	58	13	wisdom	wisdom	NOUN
ap-5030	58	14	that	that	SCONJ
ap-5030	58	15	security	security	NOUN
ap-5030	58	16	problems	problem	NOUN
ap-5030	58	17	should	should	AUX
ap-5030	58	18	best	well	ADV
ap-5030	58	19	be	be	AUX
ap-5030	58	20	solved	solve	VERB
ap-5030	58	21	at	at	ADP
ap-5030	58	22	the	the	DET
ap-5030	58	23	early	early	ADJ
ap-5030	58	24	stages	stage	NOUN
ap-5030	58	25	of	of	ADP
ap-5030	58	26	the	the	DET
ap-5030	58	27	development	development	NOUN
ap-5030	58	28	process	process	NOUN
ap-5030	58	29	,	,	PUNCT
ap-5030	58	30	ideally	ideally	ADV
ap-5030	58	31	before	before	SCONJ
ap-5030	58	32	any	any	DET
ap-5030	58	33	hardware	hardware	NOUN
ap-5030	58	34	is	be	AUX
ap-5030	58	35	built	build	VERB
ap-5030	58	36	or	or	CCONJ
ap-5030	58	37	code	code	NOUN
ap-5030	58	38	written	write	VERB
ap-5030	58	39	.	.	PUNCT
ap-5030	59	1	this	this	DET
ap-5030	59	2	approach	approach	NOUN
ap-5030	59	3	requires	require	VERB
ap-5030	59	4	a	a	DET
ap-5030	59	5	suitable	suitable	ADJ
ap-5030	59	6	model	model	NOUN
ap-5030	59	7	of	of	ADP
ap-5030	59	8	the	the	DET
ap-5030	59	9	system	system	NOUN
ap-5030	59	10	to	to	PART
ap-5030	59	11	be	be	AUX
ap-5030	59	12	created	create	VERB
ap-5030	59	13	.	.	PUNCT
ap-5030	60	1	the	the	DET
ap-5030	60	2	fpgasecml	fpgasecml	PROPN
ap-5030	60	3	metamodel	metamodel	PROPN
ap-5030	60	4	(	(	PUNCT
ap-5030	60	5	figure	figure	NOUN
ap-5030	60	6	1	1	NUM
ap-5030	60	7	)	)	PUNCT
ap-5030	60	8	defines	define	VERB
ap-5030	60	9	several	several	ADJ
ap-5030	60	10	components	component	NOUN
ap-5030	60	11	that	that	PRON
ap-5030	60	12	can	can	AUX
ap-5030	60	13	be	be	AUX
ap-5030	60	14	instantiated	instantiate	VERB
ap-5030	60	15	and	and	CCONJ
ap-5030	60	16	parameterized	parameterized	ADJ
ap-5030	60	17	for	for	ADP
ap-5030	60	18	this	this	DET
ap-5030	60	19	purpose	purpose	NOUN
ap-5030	60	20	.	.	PUNCT
ap-5030	61	1	fpgamodules	fpgamodule	NOUN
ap-5030	61	2	represent	represent	VERB
ap-5030	61	3	the	the	DET
ap-5030	61	4	computational	computational	ADJ
ap-5030	61	5	parts	part	NOUN
ap-5030	61	6	of	of	ADP
ap-5030	61	7	the	the	DET
ap-5030	61	8	design	design	NOUN
ap-5030	61	9	.	.	PUNCT
ap-5030	62	1	they	they	PRON
ap-5030	62	2	come	come	VERB
ap-5030	62	3	in	in	ADP
ap-5030	62	4	two	two	NUM
ap-5030	62	5	different	different	ADJ
ap-5030	62	6	manifestations	manifestation	NOUN
ap-5030	62	7	:	:	PUNCT
ap-5030	62	8	processing	processing	NOUN
ap-5030	62	9	blocks	block	NOUN
ap-5030	62	10	store	store	NOUN
ap-5030	62	11	and	and	CCONJ
ap-5030	62	12	process	process	NOUN
ap-5030	62	13	data	datum	NOUN
ap-5030	62	14	while	while	SCONJ
ap-5030	62	15	io	io	NOUN
ap-5030	62	16	blocks	block	NOUN
ap-5030	62	17	provide	provide	VERB
ap-5030	62	18	an	an	DET
ap-5030	62	19	additional	additional	ADJ
ap-5030	62	20	connection	connection	NOUN
ap-5030	62	21	to	to	ADP
ap-5030	62	22	the	the	DET
ap-5030	62	23	outside	outside	ADJ
ap-5030	62	24	world	world	NOUN
ap-5030	62	25	.	.	PUNCT
ap-5030	63	1	fpgamodules	fpgamodule	NOUN
ap-5030	63	2	utilize	utilize	VERB
ap-5030	63	3	slots	slot	NOUN
ap-5030	63	4	to	to	PART
ap-5030	63	5	perform	perform	VERB
ap-5030	63	6	their	their	PRON
ap-5030	63	7	tasks	task	NOUN
ap-5030	63	8	.	.	PUNCT
ap-5030	64	1	slots	slot	NOUN
ap-5030	64	2	represent	represent	VERB
ap-5030	64	3	disjunctive	disjunctive	ADJ
ap-5030	64	4	sets	set	NOUN
ap-5030	64	5	of	of	ADP
ap-5030	64	6	fpga	fpga	PROPN
ap-5030	64	7	primitives	primitive	NOUN
ap-5030	64	8	like	like	ADP
ap-5030	64	9	configurable	configurable	ADJ
ap-5030	64	10	logic	logic	NOUN
ap-5030	64	11	blocks	block	NOUN
ap-5030	64	12	and	and	CCONJ
ap-5030	64	13	embedded	embed	VERB
ap-5030	64	14	ram	ram	NOUN
ap-5030	64	15	.	.	PUNCT
ap-5030	65	1	each	each	DET
ap-5030	65	2	slot	slot	NOUN
ap-5030	65	3	can	can	AUX
ap-5030	65	4	be	be	AUX
ap-5030	65	5	used	use	VERB
ap-5030	65	6	by	by	ADP
ap-5030	65	7	only	only	ADV
ap-5030	65	8	one	one	NUM
ap-5030	65	9	fpgamodule	fpgamodule	NOUN
ap-5030	65	10	at	at	ADP
ap-5030	65	11	a	a	DET
ap-5030	65	12	time	time	NOUN
ap-5030	65	13	and	and	CCONJ
ap-5030	65	14	all	all	DET
ap-5030	65	15	fpgamodules	fpgamodule	NOUN
ap-5030	65	16	must	must	AUX
ap-5030	65	17	use	use	VERB
ap-5030	65	18	the	the	DET
ap-5030	65	19	same	same	ADJ
ap-5030	65	20	predefined	predefine	VERB
ap-5030	65	21	slot	slot	NOUN
ap-5030	65	22	every	every	DET
ap-5030	65	23	time	time	NOUN
ap-5030	65	24	.	.	PUNCT
ap-5030	66	1	communication	communication	NOUN
ap-5030	66	2	networks	network	NOUN
ap-5030	66	3	represent	represent	VERB
ap-5030	66	4	the	the	DET
ap-5030	66	5	communication	communication	NOUN
ap-5030	66	6	infrastructure	infrastructure	NOUN
ap-5030	66	7	between	between	ADP
ap-5030	66	8	the	the	DET
ap-5030	66	9	slots	slot	NOUN
ap-5030	66	10	.	.	PUNCT
ap-5030	67	1	they	they	PRON
ap-5030	67	2	are	be	AUX
ap-5030	67	3	directed	direct	VERB
ap-5030	67	4	graphs	graph	NOUN
ap-5030	67	5	with	with	ADP
ap-5030	67	6	the	the	DET
ap-5030	67	7	slots	slot	NOUN
ap-5030	67	8	as	as	ADP
ap-5030	67	9	nodes	node	NOUN
ap-5030	67	10	and	and	CCONJ
ap-5030	67	11	the	the	DET
ap-5030	67	12	possible	possible	ADJ
ap-5030	67	13	data	data	NOUN
ap-5030	67	14	flow	flow	NOUN
ap-5030	67	15	between	between	ADP
ap-5030	67	16	them	they	PRON
ap-5030	67	17	is	be	AUX
ap-5030	67	18	represented	represent	VERB
ap-5030	67	19	by	by	ADP
ap-5030	67	20	the	the	DET
ap-5030	67	21	edges	edge	NOUN
ap-5030	67	22	.	.	PUNCT
ap-5030	68	1	each	each	DET
ap-5030	68	2	change	change	NOUN
ap-5030	68	3	in	in	ADP
ap-5030	68	4	the	the	DET
ap-5030	68	5	slot	slot	NOUN
ap-5030	68	6	utilization	utilization	NOUN
ap-5030	68	7	(	(	PUNCT
ap-5030	68	8	partial	partial	ADJ
ap-5030	68	9	runtime	runtime	NOUN
ap-5030	68	10	reconfiguration	reconfiguration	NOUN
ap-5030	68	11	)	)	PUNCT
ap-5030	68	12	is	be	AUX
ap-5030	68	13	represented	represent	VERB
ap-5030	68	14	by	by	ADP
ap-5030	68	15	a	a	DET
ap-5030	68	16	reconfiguration	reconfiguration	NOUN
ap-5030	68	17	event	event	NOUN
ap-5030	68	18	.	.	PUNCT
ap-5030	69	1	each	each	DET
ap-5030	69	2	reconfiguration	reconfiguration	NOUN
ap-5030	69	3	event	event	NOUN
ap-5030	69	4	has	have	VERB
ap-5030	69	5	a	a	DET
ap-5030	69	6	set	set	NOUN
ap-5030	69	7	of	of	ADP
ap-5030	69	8	fpgamodules	fpgamodule	NOUN
ap-5030	69	9	that	that	PRON
ap-5030	69	10	replace	replace	VERB
ap-5030	69	11	the	the	DET
ap-5030	69	12	fpgamodule	fpgamodule	NOUN
ap-5030	69	13	in	in	ADP
ap-5030	69	14	their	their	PRON
ap-5030	69	15	respective	respective	ADJ
ap-5030	69	16	slot	slot	NOUN
ap-5030	69	17	.	.	PUNCT
ap-5030	70	1	the	the	DET
ap-5030	70	2	only	only	ADJ
ap-5030	70	3	mandatory	mandatory	ADJ
ap-5030	70	4	event	event	NOUN
ap-5030	70	5	evinit	evinit	NOUN
ap-5030	70	6	marks	mark	VERB
ap-5030	70	7	the	the	DET
ap-5030	70	8	initial	initial	ADJ
ap-5030	70	9	configuration	configuration	NOUN
ap-5030	70	10	of	of	ADP
ap-5030	70	11	the	the	DET
ap-5030	70	12	fpga	fpga	NOUN
ap-5030	70	13	and	and	CCONJ
ap-5030	70	14	must	must	AUX
ap-5030	70	15	provide	provide	VERB
ap-5030	70	16	one	one	NUM
ap-5030	70	17	fpgamodule	fpgamodule	NOUN
ap-5030	70	18	for	for	ADP
ap-5030	70	19	each	each	DET
ap-5030	70	20	slot	slot	NOUN
ap-5030	70	21	.	.	PUNCT
ap-5030	71	1	4	4	X
ap-5030	71	2	.	.	X
ap-5030	71	3	the	the	DET
ap-5030	71	4	mdp	mdp	PROPN
ap-5030	71	5	-	-	PUNCT
ap-5030	71	6	representation	representation	NOUN
ap-5030	71	7	of	of	ADP
ap-5030	71	8	the	the	DET
ap-5030	71	9	fpga	fpga	PROPN
ap-5030	71	10	design	design	VERB
ap-5030	71	11	our	our	PRON
ap-5030	71	12	model	model	NOUN
ap-5030	71	13	-	-	PUNCT
ap-5030	71	14	to	to	ADP
ap-5030	71	15	-	-	PUNCT
ap-5030	71	16	model	model	NOUN
ap-5030	71	17	generator	generator	NOUN
ap-5030	71	18	translates	translate	VERB
ap-5030	71	19	each	each	DET
ap-5030	71	20	valid	valid	ADJ
ap-5030	71	21	fpgasecml	fpgasecml	NOUN
ap-5030	71	22	description	description	NOUN
ap-5030	71	23	into	into	ADP
ap-5030	71	24	a	a	DET
ap-5030	71	25	computable	computable	ADJ
ap-5030	71	26	markov	markov	NOUN
ap-5030	71	27	decision	decision	NOUN
ap-5030	71	28	processes	process	NOUN
ap-5030	71	29	model	model	NOUN
ap-5030	71	30	.	.	PUNCT
ap-5030	72	1	this	this	DET
ap-5030	72	2	section	section	NOUN
ap-5030	72	3	presents	present	VERB
ap-5030	72	4	a	a	DET
ap-5030	72	5	short	short	ADJ
ap-5030	72	6	introduction	introduction	NOUN
ap-5030	72	7	into	into	ADP
ap-5030	72	8	mdps	mdps	PROPN
ap-5030	72	9	in	in	ADP
ap-5030	72	10	general	general	ADJ
ap-5030	72	11	,	,	PUNCT
ap-5030	72	12	and	and	CCONJ
ap-5030	72	13	the	the	DET
ap-5030	72	14	simplifications	simplification	NOUN
ap-5030	72	15	made	make	VERB
ap-5030	72	16	to	to	PART
ap-5030	72	17	create	create	VERB
ap-5030	72	18	a	a	DET
ap-5030	72	19	suitable	suitable	ADJ
ap-5030	72	20	representation	representation	NOUN
ap-5030	72	21	of	of	ADP
ap-5030	72	22	the	the	DET
ap-5030	72	23	problem	problem	NOUN
ap-5030	72	24	at	at	ADP
ap-5030	72	25	hand	hand	NOUN
ap-5030	72	26	.	.	PUNCT
ap-5030	73	1	we	we	PRON
ap-5030	73	2	further	far	ADV
ap-5030	73	3	discuss	discuss	VERB
ap-5030	73	4	the	the	DET
ap-5030	73	5	different	different	ADJ
ap-5030	73	6	components	component	NOUN
ap-5030	73	7	state	state	NOUN
ap-5030	73	8	,	,	PUNCT
ap-5030	73	9	action	action	NOUN
ap-5030	73	10	,	,	PUNCT
ap-5030	73	11	reward	reward	NOUN
ap-5030	73	12	and	and	CCONJ
ap-5030	73	13	terminal	terminal	ADJ
ap-5030	73	14	states	state	NOUN
ap-5030	73	15	of	of	ADP
ap-5030	73	16	our	our	PRON
ap-5030	73	17	model	model	NOUN
ap-5030	73	18	(	(	PUNCT
ap-5030	73	19	figure	figure	NOUN
ap-5030	73	20	2	2	NUM
ap-5030	73	21	)	)	PUNCT
ap-5030	73	22	and	and	CCONJ
ap-5030	73	23	its	its	PRON
ap-5030	73	24	implementation	implementation	NOUN
ap-5030	73	25	.	.	PUNCT
ap-5030	74	1	4.1	4.1	NUM
ap-5030	74	2	.	.	PUNCT
ap-5030	74	3	scope	scope	NOUN
ap-5030	74	4	and	and	CCONJ
ap-5030	74	5	constraints	constraint	NOUN
ap-5030	74	6	of	of	ADP
ap-5030	74	7	the	the	DET
ap-5030	74	8	model	model	NOUN
ap-5030	74	9	we	we	PRON
ap-5030	74	10	assume	assume	VERB
ap-5030	74	11	that	that	SCONJ
ap-5030	74	12	the	the	DET
ap-5030	74	13	attacker	attacker	NOUN
ap-5030	74	14	has	have	VERB
ap-5030	74	15	only	only	ADV
ap-5030	74	16	one	one	NUM
ap-5030	74	17	goal	goal	NOUN
ap-5030	74	18	:	:	PUNCT
ap-5030	74	19	the	the	DET
ap-5030	74	20	exfiltration	exfiltration	NOUN
ap-5030	74	21	of	of	ADP
ap-5030	74	22	all	all	DET
ap-5030	74	23	data	datum	NOUN
ap-5030	74	24	stored	store	VERB
ap-5030	74	25	within	within	ADP
ap-5030	74	26	the	the	DET
ap-5030	74	27	fpga	fpga	NOUN
ap-5030	74	28	.	.	PUNCT
ap-5030	75	1	we	we	PRON
ap-5030	75	2	further	far	ADV
ap-5030	75	3	assume	assume	VERB
ap-5030	75	4	that	that	SCONJ
ap-5030	75	5	the	the	DET
ap-5030	75	6	agent	agent	NOUN
ap-5030	75	7	has	have	VERB
ap-5030	75	8	a	a	DET
ap-5030	75	9	full	full	ADJ
ap-5030	75	10	access	access	NOUN
ap-5030	75	11	to	to	ADP
ap-5030	75	12	all	all	DET
ap-5030	75	13	peripheral	peripheral	ADJ
ap-5030	75	14	devices	device	NOUN
ap-5030	75	15	including	include	VERB
ap-5030	75	16	those	those	PRON
ap-5030	75	17	containing	contain	VERB
ap-5030	75	18	the	the	DET
ap-5030	75	19	fpgas	fpgas	NOUN
ap-5030	75	20	configuration	configuration	NOUN
ap-5030	75	21	.	.	PUNCT
ap-5030	76	1	each	each	DET
ap-5030	76	2	action	action	NOUN
ap-5030	76	3	in	in	ADP
ap-5030	76	4	the	the	DET
ap-5030	76	5	mdp	mdp	NOUN
ap-5030	76	6	-	-	PUNCT
ap-5030	76	7	model	model	NOUN
ap-5030	76	8	represents	represent	VERB
ap-5030	76	9	a	a	DET
ap-5030	76	10	whole	whole	ADJ
ap-5030	76	11	class	class	NOUN
ap-5030	76	12	of	of	ADP
ap-5030	76	13	attacks	attack	NOUN
ap-5030	76	14	,	,	PUNCT
ap-5030	76	15	many	many	ADJ
ap-5030	76	16	(	(	PUNCT
ap-5030	76	17	low	low	ADJ
ap-5030	76	18	-	-	PUNCT
ap-5030	76	19	level	level	NOUN
ap-5030	76	20	)	)	PUNCT
ap-5030	76	21	operations	operation	NOUN
ap-5030	76	22	might	might	AUX
ap-5030	76	23	be	be	AUX
ap-5030	76	24	necessary	necessary	ADJ
ap-5030	76	25	to	to	PART
ap-5030	76	26	execute	execute	VERB
ap-5030	76	27	each	each	PRON
ap-5030	76	28	of	of	ADP
ap-5030	76	29	these	these	DET
ap-5030	76	30	actions	action	NOUN
ap-5030	76	31	in	in	ADP
ap-5030	76	32	a	a	DET
ap-5030	76	33	real	real	ADJ
ap-5030	76	34	-	-	PUNCT
ap-5030	76	35	world	world	NOUN
ap-5030	76	36	case	case	NOUN
ap-5030	76	37	.	.	PUNCT
ap-5030	77	1	the	the	DET
ap-5030	77	2	model	model	NOUN
ap-5030	77	3	does	do	AUX
ap-5030	77	4	not	not	PART
ap-5030	77	5	support	support	VERB
ap-5030	77	6	side	side	ADJ
ap-5030	77	7	effects	effect	NOUN
ap-5030	77	8	of	of	ADP
ap-5030	77	9	attacks	attack	NOUN
ap-5030	77	10	like	like	ADP
ap-5030	77	11	a	a	DET
ap-5030	77	12	damaged	damage	VERB
ap-5030	77	13	configuration	configuration	NOUN
ap-5030	77	14	.	.	PUNCT
ap-5030	78	1	we	we	PRON
ap-5030	78	2	further	far	ADV
ap-5030	78	3	grant	grant	VERB
ap-5030	78	4	the	the	DET
ap-5030	78	5	attacker	attacker	NOUN
ap-5030	78	6	total	total	NOUN
ap-5030	78	7	knowledge	knowledge	NOUN
ap-5030	78	8	over	over	ADP
ap-5030	78	9	the	the	DET
ap-5030	78	10	systems	system	NOUN
ap-5030	78	11	state	state	NOUN
ap-5030	78	12	,	,	PUNCT
ap-5030	78	13	and	and	CCONJ
ap-5030	78	14	this	this	DET
ap-5030	78	15	state	state	NOUN
ap-5030	78	16	depends	depend	VERB
ap-5030	78	17	solely	solely	ADV
ap-5030	78	18	on	on	ADP
ap-5030	78	19	the	the	DET
ap-5030	78	20	actions	action	NOUN
ap-5030	78	21	of	of	ADP
ap-5030	78	22	the	the	DET
ap-5030	78	23	attacker	attacker	PROPN
ap-5030	78	24	.	.	PUNCT
ap-5030	79	1	methods	method	NOUN
ap-5030	79	2	to	to	PART
ap-5030	79	3	address	address	VERB
ap-5030	79	4	these	these	DET
ap-5030	79	5	constraints	constraint	NOUN
ap-5030	79	6	are	be	AUX
ap-5030	79	7	presented	present	VERB
ap-5030	79	8	later	later	ADV
ap-5030	79	9	.	.	PUNCT
ap-5030	80	1	4.2	4.2	NUM
ap-5030	80	2	.	.	PUNCT
ap-5030	81	1	generic	generic	ADJ
ap-5030	81	2	components	component	NOUN
ap-5030	81	3	of	of	ADP
ap-5030	81	4	markov	markov	NOUN
ap-5030	81	5	decision	decision	NOUN
ap-5030	81	6	processes	process	VERB
ap-5030	81	7	each	each	DET
ap-5030	81	8	state	state	NOUN
ap-5030	81	9	of	of	ADP
ap-5030	81	10	the	the	DET
ap-5030	81	11	markov	markov	NOUN
ap-5030	81	12	decision	decision	NOUN
ap-5030	81	13	processes	process	NOUN
ap-5030	81	14	(	(	PUNCT
ap-5030	81	15	mdp	mdp	NOUN
ap-5030	81	16	)	)	PUNCT
ap-5030	81	17	contains	contain	VERB
ap-5030	81	18	all	all	DET
ap-5030	81	19	the	the	DET
ap-5030	81	20	information	information	NOUN
ap-5030	81	21	necessary	necessary	ADJ
ap-5030	81	22	to	to	PART
ap-5030	81	23	determine	determine	VERB
ap-5030	81	24	the	the	DET
ap-5030	81	25	next	next	ADJ
ap-5030	81	26	state	state	NOUN
ap-5030	81	27	as	as	ADV
ap-5030	81	28	well	well	ADV
ap-5030	81	29	as	as	ADP
ap-5030	81	30	the	the	DET
ap-5030	81	31	agent	agent	NOUN
ap-5030	81	32	’s	’s	PART
ap-5030	81	33	reward	reward	NOUN
ap-5030	81	34	after	after	SCONJ
ap-5030	81	35	an	an	DET
ap-5030	81	36	action	action	NOUN
ap-5030	81	37	is	be	AUX
ap-5030	81	38	taken	take	VERB
ap-5030	81	39	.	.	PUNCT
ap-5030	82	1	each	each	DET
ap-5030	82	2	mdp	mdp	PROPN
ap-5030	82	3	has	have	VERB
ap-5030	82	4	one	one	NUM
ap-5030	82	5	start	start	NOUN
ap-5030	82	6	state	state	NOUN
ap-5030	82	7	and	and	CCONJ
ap-5030	82	8	at	at	ADV
ap-5030	82	9	least	least	ADV
ap-5030	82	10	one	one	NUM
ap-5030	82	11	terminal	terminal	ADJ
ap-5030	82	12	state	state	NOUN
ap-5030	82	13	.	.	PUNCT
ap-5030	83	1	the	the	DET
ap-5030	83	2	agent	agent	NOUN
ap-5030	83	3	can	can	AUX
ap-5030	83	4	choose	choose	VERB
ap-5030	83	5	from	from	ADP
ap-5030	83	6	a	a	DET
ap-5030	83	7	finite	finite	ADJ
ap-5030	83	8	set	set	NOUN
ap-5030	83	9	of	of	ADP
ap-5030	83	10	actions	action	NOUN
ap-5030	83	11	to	to	PART
ap-5030	83	12	change	change	VERB
ap-5030	83	13	the	the	DET
ap-5030	83	14	state	state	NOUN
ap-5030	83	15	of	of	ADP
ap-5030	83	16	the	the	DET
ap-5030	83	17	system	system	NOUN
ap-5030	83	18	.	.	PUNCT
ap-5030	84	1	executing	execute	VERB
ap-5030	84	2	this	this	DET
ap-5030	84	3	action	action	NOUN
ap-5030	84	4	results	result	NOUN
ap-5030	84	5	in	in	ADP
ap-5030	84	6	the	the	DET
ap-5030	84	7	transition	transition	NOUN
ap-5030	84	8	from	from	ADP
ap-5030	84	9	the	the	DET
ap-5030	84	10	state	state	NOUN
ap-5030	84	11	s	s	VERB
ap-5030	84	12	to	to	PART
ap-5030	84	13	state	state	VERB
ap-5030	84	14	s′.	s′.	PROPN
ap-5030	84	15	all	all	DET
ap-5030	84	16	actions	action	NOUN
ap-5030	84	17	are	be	AUX
ap-5030	84	18	atomic	atomic	ADJ
ap-5030	84	19	and	and	CCONJ
ap-5030	84	20	require	require	VERB
ap-5030	84	21	the	the	DET
ap-5030	84	22	same	same	ADJ
ap-5030	84	23	amount	amount	NOUN
ap-5030	84	24	of	of	ADP
ap-5030	84	25	time	time	NOUN
ap-5030	84	26	.	.	PUNCT
ap-5030	85	1	the	the	DET
ap-5030	85	2	probability	probability	NOUN
ap-5030	85	3	pa(s	pa(s	NOUN
ap-5030	85	4	,	,	PUNCT
ap-5030	85	5	s′	s′	X
ap-5030	85	6	)	)	PUNCT
ap-5030	85	7	specifies	specify	VERB
ap-5030	85	8	the	the	DET
ap-5030	85	9	likelihood	likelihood	NOUN
ap-5030	85	10	that	that	SCONJ
ap-5030	85	11	an	an	DET
ap-5030	85	12	action	action	NOUN
ap-5030	85	13	a	a	PRON
ap-5030	85	14	in	in	ADP
ap-5030	85	15	the	the	DET
ap-5030	85	16	state	state	NOUN
ap-5030	85	17	s	s	NOUN
ap-5030	85	18	leads	lead	VERB
ap-5030	85	19	to	to	ADP
ap-5030	85	20	a	a	DET
ap-5030	85	21	transition	transition	NOUN
ap-5030	85	22	into	into	ADP
ap-5030	85	23	the	the	DET
ap-5030	85	24	state	state	NOUN
ap-5030	85	25	s′.	s′.	PROPN
ap-5030	86	1	the	the	DET
ap-5030	86	2	reward	reward	NOUN
ap-5030	86	3	ra(s	ra(s	NOUN
ap-5030	86	4	,	,	PUNCT
ap-5030	86	5	s′	s′	X
ap-5030	86	6	)	)	PUNCT
ap-5030	86	7	informs	inform	VERB
ap-5030	86	8	the	the	DET
ap-5030	86	9	agent	agent	NOUN
ap-5030	86	10	which	which	PRON
ap-5030	86	11	transitions	transition	NOUN
ap-5030	86	12	are	be	AUX
ap-5030	86	13	desirable	desirable	ADJ
ap-5030	86	14	and	and	CCONJ
ap-5030	86	15	which	which	PRON
ap-5030	86	16	are	be	AUX
ap-5030	86	17	not	not	PART
ap-5030	86	18	.	.	PUNCT
ap-5030	87	1	the	the	DET
ap-5030	87	2	agent	agent	NOUN
ap-5030	87	3	receives	receive	VERB
ap-5030	87	4	its	its	PRON
ap-5030	87	5	reward	reward	NOUN
ap-5030	87	6	immediately	immediately	ADV
ap-5030	87	7	after	after	SCONJ
ap-5030	87	8	the	the	DET
ap-5030	87	9	transition	transition	NOUN
ap-5030	87	10	into	into	ADP
ap-5030	87	11	s′.	s′.	PROPN
ap-5030	87	12	rewards	reward	NOUN
ap-5030	87	13	can	can	AUX
ap-5030	87	14	also	also	ADV
ap-5030	87	15	be	be	AUX
ap-5030	87	16	negative	negative	ADJ
ap-5030	87	17	(	(	PUNCT
ap-5030	87	18	punishments	punishment	NOUN
ap-5030	87	19	)	)	PUNCT
ap-5030	87	20	or	or	CCONJ
ap-5030	87	21	zero	zero	NUM
ap-5030	87	22	.	.	PUNCT
ap-5030	88	1	γ	γ	PROPN
ap-5030	88	2	or	or	CCONJ
ap-5030	88	3	gamma	gamma	NOUN
ap-5030	88	4	is	be	AUX
ap-5030	88	5	a	a	DET
ap-5030	88	6	discount	discount	NOUN
ap-5030	88	7	factor	factor	NOUN
ap-5030	88	8	for	for	ADP
ap-5030	88	9	future	future	ADJ
ap-5030	88	10	rewards	reward	NOUN
ap-5030	88	11	(	(	PUNCT
ap-5030	88	12	we	we	PRON
ap-5030	88	13	use	use	VERB
ap-5030	88	14	the	the	DET
ap-5030	88	15	traditional	traditional	ADJ
ap-5030	88	16	value	value	NOUN
ap-5030	88	17	of	of	ADP
ap-5030	88	18	γ	γ	X
ap-5030	88	19	=	=	SYM
ap-5030	88	20	0.7	0.7	NUM
ap-5030	88	21	)	)	PUNCT
ap-5030	88	22	.	.	PUNCT
ap-5030	89	1	519	519	NUM
ap-5030	89	2	michael	michael	PROPN
ap-5030	89	3	vetter	vetter	PROPN
ap-5030	89	4	acta	acta	PROPN
ap-5030	89	5	polytechnica	polytechnica	PROPN
ap-5030	89	6	figure	figure	NOUN
ap-5030	89	7	1	1	NUM
ap-5030	89	8	.	.	PUNCT
ap-5030	90	1	fpga	fpga	PROPN
ap-5030	90	2	primitives	primitive	NOUN
ap-5030	90	3	(	(	PUNCT
ap-5030	90	4	bottom	bottom	ADJ
ap-5030	90	5	layer	layer	NOUN
ap-5030	90	6	)	)	PUNCT
ap-5030	90	7	and	and	CCONJ
ap-5030	90	8	their	their	PRON
ap-5030	90	9	abstract	abstract	ADJ
ap-5030	90	10	representation	representation	NOUN
ap-5030	90	11	in	in	ADP
ap-5030	90	12	fpgasecml	fpgasecml	NOUN
ap-5030	90	13	(	(	PUNCT
ap-5030	90	14	top	top	ADJ
ap-5030	90	15	layer	layer	NOUN
ap-5030	90	16	)	)	PUNCT
ap-5030	90	17	.	.	PUNCT
ap-5030	91	1	figure	figure	NOUN
ap-5030	91	2	2	2	NUM
ap-5030	91	3	.	.	PUNCT
ap-5030	91	4	transformation	transformation	NOUN
ap-5030	91	5	flow	flow	NOUN
ap-5030	91	6	from	from	ADP
ap-5030	91	7	the	the	DET
ap-5030	91	8	fpgasecml	fpgasecml	NOUN
ap-5030	91	9	to	to	ADP
ap-5030	91	10	the	the	DET
ap-5030	91	11	burlap	burlap	NOUN
ap-5030	91	12	based	base	VERB
ap-5030	91	13	reinforcement	reinforcement	NOUN
ap-5030	91	14	learning	learning	NOUN
ap-5030	91	15	model	model	NOUN
ap-5030	91	16	.	.	PUNCT
ap-5030	92	1	4.3	4.3	NUM
ap-5030	92	2	.	.	PUNCT
ap-5030	93	1	the	the	DET
ap-5030	93	2	domain	domain	NOUN
ap-5030	93	3	-	-	PUNCT
ap-5030	93	4	specific	specific	ADJ
ap-5030	93	5	markov	markov	NOUN
ap-5030	93	6	decision	decision	NOUN
ap-5030	93	7	process	process	NOUN
ap-5030	93	8	this	this	DET
ap-5030	93	9	section	section	NOUN
ap-5030	93	10	describes	describe	VERB
ap-5030	93	11	how	how	SCONJ
ap-5030	93	12	the	the	DET
ap-5030	93	13	fpgasecml	fpgasecml	NOUN
ap-5030	93	14	metamodel	metamodel	NOUN
ap-5030	93	15	is	be	AUX
ap-5030	93	16	translated	translate	VERB
ap-5030	93	17	into	into	ADP
ap-5030	93	18	the	the	DET
ap-5030	93	19	required	require	VERB
ap-5030	93	20	mdp	mdp	NOUN
ap-5030	93	21	components	component	NOUN
ap-5030	93	22	.	.	PUNCT
ap-5030	94	1	4.3.1	4.3.1	X
ap-5030	94	2	.	.	PUNCT
ap-5030	95	1	state	state	NOUN
ap-5030	95	2	the	the	DET
ap-5030	95	3	state	state	NOUN
ap-5030	95	4	contains	contain	VERB
ap-5030	95	5	the	the	DET
ap-5030	95	6	current	current	ADJ
ap-5030	95	7	configuration	configuration	NOUN
ap-5030	95	8	of	of	ADP
ap-5030	95	9	the	the	DET
ap-5030	95	10	fpga	fpga	NOUN
ap-5030	95	11	,	,	PUNCT
ap-5030	95	12	which	which	DET
ap-5030	95	13	fpgamodule	fpgamodule	NOUN
ap-5030	95	14	is	be	AUX
ap-5030	95	15	under	under	ADP
ap-5030	95	16	the	the	DET
ap-5030	95	17	control	control	NOUN
ap-5030	95	18	of	of	ADP
ap-5030	95	19	the	the	DET
ap-5030	95	20	attacker	attacker	NOUN
ap-5030	95	21	and	and	CCONJ
ap-5030	95	22	which	which	DET
ap-5030	95	23	slot	slot	NOUN
ap-5030	95	24	contains	contain	VERB
ap-5030	95	25	data	datum	NOUN
ap-5030	95	26	already	already	ADV
ap-5030	95	27	exfiltrated	exfiltrate	VERB
ap-5030	95	28	from	from	ADP
ap-5030	95	29	another	another	DET
ap-5030	95	30	fpgamodule	fpgamodule	NOUN
ap-5030	95	31	(	(	PUNCT
ap-5030	95	32	chain	chain	NOUN
ap-5030	95	33	exfiltration	exfiltration	NOUN
ap-5030	95	34	)	)	PUNCT
ap-5030	95	35	.	.	PUNCT
ap-5030	96	1	the	the	DET
ap-5030	96	2	state	state	NOUN
ap-5030	96	3	further	further	ADJ
ap-5030	96	4	encodes	encode	NOUN
ap-5030	96	5	which	which	PRON
ap-5030	96	6	fpgamodules	fpgamodule	VERB
ap-5030	96	7	storage	storage	NOUN
ap-5030	96	8	has	have	AUX
ap-5030	96	9	been	be	AUX
ap-5030	96	10	tampered	tamper	VERB
ap-5030	96	11	with	with	ADP
ap-5030	96	12	and	and	CCONJ
ap-5030	96	13	which	which	DET
ap-5030	96	14	data	data	NOUN
ap-5030	96	15	has	have	AUX
ap-5030	96	16	already	already	ADV
ap-5030	96	17	been	be	AUX
ap-5030	96	18	leaked	leak	VERB
ap-5030	96	19	into	into	ADP
ap-5030	96	20	the	the	DET
ap-5030	96	21	outside	outside	ADJ
ap-5030	96	22	world	world	NOUN
ap-5030	96	23	.	.	PUNCT
ap-5030	97	1	all	all	DET
ap-5030	97	2	the	the	DET
ap-5030	97	3	information	information	NOUN
ap-5030	97	4	are	be	AUX
ap-5030	97	5	stored	store	VERB
ap-5030	97	6	in	in	ADP
ap-5030	97	7	integer	integer	NOUN
ap-5030	97	8	or	or	CCONJ
ap-5030	97	9	boolean	boolean	ADJ
ap-5030	97	10	variables	variable	NOUN
ap-5030	97	11	.	.	PUNCT
ap-5030	98	1	the	the	DET
ap-5030	98	2	fpga	fpga	PROPN
ap-5030	98	3	starts	start	VERB
ap-5030	98	4	in	in	ADP
ap-5030	98	5	its	its	PRON
ap-5030	98	6	initial	initial	ADJ
ap-5030	98	7	configuration	configuration	NOUN
ap-5030	98	8	(	(	PUNCT
ap-5030	98	9	as	as	SCONJ
ap-5030	98	10	defined	define	VERB
ap-5030	98	11	by	by	ADP
ap-5030	98	12	evinit	evinit	NOUN
ap-5030	98	13	)	)	PUNCT
ap-5030	98	14	and	and	CCONJ
ap-5030	98	15	all	all	DET
ap-5030	98	16	fpgamodules	fpgamodule	NOUN
ap-5030	98	17	untampered	untampered	ADJ
ap-5030	98	18	with	with	ADP
ap-5030	98	19	.	.	PUNCT
ap-5030	99	1	terminal	terminal	ADJ
ap-5030	99	2	states	state	NOUN
ap-5030	99	3	are	be	AUX
ap-5030	99	4	,	,	PUNCT
ap-5030	99	5	by	by	ADP
ap-5030	99	6	default	default	NOUN
ap-5030	99	7	,	,	PUNCT
ap-5030	99	8	all	all	DET
ap-5030	99	9	states	state	NOUN
ap-5030	99	10	where	where	SCONJ
ap-5030	99	11	the	the	DET
ap-5030	99	12	agent	agent	NOUN
ap-5030	99	13	has	have	AUX
ap-5030	99	14	exfiltrated	exfiltrate	VERB
ap-5030	99	15	the	the	DET
ap-5030	99	16	data	datum	NOUN
ap-5030	99	17	of	of	ADP
ap-5030	99	18	each	each	DET
ap-5030	99	19	fpgamodule	fpgamodule	NOUN
ap-5030	99	20	into	into	ADP
ap-5030	99	21	the	the	DET
ap-5030	99	22	outside	outside	ADJ
ap-5030	99	23	world	world	NOUN
ap-5030	99	24	.	.	PUNCT
ap-5030	100	1	the	the	DET
ap-5030	100	2	code	code	NOUN
ap-5030	100	3	generator	generator	NOUN
ap-5030	100	4	does	do	AUX
ap-5030	100	5	not	not	PART
ap-5030	100	6	validate	validate	VERB
ap-5030	100	7	whether	whether	SCONJ
ap-5030	100	8	the	the	DET
ap-5030	100	9	agent	agent	NOUN
ap-5030	100	10	can	can	AUX
ap-5030	100	11	reach	reach	VERB
ap-5030	100	12	a	a	DET
ap-5030	100	13	terminal	terminal	ADJ
ap-5030	100	14	state	state	NOUN
ap-5030	100	15	.	.	PUNCT
ap-5030	101	1	4.3.2	4.3.2	X
ap-5030	101	2	.	.	PUNCT
ap-5030	102	1	actions	action	NOUN
ap-5030	102	2	the	the	DET
ap-5030	102	3	agent	agent	NOUN
ap-5030	102	4	can	can	AUX
ap-5030	102	5	invoke	invoke	VERB
ap-5030	102	6	one	one	NUM
ap-5030	102	7	of	of	ADP
ap-5030	102	8	the	the	DET
ap-5030	102	9	(	(	PUNCT
ap-5030	102	10	pre	pre	ADJ
ap-5030	102	11	-	-	VERB
ap-5030	102	12	defined	defined	ADJ
ap-5030	102	13	)	)	PUNCT
ap-5030	102	14	actions	action	NOUN
ap-5030	102	15	to	to	PART
ap-5030	102	16	get	get	VERB
ap-5030	102	17	and	and	CCONJ
ap-5030	102	18	extend	extend	VERB
ap-5030	102	19	control	control	NOUN
ap-5030	102	20	over	over	ADP
ap-5030	102	21	the	the	DET
ap-5030	102	22	system	system	NOUN
ap-5030	102	23	.	.	PUNCT
ap-5030	103	1	possible	possible	ADJ
ap-5030	103	2	actions	action	NOUN
ap-5030	103	3	include	include	VERB
ap-5030	103	4	the	the	DET
ap-5030	103	5	reconfiguration	reconfiguration	NOUN
ap-5030	103	6	of	of	ADP
ap-5030	103	7	the	the	DET
ap-5030	103	8	system	system	NOUN
ap-5030	103	9	,	,	PUNCT
ap-5030	103	10	tampering	tamper	VERB
ap-5030	103	11	with	with	ADP
ap-5030	103	12	a	a	DET
ap-5030	103	13	distinct	distinct	ADJ
ap-5030	103	14	fpgamodule	fpgamodule	NOUN
ap-5030	103	15	(	(	PUNCT
ap-5030	103	16	inside	inside	ADP
ap-5030	103	17	the	the	DET
ap-5030	103	18	fpga	fpga	NOUN
ap-5030	103	19	or	or	CCONJ
ap-5030	103	20	in	in	ADP
ap-5030	103	21	its	its	PRON
ap-5030	103	22	storage	storage	NOUN
ap-5030	103	23	element	element	NOUN
ap-5030	103	24	)	)	PUNCT
ap-5030	103	25	and	and	CCONJ
ap-5030	103	26	the	the	DET
ap-5030	103	27	extraction	extraction	NOUN
ap-5030	103	28	of	of	ADP
ap-5030	103	29	data	datum	NOUN
ap-5030	103	30	from	from	ADP
ap-5030	103	31	a	a	DET
ap-5030	103	32	tampered	tamper	VERB
ap-5030	103	33	-	-	PUNCT
ap-5030	103	34	with	with	ADP
ap-5030	103	35	fpgamodule	fpgamodule	NOUN
ap-5030	103	36	.	.	PUNCT
ap-5030	104	1	the	the	DET
ap-5030	104	2	attacker	attacker	NOUN
ap-5030	104	3	can	can	AUX
ap-5030	104	4	further	far	ADV
ap-5030	104	5	transfer	transfer	VERB
ap-5030	104	6	these	these	DET
ap-5030	104	7	extracted	extract	VERB
ap-5030	104	8	data	datum	NOUN
ap-5030	104	9	to	to	ADP
ap-5030	104	10	other	other	ADJ
ap-5030	104	11	slots	slot	NOUN
ap-5030	104	12	or	or	CCONJ
ap-5030	104	13	the	the	DET
ap-5030	104	14	outside	outside	ADJ
ap-5030	104	15	world	world	NOUN
ap-5030	104	16	(	(	PUNCT
ap-5030	104	17	if	if	SCONJ
ap-5030	104	18	the	the	DET
ap-5030	104	19	data	data	NOUN
ap-5030	104	20	is	be	AUX
ap-5030	104	21	in	in	ADP
ap-5030	104	22	an	an	DET
ap-5030	104	23	ioblock	ioblock	NOUN
ap-5030	104	24	that	that	SCONJ
ap-5030	104	25	the	the	DET
ap-5030	104	26	attacker	attacker	NOUN
ap-5030	104	27	has	have	AUX
ap-5030	104	28	already	already	ADV
ap-5030	104	29	tampered	tamper	VERB
ap-5030	104	30	with	with	ADP
ap-5030	104	31	.	.	PUNCT
ap-5030	104	32	)	)	PUNCT
ap-5030	105	1	the	the	DET
ap-5030	105	2	mdp	mdp	NOUN
ap-5030	105	3	moves	move	VERB
ap-5030	105	4	into	into	ADP
ap-5030	105	5	the	the	DET
ap-5030	105	6	new	new	ADJ
ap-5030	105	7	state	state	NOUN
ap-5030	105	8	if	if	SCONJ
ap-5030	105	9	the	the	DET
ap-5030	105	10	action	action	NOUN
ap-5030	105	11	is	be	AUX
ap-5030	105	12	successful	successful	ADJ
ap-5030	105	13	and	and	CCONJ
ap-5030	105	14	remains	remain	VERB
ap-5030	105	15	in	in	ADP
ap-5030	105	16	the	the	DET
ap-5030	105	17	current	current	ADJ
ap-5030	105	18	state	state	NOUN
ap-5030	105	19	(	(	PUNCT
ap-5030	105	20	s′	s′	PROPN
ap-5030	105	21	=	=	SYM
ap-5030	105	22	s	s	X
ap-5030	105	23	)	)	PUNCT
ap-5030	105	24	otherwise	otherwise	ADV
ap-5030	105	25	.	.	PUNCT
ap-5030	106	1	4.3.3	4.3.3	X
ap-5030	106	2	.	.	PUNCT
ap-5030	107	1	reward	reward	VERB
ap-5030	107	2	the	the	DET
ap-5030	107	3	agent	agent	NOUN
ap-5030	107	4	receives	receive	VERB
ap-5030	107	5	a	a	DET
ap-5030	107	6	positive	positive	ADJ
ap-5030	107	7	reward	reward	NOUN
ap-5030	107	8	for	for	ADP
ap-5030	107	9	reaching	reach	VERB
ap-5030	107	10	a	a	DET
ap-5030	107	11	terminal	terminal	ADJ
ap-5030	107	12	state	state	NOUN
ap-5030	107	13	and	and	CCONJ
ap-5030	107	14	either	either	CCONJ
ap-5030	107	15	none	none	NOUN
ap-5030	107	16	or	or	CCONJ
ap-5030	107	17	a	a	DET
ap-5030	107	18	negative	negative	ADJ
ap-5030	107	19	reward	reward	NOUN
ap-5030	107	20	520	520	NUM
ap-5030	107	21	vol	vol	NOUN
ap-5030	107	22	.	.	PUNCT
ap-5030	108	1	59	59	NUM
ap-5030	108	2	no	no	NOUN
ap-5030	108	3	.	.	PUNCT
ap-5030	109	1	5/2019	5/2019	NUM
ap-5030	109	2	security	security	NOUN
ap-5030	109	3	analysis	analysis	NOUN
ap-5030	109	4	of	of	ADP
ap-5030	109	5	fpgas	fpgas	NOUN
ap-5030	109	6	designs	design	NOUN
ap-5030	109	7	through	through	ADP
ap-5030	109	8	reinforcement	reinforcement	NOUN
ap-5030	109	9	learning	learning	NOUN
ap-5030	109	10	(	(	PUNCT
ap-5030	109	11	punishment	punishment	NOUN
ap-5030	109	12	)	)	PUNCT
ap-5030	109	13	for	for	ADP
ap-5030	109	14	every	every	DET
ap-5030	109	15	other	other	ADJ
ap-5030	109	16	step	step	NOUN
ap-5030	109	17	.	.	PUNCT
ap-5030	110	1	this	this	DET
ap-5030	110	2	penalty	penalty	NOUN
ap-5030	110	3	encourages	encourage	VERB
ap-5030	110	4	the	the	DET
ap-5030	110	5	agent	agent	NOUN
ap-5030	110	6	to	to	PART
ap-5030	110	7	use	use	VERB
ap-5030	110	8	shorter	short	ADJ
ap-5030	110	9	sequences	sequence	NOUN
ap-5030	110	10	of	of	ADP
ap-5030	110	11	low	low	ADJ
ap-5030	110	12	-	-	PUNCT
ap-5030	110	13	cost	cost	NOUN
ap-5030	110	14	actions	action	NOUN
ap-5030	110	15	and	and	CCONJ
ap-5030	110	16	constraints	constraint	VERB
ap-5030	110	17	the	the	DET
ap-5030	110	18	search	search	NOUN
ap-5030	110	19	space	space	NOUN
ap-5030	110	20	.	.	PUNCT
ap-5030	111	1	it	it	PRON
ap-5030	111	2	can	can	AUX
ap-5030	111	3	be	be	AUX
ap-5030	111	4	the	the	DET
ap-5030	111	5	same	same	ADJ
ap-5030	111	6	for	for	ADP
ap-5030	111	7	all	all	DET
ap-5030	111	8	actions	action	NOUN
ap-5030	111	9	but	but	CCONJ
ap-5030	111	10	it	it	PRON
ap-5030	111	11	is	be	AUX
ap-5030	111	12	also	also	ADV
ap-5030	111	13	possible	possible	ADJ
ap-5030	111	14	to	to	PART
ap-5030	111	15	penalize	penalize	VERB
ap-5030	111	16	the	the	DET
ap-5030	111	17	usage	usage	NOUN
ap-5030	111	18	of	of	ADP
ap-5030	111	19	distinct	distinct	ADJ
ap-5030	111	20	actions	action	NOUN
ap-5030	111	21	.	.	PUNCT
ap-5030	112	1	4.4	4.4	NUM
ap-5030	112	2	.	.	PUNCT
ap-5030	113	1	reinforcement	reinforcement	NOUN
ap-5030	113	2	learning	learning	NOUN
ap-5030	113	3	scenarios	scenario	NOUN
ap-5030	113	4	multiple	multiple	ADJ
ap-5030	113	5	experiments	experiment	NOUN
ap-5030	113	6	with	with	ADP
ap-5030	113	7	different	different	ADJ
ap-5030	113	8	parameters	parameter	NOUN
ap-5030	113	9	must	must	AUX
ap-5030	113	10	be	be	AUX
ap-5030	113	11	performed	perform	VERB
ap-5030	113	12	for	for	ADP
ap-5030	113	13	a	a	DET
ap-5030	113	14	meaningful	meaningful	ADJ
ap-5030	113	15	analysis	analysis	NOUN
ap-5030	113	16	.	.	PUNCT
ap-5030	114	1	each	each	DET
ap-5030	114	2	scenario	scenario	NOUN
ap-5030	114	3	is	be	AUX
ap-5030	114	4	a	a	DET
ap-5030	114	5	combination	combination	NOUN
ap-5030	114	6	of	of	ADP
ap-5030	114	7	mdp	mdp	NOUN
ap-5030	114	8	parameters	parameter	NOUN
ap-5030	114	9	like	like	ADP
ap-5030	114	10	the	the	DET
ap-5030	114	11	reward	reward	NOUN
ap-5030	114	12	,	,	PUNCT
ap-5030	114	13	the	the	DET
ap-5030	114	14	success	success	NOUN
ap-5030	114	15	likelihood	likelihood	NOUN
ap-5030	114	16	and	and	CCONJ
ap-5030	114	17	costs	cost	NOUN
ap-5030	114	18	of	of	ADP
ap-5030	114	19	actions	action	NOUN
ap-5030	114	20	as	as	ADV
ap-5030	114	21	well	well	ADV
ap-5030	114	22	as	as	ADP
ap-5030	114	23	a	a	DET
ap-5030	114	24	hyperparameter	hyperparameter	NOUN
ap-5030	114	25	like	like	ADP
ap-5030	114	26	the	the	DET
ap-5030	114	27	learning	learning	NOUN
ap-5030	114	28	rate	rate	NOUN
ap-5030	114	29	.	.	PUNCT
ap-5030	115	1	fpgasecml	fpgasecml	NOUN
ap-5030	115	2	allows	allow	VERB
ap-5030	115	3	the	the	DET
ap-5030	115	4	definition	definition	NOUN
ap-5030	115	5	of	of	ADP
ap-5030	115	6	scenarios	scenario	NOUN
ap-5030	115	7	that	that	PRON
ap-5030	115	8	allow	allow	VERB
ap-5030	115	9	the	the	DET
ap-5030	115	10	independent	independent	ADJ
ap-5030	115	11	definition	definition	NOUN
ap-5030	115	12	and	and	CCONJ
ap-5030	115	13	evaluation	evaluation	NOUN
ap-5030	115	14	of	of	ADP
ap-5030	115	15	these	these	DET
ap-5030	115	16	parameters	parameter	NOUN
ap-5030	115	17	.	.	PUNCT
ap-5030	116	1	all	all	DET
ap-5030	116	2	scenarios	scenario	NOUN
ap-5030	116	3	share	share	VERB
ap-5030	116	4	a	a	DET
ap-5030	116	5	set	set	NOUN
ap-5030	116	6	of	of	ADP
ap-5030	116	7	common	common	ADJ
ap-5030	116	8	parameter	parameter	NOUN
ap-5030	116	9	like	like	ADP
ap-5030	116	10	γ	γ	PROPN
ap-5030	116	11	and	and	CCONJ
ap-5030	116	12	the	the	DET
ap-5030	116	13	exploration	exploration	NOUN
ap-5030	116	14	-	-	PUNCT
ap-5030	116	15	exploitation	exploitation	NOUN
ap-5030	116	16	tradeoff	tradeoff	NOUN
ap-5030	116	17	ε	ε	PROPN
ap-5030	116	18	set	set	VERB
ap-5030	116	19	.	.	PUNCT
ap-5030	117	1	the	the	DET
ap-5030	117	2	example	example	NOUN
ap-5030	117	3	in	in	ADP
ap-5030	117	4	the	the	DET
ap-5030	117	5	next	next	ADJ
ap-5030	117	6	section	section	NOUN
ap-5030	117	7	illustrates	illustrate	VERB
ap-5030	117	8	the	the	DET
ap-5030	117	9	application	application	NOUN
ap-5030	117	10	of	of	ADP
ap-5030	117	11	scenarios	scenario	NOUN
ap-5030	117	12	.	.	PUNCT
ap-5030	118	1	4.5	4.5	NUM
ap-5030	118	2	.	.	PUNCT
ap-5030	119	1	implementation	implementation	VERB
ap-5030	119	2	our	our	PRON
ap-5030	119	3	proof	proof	NOUN
ap-5030	119	4	of	of	ADP
ap-5030	119	5	concept	concept	NOUN
ap-5030	119	6	implementation	implementation	NOUN
ap-5030	119	7	transforms	transform	VERB
ap-5030	119	8	,	,	PUNCT
ap-5030	119	9	as	as	SCONJ
ap-5030	119	10	mentioned	mention	VERB
ap-5030	119	11	earlier	early	ADV
ap-5030	119	12	,	,	PUNCT
ap-5030	119	13	any	any	DET
ap-5030	119	14	valid	valid	ADJ
ap-5030	119	15	fpgasecml	fpgasecml	NOUN
ap-5030	119	16	-	-	PUNCT
ap-5030	119	17	model	model	NOUN
ap-5030	119	18	into	into	ADP
ap-5030	119	19	java	java	PROPN
ap-5030	119	20	code	code	PROPN
ap-5030	119	21	.	.	PUNCT
ap-5030	120	1	this	this	DET
ap-5030	120	2	generated	generate	VERB
ap-5030	120	3	code	code	NOUN
ap-5030	120	4	is	be	AUX
ap-5030	120	5	meant	mean	VERB
ap-5030	120	6	to	to	PART
ap-5030	120	7	be	be	AUX
ap-5030	120	8	linked	link	VERB
ap-5030	120	9	against	against	ADP
ap-5030	120	10	the	the	DET
ap-5030	120	11	brown	brown	ADJ
ap-5030	120	12	-	-	PUNCT
ap-5030	120	13	umbc	umbc	NOUN
ap-5030	120	14	reinforcement	reinforcement	NOUN
ap-5030	120	15	learning	learning	NOUN
ap-5030	120	16	and	and	CCONJ
ap-5030	120	17	planning	planning	NOUN
ap-5030	120	18	(	(	PUNCT
ap-5030	120	19	burlap	burlap	NOUN
ap-5030	120	20	)	)	PUNCT
ap-5030	120	21	library	library	NOUN
ap-5030	121	1	[	[	X
ap-5030	121	2	33	33	NUM
ap-5030	121	3	]	]	PUNCT
ap-5030	121	4	,	,	PUNCT
ap-5030	121	5	a	a	DET
ap-5030	121	6	java	java	NOUN
ap-5030	121	7	framework	framework	NOUN
ap-5030	121	8	that	that	PRON
ap-5030	121	9	provides	provide	VERB
ap-5030	121	10	a	a	DET
ap-5030	121	11	generic	generic	ADJ
ap-5030	121	12	reinforcement	reinforcement	NOUN
ap-5030	121	13	learning	learning	NOUN
ap-5030	121	14	functionality	functionality	NOUN
ap-5030	121	15	.	.	PUNCT
ap-5030	122	1	where	where	SCONJ
ap-5030	122	2	the	the	DET
ap-5030	122	3	algorithmic	algorithmic	ADJ
ap-5030	122	4	flexibility	flexibility	NOUN
ap-5030	122	5	provided	provide	VERB
ap-5030	122	6	by	by	ADP
ap-5030	122	7	burlap	burlap	NOUN
ap-5030	122	8	is	be	AUX
ap-5030	122	9	not	not	PART
ap-5030	122	10	required	require	VERB
ap-5030	122	11	,	,	PUNCT
ap-5030	122	12	a	a	DET
ap-5030	122	13	reimplementation	reimplementation	NOUN
ap-5030	122	14	in	in	ADP
ap-5030	122	15	e.g.	e.g.	ADV
ap-5030	122	16	,	,	PUNCT
ap-5030	122	17	c++,rust	c++,rust	NOUN
ap-5030	122	18	or	or	CCONJ
ap-5030	122	19	go	go	VERB
ap-5030	122	20	should	should	AUX
ap-5030	122	21	lead	lead	VERB
ap-5030	122	22	to	to	ADP
ap-5030	122	23	a	a	DET
ap-5030	122	24	better	well	ADJ
ap-5030	122	25	performance	performance	NOUN
ap-5030	122	26	.	.	PUNCT
ap-5030	123	1	the	the	DET
ap-5030	123	2	memory	memory	NOUN
ap-5030	123	3	consumption	consumption	NOUN
ap-5030	123	4	could	could	AUX
ap-5030	123	5	be	be	AUX
ap-5030	123	6	reduced	reduce	VERB
ap-5030	123	7	through	through	ADP
ap-5030	123	8	an	an	DET
ap-5030	123	9	optimized	optimize	VERB
ap-5030	123	10	state	state	NOUN
ap-5030	123	11	encoding	encoding	NOUN
ap-5030	123	12	at	at	ADP
ap-5030	123	13	the	the	DET
ap-5030	123	14	expense	expense	NOUN
ap-5030	123	15	of	of	ADP
ap-5030	123	16	readability	readability	NOUN
ap-5030	123	17	and	and	CCONJ
ap-5030	123	18	extensible	extensible	ADJ
ap-5030	123	19	the	the	DET
ap-5030	123	20	implementation	implementation	NOUN
ap-5030	123	21	used	use	VERB
ap-5030	123	22	here	here	ADV
ap-5030	123	23	is	be	AUX
ap-5030	123	24	optimized	optimize	VERB
ap-5030	123	25	for	for	ADP
ap-5030	123	26	a	a	DET
ap-5030	123	27	simple	simple	ADJ
ap-5030	123	28	code	code	NOUN
ap-5030	123	29	generation	generation	NOUN
ap-5030	123	30	process	process	NOUN
ap-5030	123	31	.	.	PUNCT
ap-5030	124	1	5	5	X
ap-5030	124	2	.	.	NOUN
ap-5030	124	3	example	example	NOUN
ap-5030	124	4	an	an	DET
ap-5030	124	5	fpga	fpga	PROPN
ap-5030	124	6	design	design	NOUN
ap-5030	124	7	consists	consist	VERB
ap-5030	124	8	of	of	ADP
ap-5030	124	9	three	three	NUM
ap-5030	124	10	slots	slot	NOUN
ap-5030	124	11	with	with	ADP
ap-5030	124	12	two	two	NUM
ap-5030	124	13	fpgamodules	fpgamodule	NOUN
ap-5030	124	14	each	each	PRON
ap-5030	124	15	(	(	PUNCT
ap-5030	124	16	figure	figure	NOUN
ap-5030	124	17	3	3	NUM
ap-5030	124	18	)	)	PUNCT
ap-5030	124	19	.	.	PUNCT
ap-5030	125	1	fpgamodule	fpgamodule	NOUN
ap-5030	125	2	a	a	PRON
ap-5030	125	3	is	be	AUX
ap-5030	125	4	the	the	DET
ap-5030	125	5	only	only	ADJ
ap-5030	125	6	io	io	PROPN
ap-5030	125	7	block	block	NOUN
ap-5030	125	8	.	.	PUNCT
ap-5030	126	1	the	the	DET
ap-5030	126	2	run	run	NOUN
ap-5030	126	3	ends	end	VERB
ap-5030	126	4	the	the	DET
ap-5030	126	5	data	datum	NOUN
ap-5030	126	6	of	of	ADP
ap-5030	126	7	all	all	DET
ap-5030	126	8	six	six	NUM
ap-5030	126	9	modules	module	NOUN
ap-5030	126	10	(	(	PUNCT
ap-5030	126	11	a	a	DET
ap-5030	126	12	-	-	PUNCT
ap-5030	126	13	f	f	NOUN
ap-5030	126	14	)	)	PUNCT
ap-5030	126	15	is	be	AUX
ap-5030	126	16	exfiltrated	exfiltrate	VERB
ap-5030	126	17	.	.	PUNCT
ap-5030	127	1	the	the	DET
ap-5030	127	2	q	q	ADJ
ap-5030	127	3	-	-	PUNCT
ap-5030	127	4	learning	learn	VERB
ap-5030	127	5	algorithm	algorithm	NOUN
ap-5030	127	6	is	be	AUX
ap-5030	127	7	used	use	VERB
ap-5030	127	8	to	to	PART
ap-5030	127	9	find	find	VERB
ap-5030	127	10	the	the	DET
ap-5030	127	11	best	good	ADJ
ap-5030	127	12	sequence	sequence	NOUN
ap-5030	127	13	of	of	ADP
ap-5030	127	14	actions	action	NOUN
ap-5030	127	15	.	.	PUNCT
ap-5030	128	1	the	the	DET
ap-5030	128	2	heap	heap	NOUN
ap-5030	128	3	memory	memory	NOUN
ap-5030	128	4	of	of	ADP
ap-5030	128	5	the	the	DET
ap-5030	128	6	java	java	PROPN
ap-5030	128	7	virtual	virtual	ADJ
ap-5030	128	8	machine	machine	NOUN
ap-5030	128	9	(	(	PUNCT
ap-5030	128	10	jvm	jvm	PROPN
ap-5030	128	11	)	)	PUNCT
ap-5030	128	12	had	have	VERB
ap-5030	128	13	to	to	PART
ap-5030	128	14	be	be	AUX
ap-5030	128	15	extended	extend	VERB
ap-5030	128	16	to	to	ADP
ap-5030	128	17	up	up	ADP
ap-5030	128	18	to	to	PART
ap-5030	128	19	6	6	NUM
ap-5030	128	20	gb	gb	NOUN
ap-5030	128	21	to	to	PART
ap-5030	128	22	accommodate	accommodate	VERB
ap-5030	128	23	the	the	DET
ap-5030	128	24	qtable	qtable	NOUN
ap-5030	128	25	as	as	ADV
ap-5030	128	26	well	well	ADV
ap-5030	128	27	as	as	ADP
ap-5030	128	28	the	the	DET
ap-5030	128	29	telemetric	telemetric	ADJ
ap-5030	128	30	data	datum	NOUN
ap-5030	128	31	.	.	PUNCT
ap-5030	129	1	each	each	DET
ap-5030	129	2	scenario	scenario	NOUN
ap-5030	129	3	is	be	AUX
ap-5030	129	4	executed	execute	VERB
ap-5030	129	5	(	(	PUNCT
ap-5030	129	6	ran	ran	NOUN
ap-5030	129	7	)	)	PUNCT
ap-5030	129	8	once	once	ADV
ap-5030	129	9	for	for	ADP
ap-5030	129	10	each	each	DET
ap-5030	129	11	epsilon	epsilon	PROPN
ap-5030	129	12	(	(	PUNCT
ap-5030	129	13	0.1	0.1	NUM
ap-5030	129	14	,	,	PUNCT
ap-5030	129	15	0.05	0.05	NUM
ap-5030	129	16	and	and	CCONJ
ap-5030	129	17	0.25	0.25	NUM
ap-5030	129	18	in	in	ADP
ap-5030	129	19	our	our	PRON
ap-5030	129	20	example	example	NOUN
ap-5030	129	21	)	)	PUNCT
ap-5030	129	22	.	.	PUNCT
ap-5030	130	1	each	each	DET
ap-5030	130	2	run	run	NOUN
ap-5030	130	3	consists	consist	VERB
ap-5030	130	4	of	of	ADP
ap-5030	130	5	3	3	NUM
ap-5030	130	6	million	million	NUM
ap-5030	130	7	episodes	episode	NOUN
ap-5030	130	8	,	,	PUNCT
ap-5030	130	9	and	and	CCONJ
ap-5030	130	10	,	,	PUNCT
ap-5030	130	11	in	in	ADP
ap-5030	130	12	each	each	DET
ap-5030	130	13	episode	episode	NOUN
ap-5030	130	14	,	,	PUNCT
ap-5030	130	15	the	the	DET
ap-5030	130	16	agent	agent	NOUN
ap-5030	130	17	starts	start	VERB
ap-5030	130	18	at	at	ADP
ap-5030	130	19	the	the	DET
ap-5030	130	20	initial	initial	ADJ
ap-5030	130	21	state	state	NOUN
ap-5030	130	22	and	and	CCONJ
ap-5030	130	23	ends	end	VERB
ap-5030	130	24	in	in	ADP
ap-5030	130	25	a	a	DET
ap-5030	130	26	terminal	terminal	ADJ
ap-5030	130	27	state	state	NOUN
ap-5030	130	28	.	.	PUNCT
ap-5030	131	1	histograms	histogram	NOUN
ap-5030	131	2	(	(	PUNCT
ap-5030	131	3	truncated	truncate	VERB
ap-5030	131	4	to	to	ADP
ap-5030	131	5	the	the	DET
ap-5030	131	6	right	right	NOUN
ap-5030	131	7	)	)	PUNCT
ap-5030	131	8	are	be	AUX
ap-5030	131	9	used	use	VERB
ap-5030	131	10	to	to	PART
ap-5030	131	11	compare	compare	VERB
ap-5030	131	12	the	the	DET
ap-5030	131	13	agent	agent	NOUN
ap-5030	131	14	’s	’s	PART
ap-5030	131	15	learning	learn	VERB
ap-5030	131	16	performance	performance	NOUN
ap-5030	131	17	for	for	ADP
ap-5030	131	18	the	the	DET
ap-5030	131	19	different	different	ADJ
ap-5030	131	20	epsilon	epsilon	NOUN
ap-5030	131	21	and	and	CCONJ
ap-5030	131	22	scenarios	scenario	NOUN
ap-5030	131	23	.	.	PUNCT
ap-5030	132	1	the	the	DET
ap-5030	132	2	fpgasecml	fpgasecml	PROPN
ap-5030	132	3	-	-	PUNCT
ap-5030	132	4	model	model	NOUN
ap-5030	132	5	,	,	PUNCT
ap-5030	132	6	as	as	ADV
ap-5030	132	7	well	well	ADV
ap-5030	132	8	as	as	ADP
ap-5030	132	9	the	the	DET
ap-5030	132	10	generated	generate	VERB
ap-5030	132	11	java	java	PROPN
ap-5030	132	12	code	code	PROPN
ap-5030	132	13	and	and	CCONJ
ap-5030	132	14	reports	report	NOUN
ap-5030	132	15	,	,	PUNCT
ap-5030	132	16	for	for	ADP
ap-5030	132	17	this	this	DET
ap-5030	132	18	example	example	NOUN
ap-5030	132	19	,	,	PUNCT
ap-5030	132	20	can	can	AUX
ap-5030	132	21	be	be	AUX
ap-5030	132	22	found	find	VERB
ap-5030	132	23	in	in	ADP
ap-5030	132	24	the	the	DET
ap-5030	132	25	accompanying	accompanying	ADJ
ap-5030	132	26	files	file	NOUN
ap-5030	132	27	.	.	PUNCT
ap-5030	133	1	5.1	5.1	NUM
ap-5030	133	2	.	.	PUNCT
ap-5030	133	3	scenario	scenario	NOUN
ap-5030	133	4	1	1	NUM
ap-5030	133	5	:	:	PUNCT
ap-5030	133	6	baseline	baseline	PROPN
ap-5030	133	7	scenario	scenario	NOUN
ap-5030	133	8	1	1	NUM
ap-5030	133	9	establishes	establish	VERB
ap-5030	133	10	a	a	DET
ap-5030	133	11	baseline	baseline	NOUN
ap-5030	133	12	of	of	ADP
ap-5030	133	13	the	the	DET
ap-5030	133	14	model	model	NOUN
ap-5030	133	15	through	through	ADP
ap-5030	133	16	a	a	DET
ap-5030	133	17	deterministic	deterministic	ADJ
ap-5030	133	18	mdp	mdp	NOUN
ap-5030	133	19	(	(	PUNCT
ap-5030	133	20	the	the	DET
ap-5030	133	21	success	success	NOUN
ap-5030	133	22	rate	rate	NOUN
ap-5030	133	23	of	of	ADP
ap-5030	133	24	every	every	DET
ap-5030	133	25	action	action	NOUN
ap-5030	133	26	is	be	AUX
ap-5030	133	27	1	1	NUM
ap-5030	133	28	)	)	PUNCT
ap-5030	133	29	and	and	CCONJ
ap-5030	133	30	a	a	DET
ap-5030	133	31	fixed	fix	VERB
ap-5030	133	32	reward	reward	NOUN
ap-5030	133	33	/	/	SYM
ap-5030	133	34	punishment	punishment	NOUN
ap-5030	133	35	assignment	assignment	NOUN
ap-5030	133	36	.	.	PUNCT
ap-5030	134	1	the	the	DET
ap-5030	134	2	agent	agent	NOUN
ap-5030	134	3	gets	get	VERB
ap-5030	134	4	a	a	DET
ap-5030	134	5	reward	reward	NOUN
ap-5030	134	6	of	of	ADP
ap-5030	134	7	200.000	200.000	NUM
ap-5030	134	8	for	for	ADP
ap-5030	134	9	reaching	reach	VERB
ap-5030	134	10	a	a	DET
ap-5030	134	11	terminal	terminal	ADJ
ap-5030	134	12	state	state	NOUN
ap-5030	134	13	and	and	CCONJ
ap-5030	134	14	a	a	DET
ap-5030	134	15	punishment	punishment	NOUN
ap-5030	134	16	of	of	ADP
ap-5030	134	17	-1.000	-1.000	NUM
ap-5030	134	18	for	for	ADP
ap-5030	134	19	every	every	DET
ap-5030	134	20	other	other	ADJ
ap-5030	134	21	action	action	NOUN
ap-5030	134	22	.	.	PUNCT
ap-5030	135	1	all	all	DET
ap-5030	135	2	three	three	NUM
ap-5030	135	3	runs	run	NOUN
ap-5030	135	4	return	return	VERB
ap-5030	135	5	a	a	DET
ap-5030	135	6	minimal	minimal	ADJ
ap-5030	135	7	solution	solution	NOUN
ap-5030	135	8	with	with	ADP
ap-5030	135	9	24	24	NUM
ap-5030	135	10	actions	action	NOUN
ap-5030	135	11	,	,	PUNCT
ap-5030	135	12	the	the	DET
ap-5030	135	13	histogram	histogram	NOUN
ap-5030	135	14	(	(	PUNCT
ap-5030	135	15	figure	figure	NOUN
ap-5030	135	16	5	5	NUM
ap-5030	135	17	)	)	PUNCT
ap-5030	135	18	of	of	ADP
ap-5030	135	19	all	all	DET
ap-5030	135	20	three	three	NUM
ap-5030	135	21	actions	action	NOUN
ap-5030	135	22	shows	show	VERB
ap-5030	135	23	that	that	SCONJ
ap-5030	135	24	the	the	DET
ap-5030	135	25	epsilon	epsilon	PROPN
ap-5030	135	26	of	of	ADP
ap-5030	135	27	0.1	0.1	NUM
ap-5030	135	28	has	have	VERB
ap-5030	135	29	the	the	DET
ap-5030	135	30	most	most	ADJ
ap-5030	135	31	episodes	episode	NOUN
ap-5030	135	32	with	with	ADP
ap-5030	135	33	the	the	DET
ap-5030	135	34	minimal	minimal	ADJ
ap-5030	135	35	steps	step	NOUN
ap-5030	135	36	followed	follow	VERB
ap-5030	135	37	by	by	ADP
ap-5030	135	38	0.05	0.05	NUM
ap-5030	135	39	(	(	PUNCT
ap-5030	135	40	low	low	ADJ
ap-5030	135	41	exploration	exploration	NOUN
ap-5030	135	42	rate	rate	NOUN
ap-5030	135	43	requires	require	VERB
ap-5030	135	44	more	more	ADJ
ap-5030	135	45	steps	step	NOUN
ap-5030	135	46	to	to	PART
ap-5030	135	47	find	find	VERB
ap-5030	135	48	a	a	DET
ap-5030	135	49	solution	solution	NOUN
ap-5030	135	50	)	)	PUNCT
ap-5030	135	51	and	and	CCONJ
ap-5030	135	52	0.25	0.25	NUM
ap-5030	135	53	(	(	PUNCT
ap-5030	135	54	too	too	ADV
ap-5030	135	55	much	much	ADJ
ap-5030	135	56	exploration	exploration	NOUN
ap-5030	135	57	hinders	hinder	VERB
ap-5030	135	58	the	the	DET
ap-5030	135	59	efficient	efficient	ADJ
ap-5030	135	60	exploitation	exploitation	NOUN
ap-5030	135	61	of	of	ADP
ap-5030	135	62	the	the	DET
ap-5030	135	63	available	available	ADJ
ap-5030	135	64	data	datum	NOUN
ap-5030	135	65	.	.	PUNCT
ap-5030	135	66	)	)	PUNCT
ap-5030	136	1	the	the	DET
ap-5030	136	2	learning	learning	NOUN
ap-5030	136	3	process	process	NOUN
ap-5030	136	4	starts	start	VERB
ap-5030	136	5	with	with	ADP
ap-5030	136	6	episodes	episode	NOUN
ap-5030	136	7	well	well	ADV
ap-5030	136	8	over	over	ADP
ap-5030	136	9	2.000	2.000	NUM
ap-5030	136	10	steps	step	NOUN
ap-5030	136	11	(	(	PUNCT
ap-5030	136	12	figure	figure	NOUN
ap-5030	136	13	4	4	NUM
ap-5030	136	14	)	)	PUNCT
ap-5030	136	15	and	and	CCONJ
ap-5030	136	16	provides	provide	VERB
ap-5030	136	17	better	well	ADJ
ap-5030	136	18	results	result	NOUN
ap-5030	136	19	as	as	SCONJ
ap-5030	136	20	the	the	DET
ap-5030	136	21	qtable	qtable	NOUN
ap-5030	136	22	gets	get	AUX
ap-5030	136	23	filled	fill	VERB
ap-5030	136	24	.	.	PUNCT
ap-5030	137	1	for	for	ADP
ap-5030	137	2	the	the	DET
ap-5030	137	3	epsilon	epsilon	PROPN
ap-5030	137	4	of	of	ADP
ap-5030	137	5	0.1	0.1	NUM
ap-5030	137	6	,	,	PUNCT
ap-5030	137	7	the	the	DET
ap-5030	137	8	first	first	ADJ
ap-5030	137	9	minimal	minimal	ADJ
ap-5030	137	10	solution	solution	NOUN
ap-5030	137	11	is	be	AUX
ap-5030	137	12	found	find	VERB
ap-5030	137	13	after	after	ADP
ap-5030	137	14	just	just	ADV
ap-5030	137	15	303.323	303.323	NUM
ap-5030	137	16	episodes	episode	NOUN
ap-5030	137	17	indicating	indicate	VERB
ap-5030	137	18	that	that	SCONJ
ap-5030	137	19	the	the	DET
ap-5030	137	20	agent	agent	NOUN
ap-5030	137	21	has	have	AUX
ap-5030	137	22	either	either	CCONJ
ap-5030	137	23	found	find	VERB
ap-5030	137	24	the	the	DET
ap-5030	137	25	best	good	ADJ
ap-5030	137	26	possible	possible	ADJ
ap-5030	137	27	solution	solution	NOUN
ap-5030	137	28	or	or	CCONJ
ap-5030	137	29	is	be	AUX
ap-5030	137	30	stuck	stick	VERB
ap-5030	137	31	in	in	ADP
ap-5030	137	32	a	a	DET
ap-5030	137	33	local	local	ADJ
ap-5030	137	34	minimum	minimum	NOUN
ap-5030	137	35	with	with	ADP
ap-5030	137	36	little	little	ADJ
ap-5030	137	37	chances	chance	NOUN
ap-5030	137	38	to	to	PART
ap-5030	137	39	break	break	VERB
ap-5030	137	40	out	out	ADP
ap-5030	137	41	.	.	PUNCT
ap-5030	138	1	it	it	PRON
ap-5030	138	2	is	be	AUX
ap-5030	138	3	also	also	ADV
ap-5030	138	4	notable	notable	ADJ
ap-5030	138	5	that	that	SCONJ
ap-5030	138	6	all	all	DET
ap-5030	138	7	three	three	NUM
ap-5030	138	8	solutions	solution	NOUN
ap-5030	138	9	make	make	VERB
ap-5030	138	10	an	an	DET
ap-5030	138	11	extensive	extensive	ADJ
ap-5030	138	12	use	use	NOUN
ap-5030	138	13	of	of	ADP
ap-5030	138	14	the	the	DET
ap-5030	138	15	tamper−bitstream	tamper−bitstream	NOUN
ap-5030	138	16	action	action	NOUN
ap-5030	138	17	,	,	PUNCT
ap-5030	138	18	but	but	CCONJ
ap-5030	138	19	what	what	PRON
ap-5030	138	20	if	if	SCONJ
ap-5030	138	21	this	this	DET
ap-5030	138	22	action	action	NOUN
ap-5030	138	23	is	be	AUX
ap-5030	138	24	unlikely	unlikely	ADJ
ap-5030	138	25	to	to	PART
ap-5030	138	26	succeed	succeed	VERB
ap-5030	138	27	or	or	CCONJ
ap-5030	138	28	prohibitively	prohibitively	ADV
ap-5030	138	29	expensive	expensive	ADJ
ap-5030	138	30	?	?	PUNCT
ap-5030	139	1	5.2	5.2	NUM
ap-5030	139	2	.	.	PUNCT
ap-5030	139	3	scenario	scenario	NOUN
ap-5030	139	4	2	2	NUM
ap-5030	139	5	:	:	PUNCT
ap-5030	139	6	tamper	tamper	VERB
ap-5030	139	7	resistance	resistance	NOUN
ap-5030	139	8	storage	storage	NOUN
ap-5030	139	9	prevents	prevent	VERB
ap-5030	139	10	most	most	ADJ
ap-5030	139	11	attacks	attack	NOUN
ap-5030	139	12	the	the	DET
ap-5030	139	13	second	second	ADJ
ap-5030	139	14	scenario	scenario	NOUN
ap-5030	139	15	assumes	assume	VERB
ap-5030	139	16	that	that	SCONJ
ap-5030	139	17	only	only	ADV
ap-5030	139	18	one	one	NUM
ap-5030	139	19	in	in	ADP
ap-5030	139	20	a	a	DET
ap-5030	139	21	thousand	thousand	NUM
ap-5030	139	22	attacks	attack	NOUN
ap-5030	139	23	against	against	ADP
ap-5030	139	24	tamper	tamper	ADV
ap-5030	139	25	-	-	PUNCT
ap-5030	139	26	resistant	resistant	ADJ
ap-5030	139	27	bitstream	bitstream	NOUN
ap-5030	139	28	storage	storage	NOUN
ap-5030	139	29	is	be	AUX
ap-5030	139	30	successful	successful	ADJ
ap-5030	139	31	(	(	PUNCT
ap-5030	139	32	relative	relative	ADJ
ap-5030	139	33	to	to	ADP
ap-5030	139	34	all	all	DET
ap-5030	139	35	other	other	ADJ
ap-5030	139	36	attacks	attack	NOUN
ap-5030	139	37	as	as	SCONJ
ap-5030	139	38	success	success	NOUN
ap-5030	139	39	factors	factor	NOUN
ap-5030	139	40	are	be	AUX
ap-5030	139	41	normalized	normalize	VERB
ap-5030	139	42	.	.	PUNCT
ap-5030	139	43	)	)	PUNCT
ap-5030	140	1	the	the	DET
ap-5030	140	2	three	three	NUM
ap-5030	140	3	runs	run	NOUN
ap-5030	140	4	return	return	VERB
ap-5030	140	5	a	a	DET
ap-5030	140	6	minimal	minimal	ADJ
ap-5030	140	7	sequence	sequence	NOUN
ap-5030	140	8	of	of	ADP
ap-5030	140	9	25	25	NUM
ap-5030	140	10	actions	action	NOUN
ap-5030	140	11	each	each	PRON
ap-5030	140	12	and	and	CCONJ
ap-5030	140	13	all	all	DET
ap-5030	140	14	three	three	NUM
ap-5030	140	15	of	of	ADP
ap-5030	140	16	them	they	PRON
ap-5030	140	17	avoid	avoid	VERB
ap-5030	140	18	the	the	DET
ap-5030	140	19	tamper−bitsteam	tamper−bitsteam	NUM
ap-5030	140	20	actions	action	NOUN
ap-5030	140	21	completely	completely	ADV
ap-5030	140	22	.	.	PUNCT
ap-5030	141	1	the	the	DET
ap-5030	141	2	histogram	histogram	NOUN
ap-5030	141	3	(	(	PUNCT
ap-5030	141	4	figure	figure	NOUN
ap-5030	141	5	5	5	NUM
ap-5030	141	6	)	)	PUNCT
ap-5030	141	7	shows	show	VERB
ap-5030	141	8	a	a	DET
ap-5030	141	9	smaller	small	ADJ
ap-5030	141	10	spread	spread	NOUN
ap-5030	141	11	than	than	ADP
ap-5030	141	12	that	that	PRON
ap-5030	141	13	of	of	ADP
ap-5030	141	14	the	the	DET
ap-5030	141	15	first	first	ADJ
ap-5030	141	16	scenario	scenario	NOUN
ap-5030	141	17	and	and	CCONJ
ap-5030	141	18	the	the	DET
ap-5030	141	19	median	median	NOUN
ap-5030	141	20	of	of	ADP
ap-5030	141	21	the	the	DET
ap-5030	141	22	episodes	episode	NOUN
ap-5030	141	23	for	for	ADP
ap-5030	141	24	each	each	DET
ap-5030	141	25	run	run	NOUN
ap-5030	141	26	has	have	AUX
ap-5030	141	27	shrunk	shrink	VERB
ap-5030	141	28	from	from	ADP
ap-5030	141	29	(	(	PUNCT
ap-5030	141	30	29,39,44	29,39,44	NUM
ap-5030	141	31	)	)	PUNCT
ap-5030	141	32	to	to	ADP
ap-5030	141	33	(	(	PUNCT
ap-5030	141	34	28,26,35	28,26,35	NUM
ap-5030	141	35	)	)	PUNCT
ap-5030	141	36	.	.	PUNCT
ap-5030	142	1	5.3	5.3	NUM
ap-5030	142	2	.	.	PUNCT
ap-5030	142	3	scenario	scenario	NOUN
ap-5030	142	4	3	3	NUM
ap-5030	142	5	:	:	PUNCT
ap-5030	142	6	faulty	faulty	ADJ
ap-5030	142	7	encryption	encryption	NOUN
ap-5030	142	8	eases	ease	VERB
ap-5030	142	9	attacks	attack	NOUN
ap-5030	142	10	we	we	PRON
ap-5030	142	11	assume	assume	VERB
ap-5030	142	12	that	that	SCONJ
ap-5030	142	13	a	a	DET
ap-5030	142	14	weaker	weak	ADJ
ap-5030	142	15	tamper	tamper	NOUN
ap-5030	142	16	protection	protection	NOUN
ap-5030	142	17	of	of	ADP
ap-5030	142	18	the	the	DET
ap-5030	142	19	bitstream	bitstream	NOUN
ap-5030	142	20	storage	storage	NOUN
ap-5030	142	21	and	and	CCONJ
ap-5030	142	22	a	a	DET
ap-5030	142	23	weakness	weakness	NOUN
ap-5030	142	24	in	in	ADP
ap-5030	142	25	the	the	DET
ap-5030	142	26	bitstream	bitstream	NOUN
ap-5030	142	27	encryption	encryption	NOUN
ap-5030	142	28	mechanism	mechanism	NOUN
ap-5030	142	29	increases	increase	VERB
ap-5030	142	30	the	the	DET
ap-5030	142	31	success	success	NOUN
ap-5030	142	32	rate	rate	NOUN
ap-5030	142	33	of	of	ADP
ap-5030	142	34	tamper−bitstream	tamper−bitstream	NOUN
ap-5030	142	35	actions	action	NOUN
ap-5030	142	36	.	.	PUNCT
ap-5030	143	1	the	the	DET
ap-5030	143	2	minimal	minimal	ADJ
ap-5030	143	3	sequences	sequence	NOUN
ap-5030	143	4	returned	return	VERB
ap-5030	143	5	require	require	VERB
ap-5030	143	6	27	27	NUM
ap-5030	143	7	steps	step	NOUN
ap-5030	143	8	for	for	ADP
ap-5030	143	9	epsilon	epsilon	PROPN
ap-5030	143	10	of	of	ADP
ap-5030	143	11	0.1	0.1	NUM
ap-5030	143	12	,	,	PUNCT
ap-5030	143	13	26	26	NUM
ap-5030	143	14	for	for	ADP
ap-5030	143	15	0.05	0.05	NUM
ap-5030	143	16	and	and	CCONJ
ap-5030	143	17	30	30	NUM
ap-5030	143	18	for	for	ADP
ap-5030	143	19	0.25	0.25	NUM
ap-5030	143	20	a	a	DET
ap-5030	143	21	significantly	significantly	ADV
ap-5030	143	22	worse	bad	ADJ
ap-5030	143	23	performance	performance	NOUN
ap-5030	143	24	than	than	ADP
ap-5030	143	25	in	in	ADP
ap-5030	143	26	the	the	DET
ap-5030	143	27	previous	previous	ADJ
ap-5030	143	28	scenarios	scenario	NOUN
ap-5030	143	29	.	.	PUNCT
ap-5030	144	1	the	the	DET
ap-5030	144	2	histogram	histogram	NOUN
ap-5030	144	3	of	of	ADP
ap-5030	144	4	all	all	DET
ap-5030	144	5	three	three	NUM
ap-5030	144	6	runs	run	NOUN
ap-5030	144	7	also	also	ADV
ap-5030	144	8	skews	skew	VERB
ap-5030	144	9	much	much	ADV
ap-5030	144	10	more	more	ADJ
ap-5030	144	11	to	to	ADP
ap-5030	144	12	the	the	DET
ap-5030	144	13	right	right	NOUN
ap-5030	144	14	than	than	ADP
ap-5030	144	15	those	those	PRON
ap-5030	144	16	of	of	ADP
ap-5030	144	17	the	the	DET
ap-5030	144	18	preceding	precede	VERB
ap-5030	144	19	scenarios	scenario	NOUN
ap-5030	144	20	.	.	PUNCT
ap-5030	145	1	worse	bad	ADJ
ap-5030	145	2	,	,	PUNCT
ap-5030	145	3	the	the	DET
ap-5030	145	4	three	three	NUM
ap-5030	145	5	attack	attack	NOUN
ap-5030	145	6	sequences	sequence	NOUN
ap-5030	145	7	have	have	VERB
ap-5030	145	8	all	all	ADV
ap-5030	145	9	4	4	NUM
ap-5030	145	10	of	of	ADP
ap-5030	145	11	the	the	DET
ap-5030	145	12	6	6	NUM
ap-5030	145	13	bitstreams	bitstream	NOUN
ap-5030	145	14	tampered	tamper	VERB
ap-5030	145	15	with	with	ADP
ap-5030	145	16	.	.	PUNCT
ap-5030	146	1	the	the	DET
ap-5030	146	2	success	success	NOUN
ap-5030	146	3	rate	rate	NOUN
ap-5030	146	4	of	of	ADP
ap-5030	146	5	one	one	NUM
ap-5030	146	6	in	in	ADP
ap-5030	146	7	a	a	DET
ap-5030	146	8	thousand	thousand	NUM
ap-5030	146	9	was	be	AUX
ap-5030	146	10	,	,	PUNCT
ap-5030	146	11	apparently	apparently	ADV
ap-5030	146	12	,	,	PUNCT
ap-5030	146	13	low	low	ADJ
ap-5030	146	14	enough	enough	ADV
ap-5030	146	15	to	to	PART
ap-5030	146	16	discourage	discourage	VERB
ap-5030	146	17	the	the	DET
ap-5030	146	18	agent	agent	NOUN
ap-5030	146	19	from	from	ADP
ap-5030	146	20	using	use	VERB
ap-5030	146	21	these	these	DET
ap-5030	146	22	actions	action	NOUN
ap-5030	146	23	while	while	SCONJ
ap-5030	146	24	the	the	DET
ap-5030	146	25	“	"	PUNCT
ap-5030	146	26	one	one	NUM
ap-5030	146	27	in	in	ADP
ap-5030	146	28	five	five	NUM
ap-5030	146	29	”	"	PUNCT
ap-5030	146	30	-chance	-chance	NOUN
ap-5030	146	31	is	be	AUX
ap-5030	146	32	too	too	ADV
ap-5030	146	33	weak	weak	ADJ
ap-5030	146	34	to	to	PART
ap-5030	146	35	achieve	achieve	VERB
ap-5030	146	36	a	a	DET
ap-5030	146	37	similar	similar	ADJ
ap-5030	146	38	effect	effect	NOUN
ap-5030	146	39	.	.	PUNCT
ap-5030	147	1	the	the	DET
ap-5030	147	2	low	low	ADJ
ap-5030	147	3	success	success	NOUN
ap-5030	147	4	rate	rate	NOUN
ap-5030	147	5	is	be	AUX
ap-5030	147	6	,	,	PUNCT
ap-5030	147	7	however	however	ADV
ap-5030	147	8	,	,	PUNCT
ap-5030	147	9	strong	strong	ADJ
ap-5030	147	10	enough	enough	ADV
ap-5030	147	11	to	to	PART
ap-5030	147	12	prolong	prolong	VERB
ap-5030	147	13	the	the	DET
ap-5030	147	14	learning	learning	NOUN
ap-5030	147	15	process	process	NOUN
ap-5030	147	16	as	as	SCONJ
ap-5030	147	17	the	the	DET
ap-5030	147	18	median	median	ADJ
ap-5030	147	19	number	number	NOUN
ap-5030	147	20	of	of	ADP
ap-5030	147	21	steps	step	NOUN
ap-5030	147	22	per	per	ADP
ap-5030	147	23	episode	episode	NOUN
ap-5030	147	24	rises	rise	VERB
ap-5030	147	25	from	from	ADP
ap-5030	147	26	previously	previously	ADV
ap-5030	147	27	(	(	PUNCT
ap-5030	147	28	28,26,35	28,26,35	NOUN
ap-5030	147	29	)	)	PUNCT
ap-5030	147	30	to	to	ADP
ap-5030	147	31	(	(	PUNCT
ap-5030	147	32	154,155,152	154,155,152	NUM
ap-5030	147	33	)	)	PUNCT
ap-5030	147	34	.	.	PUNCT
ap-5030	148	1	the	the	DET
ap-5030	148	2	learning	learn	VERB
ap-5030	148	3	progress	progress	NOUN
ap-5030	148	4	(	(	PUNCT
ap-5030	148	5	figure	figure	NOUN
ap-5030	148	6	7	7	NUM
ap-5030	148	7	)	)	PUNCT
ap-5030	148	8	is	be	AUX
ap-5030	148	9	much	much	ADV
ap-5030	148	10	noisier	noisy	ADJ
ap-5030	148	11	than	than	SCONJ
ap-5030	148	12	it	it	PRON
ap-5030	148	13	was	be	AUX
ap-5030	148	14	in	in	ADP
ap-5030	148	15	the	the	DET
ap-5030	148	16	baseline	baseline	PROPN
ap-5030	148	17	scenario	scenario	NOUN
ap-5030	148	18	(	(	PUNCT
ap-5030	148	19	figure	figure	NOUN
ap-5030	148	20	4	4	NUM
ap-5030	148	21	.	.	PUNCT
ap-5030	148	22	)	)	PUNCT
ap-5030	149	1	5.4	5.4	NUM
ap-5030	149	2	.	.	PUNCT
ap-5030	149	3	scenario	scenario	NOUN
ap-5030	149	4	4	4	NUM
ap-5030	149	5	:	:	PUNCT
ap-5030	149	6	physical	physical	ADJ
ap-5030	149	7	attacks	attack	NOUN
ap-5030	149	8	against	against	ADP
ap-5030	149	9	storage	storage	NOUN
ap-5030	149	10	devices	device	NOUN
ap-5030	149	11	are	be	AUX
ap-5030	149	12	expensive	expensive	ADJ
ap-5030	149	13	not	not	PART
ap-5030	149	14	all	all	DET
ap-5030	149	15	attacks	attack	NOUN
ap-5030	149	16	require	require	VERB
ap-5030	149	17	the	the	DET
ap-5030	149	18	same	same	ADJ
ap-5030	149	19	amount	amount	NOUN
ap-5030	149	20	of	of	ADP
ap-5030	149	21	resources	resource	NOUN
ap-5030	149	22	some	some	PRON
ap-5030	149	23	take	take	VERB
ap-5030	149	24	longer	long	ADV
ap-5030	149	25	than	than	ADP
ap-5030	149	26	others	other	NOUN
ap-5030	149	27	,	,	PUNCT
ap-5030	149	28	a	a	DET
ap-5030	149	29	few	few	ADJ
ap-5030	149	30	require	require	VERB
ap-5030	149	31	costly	costly	ADJ
ap-5030	149	32	521	521	NUM
ap-5030	149	33	michael	michael	PROPN
ap-5030	149	34	vetter	vetter	PROPN
ap-5030	149	35	acta	acta	PROPN
ap-5030	149	36	polytechnica	polytechnica	PROPN
ap-5030	149	37	figure	figure	NOUN
ap-5030	150	1	3	3	NUM
ap-5030	150	2	.	.	PUNCT
ap-5030	150	3	schematic	schematic	ADJ
ap-5030	150	4	of	of	ADP
ap-5030	150	5	the	the	DET
ap-5030	150	6	fpgasecml	fpgasecml	NOUN
ap-5030	150	7	to	to	PART
ap-5030	150	8	burlap	burlap	VERB
ap-5030	150	9	example	example	NOUN
ap-5030	150	10	.	.	PUNCT
ap-5030	151	1	figure	figure	VERB
ap-5030	151	2	4	4	NUM
ap-5030	151	3	.	.	PUNCT
ap-5030	151	4	number	number	NOUN
ap-5030	151	5	of	of	ADP
ap-5030	151	6	actions	action	NOUN
ap-5030	151	7	necessary	necessary	ADJ
ap-5030	151	8	to	to	PART
ap-5030	151	9	reach	reach	VERB
ap-5030	151	10	a	a	DET
ap-5030	151	11	terminal	terminal	ADJ
ap-5030	151	12	state	state	NOUN
ap-5030	151	13	for	for	ADP
ap-5030	151	14	each	each	DET
ap-5030	151	15	episode	episode	NOUN
ap-5030	151	16	in	in	ADP
ap-5030	151	17	the	the	DET
ap-5030	151	18	baseline	baseline	PROPN
ap-5030	151	19	scenario	scenario	NOUN
ap-5030	151	20	(	(	PUNCT
ap-5030	151	21	the	the	DET
ap-5030	151	22	line	line	NOUN
ap-5030	151	23	at	at	ADP
ap-5030	151	24	the	the	DET
ap-5030	151	25	bottom	bottom	NOUN
ap-5030	151	26	represents	represent	VERB
ap-5030	151	27	the	the	DET
ap-5030	151	28	running	run	VERB
ap-5030	151	29	average	average	NOUN
ap-5030	151	30	)	)	PUNCT
ap-5030	151	31	.	.	PUNCT
ap-5030	152	1	522	522	NUM
ap-5030	152	2	vol	vol	NOUN
ap-5030	152	3	.	.	PUNCT
ap-5030	153	1	59	59	NUM
ap-5030	153	2	no	no	NOUN
ap-5030	153	3	.	.	PUNCT
ap-5030	154	1	5/2019	5/2019	NUM
ap-5030	154	2	security	security	NOUN
ap-5030	154	3	analysis	analysis	NOUN
ap-5030	154	4	of	of	ADP
ap-5030	154	5	fpgas	fpgas	NOUN
ap-5030	154	6	designs	design	NOUN
ap-5030	154	7	through	through	ADP
ap-5030	154	8	reinforcement	reinforcement	NOUN
ap-5030	154	9	learning	learning	NOUN
ap-5030	154	10	figure	figure	NOUN
ap-5030	154	11	5	5	NUM
ap-5030	154	12	.	.	PUNCT
ap-5030	154	13	histogram	histogram	NOUN
ap-5030	154	14	of	of	ADP
ap-5030	154	15	multiple	multiple	ADJ
ap-5030	154	16	runs	run	NOUN
ap-5030	154	17	with	with	ADP
ap-5030	154	18	different	different	ADJ
ap-5030	154	19	epsilons	epsilon	NOUN
ap-5030	154	20	for	for	ADP
ap-5030	154	21	scenario	scenario	NOUN
ap-5030	154	22	1	1	NUM
ap-5030	154	23	(	(	PUNCT
ap-5030	154	24	top	top	NOUN
ap-5030	154	25	)	)	PUNCT
ap-5030	154	26	and	and	CCONJ
ap-5030	154	27	2	2	NUM
ap-5030	154	28	(	(	PUNCT
ap-5030	154	29	bottom	bottom	NOUN
ap-5030	154	30	)	)	PUNCT
ap-5030	154	31	.	.	PUNCT
ap-5030	155	1	figure	figure	VERB
ap-5030	155	2	6	6	NUM
ap-5030	155	3	.	.	PUNCT
ap-5030	155	4	histogram	histogram	NOUN
ap-5030	155	5	of	of	ADP
ap-5030	155	6	multiple	multiple	ADJ
ap-5030	155	7	runs	run	NOUN
ap-5030	155	8	with	with	ADP
ap-5030	155	9	different	different	ADJ
ap-5030	155	10	epsilons	epsilon	NOUN
ap-5030	155	11	for	for	ADP
ap-5030	155	12	scenario	scenario	NOUN
ap-5030	155	13	3	3	NUM
ap-5030	155	14	(	(	PUNCT
ap-5030	155	15	top	top	NOUN
ap-5030	155	16	)	)	PUNCT
ap-5030	155	17	and	and	CCONJ
ap-5030	155	18	4	4	NUM
ap-5030	155	19	(	(	PUNCT
ap-5030	155	20	bottom	bottom	NOUN
ap-5030	155	21	)	)	PUNCT
ap-5030	155	22	.	.	PUNCT
ap-5030	156	1	523	523	NUM
ap-5030	156	2	michael	michael	PROPN
ap-5030	156	3	vetter	vetter	PROPN
ap-5030	156	4	acta	acta	PROPN
ap-5030	156	5	polytechnica	polytechnica	PROPN
ap-5030	156	6	figure	figure	NOUN
ap-5030	156	7	7	7	NUM
ap-5030	156	8	.	.	PUNCT
ap-5030	157	1	training	train	VERB
ap-5030	157	2	progess	progess	NOUN
ap-5030	157	3	for	for	ADP
ap-5030	157	4	scenario	scenario	NOUN
ap-5030	157	5	3	3	NUM
ap-5030	157	6	faulty	faulty	ADJ
ap-5030	157	7	encryption	encryption	NOUN
ap-5030	157	8	eases	ease	VERB
ap-5030	157	9	attacks	attack	NOUN
ap-5030	157	10	.	.	PUNCT
ap-5030	158	1	equipment	equipment	NOUN
ap-5030	158	2	and	and	CCONJ
ap-5030	158	3	highly	highly	ADV
ap-5030	158	4	specialized	specialized	ADJ
ap-5030	158	5	threat	threat	NOUN
ap-5030	158	6	agents	agent	NOUN
ap-5030	158	7	.	.	PUNCT
ap-5030	159	1	reducing	reduce	VERB
ap-5030	159	2	the	the	DET
ap-5030	159	3	success	success	NOUN
ap-5030	159	4	rate	rate	NOUN
ap-5030	159	5	of	of	ADP
ap-5030	159	6	an	an	DET
ap-5030	159	7	attack	attack	NOUN
ap-5030	159	8	makes	make	VERB
ap-5030	159	9	it	it	PRON
ap-5030	159	10	implicitly	implicitly	ADV
ap-5030	159	11	more	more	ADV
ap-5030	159	12	expensive	expensive	ADJ
ap-5030	159	13	(	(	PUNCT
ap-5030	159	14	as	as	SCONJ
ap-5030	159	15	,	,	PUNCT
ap-5030	159	16	on	on	ADP
ap-5030	159	17	average	average	ADJ
ap-5030	159	18	,	,	PUNCT
ap-5030	159	19	more	more	ADJ
ap-5030	159	20	attacks	attack	NOUN
ap-5030	159	21	must	must	AUX
ap-5030	159	22	be	be	AUX
ap-5030	159	23	performed	perform	VERB
ap-5030	159	24	before	before	ADP
ap-5030	159	25	succeeding	succeed	VERB
ap-5030	159	26	)	)	PUNCT
ap-5030	159	27	but	but	CCONJ
ap-5030	159	28	with	with	ADP
ap-5030	159	29	several	several	ADJ
ap-5030	159	30	million	million	NUM
ap-5030	159	31	trials	trial	NOUN
ap-5030	159	32	per	per	ADP
ap-5030	159	33	run	run	NOUN
ap-5030	159	34	,	,	PUNCT
ap-5030	159	35	there	there	PRON
ap-5030	159	36	is	be	VERB
ap-5030	159	37	a	a	DET
ap-5030	159	38	good	good	ADJ
ap-5030	159	39	chance	chance	NOUN
ap-5030	159	40	that	that	SCONJ
ap-5030	159	41	at	at	ADV
ap-5030	159	42	least	least	ADV
ap-5030	159	43	one	one	NUM
ap-5030	159	44	sequence	sequence	NOUN
ap-5030	159	45	of	of	ADP
ap-5030	159	46	highly	highly	ADV
ap-5030	159	47	improbable	improbable	ADJ
ap-5030	159	48	actions	action	NOUN
ap-5030	159	49	will	will	AUX
ap-5030	159	50	succeed	succeed	VERB
ap-5030	159	51	at	at	ADV
ap-5030	159	52	least	least	ADJ
ap-5030	159	53	once	once	ADV
ap-5030	159	54	.	.	PUNCT
ap-5030	160	1	a	a	DET
ap-5030	160	2	real	real	ADJ
ap-5030	160	3	attacker	attacker	NOUN
ap-5030	160	4	might	might	AUX
ap-5030	160	5	not	not	PART
ap-5030	160	6	want	want	VERB
ap-5030	160	7	to	to	PART
ap-5030	160	8	take	take	VERB
ap-5030	160	9	this	this	DET
ap-5030	160	10	risk	risk	NOUN
ap-5030	160	11	.	.	PUNCT
ap-5030	161	1	we	we	PRON
ap-5030	161	2	convert	convert	VERB
ap-5030	161	3	our	our	PRON
ap-5030	161	4	mdp	mdp	NOUN
ap-5030	161	5	back	back	ADV
ap-5030	161	6	into	into	ADP
ap-5030	161	7	a	a	DET
ap-5030	161	8	dmdp	dmdp	NOUN
ap-5030	161	9	and	and	CCONJ
ap-5030	161	10	assign	assign	VERB
ap-5030	161	11	the	the	DET
ap-5030	161	12	tamper−bitstream	tamper−bitstream	NOUN
ap-5030	161	13	action	action	NOUN
ap-5030	161	14	an	an	DET
ap-5030	161	15	additional	additional	ADJ
ap-5030	161	16	cost	cost	NOUN
ap-5030	161	17	weight	weight	NOUN
ap-5030	161	18	of	of	ADP
ap-5030	161	19	100.000	100.000	NUM
ap-5030	161	20	(	(	PUNCT
ap-5030	161	21	this	this	DET
ap-5030	161	22	cost	cost	NOUN
ap-5030	161	23	is	be	AUX
ap-5030	161	24	subtracted	subtract	VERB
ap-5030	161	25	from	from	ADP
ap-5030	161	26	the	the	DET
ap-5030	161	27	default	default	NOUN
ap-5030	161	28	reward	reward	NOUN
ap-5030	161	29	of	of	ADP
ap-5030	161	30	-1.000	-1.000	PROPN
ap-5030	161	31	.	.	PUNCT
ap-5030	161	32	)	)	PUNCT
ap-5030	162	1	the	the	DET
ap-5030	162	2	minimal	minimal	ADJ
ap-5030	162	3	sequence	sequence	NOUN
ap-5030	162	4	for	for	ADP
ap-5030	162	5	the	the	DET
ap-5030	162	6	epsilons	epsilon	NOUN
ap-5030	162	7	0.1	0.1	NUM
ap-5030	162	8	and	and	CCONJ
ap-5030	162	9	0.25	0.25	NUM
ap-5030	162	10	are	be	AUX
ap-5030	162	11	24	24	NUM
ap-5030	162	12	,	,	PUNCT
ap-5030	162	13	but	but	CCONJ
ap-5030	162	14	the	the	DET
ap-5030	162	15	low	low	ADJ
ap-5030	162	16	exploration	exploration	NOUN
ap-5030	162	17	rate	rate	NOUN
ap-5030	162	18	of	of	ADP
ap-5030	162	19	0.05	0.05	NUM
ap-5030	162	20	has	have	AUX
ap-5030	162	21	led	lead	VERB
ap-5030	162	22	to	to	ADP
ap-5030	162	23	a	a	DET
ap-5030	162	24	global	global	ADJ
ap-5030	162	25	minimum	minimum	NOUN
ap-5030	162	26	of	of	ADP
ap-5030	162	27	23	23	NUM
ap-5030	162	28	.	.	PUNCT
ap-5030	163	1	it	it	PRON
ap-5030	163	2	is	be	AUX
ap-5030	163	3	also	also	ADV
ap-5030	163	4	notable	notable	ADJ
ap-5030	163	5	that	that	SCONJ
ap-5030	163	6	this	this	DET
ap-5030	163	7	global	global	ADJ
ap-5030	163	8	minimum	minimum	NOUN
ap-5030	163	9	is	be	AUX
ap-5030	163	10	not	not	PART
ap-5030	163	11	reached	reach	VERB
ap-5030	163	12	until	until	ADP
ap-5030	163	13	the	the	DET
ap-5030	163	14	2.543.660th	2.543.660th	NUM
ap-5030	163	15	episode	episode	NOUN
ap-5030	163	16	and	and	CCONJ
ap-5030	163	17	that	that	SCONJ
ap-5030	163	18	the	the	DET
ap-5030	163	19	agent	agent	NOUN
ap-5030	163	20	tampers	tamper	VERB
ap-5030	163	21	with	with	ADP
ap-5030	163	22	two	two	NUM
ap-5030	163	23	bitstream	bitstream	NOUN
ap-5030	163	24	storages	storage	NOUN
ap-5030	163	25	while	while	SCONJ
ap-5030	163	26	the	the	DET
ap-5030	163	27	other	other	ADJ
ap-5030	163	28	two	two	NUM
ap-5030	163	29	use	use	NOUN
ap-5030	163	30	this	this	DET
ap-5030	163	31	expensive	expensive	ADJ
ap-5030	163	32	operation	operation	NOUN
ap-5030	163	33	only	only	ADV
ap-5030	163	34	once	once	ADV
ap-5030	163	35	.	.	PUNCT
ap-5030	164	1	imposing	impose	VERB
ap-5030	164	2	even	even	ADV
ap-5030	164	3	higher	high	ADJ
ap-5030	164	4	costs	cost	NOUN
ap-5030	164	5	(	(	PUNCT
ap-5030	164	6	bitstream	bitstream	NOUN
ap-5030	164	7	tampering	tampering	NOUN
ap-5030	164	8	is	be	AUX
ap-5030	164	9	extremely	extremely	ADV
ap-5030	164	10	expensive	expensive	ADJ
ap-5030	164	11	)	)	PUNCT
ap-5030	164	12	on	on	ADP
ap-5030	164	13	the	the	DET
ap-5030	164	14	tamper−bitstream	tamper−bitstream	NOUN
ap-5030	164	15	operation	operation	NOUN
ap-5030	164	16	had	have	AUX
ap-5030	164	17	no	no	DET
ap-5030	164	18	positive	positive	ADJ
ap-5030	164	19	effect	effect	NOUN
ap-5030	164	20	with	with	ADP
ap-5030	164	21	all	all	DET
ap-5030	164	22	three	three	NUM
ap-5030	164	23	runs	run	NOUN
ap-5030	164	24	returning	return	VERB
ap-5030	164	25	a	a	DET
ap-5030	164	26	sequence	sequence	NOUN
ap-5030	164	27	of	of	ADP
ap-5030	164	28	the	the	DET
ap-5030	164	29	length	length	NOUN
ap-5030	164	30	24	24	NUM
ap-5030	164	31	.	.	PUNCT
ap-5030	165	1	5.5	5.5	NUM
ap-5030	165	2	.	.	PUNCT
ap-5030	165	3	results	result	NOUN
ap-5030	165	4	we	we	PRON
ap-5030	165	5	can	can	AUX
ap-5030	165	6	conclude	conclude	VERB
ap-5030	165	7	that	that	SCONJ
ap-5030	165	8	our	our	PRON
ap-5030	165	9	approach	approach	NOUN
ap-5030	165	10	found	find	VERB
ap-5030	165	11	a	a	DET
ap-5030	165	12	reasonable	reasonable	ADJ
ap-5030	165	13	solution	solution	NOUN
ap-5030	165	14	to	to	ADP
ap-5030	165	15	this	this	DET
ap-5030	165	16	problem	problem	NOUN
ap-5030	165	17	within	within	ADP
ap-5030	165	18	a	a	DET
ap-5030	165	19	feasible	feasible	ADJ
ap-5030	165	20	amount	amount	NOUN
ap-5030	165	21	of	of	ADP
ap-5030	165	22	computations	computation	NOUN
ap-5030	165	23	.	.	PUNCT
ap-5030	166	1	the	the	DET
ap-5030	166	2	noisy	noisy	ADJ
ap-5030	166	3	progress	progress	NOUN
ap-5030	166	4	of	of	ADP
ap-5030	166	5	the	the	DET
ap-5030	166	6	rl	rl	PROPN
ap-5030	166	7	algorithm	algorithm	NOUN
ap-5030	166	8	makes	make	VERB
ap-5030	166	9	it	it	PRON
ap-5030	166	10	impossible	impossible	ADJ
ap-5030	166	11	to	to	PART
ap-5030	166	12	determine	determine	VERB
ap-5030	166	13	whether	whether	SCONJ
ap-5030	166	14	a	a	DET
ap-5030	166	15	local	local	ADJ
ap-5030	166	16	or	or	CCONJ
ap-5030	166	17	global	global	ADJ
ap-5030	166	18	minimum	minimum	NOUN
ap-5030	166	19	has	have	AUX
ap-5030	166	20	been	be	AUX
ap-5030	166	21	found	find	VERB
ap-5030	166	22	.	.	PUNCT
ap-5030	167	1	scenarios	scenario	NOUN
ap-5030	167	2	can	can	AUX
ap-5030	167	3	be	be	AUX
ap-5030	167	4	used	use	VERB
ap-5030	167	5	to	to	PART
ap-5030	167	6	assess	assess	VERB
ap-5030	167	7	the	the	DET
ap-5030	167	8	impact	impact	NOUN
ap-5030	167	9	of	of	ADP
ap-5030	167	10	high	high	ADJ
ap-5030	167	11	costs	cost	NOUN
ap-5030	167	12	and	and	CCONJ
ap-5030	167	13	low	low	ADJ
ap-5030	167	14	success	success	NOUN
ap-5030	167	15	rates	rate	NOUN
ap-5030	167	16	and	and	CCONJ
ap-5030	167	17	to	to	PART
ap-5030	167	18	find	find	VERB
ap-5030	167	19	better	well	ADJ
ap-5030	167	20	solutions	solution	NOUN
ap-5030	167	21	by	by	ADP
ap-5030	167	22	constraining	constrain	VERB
ap-5030	167	23	the	the	DET
ap-5030	167	24	size	size	NOUN
ap-5030	167	25	of	of	ADP
ap-5030	167	26	the	the	DET
ap-5030	167	27	search	search	NOUN
ap-5030	167	28	space	space	NOUN
ap-5030	167	29	.	.	PUNCT
ap-5030	168	1	a	a	DET
ap-5030	168	2	plausibility	plausibility	NOUN
ap-5030	168	3	check	check	NOUN
ap-5030	168	4	of	of	ADP
ap-5030	168	5	the	the	DET
ap-5030	168	6	results	result	NOUN
ap-5030	168	7	(	(	PUNCT
ap-5030	168	8	here	here	ADV
ap-5030	168	9	performed	perform	VERB
ap-5030	168	10	on	on	ADP
ap-5030	168	11	the	the	DET
ap-5030	168	12	generated	generate	VERB
ap-5030	168	13	sequence	sequence	NOUN
ap-5030	168	14	)	)	PUNCT
ap-5030	168	15	is	be	AUX
ap-5030	168	16	mandatory	mandatory	ADJ
ap-5030	168	17	,	,	PUNCT
ap-5030	168	18	as	as	ADP
ap-5030	168	19	for	for	ADP
ap-5030	168	20	all	all	DET
ap-5030	168	21	machine	machine	NOUN
ap-5030	168	22	learning	learning	NOUN
ap-5030	168	23	methods	method	NOUN
ap-5030	168	24	.	.	PUNCT
ap-5030	169	1	the	the	DET
ap-5030	169	2	number	number	NOUN
ap-5030	169	3	of	of	ADP
ap-5030	169	4	training	training	NOUN
ap-5030	169	5	episodes	episode	NOUN
ap-5030	169	6	required	require	VERB
ap-5030	169	7	for	for	ADP
ap-5030	169	8	this	this	DET
ap-5030	169	9	small	small	ADJ
ap-5030	169	10	example	example	NOUN
ap-5030	169	11	indicates	indicate	VERB
ap-5030	169	12	challenges	challenge	NOUN
ap-5030	169	13	for	for	ADP
ap-5030	169	14	intricate	intricate	ADJ
ap-5030	169	15	designs	design	NOUN
ap-5030	169	16	and	and	CCONJ
ap-5030	169	17	models	model	NOUN
ap-5030	169	18	with	with	ADP
ap-5030	169	19	a	a	DET
ap-5030	169	20	richer	rich	ADJ
ap-5030	169	21	state	state	NOUN
ap-5030	169	22	,	,	PUNCT
ap-5030	169	23	action	action	NOUN
ap-5030	169	24	representation	representation	NOUN
ap-5030	169	25	.	.	PUNCT
ap-5030	170	1	the	the	DET
ap-5030	170	2	memory	memory	NOUN
ap-5030	170	3	consumption	consumption	NOUN
ap-5030	170	4	of	of	ADP
ap-5030	170	5	the	the	DET
ap-5030	170	6	program	program	NOUN
ap-5030	170	7	,	,	PUNCT
ap-5030	170	8	presumably	presumably	ADV
ap-5030	170	9	driven	drive	VERB
ap-5030	170	10	by	by	ADP
ap-5030	170	11	the	the	DET
ap-5030	170	12	expanding	expand	VERB
ap-5030	170	13	qtable	qtable	NOUN
ap-5030	170	14	,	,	PUNCT
ap-5030	170	15	supports	support	VERB
ap-5030	170	16	this	this	DET
ap-5030	170	17	assumption	assumption	NOUN
ap-5030	170	18	.	.	PUNCT
ap-5030	171	1	finding	find	VERB
ap-5030	171	2	the	the	DET
ap-5030	171	3	appropriate	appropriate	ADJ
ap-5030	171	4	cost	cost	NOUN
ap-5030	171	5	/	/	SYM
ap-5030	171	6	reward	reward	NOUN
ap-5030	171	7	structure	structure	NOUN
ap-5030	171	8	remains	remain	VERB
ap-5030	171	9	a	a	DET
ap-5030	171	10	challenge	challenge	NOUN
ap-5030	171	11	and	and	CCONJ
ap-5030	171	12	any	any	DET
ap-5030	171	13	analysis	analysis	NOUN
ap-5030	171	14	should	should	AUX
ap-5030	171	15	include	include	VERB
ap-5030	171	16	multiple	multiple	ADJ
ap-5030	171	17	scenarios	scenario	NOUN
ap-5030	171	18	with	with	ADP
ap-5030	171	19	different	different	ADJ
ap-5030	171	20	cost	cost	NOUN
ap-5030	171	21	and	and	CCONJ
ap-5030	171	22	success	success	NOUN
ap-5030	171	23	rate	rate	NOUN
ap-5030	171	24	values	value	NOUN
ap-5030	171	25	.	.	PUNCT
ap-5030	172	1	6	6	X
ap-5030	172	2	.	.	NOUN
ap-5030	172	3	limits	limit	NOUN
ap-5030	172	4	and	and	CCONJ
ap-5030	172	5	restrictions	restriction	NOUN
ap-5030	172	6	of	of	ADP
ap-5030	172	7	the	the	DET
ap-5030	172	8	mdp	mdp	NOUN
ap-5030	172	9	based	base	VERB
ap-5030	172	10	approach	approach	NOUN
ap-5030	172	11	and	and	CCONJ
ap-5030	172	12	possible	possible	ADJ
ap-5030	172	13	solutions	solution	NOUN
ap-5030	172	14	the	the	DET
ap-5030	172	15	mdp	mdp	PROPN
ap-5030	172	16	based	base	VERB
ap-5030	172	17	model	model	NOUN
ap-5030	172	18	presented	present	VERB
ap-5030	172	19	here	here	ADV
ap-5030	172	20	relies	rely	VERB
ap-5030	172	21	on	on	ADP
ap-5030	172	22	certain	certain	ADJ
ap-5030	172	23	constraints	constraint	NOUN
ap-5030	172	24	that	that	PRON
ap-5030	172	25	could	could	AUX
ap-5030	172	26	be	be	AUX
ap-5030	172	27	lifted	lift	VERB
ap-5030	172	28	by	by	ADP
ap-5030	172	29	supplementing	supplement	VERB
ap-5030	172	30	the	the	DET
ap-5030	172	31	proposed	propose	VERB
ap-5030	172	32	model	model	NOUN
ap-5030	172	33	.	.	PUNCT
ap-5030	173	1	hierarchical	hierarchical	ADJ
ap-5030	173	2	learning	learning	NOUN
ap-5030	173	3	[	[	X
ap-5030	173	4	34	34	NUM
ap-5030	173	5	]	]	PUNCT
ap-5030	173	6	can	can	AUX
ap-5030	173	7	be	be	AUX
ap-5030	173	8	used	use	VERB
ap-5030	173	9	to	to	PART
ap-5030	173	10	combine	combine	VERB
ap-5030	173	11	this	this	DET
ap-5030	173	12	high	high	ADJ
ap-5030	173	13	-	-	PUNCT
ap-5030	173	14	level	level	NOUN
ap-5030	173	15	model	model	NOUN
ap-5030	173	16	with	with	ADP
ap-5030	173	17	the	the	DET
ap-5030	173	18	low	low	ADJ
ap-5030	173	19	-	-	PUNCT
ap-5030	173	20	level	level	NOUN
ap-5030	173	21	activities	activity	NOUN
ap-5030	173	22	needed	need	VERB
ap-5030	173	23	to	to	PART
ap-5030	173	24	execute	execute	VERB
ap-5030	173	25	the	the	DET
ap-5030	173	26	strategy	strategy	NOUN
ap-5030	173	27	.	.	PUNCT
ap-5030	174	1	it	it	PRON
ap-5030	174	2	is	be	AUX
ap-5030	174	3	also	also	ADV
ap-5030	174	4	plausible	plausible	ADJ
ap-5030	174	5	that	that	SCONJ
ap-5030	174	6	the	the	DET
ap-5030	174	7	attacker	attacker	NOUN
ap-5030	174	8	does	do	AUX
ap-5030	174	9	not	not	PART
ap-5030	174	10	know	know	VERB
ap-5030	174	11	the	the	DET
ap-5030	174	12	state	state	NOUN
ap-5030	174	13	of	of	ADP
ap-5030	174	14	the	the	DET
ap-5030	174	15	system	system	NOUN
ap-5030	174	16	but	but	CCONJ
ap-5030	174	17	has	have	VERB
ap-5030	174	18	to	to	PART
ap-5030	174	19	guess	guess	VERB
ap-5030	174	20	it	it	PRON
ap-5030	174	21	from	from	ADP
ap-5030	174	22	a	a	DET
ap-5030	174	23	limited	limited	ADJ
ap-5030	174	24	set	set	NOUN
ap-5030	174	25	of	of	ADP
ap-5030	174	26	,	,	PUNCT
ap-5030	174	27	presumably	presumably	ADV
ap-5030	174	28	noisy	noisy	ADJ
ap-5030	174	29	,	,	PUNCT
ap-5030	174	30	indicators	indicator	NOUN
ap-5030	174	31	available	available	ADJ
ap-5030	174	32	(	(	PUNCT
ap-5030	174	33	like	like	ADP
ap-5030	174	34	physical	physical	ADJ
ap-5030	174	35	side	side	NOUN
ap-5030	174	36	channels	channel	NOUN
ap-5030	174	37	[	[	X
ap-5030	174	38	35][36	35][36	PROPN
ap-5030	174	39	]	]	PUNCT
ap-5030	174	40	)	)	PUNCT
ap-5030	174	41	.	.	PUNCT
ap-5030	175	1	the	the	DET
ap-5030	175	2	reinforcement	reinforcement	NOUN
ap-5030	175	3	learning	learning	NOUN
ap-5030	175	4	equivalent	equivalent	ADJ
ap-5030	175	5	of	of	ADP
ap-5030	175	6	this	this	DET
ap-5030	175	7	effect	effect	NOUN
ap-5030	175	8	is	be	AUX
ap-5030	175	9	a	a	DET
ap-5030	175	10	partially	partially	ADV
ap-5030	175	11	observable	observable	ADJ
ap-5030	175	12	mdp	mdp	NOUN
ap-5030	175	13	(	(	PUNCT
ap-5030	175	14	pomdp	pomdp	NOUN
ap-5030	176	1	[	[	X
ap-5030	176	2	37	37	NUM
ap-5030	176	3	]	]	X
ap-5030	176	4	burlap	burlap	NOUN
ap-5030	176	5	provides	provide	VERB
ap-5030	176	6	limited	limited	ADJ
ap-5030	176	7	support	support	NOUN
ap-5030	176	8	for	for	ADP
ap-5030	176	9	pomdp	pomdp	NOUN
ap-5030	176	10	)	)	PUNCT
ap-5030	176	11	transforming	transform	VERB
ap-5030	176	12	the	the	DET
ap-5030	176	13	mdp	mdp	NOUN
ap-5030	176	14	into	into	ADP
ap-5030	176	15	a	a	DET
ap-5030	176	16	stochastic	stochastic	ADJ
ap-5030	176	17	game	game	NOUN
ap-5030	176	18	allows	allow	VERB
ap-5030	176	19	the	the	DET
ap-5030	176	20	integration	integration	NOUN
ap-5030	176	21	of	of	ADP
ap-5030	176	22	multiple	multiple	ADJ
ap-5030	176	23	actors	actor	NOUN
ap-5030	176	24	(	(	PUNCT
ap-5030	176	25	e.g.	e.g.	ADV
ap-5030	176	26	an	an	DET
ap-5030	176	27	active	active	ADJ
ap-5030	176	28	defence	defence	NOUN
ap-5030	176	29	mechanism	mechanism	NOUN
ap-5030	176	30	.	.	PUNCT
ap-5030	176	31	)	)	PUNCT
ap-5030	177	1	navigating	navigate	VERB
ap-5030	177	2	the	the	DET
ap-5030	177	3	vast	vast	ADJ
ap-5030	177	4	search	search	NOUN
ap-5030	177	5	space	space	NOUN
ap-5030	177	6	remains	remain	VERB
ap-5030	177	7	the	the	DET
ap-5030	177	8	main	main	ADJ
ap-5030	177	9	problem	problem	NOUN
ap-5030	177	10	of	of	ADP
ap-5030	177	11	the	the	DET
ap-5030	177	12	reinforcement	reinforcement	NOUN
ap-5030	177	13	learning	learning	NOUN
ap-5030	177	14	.	.	PUNCT
ap-5030	178	1	limiting	limit	VERB
ap-5030	178	2	the	the	DET
ap-5030	178	3	content	content	NOUN
ap-5030	178	4	of	of	ADP
ap-5030	178	5	the	the	DET
ap-5030	178	6	state	state	NOUN
ap-5030	178	7	and	and	CCONJ
ap-5030	178	8	the	the	DET
ap-5030	178	9	number	number	NOUN
ap-5030	178	10	of	of	ADP
ap-5030	178	11	actions	action	NOUN
ap-5030	178	12	eases	ease	VERB
ap-5030	178	13	this	this	DET
ap-5030	178	14	problem	problem	NOUN
ap-5030	178	15	but	but	CCONJ
ap-5030	178	16	restricts	restrict	VERB
ap-5030	178	17	the	the	DET
ap-5030	178	18	expressiveness	expressiveness	NOUN
ap-5030	178	19	of	of	ADP
ap-5030	178	20	the	the	DET
ap-5030	178	21	model	model	NOUN
ap-5030	178	22	.	.	PUNCT
ap-5030	179	1	constraining	constrain	VERB
ap-5030	179	2	the	the	DET
ap-5030	179	3	search	search	NOUN
ap-5030	179	4	space	space	NOUN
ap-5030	179	5	through	through	ADP
ap-5030	179	6	transition	transition	NOUN
ap-5030	179	7	probabilities	probability	NOUN
ap-5030	179	8	and	and	CCONJ
ap-5030	179	9	costs	cost	NOUN
ap-5030	179	10	is	be	AUX
ap-5030	179	11	another	another	DET
ap-5030	179	12	method	method	NOUN
ap-5030	179	13	but	but	CCONJ
ap-5030	179	14	increases	increase	VERB
ap-5030	179	15	the	the	DET
ap-5030	179	16	number	number	NOUN
ap-5030	179	17	of	of	ADP
ap-5030	179	18	experiments	experiment	NOUN
ap-5030	179	19	.	.	PUNCT
ap-5030	180	1	a	a	DET
ap-5030	180	2	functional	functional	ADJ
ap-5030	180	3	approximation	approximation	NOUN
ap-5030	180	4	of	of	ADP
ap-5030	180	5	the	the	DET
ap-5030	180	6	qtable	qtable	NOUN
ap-5030	180	7	could	could	AUX
ap-5030	180	8	provide	provide	VERB
ap-5030	180	9	an	an	DET
ap-5030	180	10	avenue	avenue	NOUN
ap-5030	180	11	towards	towards	ADP
ap-5030	180	12	an	an	DET
ap-5030	180	13	improved	improved	ADJ
ap-5030	180	14	rl	rl	NOUN
ap-5030	180	15	based	base	VERB
ap-5030	180	16	weakness	weakness	NOUN
ap-5030	180	17	analysis	analysis	NOUN
ap-5030	180	18	.	.	PUNCT
ap-5030	181	1	one	one	NUM
ap-5030	181	2	candidate	candidate	NOUN
ap-5030	181	3	for	for	ADP
ap-5030	181	4	this	this	DET
ap-5030	181	5	approximation	approximation	NOUN
ap-5030	181	6	are	be	AUX
ap-5030	181	7	deep	deep	ADJ
ap-5030	181	8	neural	neural	ADJ
ap-5030	181	9	networks	network	NOUN
ap-5030	181	10	,	,	PUNCT
ap-5030	181	11	whose	whose	DET
ap-5030	181	12	general	general	ADJ
ap-5030	181	13	feasibility	feasibility	NOUN
ap-5030	181	14	has	have	AUX
ap-5030	181	15	been	be	AUX
ap-5030	181	16	demon524	demon524	PROPN
ap-5030	181	17	vol	vol	NOUN
ap-5030	181	18	.	.	PROPN
ap-5030	182	1	59	59	NUM
ap-5030	182	2	no	no	NOUN
ap-5030	182	3	.	.	PUNCT
ap-5030	183	1	5/2019	5/2019	NUM
ap-5030	183	2	security	security	NOUN
ap-5030	183	3	analysis	analysis	NOUN
ap-5030	183	4	of	of	ADP
ap-5030	183	5	fpgas	fpgas	NOUN
ap-5030	183	6	designs	design	NOUN
ap-5030	183	7	through	through	ADP
ap-5030	183	8	reinforcement	reinforcement	NOUN
ap-5030	183	9	learning	learning	NOUN
ap-5030	183	10	strated	strate	VERB
ap-5030	183	11	by	by	ADP
ap-5030	183	12	the	the	DET
ap-5030	183	13	deepq	deepq	NOUN
ap-5030	183	14	[	[	X
ap-5030	183	15	38	38	NUM
ap-5030	183	16	]	]	PUNCT
ap-5030	183	17	algorithm	algorithm	NOUN
ap-5030	183	18	.	.	PUNCT
ap-5030	184	1	this	this	DET
ap-5030	184	2	more	more	ADV
ap-5030	184	3	recent	recent	ADJ
ap-5030	184	4	technique	technique	NOUN
ap-5030	184	5	replaces	replace	VERB
ap-5030	184	6	the	the	DET
ap-5030	184	7	qtable	qtable	NOUN
ap-5030	184	8	with	with	ADP
ap-5030	184	9	a	a	DET
ap-5030	184	10	neuronal	neuronal	ADJ
ap-5030	184	11	network	network	NOUN
ap-5030	184	12	that	that	PRON
ap-5030	184	13	approximates	approximate	VERB
ap-5030	184	14	the	the	DET
ap-5030	184	15	value	value	NOUN
ap-5030	184	16	of	of	ADP
ap-5030	184	17	a	a	DET
ap-5030	184	18	given	give	VERB
ap-5030	184	19	state	state	NOUN
ap-5030	184	20	-	-	PUNCT
ap-5030	184	21	action	action	NOUN
ap-5030	184	22	combination	combination	NOUN
ap-5030	184	23	.	.	PUNCT
ap-5030	185	1	suitable	suitable	ADJ
ap-5030	185	2	neural	neural	ADJ
ap-5030	185	3	networks	network	NOUN
ap-5030	185	4	mitigate	mitigate	VERB
ap-5030	185	5	the	the	DET
ap-5030	185	6	state	state	NOUN
ap-5030	185	7	explosion	explosion	NOUN
ap-5030	185	8	problem	problem	NOUN
ap-5030	185	9	and	and	CCONJ
ap-5030	185	10	should	should	AUX
ap-5030	185	11	be	be	AUX
ap-5030	185	12	able	able	ADJ
ap-5030	185	13	to	to	PART
ap-5030	185	14	detect	detect	VERB
ap-5030	185	15	patterns	pattern	NOUN
ap-5030	185	16	.	.	PUNCT
ap-5030	186	1	they	they	PRON
ap-5030	186	2	may	may	AUX
ap-5030	186	3	,	,	PUNCT
ap-5030	186	4	therefore	therefore	ADV
ap-5030	186	5	,	,	PUNCT
ap-5030	186	6	be	be	AUX
ap-5030	186	7	used	use	VERB
ap-5030	186	8	to	to	PART
ap-5030	186	9	extract	extract	VERB
ap-5030	186	10	and	and	CCONJ
ap-5030	186	11	abstract	abstract	ADJ
ap-5030	186	12	information	information	NOUN
ap-5030	186	13	from	from	ADP
ap-5030	186	14	a	a	DET
ap-5030	186	15	feature	feature	NOUN
ap-5030	186	16	-	-	PUNCT
ap-5030	186	17	rich	rich	ADJ
ap-5030	186	18	state	state	NOUN
ap-5030	186	19	,	,	PUNCT
ap-5030	186	20	action	action	NOUN
ap-5030	186	21	representation	representation	NOUN
ap-5030	186	22	in	in	ADP
ap-5030	186	23	the	the	DET
ap-5030	186	24	same	same	ADJ
ap-5030	186	25	way	way	NOUN
ap-5030	186	26	convolution	convolution	NOUN
ap-5030	186	27	networks	network	NOUN
ap-5030	186	28	extract	extract	VERB
ap-5030	186	29	information	information	NOUN
ap-5030	186	30	e.g	e.g	VERB
ap-5030	186	31	from	from	ADP
ap-5030	186	32	a	a	DET
ap-5030	186	33	picture	picture	NOUN
ap-5030	187	1	[	[	X
ap-5030	187	2	16	16	NUM
ap-5030	187	3	]	]	PUNCT
ap-5030	187	4	.	.	PUNCT
ap-5030	188	1	policy	policy	NOUN
ap-5030	188	2	shaping	shaping	NOUN
ap-5030	188	3	[	[	X
ap-5030	188	4	39	39	NUM
ap-5030	188	5	]	]	PUNCT
ap-5030	188	6	or	or	CCONJ
ap-5030	188	7	reward	reward	NOUN
ap-5030	188	8	learning	learn	VERB
ap-5030	188	9	[	[	X
ap-5030	188	10	40	40	NUM
ap-5030	188	11	]	]	PUNCT
ap-5030	188	12	provide	provide	VERB
ap-5030	188	13	other	other	ADJ
ap-5030	188	14	avenues	avenue	NOUN
ap-5030	188	15	to	to	PART
ap-5030	188	16	explore	explore	VERB
ap-5030	188	17	.	.	PUNCT
ap-5030	189	1	a	a	DET
ap-5030	189	2	sufficiently	sufficiently	ADV
ap-5030	189	3	competent	competent	ADJ
ap-5030	189	4	and	and	CCONJ
ap-5030	189	5	autonomous	autonomous	ADJ
ap-5030	189	6	reinforcement	reinforcement	NOUN
ap-5030	189	7	learning	learning	NOUN
ap-5030	189	8	system	system	NOUN
ap-5030	189	9	could	could	AUX
ap-5030	189	10	be	be	AUX
ap-5030	189	11	pitted	pit	VERB
ap-5030	189	12	against	against	ADP
ap-5030	189	13	real	real	ADJ
ap-5030	189	14	-	-	PUNCT
ap-5030	189	15	life	life	NOUN
ap-5030	189	16	systems	system	NOUN
ap-5030	189	17	(	(	PUNCT
ap-5030	189	18	preferably	preferably	ADV
ap-5030	189	19	in	in	ADP
ap-5030	189	20	a	a	DET
ap-5030	189	21	laboratory	laboratory	NOUN
ap-5030	189	22	environment	environment	NOUN
ap-5030	189	23	)	)	PUNCT
ap-5030	189	24	for	for	ADP
ap-5030	189	25	further	further	ADJ
ap-5030	189	26	refinement	refinement	NOUN
ap-5030	189	27	without	without	ADP
ap-5030	189	28	a	a	DET
ap-5030	189	29	human	human	ADJ
ap-5030	189	30	interference	interference	NOUN
ap-5030	189	31	(	(	PUNCT
ap-5030	189	32	similar	similar	ADJ
ap-5030	189	33	to	to	ADP
ap-5030	189	34	alphazero	alphazero	NOUN
ap-5030	189	35	[	[	X
ap-5030	189	36	41	41	NUM
ap-5030	189	37	]	]	PUNCT
ap-5030	189	38	.	.	PUNCT
ap-5030	189	39	)	)	PUNCT
ap-5030	190	1	these	these	DET
ap-5030	190	2	proposals	proposal	NOUN
ap-5030	190	3	require	require	VERB
ap-5030	190	4	at	at	ADP
ap-5030	190	5	least	least	ADJ
ap-5030	190	6	a	a	DET
ap-5030	190	7	rudimentary	rudimentary	ADJ
ap-5030	190	8	implementation	implementation	NOUN
ap-5030	190	9	of	of	ADP
ap-5030	190	10	the	the	DET
ap-5030	190	11	design	design	NOUN
ap-5030	190	12	and	and	CCONJ
ap-5030	190	13	the	the	DET
ap-5030	190	14	insight	insight	NOUN
ap-5030	190	15	gained	gain	VERB
ap-5030	190	16	from	from	ADP
ap-5030	190	17	attacking	attack	VERB
ap-5030	190	18	this	this	DET
ap-5030	190	19	system	system	NOUN
ap-5030	190	20	could	could	AUX
ap-5030	190	21	be	be	AUX
ap-5030	190	22	transferred	transfer	VERB
ap-5030	190	23	to	to	ADP
ap-5030	190	24	other	other	ADJ
ap-5030	190	25	systems	system	NOUN
ap-5030	190	26	still	still	ADV
ap-5030	190	27	in	in	ADP
ap-5030	190	28	the	the	DET
ap-5030	190	29	design	design	NOUN
ap-5030	190	30	and	and	CCONJ
ap-5030	190	31	analysis	analysis	NOUN
ap-5030	190	32	phase	phase	NOUN
ap-5030	190	33	.	.	PUNCT
ap-5030	191	1	finding	find	VERB
ap-5030	191	2	the	the	DET
ap-5030	191	3	best	good	ADJ
ap-5030	191	4	parameters	parameter	NOUN
ap-5030	191	5	for	for	ADP
ap-5030	191	6	the	the	DET
ap-5030	191	7	reward	reward	NOUN
ap-5030	191	8	and	and	CCONJ
ap-5030	191	9	success	success	NOUN
ap-5030	191	10	rates	rate	NOUN
ap-5030	191	11	represents	represent	VERB
ap-5030	191	12	an	an	DET
ap-5030	191	13	additional	additional	ADJ
ap-5030	191	14	challenge	challenge	NOUN
ap-5030	191	15	.	.	PUNCT
ap-5030	192	1	different	different	ADJ
ap-5030	192	2	domain	domain	NOUN
ap-5030	192	3	experts	expert	NOUN
ap-5030	192	4	may	may	AUX
ap-5030	192	5	have	have	VERB
ap-5030	192	6	different	different	ADJ
ap-5030	192	7	assessments	assessment	NOUN
ap-5030	192	8	of	of	ADP
ap-5030	192	9	the	the	DET
ap-5030	192	10	threat	threat	NOUN
ap-5030	192	11	landscape	landscape	NOUN
ap-5030	192	12	or	or	CCONJ
ap-5030	192	13	varying	vary	VERB
ap-5030	192	14	confidence	confidence	NOUN
ap-5030	192	15	in	in	ADP
ap-5030	192	16	their	their	PRON
ap-5030	192	17	assumptions	assumption	NOUN
ap-5030	192	18	.	.	PUNCT
ap-5030	193	1	running	run	VERB
ap-5030	193	2	a	a	DET
ap-5030	193	3	large	large	ADJ
ap-5030	193	4	number	number	NOUN
ap-5030	193	5	of	of	ADP
ap-5030	193	6	scenarios	scenario	NOUN
ap-5030	193	7	with	with	ADP
ap-5030	193	8	different	different	ADJ
ap-5030	193	9	parameters	parameter	NOUN
ap-5030	193	10	also	also	ADV
ap-5030	193	11	increases	increase	VERB
ap-5030	193	12	the	the	DET
ap-5030	193	13	chance	chance	NOUN
ap-5030	193	14	to	to	PART
ap-5030	193	15	find	find	VERB
ap-5030	193	16	an	an	DET
ap-5030	193	17	efficient	efficient	ADJ
ap-5030	193	18	and	and	CCONJ
ap-5030	193	19	robust	robust	ADJ
ap-5030	193	20	solution	solution	NOUN
ap-5030	193	21	.	.	PUNCT
ap-5030	194	1	in	in	ADP
ap-5030	194	2	later	later	ADJ
ap-5030	194	3	stages	stage	NOUN
ap-5030	194	4	of	of	ADP
ap-5030	194	5	the	the	DET
ap-5030	194	6	development	development	NOUN
ap-5030	194	7	cycle	cycle	NOUN
ap-5030	194	8	,	,	PUNCT
ap-5030	194	9	data	datum	NOUN
ap-5030	194	10	from	from	ADP
ap-5030	194	11	penetration	penetration	NOUN
ap-5030	194	12	tests	test	NOUN
ap-5030	194	13	and	and	CCONJ
ap-5030	194	14	from	from	ADP
ap-5030	194	15	security	security	NOUN
ap-5030	194	16	incidents	incident	NOUN
ap-5030	194	17	can	can	AUX
ap-5030	194	18	be	be	AUX
ap-5030	194	19	used	use	VERB
ap-5030	194	20	to	to	PART
ap-5030	194	21	verify	verify	VERB
ap-5030	194	22	the	the	DET
ap-5030	194	23	validity	validity	NOUN
ap-5030	194	24	of	of	ADP
ap-5030	194	25	the	the	DET
ap-5030	194	26	assumptions	assumption	NOUN
ap-5030	194	27	made	make	VERB
ap-5030	194	28	.	.	PUNCT
ap-5030	195	1	to	to	PART
ap-5030	195	2	simplify	simplify	VERB
ap-5030	195	3	the	the	DET
ap-5030	195	4	creation	creation	NOUN
ap-5030	195	5	of	of	ADP
ap-5030	195	6	scenarios	scenario	NOUN
ap-5030	195	7	,	,	PUNCT
ap-5030	195	8	the	the	DET
ap-5030	195	9	point	point	NOUN
ap-5030	195	10	values	value	NOUN
ap-5030	195	11	for	for	ADP
ap-5030	195	12	the	the	DET
ap-5030	195	13	cost	cost	NOUN
ap-5030	195	14	and	and	CCONJ
ap-5030	195	15	success	success	NOUN
ap-5030	195	16	rate	rate	NOUN
ap-5030	195	17	could	could	AUX
ap-5030	195	18	be	be	AUX
ap-5030	195	19	replaced	replace	VERB
ap-5030	195	20	with	with	ADP
ap-5030	195	21	distributions	distribution	NOUN
ap-5030	195	22	(	(	PUNCT
ap-5030	195	23	similar	similar	ADJ
ap-5030	195	24	to	to	ADP
ap-5030	195	25	hierarchical	hierarchical	ADJ
ap-5030	195	26	learning	learning	NOUN
ap-5030	195	27	in	in	ADP
ap-5030	195	28	probabilistic	probabilistic	ADJ
ap-5030	195	29	programming	programming	NOUN
ap-5030	195	30	)	)	PUNCT
ap-5030	195	31	where	where	SCONJ
ap-5030	195	32	the	the	DET
ap-5030	195	33	actual	actual	ADJ
ap-5030	195	34	values	value	NOUN
ap-5030	195	35	for	for	ADP
ap-5030	195	36	each	each	DET
ap-5030	195	37	run	run	NOUN
ap-5030	195	38	are	be	AUX
ap-5030	195	39	drawn	draw	VERB
ap-5030	195	40	from	from	ADP
ap-5030	195	41	.	.	PROPN
ap-5030	196	1	7	7	X
ap-5030	196	2	.	.	X
ap-5030	196	3	conclusion	conclusion	NOUN
ap-5030	196	4	reinforcement	reinforcement	NOUN
ap-5030	196	5	learning	learning	NOUN
ap-5030	196	6	can	can	AUX
ap-5030	196	7	be	be	AUX
ap-5030	196	8	used	use	VERB
ap-5030	196	9	to	to	PART
ap-5030	196	10	identify	identify	VERB
ap-5030	196	11	potential	potential	ADJ
ap-5030	196	12	weaknesses	weakness	NOUN
ap-5030	196	13	in	in	ADP
ap-5030	196	14	an	an	DET
ap-5030	196	15	fpga	fpga	PROPN
ap-5030	196	16	design	design	NOUN
ap-5030	196	17	and	and	CCONJ
ap-5030	196	18	the	the	DET
ap-5030	196	19	steps	step	NOUN
ap-5030	196	20	a	a	DET
ap-5030	196	21	reasonable	reasonable	ADJ
ap-5030	196	22	attacker	attacker	NOUN
ap-5030	196	23	may	may	AUX
ap-5030	196	24	take	take	VERB
ap-5030	196	25	to	to	PART
ap-5030	196	26	exploit	exploit	VERB
ap-5030	196	27	them	they	PRON
ap-5030	196	28	.	.	PUNCT
ap-5030	197	1	we	we	PRON
ap-5030	197	2	presented	present	VERB
ap-5030	197	3	a	a	DET
ap-5030	197	4	method	method	NOUN
ap-5030	197	5	to	to	PART
ap-5030	197	6	generate	generate	VERB
ap-5030	197	7	a	a	DET
ap-5030	197	8	markov	markov	NOUN
ap-5030	197	9	decision	decision	NOUN
ap-5030	197	10	process	process	NOUN
ap-5030	197	11	based	base	VERB
ap-5030	197	12	reinforcement	reinforcement	NOUN
ap-5030	197	13	learning	learning	NOUN
ap-5030	197	14	model	model	NOUN
ap-5030	197	15	from	from	ADP
ap-5030	197	16	a	a	DET
ap-5030	197	17	formal	formal	ADJ
ap-5030	197	18	,	,	PUNCT
ap-5030	197	19	high	high	ADJ
ap-5030	197	20	-	-	PUNCT
ap-5030	197	21	level	level	NOUN
ap-5030	197	22	system	system	NOUN
ap-5030	197	23	description	description	NOUN
ap-5030	197	24	(	(	PUNCT
ap-5030	197	25	formulated	formulate	VERB
ap-5030	197	26	in	in	ADP
ap-5030	197	27	the	the	DET
ap-5030	197	28	domain	domain	NOUN
ap-5030	197	29	-	-	PUNCT
ap-5030	197	30	specific	specific	ADJ
ap-5030	197	31	language	language	NOUN
ap-5030	197	32	fpgasecml	fpgasecml	NOUN
ap-5030	197	33	.	.	PUNCT
ap-5030	197	34	)	)	PUNCT
ap-5030	198	1	the	the	DET
ap-5030	198	2	automatic	automatic	ADJ
ap-5030	198	3	model	model	NOUN
ap-5030	198	4	-	-	PUNCT
ap-5030	198	5	to	to	ADP
ap-5030	198	6	-	-	PUNCT
ap-5030	198	7	model	model	NOUN
ap-5030	198	8	translation	translation	NOUN
ap-5030	198	9	reduces	reduce	VERB
ap-5030	198	10	the	the	DET
ap-5030	198	11	developer	developer	NOUN
ap-5030	198	12	’s	’s	PART
ap-5030	198	13	workload	workload	NOUN
ap-5030	198	14	,	,	PUNCT
ap-5030	198	15	decreases	decrease	VERB
ap-5030	198	16	the	the	DET
ap-5030	198	17	risk	risk	NOUN
ap-5030	198	18	of	of	ADP
ap-5030	198	19	errors	error	NOUN
ap-5030	198	20	and	and	CCONJ
ap-5030	198	21	helps	help	VERB
ap-5030	198	22	to	to	PART
ap-5030	198	23	keep	keep	VERB
ap-5030	198	24	both	both	DET
ap-5030	198	25	models	model	NOUN
ap-5030	198	26	in	in	ADP
ap-5030	198	27	sync	sync	NOUN
ap-5030	198	28	.	.	PUNCT
ap-5030	199	1	probabilistic	probabilistic	ADJ
ap-5030	199	2	transitions	transition	NOUN
ap-5030	199	3	and	and	CCONJ
ap-5030	199	4	the	the	DET
ap-5030	199	5	reward	reward	NOUN
ap-5030	199	6	function	function	NOUN
ap-5030	199	7	can	can	AUX
ap-5030	199	8	be	be	AUX
ap-5030	199	9	used	use	VERB
ap-5030	199	10	to	to	PART
ap-5030	199	11	model	model	VERB
ap-5030	199	12	the	the	DET
ap-5030	199	13	varying	vary	VERB
ap-5030	199	14	resilience	resilience	NOUN
ap-5030	199	15	of	of	ADP
ap-5030	199	16	different	different	ADJ
ap-5030	199	17	elements	element	NOUN
ap-5030	199	18	against	against	ADP
ap-5030	199	19	attacks	attack	NOUN
ap-5030	199	20	and	and	CCONJ
ap-5030	199	21	the	the	DET
ap-5030	199	22	capabilities	capability	NOUN
ap-5030	199	23	of	of	ADP
ap-5030	199	24	an	an	DET
ap-5030	199	25	attacker	attacker	NOUN
ap-5030	199	26	.	.	PUNCT
ap-5030	200	1	supplementary	supplementary	ADJ
ap-5030	200	2	techniques	technique	NOUN
ap-5030	200	3	like	like	ADP
ap-5030	200	4	hierarchical	hierarchical	ADJ
ap-5030	200	5	learning	learning	NOUN
ap-5030	200	6	,	,	PUNCT
ap-5030	200	7	partially	partially	ADV
ap-5030	200	8	observable	observable	ADJ
ap-5030	200	9	mdps	mdps	NOUN
ap-5030	200	10	,	,	PUNCT
ap-5030	200	11	and	and	CCONJ
ap-5030	200	12	stochastic	stochastic	ADJ
ap-5030	200	13	games	game	NOUN
ap-5030	200	14	can	can	AUX
ap-5030	200	15	be	be	AUX
ap-5030	200	16	used	use	VERB
ap-5030	200	17	to	to	PART
ap-5030	200	18	extend	extend	VERB
ap-5030	200	19	the	the	DET
ap-5030	200	20	scope	scope	NOUN
ap-5030	200	21	of	of	ADP
ap-5030	200	22	the	the	DET
ap-5030	200	23	model	model	NOUN
ap-5030	200	24	.	.	PUNCT
ap-5030	201	1	references	reference	NOUN
ap-5030	201	2	[	[	X
ap-5030	201	3	1	1	NUM
ap-5030	201	4	]	]	X
ap-5030	201	5	b.	b.	PROPN
ap-5030	201	6	schneier	schneier	PROPN
ap-5030	201	7	.	.	PUNCT
ap-5030	202	1	academic	academic	ADJ
ap-5030	202	2	:	:	PUNCT
ap-5030	202	3	attack	attack	NOUN
ap-5030	202	4	trees	tree	NOUN
ap-5030	202	5	:	:	PUNCT
ap-5030	202	6	schneier	schneier	PROPN
ap-5030	202	7	on	on	ADP
ap-5030	202	8	security	security	NOUN
ap-5030	202	9	.	.	PUNCT
ap-5030	203	1	https://www.schneier.com/academic/	https://www.schneier.com/academic/	X
ap-5030	203	2	archives/1999/12	archives/1999/12	PROPN
ap-5030	203	3	/	/	SYM
ap-5030	203	4	attack_trees.html	attack_trees.html	PUNCT
ap-5030	203	5	.	.	PUNCT
ap-5030	204	1	[	[	X
ap-5030	204	2	2	2	X
ap-5030	204	3	]	]	PUNCT
ap-5030	204	4	s.	s.	PROPN
ap-5030	204	5	m.	m.	PROPN
ap-5030	204	6	trimberger	trimberger	PROPN
ap-5030	204	7	,	,	PUNCT
ap-5030	204	8	j.	j.	PROPN
ap-5030	204	9	j.	j.	PROPN
ap-5030	204	10	moore	moore	PROPN
ap-5030	204	11	.	.	PUNCT
ap-5030	205	1	fpga	fpga	PROPN
ap-5030	205	2	security	security	PROPN
ap-5030	205	3	:	:	PUNCT
ap-5030	205	4	motivations	motivation	NOUN
ap-5030	205	5	,	,	PUNCT
ap-5030	205	6	features	feature	NOUN
ap-5030	205	7	,	,	PUNCT
ap-5030	205	8	and	and	CCONJ
ap-5030	205	9	applications	application	NOUN
ap-5030	205	10	.	.	PUNCT
ap-5030	206	1	proceedings	proceeding	NOUN
ap-5030	206	2	of	of	ADP
ap-5030	206	3	the	the	DET
ap-5030	206	4	ieee	ieee	NOUN
ap-5030	206	5	102(8):1248–1265	102(8):1248–1265	NUM
ap-5030	206	6	,	,	PUNCT
ap-5030	206	7	2014	2014	NUM
ap-5030	206	8	.	.	PUNCT
ap-5030	207	1	doi:10.1109	doi:10.1109	VERB
ap-5030	207	2	/	/	SYM
ap-5030	207	3	jproc.2014.2331672	jproc.2014.2331672	PROPN
ap-5030	207	4	.	.	PUNCT
ap-5030	208	1	[	[	X
ap-5030	208	2	3	3	X
ap-5030	208	3	]	]	X
ap-5030	208	4	c.	c.	PROPN
ap-5030	208	5	isbell	isbell	PROPN
ap-5030	208	6	,	,	PUNCT
ap-5030	208	7	m.	m.	PROPN
ap-5030	208	8	littman	littman	PROPN
ap-5030	208	9	.	.	PUNCT
ap-5030	209	1	reinforcement	reinforcement	NOUN
ap-5030	209	2	learning	learning	NOUN
ap-5030	209	3	udacity	udacity	NOUN
ap-5030	209	4	.	.	PUNCT
ap-5030	210	1	https://classroom.udacity.com/courses/ud600	https://classroom.udacity.com/courses/ud600	NOUN
ap-5030	210	2	.	.	PUNCT
ap-5030	211	1	[	[	X
ap-5030	211	2	4	4	X
ap-5030	211	3	]	]	PUNCT
ap-5030	211	4	f.	f.	PROPN
ap-5030	211	5	akhtar	akhtar	PROPN
ap-5030	211	6	.	.	PUNCT
ap-5030	212	1	practical	practical	ADJ
ap-5030	212	2	reinforcment	reinforcment	PROPN
ap-5030	212	3	learning	learning	NOUN
ap-5030	212	4	.	.	PUNCT
ap-5030	213	1	packt	packt	PROPN
ap-5030	213	2	publishing	publishing	PROPN
ap-5030	213	3	limited	limit	VERB
ap-5030	213	4	,	,	PUNCT
ap-5030	213	5	2017	2017	NUM
ap-5030	213	6	.	.	PUNCT
ap-5030	214	1	[	[	X
ap-5030	214	2	5	5	NUM
ap-5030	214	3	]	]	PUNCT
ap-5030	214	4	m.	m.	NOUN
ap-5030	214	5	brambilla	brambilla	NOUN
ap-5030	214	6	,	,	PUNCT
ap-5030	214	7	j.	j.	PROPN
ap-5030	214	8	cabot	cabot	PROPN
ap-5030	214	9	,	,	PUNCT
ap-5030	214	10	m.	m.	PROPN
ap-5030	214	11	wimmer	wimmer	PROPN
ap-5030	214	12	.	.	PUNCT
ap-5030	214	13	model	model	NOUN
ap-5030	214	14	-	-	PUNCT
ap-5030	214	15	driven	drive	VERB
ap-5030	214	16	software	software	NOUN
ap-5030	214	17	engineering	engineering	NOUN
ap-5030	214	18	in	in	ADP
ap-5030	214	19	practice	practice	NOUN
ap-5030	214	20	.	.	PUNCT
ap-5030	215	1	synthesis	synthesis	NOUN
ap-5030	215	2	lectures	lecture	NOUN
ap-5030	215	3	on	on	ADP
ap-5030	215	4	software	software	NOUN
ap-5030	215	5	engineering	engineering	NOUN
ap-5030	215	6	.	.	PUNCT
ap-5030	216	1	morgan	morgan	PROPN
ap-5030	216	2	&	&	CCONJ
ap-5030	216	3	claypool	claypool	PROPN
ap-5030	216	4	publishers	publisher	NOUN
ap-5030	216	5	,	,	PUNCT
ap-5030	216	6	second	second	ADJ
ap-5030	216	7	edition	edition	NOUN
ap-5030	216	8	edn	edn	PROPN
ap-5030	216	9	.	.	PROPN
ap-5030	216	10	,	,	PUNCT
ap-5030	216	11	2017	2017	NUM
ap-5030	216	12	.	.	PUNCT
ap-5030	217	1	[	[	X
ap-5030	217	2	6	6	NUM
ap-5030	217	3	]	]	PUNCT
ap-5030	217	4	l.	l.	PROPN
ap-5030	217	5	delligatti	delligatti	PROPN
ap-5030	217	6	.	.	PUNCT
ap-5030	218	1	sysml	sysml	NOUN
ap-5030	218	2	distilled	distil	VERB
ap-5030	218	3	:	:	PUNCT
ap-5030	218	4	a	a	DET
ap-5030	218	5	brief	brief	ADJ
ap-5030	218	6	guide	guide	NOUN
ap-5030	218	7	to	to	ADP
ap-5030	218	8	the	the	DET
ap-5030	218	9	systems	system	NOUN
ap-5030	218	10	modeling	modeling	NOUN
ap-5030	218	11	language	language	NOUN
ap-5030	218	12	.	.	PUNCT
ap-5030	219	1	addison	addison	PROPN
ap-5030	219	2	-	-	PUNCT
ap-5030	219	3	wesley	wesley	PROPN
ap-5030	219	4	professional	professional	NOUN
ap-5030	219	5	,	,	PUNCT
ap-5030	219	6	2014	2014	NUM
ap-5030	219	7	.	.	PUNCT
ap-5030	220	1	[	[	X
ap-5030	220	2	7	7	X
ap-5030	220	3	]	]	X
ap-5030	220	4	j.	j.	PROPN
ap-5030	220	5	delange	delange	PROPN
ap-5030	220	6	.	.	PUNCT
ap-5030	221	1	aadl	aadl	NOUN
ap-5030	221	2	in	in	ADP
ap-5030	221	3	practice	practice	NOUN
ap-5030	221	4	.	.	PUNCT
ap-5030	222	1	reblechon	reblechon	PROPN
ap-5030	222	2	development	development	PROPN
ap-5030	222	3	co	co	PROPN
ap-5030	222	4	,	,	PUNCT
ap-5030	222	5	2017	2017	NUM
ap-5030	222	6	.	.	PUNCT
ap-5030	223	1	[	[	X
ap-5030	223	2	8	8	NUM
ap-5030	223	3	]	]	X
ap-5030	223	4	robert	robert	PROPN
ap-5030	223	5	ellison	ellison	PROPN
ap-5030	223	6	,	,	PUNCT
ap-5030	223	7	allen	allen	PROPN
ap-5030	223	8	householder	householder	PROPN
ap-5030	223	9	,	,	PUNCT
ap-5030	223	10	john	john	PROPN
ap-5030	223	11	hudak	hudak	PROPN
ap-5030	223	12	,	,	PUNCT
ap-5030	223	13	rik	rik	PROPN
ap-5030	223	14	kazman	kazman	PROPN
ap-5030	223	15	,	,	PUNCT
ap-5030	223	16	carol	carol	NOUN
ap-5030	223	17	woody	woody	NOUN
ap-5030	223	18	.	.	PUNCT
ap-5030	224	1	extending	extend	VERB
ap-5030	224	2	aadl	aadl	NOUN
ap-5030	224	3	for	for	ADP
ap-5030	224	4	security	security	NOUN
ap-5030	224	5	design	design	NOUN
ap-5030	224	6	assurance	assurance	NOUN
ap-5030	224	7	of	of	ADP
ap-5030	224	8	cyber	cyber	NOUN
ap-5030	224	9	-	-	PUNCT
ap-5030	224	10	physical	physical	ADJ
ap-5030	224	11	systems	system	NOUN
ap-5030	224	12	.	.	PUNCT
ap-5030	225	1	https://resources.sei.cmu.edu/asset_files/	https://resources.sei.cmu.edu/asset_files/	PROPN
ap-5030	225	2	technicalreport/2015_005_001_449522.pdf	technicalreport/2015_005_001_449522.pdf	NOUN
ap-5030	225	3	.	.	PUNCT
ap-5030	226	1	[	[	X
ap-5030	226	2	9	9	NUM
ap-5030	226	3	]	]	PUNCT
ap-5030	226	4	f.	f.	NOUN
ap-5030	226	5	swiderski	swiderski	PROPN
ap-5030	226	6	,	,	PUNCT
ap-5030	226	7	w.	w.	PROPN
ap-5030	226	8	snyder	snyder	PROPN
ap-5030	226	9	.	.	PUNCT
ap-5030	227	1	threat	threat	NOUN
ap-5030	227	2	modeling	modeling	NOUN
ap-5030	227	3	.	.	PUNCT
ap-5030	228	1	microsoft	microsoft	PROPN
ap-5030	228	2	press	press	PROPN
ap-5030	228	3	,	,	PUNCT
ap-5030	228	4	redmond	redmond	NOUN
ap-5030	228	5	and	and	CCONJ
ap-5030	228	6	wash	wash	NOUN
ap-5030	228	7	,	,	PUNCT
ap-5030	228	8	2004	2004	NUM
ap-5030	228	9	.	.	PUNCT
ap-5030	229	1	[	[	X
ap-5030	229	2	10	10	NUM
ap-5030	229	3	]	]	X
ap-5030	229	4	jegeib	jegeib	PROPN
ap-5030	229	5	.	.	PUNCT
ap-5030	230	1	getting	getting	AUX
ap-5030	230	2	started	start	VERB
ap-5030	230	3	microsoft	microsoft	PROPN
ap-5030	230	4	threat	threat	NOUN
ap-5030	230	5	modeling	model	VERB
ap-5030	230	6	tool	tool	NOUN
ap-5030	230	7	azure	azure	NOUN
ap-5030	230	8	.	.	PUNCT
ap-5030	231	1	https://docs.microsoft.com/en-us/azure/	https://docs.microsoft.com/en-us/azure/	PROPN
ap-5030	231	2	security	security	NOUN
ap-5030	231	3	/	/	SYM
ap-5030	231	4	azure	azure	NOUN
ap-5030	231	5	-	-	PUNCT
ap-5030	231	6	security	security	NOUN
ap-5030	231	7	-	-	PUNCT
ap-5030	231	8	threat	threat	NOUN
ap-5030	231	9	-	-	PUNCT
ap-5030	231	10	modelingtool	modelingtool	NOUN
ap-5030	231	11	-	-	PUNCT
ap-5030	231	12	getting	getting	NOUN
ap-5030	231	13	-	-	PUNCT
ap-5030	231	14	started	start	VERB
ap-5030	231	15	.	.	PUNCT
ap-5030	232	1	[	[	X
ap-5030	232	2	11	11	NUM
ap-5030	232	3	]	]	PUNCT
ap-5030	232	4	threat	threat	NOUN
ap-5030	232	5	dragon	dragon	NOUN
ap-5030	232	6	.	.	PUNCT
ap-5030	233	1	https://threatdragon.org/login	https://threatdragon.org/login	PROPN
ap-5030	233	2	.	.	PUNCT
ap-5030	234	1	[	[	X
ap-5030	234	2	12	12	NUM
ap-5030	234	3	]	]	X
ap-5030	234	4	j.	j.	PROPN
ap-5030	234	5	saxe	saxe	PROPN
ap-5030	234	6	,	,	PUNCT
ap-5030	234	7	h.	h.	PROPN
ap-5030	234	8	sanders	sanders	PROPN
ap-5030	234	9	.	.	PUNCT
ap-5030	235	1	malware	malware	PROPN
ap-5030	235	2	data	data	PROPN
ap-5030	235	3	science	science	NOUN
ap-5030	235	4	:	:	PUNCT
ap-5030	235	5	attack	attack	NOUN
ap-5030	235	6	detection	detection	NOUN
ap-5030	235	7	and	and	CCONJ
ap-5030	235	8	attribution	attribution	NOUN
ap-5030	235	9	.	.	PUNCT
ap-5030	236	1	no	no	DET
ap-5030	236	2	starch	starch	NOUN
ap-5030	236	3	press	press	NOUN
ap-5030	236	4	incorporated	incorporate	VERB
ap-5030	236	5	,	,	PUNCT
ap-5030	236	6	san	san	PROPN
ap-5030	236	7	francisco	francisco	PROPN
ap-5030	236	8	,	,	PUNCT
ap-5030	236	9	ca	ca	NOUN
ap-5030	236	10	,	,	PUNCT
ap-5030	236	11	2018	2018	NUM
ap-5030	236	12	.	.	PUNCT
ap-5030	237	1	[	[	X
ap-5030	237	2	13	13	NUM
ap-5030	237	3	]	]	PUNCT
ap-5030	237	4	s.	s.	PROPN
ap-5030	237	5	parkinson	parkinson	PROPN
ap-5030	237	6	,	,	PUNCT
ap-5030	237	7	a.	a.	PROPN
ap-5030	237	8	crampton	crampton	PROPN
ap-5030	237	9	,	,	PUNCT
ap-5030	237	10	r.	r.	PROPN
ap-5030	237	11	hill	hill	PROPN
ap-5030	237	12	.	.	PUNCT
ap-5030	238	1	guide	guide	VERB
ap-5030	238	2	to	to	PART
ap-5030	238	3	vulnerability	vulnerability	VERB
ap-5030	238	4	analysis	analysis	NOUN
ap-5030	238	5	for	for	ADP
ap-5030	238	6	computer	computer	NOUN
ap-5030	238	7	networks	network	NOUN
ap-5030	238	8	and	and	CCONJ
ap-5030	238	9	systems	system	NOUN
ap-5030	238	10	:	:	PUNCT
ap-5030	238	11	an	an	DET
ap-5030	238	12	artificial	artificial	ADJ
ap-5030	238	13	intelligence	intelligence	NOUN
ap-5030	238	14	approach	approach	NOUN
ap-5030	238	15	.	.	PUNCT
ap-5030	239	1	computer	computer	NOUN
ap-5030	239	2	communications	communication	NOUN
ap-5030	239	3	and	and	CCONJ
ap-5030	239	4	networks	network	NOUN
ap-5030	239	5	.	.	PUNCT
ap-5030	240	1	springer	springer	NOUN
ap-5030	240	2	,	,	PUNCT
ap-5030	240	3	2018	2018	NUM
ap-5030	240	4	.	.	PUNCT
ap-5030	241	1	[	[	X
ap-5030	241	2	14	14	NUM
ap-5030	241	3	]	]	PUNCT
ap-5030	241	4	m.	m.	NOUN
ap-5030	241	5	sutton	sutton	PROPN
ap-5030	241	6	,	,	PUNCT
ap-5030	241	7	a.	a.	PROPN
ap-5030	241	8	greene	greene	PROPN
ap-5030	241	9	,	,	PUNCT
ap-5030	241	10	p.	p.	PROPN
ap-5030	241	11	amini	amini	PROPN
ap-5030	241	12	.	.	PUNCT
ap-5030	242	1	fuzzing	fuzzing	NOUN
ap-5030	242	2	:	:	PUNCT
ap-5030	242	3	brute	brute	ADJ
ap-5030	242	4	force	force	NOUN
ap-5030	242	5	vulnerabilty	vulnerabilty	NOUN
ap-5030	242	6	discovery	discovery	PROPN
ap-5030	242	7	.	.	PUNCT
ap-5030	243	1	addison	addison	PROPN
ap-5030	243	2	-	-	PUNCT
ap-5030	243	3	wesley	wesley	PROPN
ap-5030	243	4	,	,	PUNCT
ap-5030	243	5	upper	upper	ADJ
ap-5030	243	6	saddle	saddle	NOUN
ap-5030	243	7	river	river	PROPN
ap-5030	243	8	,	,	PUNCT
ap-5030	243	9	n.j	n.j	PROPN
ap-5030	243	10	.	.	PROPN
ap-5030	243	11	,	,	PUNCT
ap-5030	243	12	2007	2007	NUM
ap-5030	243	13	.	.	PUNCT
ap-5030	244	1	[	[	X
ap-5030	244	2	15	15	NUM
ap-5030	244	3	]	]	X
ap-5030	244	4	p.	p.	NOUN
ap-5030	244	5	bontrager	bontrager	PROPN
ap-5030	244	6	,	,	PUNCT
ap-5030	244	7	a.	a.	PROPN
ap-5030	244	8	roy	roy	PROPN
ap-5030	244	9	,	,	PUNCT
ap-5030	244	10	j.	j.	PROPN
ap-5030	244	11	togelius	togelius	PROPN
ap-5030	244	12	,	,	PUNCT
ap-5030	244	13	et	et	PROPN
ap-5030	244	14	al	al	PROPN
ap-5030	244	15	.	.	PROPN
ap-5030	244	16	deepmasterprints	deepmasterprint	NOUN
ap-5030	244	17	:	:	PUNCT
ap-5030	244	18	generating	generate	VERB
ap-5030	244	19	masterprints	masterprint	NOUN
ap-5030	244	20	for	for	ADP
ap-5030	244	21	dictionary	dictionary	ADJ
ap-5030	244	22	attacks	attack	NOUN
ap-5030	244	23	via	via	ADP
ap-5030	244	24	latent	latent	ADJ
ap-5030	244	25	variable	variable	ADJ
ap-5030	244	26	evolution	evolution	NOUN
ap-5030	244	27	.	.	PUNCT
ap-5030	245	1	https://arxiv.org/pdf/1705.07386.pdf	https://arxiv.org/pdf/1705.07386.pdf	NOUN
ap-5030	245	2	.	.	PUNCT
ap-5030	246	1	[	[	X
ap-5030	246	2	16	16	NUM
ap-5030	246	3	]	]	X
ap-5030	246	4	i.	i.	PROPN
ap-5030	246	5	goodfellow	goodfellow	PROPN
ap-5030	246	6	,	,	PUNCT
ap-5030	246	7	y.	y.	PROPN
ap-5030	246	8	bengio	bengio	PROPN
ap-5030	246	9	,	,	PUNCT
ap-5030	246	10	a.	a.	NOUN
ap-5030	246	11	courville	courville	NOUN
ap-5030	246	12	.	.	PUNCT
ap-5030	247	1	deep	deep	ADJ
ap-5030	247	2	learning	learning	NOUN
ap-5030	247	3	.	.	PUNCT
ap-5030	248	1	mit	mit	PROPN
ap-5030	248	2	press	press	PROPN
ap-5030	248	3	,	,	PUNCT
ap-5030	248	4	cambridge	cambridge	PROPN
ap-5030	248	5	,	,	PUNCT
ap-5030	248	6	massachusetts	massachusetts	PROPN
ap-5030	248	7	and	and	CCONJ
ap-5030	248	8	london	london	PROPN
ap-5030	248	9	,	,	PUNCT
ap-5030	248	10	england	england	PROPN
ap-5030	248	11	,	,	PUNCT
ap-5030	248	12	2016	2016	NUM
ap-5030	248	13	.	.	PUNCT
ap-5030	249	1	[	[	X
ap-5030	249	2	17	17	NUM
ap-5030	249	3	]	]	PUNCT
ap-5030	249	4	tencent	tencent	NOUN
ap-5030	249	5	keen	keen	ADJ
ap-5030	249	6	security	security	NOUN
ap-5030	249	7	lab	lab	NOUN
ap-5030	249	8	.	.	PUNCT
ap-5030	250	1	autopilot	autopilot	PROPN
ap-5030	250	2	.	.	PUNCT
ap-5030	251	1	https://keenlab.tencent.com/en/whitepapers/	https://keenlab.tencent.com/en/whitepapers/	PROPN
ap-5030	251	2	experimental_security_research_of_tesla	experimental_security_research_of_tesla	NOUN
ap-5030	251	3	_	_	X
ap-5030	252	1	autopilot.pdf	autopilot.pdf	NOUN
ap-5030	252	2	.	.	PUNCT
ap-5030	253	1	[	[	X
ap-5030	253	2	18	18	NUM
ap-5030	253	3	]	]	X
ap-5030	253	4	s.	s.	PROPN
ap-5030	253	5	lakshminarayana	lakshminarayana	PROPN
ap-5030	253	6	,	,	PUNCT
ap-5030	253	7	t.	t.	PROPN
ap-5030	253	8	z.	z.	PROPN
ap-5030	253	9	teng	teng	PROPN
ap-5030	253	10	,	,	PUNCT
ap-5030	254	1	d.	d.	PROPN
ap-5030	254	2	k.	k.	PROPN
ap-5030	255	1	y.	y.	PROPN
ap-5030	255	2	yau	yau	PROPN
ap-5030	255	3	,	,	PUNCT
ap-5030	255	4	r.	r.	PROPN
ap-5030	255	5	tan	tan	PROPN
ap-5030	255	6	.	.	PUNCT
ap-5030	256	1	optimal	optimal	ADJ
ap-5030	256	2	attack	attack	NOUN
ap-5030	256	3	against	against	ADP
ap-5030	256	4	cyber	cyber	NOUN
ap-5030	256	5	-	-	PUNCT
ap-5030	256	6	physical	physical	PROPN
ap-5030	256	7	control	control	NOUN
ap-5030	256	8	systems	system	NOUN
ap-5030	256	9	with	with	ADP
ap-5030	256	10	reactive	reactive	ADJ
ap-5030	256	11	attack	attack	NOUN
ap-5030	256	12	mitigation	mitigation	NOUN
ap-5030	256	13	.	.	PUNCT
ap-5030	257	1	proceedings	proceeding	NOUN
ap-5030	257	2	of	of	ADP
ap-5030	257	3	the	the	DET
ap-5030	257	4	eighth	eighth	ADJ
ap-5030	257	5	international	international	ADJ
ap-5030	257	6	conference	conference	NOUN
ap-5030	257	7	on	on	ADP
ap-5030	257	8	future	future	ADJ
ap-5030	257	9	energy	energy	NOUN
ap-5030	257	10	systems	system	NOUN
ap-5030	257	11	2017	2017	NUM
ap-5030	257	12	,	,	PUNCT
ap-5030	257	13	2017	2017	NUM
ap-5030	257	14	.	.	PUNCT
ap-5030	257	15	doi:10.1145/3077839.3077852	doi:10.1145/3077839.3077852	NOUN
ap-5030	257	16	.	.	PUNCT
ap-5030	258	1	[	[	X
ap-5030	258	2	19	19	NUM
ap-5030	258	3	]	]	X
ap-5030	258	4	h.	h.	PROPN
ap-5030	258	5	koduvely	koduvely	PROPN
ap-5030	258	6	.	.	PUNCT
ap-5030	259	1	anomaly	anomaly	PROPN
ap-5030	259	2	detection	detection	NOUN
ap-5030	259	3	through	through	ADP
ap-5030	259	4	reinforcement	reinforcement	NOUN
ap-5030	259	5	learning	learning	NOUN
ap-5030	259	6	.	.	PUNCT
ap-5030	260	1	http://blog.zighra.com/	http://blog.zighra.com/	PROPN
ap-5030	260	2	anomaly	anomaly	NOUN
ap-5030	260	3	-	-	PUNCT
ap-5030	260	4	detection	detection	NOUN
ap-5030	260	5	-	-	PUNCT
ap-5030	260	6	and	and	CCONJ
ap-5030	260	7	-	-	PUNCT
ap-5030	260	8	reinforcement	reinforcement	NOUN
ap-5030	260	9	-	-	PUNCT
ap-5030	260	10	learning	learning	NOUN
ap-5030	260	11	.	.	PUNCT
ap-5030	261	1	525	525	NUM
ap-5030	261	2	https://www.schneier.com/academic/archives/1999/12/attack_trees.html	https://www.schneier.com/academic/archives/1999/12/attack_trees.html	PROPN
ap-5030	261	3	https://www.schneier.com/academic/archives/1999/12/attack_trees.html	https://www.schneier.com/academic/archives/1999/12/attack_trees.html	PROPN
ap-5030	261	4	http://dx.doi.org/10.1109/jproc.2014.2331672	http://dx.doi.org/10.1109/jproc.2014.2331672	NOUN
ap-5030	261	5	https://classroom.udacity.com/courses/ud600	https://classroom.udacity.com/courses/ud600	ADP
ap-5030	261	6	https://resources.sei.cmu.edu/asset_files/technicalreport/2015_005_001_449522.pdf	https://resources.sei.cmu.edu/asset_files/technicalreport/2015_005_001_449522.pdf	NOUN
ap-5030	261	7	https://resources.sei.cmu.edu/asset_files/technicalreport/2015_005_001_449522.pdf	https://resources.sei.cmu.edu/asset_files/technicalreport/2015_005_001_449522.pdf	NOUN
ap-5030	261	8	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	PROPN
ap-5030	261	9	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	PROPN
ap-5030	261	10	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	https://docs.microsoft.com/en-us/azure/security/azure-security-threat-modeling-tool-getting-started	PROPN
ap-5030	261	11	https://threatdragon.org/login	https://threatdragon.org/login	PROPN
ap-5030	261	12	https://arxiv.org/pdf/1705.07386.pdf	https://arxiv.org/pdf/1705.07386.pdf	PROPN
ap-5030	261	13	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	PROPN
ap-5030	261	14	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	PROPN
ap-5030	261	15	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	https://keenlab.tencent.com/en/whitepapers/experimental_security_research_of_tesla_autopilot.pdf	NOUN
ap-5030	261	16	http://dx.doi.org/10.1145/3077839.3077852	http://dx.doi.org/10.1145/3077839.3077852	NOUN
ap-5030	261	17	http://blog.zighra.com/anomaly-detection-and-reinforcement-learning	http://blog.zighra.com/anomaly-detection-and-reinforcement-learne	VERB
ap-5030	261	18	http://blog.zighra.com/anomaly-detection-and-reinforcement-learning	http://blog.zighra.com/anomaly-detection-and-reinforcement-learne	VERB
ap-5030	261	19	michael	michael	PROPN
ap-5030	261	20	vetter	vetter	PROPN
ap-5030	261	21	acta	acta	PROPN
ap-5030	261	22	polytechnica	polytechnica	PROPN
ap-5030	262	1	[	[	X
ap-5030	262	2	20	20	NUM
ap-5030	262	3	]	]	X
ap-5030	262	4	cyber	cyber	ADJ
ap-5030	262	5	grand	grand	ADJ
ap-5030	262	6	challenge	challenge	NOUN
ap-5030	262	7	(	(	PUNCT
ap-5030	262	8	cgc	cgc	PROPN
ap-5030	262	9	)	)	PUNCT
ap-5030	262	10	.	.	PUNCT
ap-5030	263	1	https	https	NOUN
ap-5030	263	2	:	:	PUNCT
ap-5030	263	3	//www.darpa.mil	//www.darpa.mil	PUNCT
ap-5030	263	4	/	/	SYM
ap-5030	263	5	program	program	NOUN
ap-5030	263	6	/	/	SYM
ap-5030	263	7	cyber	cyber	NOUN
ap-5030	263	8	-	-	PUNCT
ap-5030	263	9	grand	grand	ADJ
ap-5030	263	10	-	-	PUNCT
ap-5030	263	11	challenge	challenge	NOUN
ap-5030	263	12	.	.	PUNCT
ap-5030	264	1	[	[	X
ap-5030	264	2	21	21	NUM
ap-5030	264	3	]	]	X
ap-5030	264	4	mayhem	mayhem	NOUN
ap-5030	264	5	,	,	PUNCT
ap-5030	264	6	the	the	DET
ap-5030	264	7	tech	tech	NOUN
ap-5030	264	8	behind	behind	ADP
ap-5030	264	9	the	the	DET
ap-5030	264	10	darpa	darpa	PROPN
ap-5030	264	11	grand	grand	ADJ
ap-5030	264	12	challenge	challenge	NOUN
ap-5030	264	13	winner	winner	NOUN
ap-5030	264	14	,	,	PUNCT
ap-5030	264	15	now	now	ADV
ap-5030	264	16	used	use	VERB
ap-5030	264	17	by	by	ADP
ap-5030	264	18	the	the	DET
ap-5030	264	19	pentagon	pentagon	PROPN
ap-5030	264	20	cyberscoop	cyberscoop	NOUN
ap-5030	264	21	.	.	PUNCT
ap-5030	265	1	https://www.cyberscoop.com/	https://www.cyberscoop.com/	PROPN
ap-5030	265	2	mayhem	mayhem	NOUN
ap-5030	265	3	-	-	PUNCT
ap-5030	265	4	darpa	darpa	PROPN
ap-5030	265	5	-	-	PUNCT
ap-5030	265	6	cyber	cyber	NOUN
ap-5030	265	7	-	-	PUNCT
ap-5030	265	8	grand	grand	ADJ
ap-5030	265	9	-	-	PUNCT
ap-5030	265	10	challenge	challenge	NOUN
ap-5030	265	11	-	-	PUNCT
ap-5030	265	12	dod	dod	NOUN
ap-5030	265	13	-	-	PUNCT
ap-5030	265	14	voltron/.	voltron/.	PROPN
ap-5030	266	1	[	[	X
ap-5030	266	2	22	22	NUM
ap-5030	266	3	]	]	PUNCT
ap-5030	266	4	chandrasekaran	chandrasekaran	ADJ
ap-5030	266	5	,	,	PUNCT
ap-5030	266	6	rajiv	rajiv	PROPN
ap-5030	266	7	.	.	PUNCT
ap-5030	267	1	washingtonpost.com	washingtonpost.com	X
ap-5030	267	2	:	:	PUNCT
ap-5030	267	3	deep	deep	ADJ
ap-5030	267	4	blue	blue	ADJ
ap-5030	267	5	defeats	defeat	VERB
ap-5030	267	6	kasparov	kasparov	PROPN
ap-5030	267	7	in	in	ADP
ap-5030	267	8	game	game	NOUN
ap-5030	267	9	2	2	NUM
ap-5030	267	10	.	.	PUNCT
ap-5030	268	1	https://www.washingtonpost.com/wp-srv/tech/	https://www.washingtonpost.com/wp-srv/tech/	ADJ
ap-5030	268	2	analysis	analysis	NOUN
ap-5030	268	3	/	/	SYM
ap-5030	268	4	kasparov	kasparov	NOUN
ap-5030	268	5	/	/	SYM
ap-5030	268	6	kasparov.htm	kasparov.htm	X
ap-5030	268	7	?	?	PUNCT
ap-5030	269	1	[	[	X
ap-5030	269	2	23	23	NUM
ap-5030	269	3	]	]	X
ap-5030	269	4	n.	n.	PROPN
ap-5030	269	5	jones	jones	PROPN
ap-5030	269	6	.	.	PUNCT
ap-5030	269	7	quiz	quiz	PROPN
ap-5030	269	8	-	-	PUNCT
ap-5030	269	9	playing	play	VERB
ap-5030	269	10	computer	computer	NOUN
ap-5030	269	11	system	system	NOUN
ap-5030	269	12	could	could	AUX
ap-5030	269	13	revolutionize	revolutionize	VERB
ap-5030	269	14	research	research	NOUN
ap-5030	269	15	.	.	PUNCT
ap-5030	270	1	https://www.nature.com/news/	https://www.nature.com/news/	X
ap-5030	271	1	2011/110215	2011/110215	NUM
ap-5030	271	2	/	/	SYM
ap-5030	271	3	full	full	ADJ
ap-5030	271	4	/	/	SYM
ap-5030	271	5	news.2011.95.html	news.2011.95.html	NOUN
ap-5030	271	6	.	.	PUNCT
ap-5030	272	1	[	[	X
ap-5030	272	2	24	24	NUM
ap-5030	272	3	]	]	X
ap-5030	272	4	h.	h.	PROPN
ap-5030	272	5	van	van	PROPN
ap-5030	272	6	seijen	seijen	PROPN
ap-5030	272	7	,	,	PUNCT
ap-5030	272	8	m.	m.	NOUN
ap-5030	272	9	fatemi	fatemi	PROPN
ap-5030	272	10	,	,	PUNCT
ap-5030	272	11	j.	j.	PROPN
ap-5030	272	12	romoff	romoff	PROPN
ap-5030	272	13	,	,	PUNCT
ap-5030	272	14	et	et	PROPN
ap-5030	272	15	al	al	PROPN
ap-5030	272	16	.	.	PROPN
ap-5030	272	17	hybrid	hybrid	ADJ
ap-5030	272	18	reward	reward	NOUN
ap-5030	272	19	architecture	architecture	NOUN
ap-5030	272	20	for	for	ADP
ap-5030	272	21	reinforcement	reinforcement	NOUN
ap-5030	272	22	learning	learning	NOUN
ap-5030	272	23	.	.	PUNCT
ap-5030	273	1	corr	corr	PROPN
ap-5030	273	2	abs/1706.04208	abs/1706.04208	NOUN
ap-5030	273	3	,	,	PUNCT
ap-5030	273	4	2017	2017	NUM
ap-5030	273	5	.	.	PUNCT
ap-5030	274	1	[	[	X
ap-5030	274	2	25	25	NUM
ap-5030	274	3	]	]	X
ap-5030	274	4	v.	v.	PROPN
ap-5030	274	5	mnih	mnih	PROPN
ap-5030	274	6	,	,	PUNCT
ap-5030	274	7	k.	k.	PROPN
ap-5030	274	8	kavukcuoglu	kavukcuoglu	PROPN
ap-5030	274	9	,	,	PUNCT
ap-5030	274	10	d.	d.	PROPN
ap-5030	274	11	silver	silver	PROPN
ap-5030	274	12	,	,	PUNCT
ap-5030	274	13	et	et	PROPN
ap-5030	274	14	al	al	PROPN
ap-5030	274	15	.	.	PUNCT
ap-5030	274	16	playing	play	VERB
ap-5030	274	17	atari	atari	NOUN
ap-5030	274	18	with	with	ADP
ap-5030	274	19	deep	deep	ADJ
ap-5030	274	20	reinforcement	reinforcement	NOUN
ap-5030	274	21	learning	learning	NOUN
ap-5030	274	22	.	.	PUNCT
ap-5030	275	1	[	[	X
ap-5030	275	2	26	26	NUM
ap-5030	275	3	]	]	PUNCT
ap-5030	275	4	alphago	alphago	NOUN
ap-5030	275	5	:	:	PUNCT
ap-5030	275	6	using	use	VERB
ap-5030	275	7	machine	machine	NOUN
ap-5030	275	8	learning	learn	VERB
ap-5030	275	9	to	to	PART
ap-5030	275	10	master	master	VERB
ap-5030	275	11	the	the	DET
ap-5030	275	12	ancient	ancient	ADJ
ap-5030	275	13	game	game	NOUN
ap-5030	275	14	of	of	ADP
ap-5030	275	15	go	go	NOUN
ap-5030	275	16	.	.	PUNCT
ap-5030	276	1	https://blog.google/topics/machine-learning/	https://blog.google/topics/machine-learning/	PROPN
ap-5030	276	2	alphago	alphago	NOUN
ap-5030	276	3	-	-	PUNCT
ap-5030	276	4	machine	machine	NOUN
ap-5030	276	5	-	-	PUNCT
ap-5030	276	6	learning	learning	NOUN
ap-5030	276	7	-	-	PUNCT
ap-5030	276	8	game	game	NOUN
ap-5030	276	9	-	-	PUNCT
ap-5030	276	10	go/.	go/.	NOUN
ap-5030	276	11	[	[	NOUN
ap-5030	276	12	27	27	NUM
ap-5030	276	13	]	]	X
ap-5030	276	14	o.	o.	PROPN
ap-5030	276	15	vinyals	vinyal	NOUN
ap-5030	276	16	,	,	PUNCT
ap-5030	276	17	i.	i.	NOUN
ap-5030	276	18	babuschkin	babuschkin	PROPN
ap-5030	276	19	,	,	PUNCT
ap-5030	276	20	j.	j.	PROPN
ap-5030	276	21	chung	chung	PROPN
ap-5030	276	22	,	,	PUNCT
ap-5030	276	23	et	et	PROPN
ap-5030	276	24	al	al	PROPN
ap-5030	276	25	.	.	PROPN
ap-5030	276	26	alphastar	alphastar	PROPN
ap-5030	276	27	:	:	PUNCT
ap-5030	276	28	mastering	master	VERB
ap-5030	276	29	the	the	DET
ap-5030	276	30	real	real	ADJ
ap-5030	276	31	-	-	PUNCT
ap-5030	276	32	time	time	NOUN
ap-5030	276	33	strategy	strategy	NOUN
ap-5030	276	34	game	game	NOUN
ap-5030	276	35	starcraft	starcraft	PROPN
ap-5030	276	36	ii	ii	PROPN
ap-5030	276	37	.	.	PUNCT
ap-5030	277	1	https://deepmind.com/blog/alphastarmastering-real-time-strategy-game-starcraft-ii/.	https://deepmind.com/blog/alphastarmastering-real-time-strategy-game-starcraft-ii/.	X
ap-5030	278	1	[	[	X
ap-5030	278	2	28	28	NUM
ap-5030	278	3	]	]	PUNCT
ap-5030	278	4	capture	capture	VERB
ap-5030	278	5	the	the	DET
ap-5030	278	6	flag	flag	NOUN
ap-5030	278	7	:	:	PUNCT
ap-5030	278	8	the	the	DET
ap-5030	278	9	emergence	emergence	NOUN
ap-5030	278	10	of	of	ADP
ap-5030	278	11	complex	complex	ADJ
ap-5030	278	12	cooperative	cooperative	ADJ
ap-5030	278	13	agents	agent	NOUN
ap-5030	278	14	|	|	ADV
ap-5030	278	15	deepmind	deepmind	NOUN
ap-5030	278	16	.	.	PUNCT
ap-5030	279	1	https	https	NOUN
ap-5030	279	2	:	:	PUNCT
ap-5030	280	1	//deepmind.com	//deepmind.com	ADJ
ap-5030	280	2	/	/	SYM
ap-5030	280	3	blog	blog	NOUN
ap-5030	280	4	/	/	SYM
ap-5030	280	5	capture	capture	NOUN
ap-5030	280	6	-	-	PUNCT
ap-5030	280	7	the	the	DET
ap-5030	280	8	-	-	PUNCT
ap-5030	280	9	flag	flag	NOUN
ap-5030	280	10	-	-	PUNCT
ap-5030	280	11	science/.	science/.	ADV
ap-5030	280	12	[	[	X
ap-5030	280	13	29	29	NUM
ap-5030	280	14	]	]	PUNCT
ap-5030	280	15	j.	j.	PROPN
ap-5030	280	16	pearl	pearl	PROPN
ap-5030	280	17	.	.	PUNCT
ap-5030	281	1	causal	causal	PROPN
ap-5030	281	2	inference	inference	NOUN
ap-5030	281	3	in	in	ADP
ap-5030	281	4	statistics	statistic	NOUN
ap-5030	281	5	:	:	PUNCT
ap-5030	281	6	an	an	DET
ap-5030	281	7	overview	overview	NOUN
ap-5030	281	8	.	.	PUNCT
ap-5030	282	1	statistics	statistic	NOUN
ap-5030	282	2	surveys	survey	NOUN
ap-5030	282	3	3(0):96–146	3(0):96–146	NUM
ap-5030	282	4	,	,	PUNCT
ap-5030	282	5	2009	2009	NUM
ap-5030	282	6	.	.	PUNCT
ap-5030	283	1	doi:10.1214/09	doi:10.1214/09	NOUN
ap-5030	283	2	-	-	PUNCT
ap-5030	283	3	ss057	ss057	NOUN
ap-5030	283	4	.	.	PUNCT
ap-5030	284	1	[	[	X
ap-5030	284	2	30	30	NUM
ap-5030	284	3	]	]	X
ap-5030	284	4	j.	j.	PROPN
ap-5030	284	5	pearl	pearl	PROPN
ap-5030	284	6	,	,	PUNCT
ap-5030	284	7	d.	d.	PROPN
ap-5030	284	8	mackenzie	mackenzie	PROPN
ap-5030	284	9	.	.	PUNCT
ap-5030	285	1	the	the	DET
ap-5030	285	2	book	book	NOUN
ap-5030	285	3	of	of	ADP
ap-5030	285	4	why	why	SCONJ
ap-5030	285	5	:	:	PUNCT
ap-5030	285	6	the	the	DET
ap-5030	285	7	new	new	ADJ
ap-5030	285	8	science	science	NOUN
ap-5030	285	9	of	of	ADP
ap-5030	285	10	cause	cause	NOUN
ap-5030	285	11	and	and	CCONJ
ap-5030	285	12	effect	effect	NOUN
ap-5030	285	13	.	.	PUNCT
ap-5030	286	1	first	first	PROPN
ap-5030	286	2	edition	edition	PROPN
ap-5030	286	3	edn	edn	PROPN
ap-5030	286	4	.	.	PROPN
ap-5030	286	5	,	,	PUNCT
ap-5030	286	6	2018	2018	NUM
ap-5030	286	7	.	.	PUNCT
ap-5030	287	1	[	[	X
ap-5030	287	2	31	31	NUM
ap-5030	287	3	]	]	X
ap-5030	287	4	c.	c.	PROPN
ap-5030	287	5	davidson	davidson	PROPN
ap-5030	287	6	-	-	PUNCT
ap-5030	287	7	pilon	pilon	PROPN
ap-5030	287	8	.	.	PUNCT
ap-5030	287	9	bayesian	bayesian	NOUN
ap-5030	287	10	methods	method	NOUN
ap-5030	287	11	for	for	ADP
ap-5030	287	12	hackers	hacker	NOUN
ap-5030	287	13	:	:	PUNCT
ap-5030	287	14	probabilistic	probabilistic	ADJ
ap-5030	287	15	programming	programming	NOUN
ap-5030	287	16	and	and	CCONJ
ap-5030	287	17	bayesian	bayesian	NOUN
ap-5030	287	18	methods	method	NOUN
ap-5030	287	19	.	.	PUNCT
ap-5030	288	1	addison	addison	PROPN
ap-5030	288	2	-	-	PUNCT
ap-5030	288	3	wesley	wesley	PROPN
ap-5030	288	4	data	data	PROPN
ap-5030	288	5	and	and	CCONJ
ap-5030	288	6	analytics	analytic	NOUN
ap-5030	288	7	series	series	NOUN
ap-5030	288	8	.	.	PUNCT
ap-5030	289	1	addison	addison	PROPN
ap-5030	289	2	-	-	PUNCT
ap-5030	289	3	wesley	wesley	PROPN
ap-5030	289	4	,	,	PUNCT
ap-5030	289	5	new	new	PROPN
ap-5030	289	6	york	york	PROPN
ap-5030	289	7	,	,	PUNCT
ap-5030	289	8	2016	2016	NUM
ap-5030	289	9	.	.	PUNCT
ap-5030	290	1	[	[	X
ap-5030	290	2	32	32	NUM
ap-5030	290	3	]	]	PUNCT
ap-5030	290	4	o.	o.	PROPN
ap-5030	290	5	martin	martin	PROPN
ap-5030	290	6	.	.	PUNCT
ap-5030	291	1	bayesian	bayesian	NOUN
ap-5030	291	2	analysis	analysis	NOUN
ap-5030	291	3	with	with	ADP
ap-5030	291	4	python	python	NOUN
ap-5030	291	5	:	:	PUNCT
ap-5030	291	6	introduction	introduction	NOUN
ap-5030	291	7	to	to	ADP
ap-5030	291	8	statistical	statistical	ADJ
ap-5030	291	9	modeling	modeling	NOUN
ap-5030	291	10	and	and	CCONJ
ap-5030	291	11	probabilistic	probabilistic	ADJ
ap-5030	291	12	programming	programming	NOUN
ap-5030	291	13	using	use	VERB
ap-5030	291	14	pymc3	pymc3	NOUN
ap-5030	291	15	and	and	CCONJ
ap-5030	291	16	arviz	arviz	PROPN
ap-5030	291	17	,	,	PUNCT
ap-5030	291	18	2nd	2nd	PROPN
ap-5030	291	19	edition	edition	NOUN
ap-5030	291	20	.	.	PUNCT
ap-5030	292	1	packt	packt	PROPN
ap-5030	292	2	publishing	publishing	PROPN
ap-5030	292	3	ltd	ltd	PROPN
ap-5030	292	4	,	,	PUNCT
ap-5030	292	5	birmingham	birmingham	PROPN
ap-5030	292	6	,	,	PUNCT
ap-5030	292	7	2nd	2nd	ADJ
ap-5030	292	8	edn	edn	PROPN
ap-5030	292	9	.	.	PUNCT
ap-5030	292	10	,	,	PUNCT
ap-5030	292	11	2018	2018	NUM
ap-5030	292	12	.	.	PUNCT
ap-5030	293	1	[	[	X
ap-5030	293	2	33	33	NUM
ap-5030	293	3	]	]	PUNCT
ap-5030	293	4	burlap	burlap	NOUN
ap-5030	293	5	.	.	PUNCT
ap-5030	294	1	http://burlap.cs.brown.edu/.	http://burlap.cs.brown.edu/.	PROPN
ap-5030	294	2	[	[	X
ap-5030	294	3	34	34	NUM
ap-5030	294	4	]	]	PUNCT
ap-5030	294	5	r.	r.	PROPN
ap-5030	294	6	s.	s.	PROPN
ap-5030	294	7	sutton	sutton	PROPN
ap-5030	294	8	,	,	PUNCT
ap-5030	294	9	d.	d.	PROPN
ap-5030	294	10	precup	precup	PROPN
ap-5030	294	11	,	,	PUNCT
ap-5030	294	12	s.	s.	PROPN
ap-5030	294	13	singh	singh	PROPN
ap-5030	294	14	.	.	PUNCT
ap-5030	295	1	between	between	ADP
ap-5030	295	2	mdps	mdps	PROPN
ap-5030	295	3	and	and	CCONJ
ap-5030	295	4	semi	semi	ADJ
ap-5030	295	5	-	-	ADJ
ap-5030	295	6	mdps	mdps	ADJ
ap-5030	295	7	:	:	PUNCT
ap-5030	295	8	a	a	DET
ap-5030	295	9	framework	framework	NOUN
ap-5030	295	10	for	for	ADP
ap-5030	295	11	temporal	temporal	ADJ
ap-5030	295	12	abstraction	abstraction	NOUN
ap-5030	295	13	in	in	ADP
ap-5030	295	14	reinforcement	reinforcement	NOUN
ap-5030	295	15	learning	learning	NOUN
ap-5030	295	16	.	.	PUNCT
ap-5030	296	1	artificial	artificial	ADJ
ap-5030	296	2	intelligence	intelligence	NOUN
ap-5030	296	3	112(12):181–211	112(12):181–211	NUM
ap-5030	296	4	,	,	PUNCT
ap-5030	296	5	1999	1999	NUM
ap-5030	296	6	.	.	PUNCT
ap-5030	297	1	doi:10.1016	doi:10.1016	PROPN
ap-5030	297	2	/	/	SYM
ap-5030	297	3	s0004	s0004	PROPN
ap-5030	297	4	-	-	PUNCT
ap-5030	297	5	3702(99)00052	3702(99)00052	NOUN
ap-5030	297	6	-	-	SYM
ap-5030	297	7	1	1	NUM
ap-5030	297	8	.	.	PUNCT
ap-5030	298	1	[	[	X
ap-5030	298	2	35	35	NUM
ap-5030	298	3	]	]	X
ap-5030	298	4	s.	s.	PROPN
ap-5030	298	5	mangard	mangard	PROPN
ap-5030	298	6	,	,	PUNCT
ap-5030	298	7	e.	e.	PROPN
ap-5030	298	8	oswald	oswald	PROPN
ap-5030	298	9	,	,	PUNCT
ap-5030	298	10	t.	t.	PROPN
ap-5030	298	11	popp	popp	PROPN
ap-5030	298	12	.	.	PUNCT
ap-5030	299	1	power	power	NOUN
ap-5030	299	2	analysis	analysis	NOUN
ap-5030	299	3	attacks	attack	NOUN
ap-5030	299	4	:	:	PUNCT
ap-5030	299	5	revealing	reveal	VERB
ap-5030	299	6	the	the	DET
ap-5030	299	7	secrets	secret	NOUN
ap-5030	299	8	of	of	ADP
ap-5030	299	9	smart	smart	ADJ
ap-5030	299	10	cards	card	NOUN
ap-5030	299	11	.	.	PUNCT
ap-5030	300	1	springer	springer	NOUN
ap-5030	300	2	ebook	ebook	NOUN
ap-5030	300	3	collection	collection	NOUN
ap-5030	300	4	,	,	PUNCT
ap-5030	300	5	computer	computer	NOUN
ap-5030	300	6	science	science	NOUN
ap-5030	301	1	[	[	X
ap-5030	301	2	dig	dig	X
ap-5030	301	3	.	.	PUNCT
ap-5030	302	1	serial	serial	NOUN
ap-5030	302	2	]	]	PUNCT
ap-5030	302	3	,	,	PUNCT
ap-5030	302	4	springer-11645	springer-11645	ADJ
ap-5030	303	1	[	[	X
ap-5030	303	2	dig	dig	X
ap-5030	303	3	.	.	PUNCT
ap-5030	304	1	serial	serial	NOUN
ap-5030	304	2	]	]	PUNCT
ap-5030	304	3	.	.	PUNCT
ap-5030	305	1	springer	springer	NOUN
ap-5030	305	2	science+business	science+business	PROPN
ap-5030	305	3	media	medium	NOUN
ap-5030	305	4	,	,	PUNCT
ap-5030	305	5	llc	llc	PROPN
ap-5030	305	6	,	,	PUNCT
ap-5030	305	7	boston	boston	PROPN
ap-5030	305	8	,	,	PUNCT
ap-5030	305	9	ma	ma	PROPN
ap-5030	305	10	,	,	PUNCT
ap-5030	305	11	2007	2007	NUM
ap-5030	305	12	.	.	PUNCT
ap-5030	306	1	doi:10.1007/978	doi:10.1007/978	ADJ
ap-5030	306	2	-	-	PUNCT
ap-5030	306	3	0	0	NUM
ap-5030	306	4	-	-	PUNCT
ap-5030	306	5	387	387	NUM
ap-5030	306	6	-	-	PUNCT
ap-5030	306	7	38162	38162	NUM
ap-5030	306	8	-	-	SYM
ap-5030	306	9	6	6	NUM
ap-5030	306	10	.	.	PUNCT
ap-5030	307	1	[	[	X
ap-5030	307	2	36	36	NUM
ap-5030	307	3	]	]	X
ap-5030	307	4	meltdown	meltdown	NOUN
ap-5030	307	5	and	and	CCONJ
ap-5030	307	6	spectre	spectre	NOUN
ap-5030	307	7	.	.	PUNCT
ap-5030	308	1	https://meltdownattack.com/.	https://meltdownattack.com/.	PROPN
ap-5030	309	1	[	[	X
ap-5030	309	2	37	37	NUM
ap-5030	309	3	]	]	PUNCT
ap-5030	309	4	planning	planning	NOUN
ap-5030	309	5	and	and	CCONJ
ap-5030	309	6	acting	act	VERB
ap-5030	309	7	in	in	ADP
ap-5030	309	8	partially	partially	ADV
ap-5030	309	9	observable	observable	ADJ
ap-5030	309	10	stochastic	stochastic	ADJ
ap-5030	309	11	domains	domain	NOUN
ap-5030	309	12	.	.	PUNCT
ap-5030	310	1	https://www.sciencedirect.com/science/	https://www.sciencedirect.com/science/	X
ap-5030	310	2	article	article	PROPN
ap-5030	310	3	/	/	SYM
ap-5030	310	4	pii	pii	PROPN
ap-5030	310	5	/	/	SYM
ap-5030	310	6	s000437029800023x?via%3dihub	s000437029800023x?via%3dihub	PROPN
ap-5030	310	7	.	.	PUNCT
ap-5030	311	1	[	[	X
ap-5030	311	2	38	38	NUM
ap-5030	311	3	]	]	X
ap-5030	311	4	v.	v.	PROPN
ap-5030	311	5	mnih	mnih	PROPN
ap-5030	311	6	,	,	PUNCT
ap-5030	311	7	k.	k.	PROPN
ap-5030	311	8	kavukcuoglu	kavukcuoglu	PROPN
ap-5030	311	9	,	,	PUNCT
ap-5030	311	10	d.	d.	PROPN
ap-5030	311	11	silver	silver	PROPN
ap-5030	311	12	,	,	PUNCT
ap-5030	311	13	et	et	PROPN
ap-5030	311	14	al	al	PROPN
ap-5030	311	15	.	.	PROPN
ap-5030	311	16	humanlevel	humanlevel	PROPN
ap-5030	311	17	control	control	NOUN
ap-5030	311	18	through	through	ADP
ap-5030	311	19	deep	deep	ADJ
ap-5030	311	20	reinforcement	reinforcement	NOUN
ap-5030	311	21	learning	learning	NOUN
ap-5030	311	22	.	.	PUNCT
ap-5030	312	1	nature	nature	NOUN
ap-5030	312	2	518(7540):529–533	518(7540):529–533	NUM
ap-5030	312	3	,	,	PUNCT
ap-5030	312	4	2015	2015	NUM
ap-5030	312	5	.	.	PUNCT
ap-5030	313	1	doi:10.1038	doi:10.1038	NOUN
ap-5030	313	2	/	/	SYM
ap-5030	313	3	nature14236	nature14236	PROPN
ap-5030	313	4	.	.	PUNCT
ap-5030	314	1	[	[	X
ap-5030	314	2	39	39	NUM
ap-5030	314	3	]	]	PUNCT
ap-5030	314	4	s.	s.	PROPN
ap-5030	314	5	griffith	griffith	PROPN
ap-5030	314	6	,	,	PUNCT
ap-5030	314	7	k.	k.	PROPN
ap-5030	314	8	subramanian	subramanian	PROPN
ap-5030	314	9	,	,	PUNCT
ap-5030	314	10	j.	j.	PROPN
ap-5030	314	11	scholz	scholz	PROPN
ap-5030	314	12	,	,	PUNCT
ap-5030	315	1	et	et	PROPN
ap-5030	315	2	al	al	PROPN
ap-5030	315	3	.	.	PUNCT
ap-5030	315	4	policy	policy	NOUN
ap-5030	315	5	shaping	shaping	NOUN
ap-5030	315	6	:	:	PUNCT
ap-5030	315	7	integrating	integrate	VERB
ap-5030	315	8	human	human	ADJ
ap-5030	315	9	feedback	feedback	NOUN
ap-5030	315	10	with	with	ADP
ap-5030	315	11	reinforcement	reinforcement	NOUN
ap-5030	315	12	learning	learning	NOUN
ap-5030	315	13	.	.	PUNCT
ap-5030	316	1	in	in	ADP
ap-5030	316	2	in	in	ADP
ap-5030	316	3	advances	advance	NOUN
ap-5030	316	4	in	in	ADP
ap-5030	316	5	neural	neural	ADJ
ap-5030	316	6	information	information	NOUN
ap-5030	316	7	processing	processing	NOUN
ap-5030	316	8	systems	system	NOUN
ap-5030	316	9	.	.	PUNCT
ap-5030	317	1	2013	2013	NUM
ap-5030	317	2	.	.	PUNCT
ap-5030	318	1	[	[	X
ap-5030	318	2	40	40	NUM
ap-5030	318	3	]	]	PUNCT
ap-5030	318	4	a.	a.	NOUN
ap-5030	318	5	gonfalonieri	gonfalonieri	PROPN
ap-5030	318	6	.	.	PUNCT
ap-5030	319	1	inverse	inverse	ADJ
ap-5030	319	2	reinforcement	reinforcement	NOUN
ap-5030	319	3	learning	learning	NOUN
ap-5030	319	4	–	–	PUNCT
ap-5030	319	5	towards	towards	ADP
ap-5030	319	6	data	data	NOUN
ap-5030	319	7	science	science	NOUN
ap-5030	319	8	.	.	PUNCT
ap-5030	320	1	https://towardsdatascience.com/	https://towardsdatascience.com/	ADJ
ap-5030	320	2	inverse	inverse	NOUN
ap-5030	320	3	-	-	PUNCT
ap-5030	320	4	reinforcement	reinforcement	NOUN
ap-5030	320	5	-	-	PUNCT
ap-5030	320	6	learning-6453b7cdc90d	learning-6453b7cdc90d	NOUN
ap-5030	320	7	.	.	PUNCT
ap-5030	321	1	[	[	X
ap-5030	321	2	41	41	NUM
ap-5030	321	3	]	]	X
ap-5030	321	4	alphazero	alphazero	NOUN
ap-5030	321	5	:	:	PUNCT
ap-5030	321	6	shedding	shed	VERB
ap-5030	321	7	new	new	ADJ
ap-5030	321	8	light	light	NOUN
ap-5030	321	9	on	on	ADP
ap-5030	321	10	the	the	DET
ap-5030	321	11	grand	grand	ADJ
ap-5030	321	12	games	game	NOUN
ap-5030	321	13	of	of	ADP
ap-5030	321	14	chess	chess	NOUN
ap-5030	321	15	,	,	PUNCT
ap-5030	321	16	shogi	shogi	NOUN
ap-5030	321	17	and	and	CCONJ
ap-5030	321	18	go	go	VERB
ap-5030	321	19	|	|	ADV
ap-5030	321	20	deepmind	deepmind	NOUN
ap-5030	321	21	.	.	PUNCT
ap-5030	322	1	https://deepmind.com/blog/	https://deepmind.com/blog/	VERB
ap-5030	322	2	alphazero	alphazero	NOUN
ap-5030	322	3	-	-	PUNCT
ap-5030	322	4	shedding	shed	VERB
ap-5030	322	5	-	-	PUNCT
ap-5030	322	6	new	new	ADJ
ap-5030	322	7	-	-	PUNCT
ap-5030	322	8	light	light	ADJ
ap-5030	322	9	-	-	PUNCT
ap-5030	322	10	grand	grand	ADJ
ap-5030	322	11	-	-	PUNCT
ap-5030	322	12	gameschess	gameschess	NOUN
ap-5030	322	13	-	-	PUNCT
ap-5030	322	14	shogi	shogi	NOUN
ap-5030	322	15	-	-	PUNCT
ap-5030	322	16	and	and	CCONJ
ap-5030	322	17	-	-	PUNCT
ap-5030	322	18	go/.	go/.	NOUN
ap-5030	322	19	526	526	NUM
ap-5030	323	1	https://www.darpa.mil/program/cyber-grand-challenge	https://www.darpa.mil/program/cyber-grand-challenge	ADP
ap-5030	323	2	https://www.darpa.mil/program/cyber-grand-challenge	https://www.darpa.mil/program/cyber-grand-challenge	ADP
ap-5030	323	3	https://www.cyberscoop.com/mayhem-darpa-cyber-grand-challenge-dod-voltron/	https://www.cyberscoop.com/mayhem-darpa-cyber-grand-challenge-dod-voltron/	NOUN
ap-5030	323	4	https://www.cyberscoop.com/mayhem-darpa-cyber-grand-challenge-dod-voltron/	https://www.cyberscoop.com/mayhem-darpa-cyber-grand-challenge-dod-voltron/	NOUN
ap-5030	323	5	https://www.washingtonpost.com/wp-srv/tech/analysis/kasparov/kasparov.htm	https://www.washingtonpost.com/wp-srv/tech/analysis/kasparov/kasparov.htm	PROPN
ap-5030	323	6	?	?	PUNCT
ap-5030	324	1	https://www.washingtonpost.com/wp-srv/tech/analysis/kasparov/kasparov.htm	https://www.washingtonpost.com/wp-srv/tech/analysis/kasparov/kasparov.htm	PROPN
ap-5030	324	2	?	?	PUNCT
ap-5030	325	1	https://www.nature.com/news/2011/110215/full/news.2011.95.html	https://www.nature.com/news/2011/110215/full/news.2011.95.html	PROPN
ap-5030	325	2	https://www.nature.com/news/2011/110215/full/news.2011.95.html	https://www.nature.com/news/2011/110215/full/news.2011.95.html	PROPN
ap-5030	325	3	https://blog.google/topics/machine-learning/alphago-machine-learning-game-go/	https://blog.google/topics/machine-learning/alphago-machine-learning-game-go/	ADV
ap-5030	325	4	https://blog.google/topics/machine-learning/alphago-machine-learning-game-go/	https://blog.google/topics/machine-learning/alphago-machine-learning-game-go/	ADV
ap-5030	325	5	https://deepmind.com/blog/alphastar-mastering-real-time-strategy-game-starcraft-ii/	https://deepmind.com/blog/alphastar-mastering-real-time-strategy-game-starcraft-ii/	PUNCT
ap-5030	325	6	https://deepmind.com/blog/alphastar-mastering-real-time-strategy-game-starcraft-ii/	https://deepmind.com/blog/alphastar-mastering-real-time-strategy-game-starcraft-ii/	PROPN
ap-5030	326	1	https://deepmind.com/blog/capture-the-flag-science/	https://deepmind.com/blog/capture-the-flag-science/	X
ap-5030	326	2	https://deepmind.com/blog/capture-the-flag-science/	https://deepmind.com/blog/capture-the-flag-science/	X
ap-5030	326	3	http://dx.doi.org/10.1214/09-ss057	http://dx.doi.org/10.1214/09-ss057	VERB
ap-5030	326	4	http://burlap.cs.brown.edu/	http://burlap.cs.brown.edu/	PROPN
ap-5030	326	5	http://dx.doi.org/10.1016/s0004-3702(99)00052-1	http://dx.doi.org/10.1016/s0004-3702(99)00052-1	ADP
ap-5030	326	6	http://dx.doi.org/10.1007/978-0-387-38162-6	http://dx.doi.org/10.1007/978-0-387-38162-6	PROPN
ap-5030	326	7	https://meltdownattack.com/	https://meltdownattack.com/	ADJ
ap-5030	326	8	https://www.sciencedirect.com/science/article/pii/s000437029800023x?via%3dihub	https://www.sciencedirect.com/science/article/pii/s000437029800023x?via%3dihub	NOUN
ap-5030	326	9	https://www.sciencedirect.com/science/article/pii/s000437029800023x?via%3dihub	https://www.sciencedirect.com/science/article/pii/s000437029800023x?via%3dihub	PROPN
ap-5030	326	10	http://dx.doi.org/10.1038/nature14236	http://dx.doi.org/10.1038/nature14236	ADP
ap-5030	326	11	https://towardsdatascience.com/inverse-reinforcement-learning-6453b7cdc90d	https://towardsdatascience.com/inverse-reinforcement-learning-6453b7cdc90d	PROPN
ap-5030	326	12	https://towardsdatascience.com/inverse-reinforcement-learning-6453b7cdc90d	https://towardsdatascience.com/inverse-reinforcement-learning-6453b7cdc90d	PROPN
ap-5030	326	13	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	ADV
ap-5030	326	14	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	ADV
ap-5030	326	15	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	https://deepmind.com/blog/alphazero-shedding-new-light-grand-games-chess-shogi-and-go/	ADP
ap-5030	326	16	acta	acta	PROPN
ap-5030	326	17	polytechnica	polytechnica	PROPN
ap-5030	326	18	59(5):518–526	59(5):518–526	PROPN
ap-5030	326	19	,	,	PUNCT
ap-5030	326	20	2019	2019	NUM
ap-5030	326	21	1	1	NUM
ap-5030	326	22	introduction	introduction	NOUN
ap-5030	326	23	2	2	NUM
ap-5030	326	24	related	relate	VERB
ap-5030	326	25	work	work	NOUN
ap-5030	326	26	3	3	NUM
ap-5030	326	27	the	the	DET
ap-5030	326	28	descriptive	descriptive	ADJ
ap-5030	326	29	model	model	NOUN
ap-5030	326	30	of	of	ADP
ap-5030	326	31	the	the	DET
ap-5030	326	32	fpga	fpga	PROPN
ap-5030	326	33	design	design	NOUN
ap-5030	326	34	4	4	NUM
ap-5030	326	35	the	the	DET
ap-5030	326	36	mdp	mdp	PROPN
ap-5030	326	37	-	-	PUNCT
ap-5030	326	38	representation	representation	NOUN
ap-5030	326	39	of	of	ADP
ap-5030	326	40	the	the	DET
ap-5030	326	41	fpga	fpga	PROPN
ap-5030	326	42	design	design	VERB
ap-5030	326	43	4.1	4.1	NUM
ap-5030	326	44	scope	scope	NOUN
ap-5030	326	45	and	and	CCONJ
ap-5030	326	46	constraints	constraint	NOUN
ap-5030	326	47	of	of	ADP
ap-5030	326	48	the	the	DET
ap-5030	326	49	model	model	NOUN
ap-5030	326	50	4.2	4.2	NUM
ap-5030	326	51	generic	generic	ADJ
ap-5030	326	52	components	component	NOUN
ap-5030	326	53	of	of	ADP
ap-5030	326	54	markov	markov	NOUN
ap-5030	326	55	decision	decision	NOUN
ap-5030	326	56	processes	process	VERB
ap-5030	326	57	4.3	4.3	NUM
ap-5030	326	58	the	the	DET
ap-5030	326	59	domain	domain	NOUN
ap-5030	326	60	-	-	PUNCT
ap-5030	326	61	specific	specific	ADJ
ap-5030	326	62	markov	markov	NOUN
ap-5030	326	63	decision	decision	NOUN
ap-5030	326	64	process	process	NOUN
ap-5030	326	65	4.3.1	4.3.1	NUM
ap-5030	326	66	state	state	NOUN
ap-5030	326	67	4.3.2	4.3.2	NUM
ap-5030	326	68	actions	action	NOUN
ap-5030	326	69	4.3.3	4.3.3	PRON
ap-5030	326	70	reward	reward	VERB
ap-5030	326	71	4.4	4.4	NUM
ap-5030	326	72	reinforcement	reinforcement	NOUN
ap-5030	326	73	learning	learning	NOUN
ap-5030	326	74	scenarios	scenario	NOUN
ap-5030	326	75	4.5	4.5	NUM
ap-5030	326	76	implementation	implementation	NOUN
ap-5030	326	77	5	5	NUM
ap-5030	326	78	example	example	NOUN
ap-5030	326	79	5.1	5.1	NUM
ap-5030	326	80	scenario	scenario	NOUN
ap-5030	326	81	1	1	NUM
ap-5030	326	82	:	:	PUNCT
ap-5030	326	83	baseline	baseline	VERB
ap-5030	326	84	5.2	5.2	NUM
ap-5030	326	85	scenario	scenario	NOUN
ap-5030	326	86	2	2	NUM
ap-5030	326	87	:	:	PUNCT
ap-5030	326	88	tamper	tamper	VERB
ap-5030	326	89	resistance	resistance	NOUN
ap-5030	326	90	storage	storage	NOUN
ap-5030	326	91	prevents	prevent	VERB
ap-5030	326	92	most	most	ADJ
ap-5030	326	93	attacks	attack	NOUN
ap-5030	326	94	5.3	5.3	NUM
ap-5030	326	95	scenario	scenario	NOUN
ap-5030	326	96	3	3	NUM
ap-5030	326	97	:	:	PUNCT
ap-5030	326	98	faulty	faulty	ADJ
ap-5030	326	99	encryption	encryption	NOUN
ap-5030	326	100	eases	ease	VERB
ap-5030	326	101	attacks	attack	NOUN
ap-5030	326	102	5.4	5.4	NUM
ap-5030	326	103	scenario	scenario	NOUN
ap-5030	326	104	4	4	NUM
ap-5030	326	105	:	:	PUNCT
ap-5030	326	106	physical	physical	ADJ
ap-5030	326	107	attacks	attack	NOUN
ap-5030	326	108	against	against	ADP
ap-5030	326	109	storage	storage	NOUN
ap-5030	326	110	devices	device	NOUN
ap-5030	326	111	are	be	AUX
ap-5030	326	112	expensive	expensive	ADJ
ap-5030	326	113	5.5	5.5	NUM
ap-5030	326	114	results	result	NOUN
ap-5030	326	115	6	6	NUM
ap-5030	326	116	limits	limit	NOUN
ap-5030	326	117	and	and	CCONJ
ap-5030	326	118	restrictions	restriction	NOUN
ap-5030	326	119	of	of	ADP
ap-5030	326	120	the	the	DET
ap-5030	326	121	mdp	mdp	NOUN
ap-5030	326	122	based	base	VERB
ap-5030	326	123	approach	approach	NOUN
ap-5030	326	124	and	and	CCONJ
ap-5030	326	125	possible	possible	ADJ
ap-5030	326	126	solutions	solution	NOUN
ap-5030	326	127	7	7	NUM
ap-5030	326	128	conclusion	conclusion	NOUN
ap-5030	326	129	references	reference	NOUN