id	sid	tid	token	lemma	pos
ajst-20197	1	1	academic	academic	ADJ
ajst-20197	1	2	journal	journal	NOUN
ajst-20197	1	3	of	of	ADP
ajst-20197	1	4	science	science	NOUN
ajst-20197	1	5	and	and	CCONJ
ajst-20197	1	6	technology	technology	NOUN
ajst-20197	1	7	issn	issn	NOUN
ajst-20197	1	8	:	:	PUNCT
ajst-20197	1	9	2771	2771	NUM
ajst-20197	1	10	-	-	SYM
ajst-20197	1	11	3032	3032	NUM
ajst-20197	1	12	|	|	NOUN
ajst-20197	1	13	vol	vol	NOUN
ajst-20197	1	14	.	.	PROPN
ajst-20197	2	1	10	10	NUM
ajst-20197	2	2	,	,	PUNCT
ajst-20197	2	3	no	no	INTJ
ajst-20197	2	4	.	.	NOUN
ajst-20197	2	5	2	2	NUM
ajst-20197	2	6	,	,	PUNCT
ajst-20197	2	7	2024	2024	NUM
ajst-20197	2	8	109	109	NUM
ajst-20197	2	9	memory	memory	NOUN
ajst-20197	2	10	and	and	CCONJ
ajst-20197	2	11	attention	attention	NOUN
ajst-20197	2	12	in	in	ADP
ajst-20197	2	13	deep	deep	ADJ
ajst-20197	2	14	learning	learn	VERB
ajst-20197	2	15	yi	yi	PROPN
ajst-20197	2	16	zhang1	zhang1	PROPN
ajst-20197	2	17	,	,	PUNCT
ajst-20197	2	18	a	a	DET
ajst-20197	2	19	,	,	PUNCT
ajst-20197	2	20	ziying	ziying	NOUN
ajst-20197	2	21	fan2	fan2	PROPN
ajst-20197	2	22	,	,	PUNCT
ajst-20197	2	23	b	b	PROPN
ajst-20197	2	24	1intellifusion	1intellifusion	PROPN
ajst-20197	2	25	pty	pty	PROPN
ajst-20197	2	26	ltd	ltd	PROPN
ajst-20197	2	27	,	,	PUNCT
ajst-20197	2	28	melbourne	melbourne	PROPN
ajst-20197	2	29	,	,	PUNCT
ajst-20197	2	30	aussie	aussie	PROPN
ajst-20197	2	31	ayizhang@xs.ustb.edu.cn	ayizhang@xs.ustb.edu.cn	PROPN
ajst-20197	2	32	,	,	PUNCT
ajst-20197	2	33	bziyingfan1@gmail.com	bziyingfan1@gmail.com	X
ajst-20197	2	34	abstract	abstract	NOUN
ajst-20197	2	35	:	:	PUNCT
ajst-20197	2	36	this	this	DET
ajst-20197	2	37	paper	paper	NOUN
ajst-20197	2	38	has	have	AUX
ajst-20197	2	39	highlighted	highlight	VERB
ajst-20197	2	40	the	the	DET
ajst-20197	2	41	advancements	advancement	NOUN
ajst-20197	2	42	made	make	VERB
ajst-20197	2	43	within	within	ADP
ajst-20197	2	44	the	the	DET
ajst-20197	2	45	deep	deep	ADJ
ajst-20197	2	46	learning	learning	NOUN
ajst-20197	2	47	approached	approach	VERB
ajst-20197	2	48	through	through	ADP
ajst-20197	2	49	the	the	DET
ajst-20197	2	50	evolution	evolution	NOUN
ajst-20197	2	51	of	of	ADP
ajst-20197	2	52	attention	attention	NOUN
ajst-20197	2	53	and	and	CCONJ
ajst-20197	2	54	memory	memory	NOUN
ajst-20197	2	55	algorithms	algorithm	NOUN
ajst-20197	2	56	.	.	PUNCT
ajst-20197	3	1	there	there	PRON
ajst-20197	3	2	has	have	AUX
ajst-20197	3	3	been	be	AUX
ajst-20197	3	4	a	a	DET
ajst-20197	3	5	gradual	gradual	ADJ
ajst-20197	3	6	replacement	replacement	NOUN
ajst-20197	3	7	of	of	ADP
ajst-20197	3	8	the	the	DET
ajst-20197	3	9	traditional	traditional	ADJ
ajst-20197	3	10	approaches	approach	NOUN
ajst-20197	3	11	in	in	ADP
ajst-20197	3	12	deep	deep	ADJ
ajst-20197	3	13	learning	learning	NOUN
ajst-20197	3	14	through	through	ADP
ajst-20197	3	15	the	the	DET
ajst-20197	3	16	inclusion	inclusion	NOUN
ajst-20197	3	17	of	of	ADP
ajst-20197	3	18	these	these	DET
ajst-20197	3	19	algorithms	algorithm	NOUN
ajst-20197	3	20	that	that	PRON
ajst-20197	3	21	helps	help	VERB
ajst-20197	3	22	in	in	ADP
ajst-20197	3	23	capturing	capture	VERB
ajst-20197	3	24	the	the	DET
ajst-20197	3	25	data	datum	NOUN
ajst-20197	3	26	based	base	VERB
ajst-20197	3	27	on	on	ADP
ajst-20197	3	28	time	time	NOUN
ajst-20197	3	29	and	and	CCONJ
ajst-20197	3	30	sequence	sequence	NOUN
ajst-20197	3	31	.	.	PUNCT
ajst-20197	4	1	the	the	DET
ajst-20197	4	2	model	model	NOUN
ajst-20197	4	3	performance	performance	NOUN
ajst-20197	4	4	is	be	AUX
ajst-20197	4	5	greatly	greatly	ADV
ajst-20197	4	6	enhanced	enhance	VERB
ajst-20197	4	7	through	through	ADP
ajst-20197	4	8	the	the	DET
ajst-20197	4	9	usage	usage	NOUN
ajst-20197	4	10	of	of	ADP
ajst-20197	4	11	rnns	rnn	NOUN
ajst-20197	4	12	that	that	PRON
ajst-20197	4	13	uses	use	VERB
ajst-20197	4	14	these	these	DET
ajst-20197	4	15	algorithms	algorithm	NOUN
ajst-20197	4	16	for	for	ADP
ajst-20197	4	17	developing	develop	VERB
ajst-20197	4	18	sequential	sequential	ADJ
ajst-20197	4	19	modelling	modelling	NOUN
ajst-20197	4	20	of	of	ADP
ajst-20197	4	21	data	datum	NOUN
ajst-20197	4	22	.	.	PUNCT
ajst-20197	5	1	this	this	DET
ajst-20197	5	2	paper	paper	NOUN
ajst-20197	5	3	has	have	AUX
ajst-20197	5	4	performed	perform	VERB
ajst-20197	5	5	a	a	DET
ajst-20197	5	6	peer	peer	NOUN
ajst-20197	5	7	review	review	NOUN
ajst-20197	5	8	to	to	PART
ajst-20197	5	9	understand	understand	VERB
ajst-20197	5	10	the	the	DET
ajst-20197	5	11	different	different	ADJ
ajst-20197	5	12	mechanisms	mechanism	NOUN
ajst-20197	5	13	that	that	PRON
ajst-20197	5	14	include	include	VERB
ajst-20197	5	15	grus	grus	NOUN
ajst-20197	5	16	,	,	PUNCT
ajst-20197	5	17	mann	mann	PROPN
ajst-20197	5	18	,	,	PUNCT
ajst-20197	5	19	lstm	lstm	NOUN
ajst-20197	5	20	and	and	CCONJ
ajst-20197	5	21	self	self	NOUN
ajst-20197	5	22	-	-	PUNCT
ajst-20197	5	23	attention	attention	NOUN
ajst-20197	5	24	mechanisms	mechanism	NOUN
ajst-20197	5	25	.	.	PUNCT
ajst-20197	6	1	memory	memory	NOUN
ajst-20197	6	2	mechanism	mechanism	NOUN
ajst-20197	6	3	assists	assist	VERB
ajst-20197	6	4	in	in	ADP
ajst-20197	6	5	capturing	capture	VERB
ajst-20197	6	6	the	the	DET
ajst-20197	6	7	past	past	ADJ
ajst-20197	6	8	sequence	sequence	NOUN
ajst-20197	6	9	of	of	ADP
ajst-20197	6	10	datasets	dataset	NOUN
ajst-20197	6	11	for	for	ADP
ajst-20197	6	12	analysing	analyse	VERB
ajst-20197	6	13	the	the	DET
ajst-20197	6	14	hidden	hidden	ADJ
ajst-20197	6	15	state	state	NOUN
ajst-20197	6	16	within	within	ADP
ajst-20197	6	17	the	the	DET
ajst-20197	6	18	datasets	dataset	NOUN
ajst-20197	6	19	.	.	PUNCT
ajst-20197	7	1	attention	attention	NOUN
ajst-20197	7	2	mechanisms	mechanism	NOUN
ajst-20197	7	3	help	help	VERB
ajst-20197	7	4	in	in	ADP
ajst-20197	7	5	capturing	capture	VERB
ajst-20197	7	6	a	a	DET
ajst-20197	7	7	particular	particular	ADJ
ajst-20197	7	8	location	location	NOUN
ajst-20197	7	9	within	within	ADP
ajst-20197	7	10	a	a	DET
ajst-20197	7	11	video	video	NOUN
ajst-20197	7	12	dataset	dataset	NOUN
ajst-20197	7	13	to	to	PART
ajst-20197	7	14	understand	understand	VERB
ajst-20197	7	15	the	the	DET
ajst-20197	7	16	patterns	pattern	NOUN
ajst-20197	7	17	of	of	ADP
ajst-20197	7	18	a	a	DET
ajst-20197	7	19	data	datum	NOUN
ajst-20197	7	20	.	.	PUNCT
ajst-20197	8	1	there	there	PRON
ajst-20197	8	2	is	be	VERB
ajst-20197	8	3	an	an	DET
ajst-20197	8	4	accurate	accurate	ADJ
ajst-20197	8	5	recognition	recognition	NOUN
ajst-20197	8	6	of	of	ADP
ajst-20197	8	7	human	human	ADJ
ajst-20197	8	8	actions	action	NOUN
ajst-20197	8	9	in	in	ADP
ajst-20197	8	10	the	the	DET
ajst-20197	8	11	datasets	dataset	NOUN
ajst-20197	8	12	through	through	ADP
ajst-20197	8	13	the	the	DET
ajst-20197	8	14	implementation	implementation	NOUN
ajst-20197	8	15	of	of	ADP
ajst-20197	8	16	attention	attention	NOUN
ajst-20197	8	17	mechanisms	mechanism	NOUN
ajst-20197	8	18	.	.	PUNCT
ajst-20197	9	1	this	this	PRON
ajst-20197	9	2	helps	help	VERB
ajst-20197	9	3	in	in	ADP
ajst-20197	9	4	increasing	increase	VERB
ajst-20197	9	5	the	the	DET
ajst-20197	9	6	model	model	NOUN
ajst-20197	9	7	performance	performance	NOUN
ajst-20197	9	8	by	by	ADP
ajst-20197	9	9	enhancing	enhance	VERB
ajst-20197	9	10	the	the	DET
ajst-20197	9	11	prediction	prediction	NOUN
ajst-20197	9	12	accuracy	accuracy	NOUN
ajst-20197	9	13	and	and	CCONJ
ajst-20197	9	14	visibility	visibility	NOUN
ajst-20197	9	15	within	within	ADP
ajst-20197	9	16	a	a	DET
ajst-20197	9	17	particular	particular	ADJ
ajst-20197	9	18	dataset	dataset	NOUN
ajst-20197	9	19	.	.	PUNCT
ajst-20197	10	1	keywords	keyword	NOUN
ajst-20197	10	2	:	:	PUNCT
ajst-20197	10	3	lstm	lstm	ADJ
ajst-20197	10	4	,	,	PUNCT
ajst-20197	10	5	deep	deep	ADJ
ajst-20197	10	6	learning	learning	NOUN
ajst-20197	10	7	,	,	PUNCT
ajst-20197	10	8	memory	memory	NOUN
ajst-20197	10	9	mechanisms	mechanism	NOUN
ajst-20197	10	10	.	.	PUNCT
ajst-20197	11	1	1	1	X
ajst-20197	11	2	.	.	X
ajst-20197	11	3	introduction	introduction	NOUN
ajst-20197	11	4	the	the	DET
ajst-20197	11	5	two	two	NUM
ajst-20197	11	6	fundamental	fundamental	ADJ
ajst-20197	11	7	components	component	NOUN
ajst-20197	11	8	of	of	ADP
ajst-20197	11	9	deep	deep	ADJ
ajst-20197	11	10	learning	learning	NOUN
ajst-20197	11	11	are	be	AUX
ajst-20197	11	12	considered	consider	VERB
ajst-20197	11	13	as	as	ADP
ajst-20197	11	14	attention	attention	NOUN
ajst-20197	11	15	and	and	CCONJ
ajst-20197	11	16	memory	memory	NOUN
ajst-20197	11	17	mechanisms	mechanism	NOUN
ajst-20197	11	18	.	.	PUNCT
ajst-20197	12	1	data	datum	NOUN
ajst-20197	12	2	complexity	complexity	NOUN
ajst-20197	12	3	is	be	AUX
ajst-20197	12	4	clearly	clearly	ADV
ajst-20197	12	5	diagnosed	diagnose	VERB
ajst-20197	12	6	through	through	ADP
ajst-20197	12	7	the	the	DET
ajst-20197	12	8	implementation	implementation	NOUN
ajst-20197	12	9	of	of	ADP
ajst-20197	12	10	these	these	DET
ajst-20197	12	11	mechanisms	mechanism	NOUN
ajst-20197	12	12	and	and	CCONJ
ajst-20197	12	13	helps	help	VERB
ajst-20197	12	14	in	in	ADP
ajst-20197	12	15	increasing	increase	VERB
ajst-20197	12	16	prediction	prediction	NOUN
ajst-20197	12	17	accuracy	accuracy	NOUN
ajst-20197	12	18	on	on	ADP
ajst-20197	12	19	a	a	DET
ajst-20197	12	20	large	large	ADJ
ajst-20197	12	21	scale	scale	NOUN
ajst-20197	12	22	.	.	PUNCT
ajst-20197	13	1	the	the	DET
ajst-20197	13	2	performance	performance	NOUN
ajst-20197	13	3	of	of	ADP
ajst-20197	13	4	“	"	PUNCT
ajst-20197	13	5	deep	deep	ADJ
ajst-20197	13	6	learning	learning	NOUN
ajst-20197	13	7	models	model	NOUN
ajst-20197	13	8	”	"	PUNCT
ajst-20197	13	9	are	be	AUX
ajst-20197	13	10	greatly	greatly	ADV
ajst-20197	13	11	enhanced	enhance	VERB
ajst-20197	13	12	through	through	ADP
ajst-20197	13	13	the	the	DET
ajst-20197	13	14	implementation	implementation	NOUN
ajst-20197	13	15	of	of	ADP
ajst-20197	13	16	attention	attention	NOUN
ajst-20197	13	17	and	and	CCONJ
ajst-20197	13	18	memory	memory	NOUN
ajst-20197	13	19	mechanisms	mechanism	NOUN
ajst-20197	13	20	.	.	PUNCT
ajst-20197	14	1	cognition	cognition	NOUN
ajst-20197	14	2	guidance	guidance	NOUN
ajst-20197	14	3	is	be	AUX
ajst-20197	14	4	considered	consider	VERB
ajst-20197	14	5	to	to	PART
ajst-20197	14	6	be	be	AUX
ajst-20197	14	7	a	a	DET
ajst-20197	14	8	necessary	necessary	ADJ
ajst-20197	14	9	aspect	aspect	NOUN
ajst-20197	14	10	in	in	ADP
ajst-20197	14	11	the	the	DET
ajst-20197	14	12	development	development	NOUN
ajst-20197	14	13	of	of	ADP
ajst-20197	14	14	focus	focus	NOUN
ajst-20197	14	15	over	over	ADP
ajst-20197	14	16	a	a	DET
ajst-20197	14	17	particular	particular	ADJ
ajst-20197	14	18	aspect	aspect	NOUN
ajst-20197	14	19	by	by	ADP
ajst-20197	14	20	ignoring	ignore	VERB
ajst-20197	14	21	others	other	NOUN
ajst-20197	14	22	.	.	PUNCT
ajst-20197	15	1	this	this	DET
ajst-20197	15	2	aspect	aspect	NOUN
ajst-20197	15	3	is	be	AUX
ajst-20197	15	4	provided	provide	VERB
ajst-20197	15	5	through	through	ADP
ajst-20197	15	6	these	these	DET
ajst-20197	15	7	mechanisms	mechanism	NOUN
ajst-20197	15	8	in	in	ADP
ajst-20197	15	9	deep	deep	ADJ
ajst-20197	15	10	learning	learning	NOUN
ajst-20197	15	11	that	that	PRON
ajst-20197	15	12	helps	help	VERB
ajst-20197	15	13	in	in	ADP
ajst-20197	15	14	picking	pick	VERB
ajst-20197	15	15	salient	salient	ADJ
ajst-20197	15	16	information	information	NOUN
ajst-20197	15	17	that	that	PRON
ajst-20197	15	18	are	be	AUX
ajst-20197	15	19	available	available	ADJ
ajst-20197	15	20	within	within	ADP
ajst-20197	15	21	noisy	noisy	ADJ
ajst-20197	15	22	data	datum	NOUN
ajst-20197	15	23	.	.	PUNCT
ajst-20197	16	1	this	this	PRON
ajst-20197	16	2	is	be	AUX
ajst-20197	16	3	effective	effective	ADJ
ajst-20197	16	4	in	in	ADP
ajst-20197	16	5	memorizing	memorize	VERB
ajst-20197	16	6	one	one	NUM
ajst-20197	16	7	particular	particular	ADJ
ajst-20197	16	8	event	event	NOUN
ajst-20197	16	9	at	at	ADP
ajst-20197	16	10	a	a	DET
ajst-20197	16	11	time	time	NOUN
ajst-20197	16	12	and	and	CCONJ
ajst-20197	16	13	increases	increase	VERB
ajst-20197	16	14	the	the	DET
ajst-20197	16	15	model	model	NOUN
ajst-20197	16	16	performance	performance	NOUN
ajst-20197	16	17	by	by	ADP
ajst-20197	16	18	analyzing	analyze	VERB
ajst-20197	16	19	the	the	DET
ajst-20197	16	20	dataset	dataset	NOUN
ajst-20197	16	21	features	feature	NOUN
ajst-20197	16	22	.	.	PUNCT
ajst-20197	17	1	this	this	DET
ajst-20197	17	2	paper	paper	NOUN
ajst-20197	17	3	has	have	AUX
ajst-20197	17	4	identified	identify	VERB
ajst-20197	17	5	an	an	DET
ajst-20197	17	6	issue	issue	NOUN
ajst-20197	17	7	within	within	ADP
ajst-20197	17	8	the	the	DET
ajst-20197	17	9	“	"	PUNCT
ajst-20197	17	10	traditional	traditional	ADJ
ajst-20197	17	11	deep	deep	ADJ
ajst-20197	17	12	learning	learning	NOUN
ajst-20197	17	13	approaches	approach	NOUN
ajst-20197	17	14	”	"	PUNCT
ajst-20197	17	15	in	in	ADP
ajst-20197	17	16	processing	processing	NOUN
ajst-20197	17	17	and	and	CCONJ
ajst-20197	17	18	analyzing	analyze	VERB
ajst-20197	17	19	massive	massive	ADJ
ajst-20197	17	20	three	three	NUM
ajst-20197	17	21	-	-	PUNCT
ajst-20197	17	22	dimensional	dimensional	ADJ
ajst-20197	17	23	volumes	volume	NOUN
ajst-20197	17	24	by	by	ADP
ajst-20197	17	25	achieving	achieve	VERB
ajst-20197	17	26	fine	fine	ADV
ajst-20197	17	27	-	-	PUNCT
ajst-20197	17	28	grained	grain	VERB
ajst-20197	17	29	details	detail	NOUN
ajst-20197	17	30	and	and	CCONJ
ajst-20197	17	31	spatial	spatial	ADJ
ajst-20197	17	32	context	context	NOUN
ajst-20197	17	33	.	.	PUNCT
ajst-20197	18	1	this	this	PRON
ajst-20197	18	2	indicates	indicate	VERB
ajst-20197	18	3	the	the	DET
ajst-20197	18	4	generation	generation	NOUN
ajst-20197	18	5	of	of	ADP
ajst-20197	18	6	false	false	ADJ
ajst-20197	18	7	positives	positive	NOUN
ajst-20197	18	8	and	and	CCONJ
ajst-20197	18	9	missed	miss	VERB
ajst-20197	18	10	detections	detection	NOUN
ajst-20197	18	11	through	through	ADP
ajst-20197	18	12	the	the	DET
ajst-20197	18	13	formation	formation	NOUN
ajst-20197	18	14	of	of	ADP
ajst-20197	18	15	“	"	PUNCT
ajst-20197	18	16	suboptimal	suboptimal	ADJ
ajst-20197	18	17	model	model	NOUN
ajst-20197	18	18	performance	performance	NOUN
ajst-20197	18	19	”	"	PUNCT
ajst-20197	18	20	leading	lead	VERB
ajst-20197	18	21	to	to	ADP
ajst-20197	18	22	the	the	DET
ajst-20197	18	23	decline	decline	NOUN
ajst-20197	18	24	in	in	ADP
ajst-20197	18	25	prediction	prediction	NOUN
ajst-20197	18	26	accuracy	accuracy	NOUN
ajst-20197	18	27	and	and	CCONJ
ajst-20197	18	28	reliability	reliability	NOUN
ajst-20197	18	29	.	.	PUNCT
ajst-20197	19	1	based	base	VERB
ajst-20197	19	2	on	on	ADP
ajst-20197	19	3	this	this	DET
ajst-20197	19	4	issue	issue	NOUN
ajst-20197	19	5	,	,	PUNCT
ajst-20197	19	6	this	this	DET
ajst-20197	19	7	paper	paper	NOUN
ajst-20197	19	8	has	have	AUX
ajst-20197	19	9	conducted	conduct	VERB
ajst-20197	19	10	a	a	DET
ajst-20197	19	11	research	research	NOUN
ajst-20197	19	12	on	on	ADP
ajst-20197	19	13	attention	attention	NOUN
ajst-20197	19	14	and	and	CCONJ
ajst-20197	19	15	memory	memory	NOUN
ajst-20197	19	16	mechanisms	mechanism	NOUN
ajst-20197	19	17	to	to	PART
ajst-20197	19	18	understand	understand	VERB
ajst-20197	19	19	their	their	PRON
ajst-20197	19	20	contributions	contribution	NOUN
ajst-20197	19	21	made	make	VERB
ajst-20197	19	22	in	in	ADP
ajst-20197	19	23	enhancing	enhance	VERB
ajst-20197	19	24	the	the	DET
ajst-20197	19	25	suboptimal	suboptimal	ADJ
ajst-20197	19	26	performance	performance	NOUN
ajst-20197	19	27	of	of	ADP
ajst-20197	19	28	deep	deep	ADJ
ajst-20197	19	29	learning	learning	NOUN
ajst-20197	19	30	techniques	technique	NOUN
ajst-20197	19	31	.	.	PUNCT
ajst-20197	20	1	this	this	DET
ajst-20197	20	2	paper	paper	NOUN
ajst-20197	20	3	has	have	AUX
ajst-20197	20	4	made	make	VERB
ajst-20197	20	5	a	a	DET
ajst-20197	20	6	detailed	detailed	ADJ
ajst-20197	20	7	investigation	investigation	NOUN
ajst-20197	20	8	about	about	ADP
ajst-20197	20	9	different	different	ADJ
ajst-20197	20	10	attention	attention	NOUN
ajst-20197	20	11	and	and	CCONJ
ajst-20197	20	12	memory	memory	NOUN
ajst-20197	20	13	mechanisms	mechanism	NOUN
ajst-20197	20	14	used	use	VERB
ajst-20197	20	15	in	in	ADP
ajst-20197	20	16	increasing	increase	VERB
ajst-20197	20	17	the	the	DET
ajst-20197	20	18	quality	quality	NOUN
ajst-20197	20	19	and	and	CCONJ
ajst-20197	20	20	prediction	prediction	NOUN
ajst-20197	20	21	accuracy	accuracy	NOUN
ajst-20197	20	22	.	.	PUNCT
ajst-20197	21	1	further	far	ADV
ajst-20197	21	2	,	,	PUNCT
ajst-20197	21	3	this	this	DET
ajst-20197	21	4	paper	paper	NOUN
ajst-20197	21	5	has	have	AUX
ajst-20197	21	6	elaborated	elaborate	VERB
ajst-20197	21	7	the	the	DET
ajst-20197	21	8	methods	method	NOUN
ajst-20197	21	9	that	that	PRON
ajst-20197	21	10	are	be	AUX
ajst-20197	21	11	used	use	VERB
ajst-20197	21	12	for	for	ADP
ajst-20197	21	13	conducting	conduct	VERB
ajst-20197	21	14	this	this	DET
ajst-20197	21	15	research	research	NOUN
ajst-20197	21	16	and	and	CCONJ
ajst-20197	21	17	has	have	AUX
ajst-20197	21	18	provided	provide	VERB
ajst-20197	21	19	a	a	DET
ajst-20197	21	20	peer	peer	NOUN
ajst-20197	21	21	review	review	NOUN
ajst-20197	21	22	from	from	ADP
ajst-20197	21	23	different	different	ADJ
ajst-20197	21	24	journals	journal	NOUN
ajst-20197	21	25	to	to	PART
ajst-20197	21	26	make	make	VERB
ajst-20197	21	27	a	a	DET
ajst-20197	21	28	proper	proper	ADJ
ajst-20197	21	29	investigation	investigation	NOUN
ajst-20197	21	30	about	about	ADP
ajst-20197	21	31	attention	attention	NOUN
ajst-20197	21	32	and	and	CCONJ
ajst-20197	21	33	memory	memory	NOUN
ajst-20197	21	34	mechanisms	mechanism	NOUN
ajst-20197	21	35	.	.	PUNCT
ajst-20197	22	1	2	2	X
ajst-20197	22	2	.	.	X
ajst-20197	22	3	background	background	NOUN
ajst-20197	22	4	neural	neural	ADJ
ajst-20197	22	5	networks	network	NOUN
ajst-20197	22	6	help	help	VERB
ajst-20197	22	7	in	in	ADP
ajst-20197	22	8	the	the	DET
ajst-20197	22	9	representation	representation	NOUN
ajst-20197	22	10	of	of	ADP
ajst-20197	22	11	deep	deep	ADJ
ajst-20197	22	12	learning	learning	NOUN
ajst-20197	22	13	through	through	ADP
ajst-20197	22	14	the	the	DET
ajst-20197	22	15	usage	usage	NOUN
ajst-20197	22	16	of	of	ADP
ajst-20197	22	17	artificial	artificial	ADJ
ajst-20197	22	18	intelligence	intelligence	NOUN
ajst-20197	22	19	[	[	X
ajst-20197	22	20	23	23	NUM
ajst-20197	22	21	]	]	X
ajst-20197	22	22	(	(	PUNCT
ajst-20197	22	23	smys	smys	PROPN
ajst-20197	22	24	,	,	PUNCT
ajst-20197	22	25	chen	chen	PROPN
ajst-20197	22	26	and	and	CCONJ
ajst-20197	22	27	shakya	shakya	PROPN
ajst-20197	22	28	2020	2020	NUM
ajst-20197	22	29	)	)	PUNCT
ajst-20197	22	30	.	.	PUNCT
ajst-20197	23	1	these	these	DET
ajst-20197	23	2	networks	network	NOUN
ajst-20197	23	3	possess	possess	VERB
ajst-20197	23	4	interconnected	interconnected	ADJ
ajst-20197	23	5	nodes	node	NOUN
ajst-20197	23	6	referred	refer	VERB
ajst-20197	23	7	to	to	ADP
ajst-20197	23	8	as	as	ADP
ajst-20197	23	9	neurons	neuron	NOUN
ajst-20197	23	10	that	that	PRON
ajst-20197	23	11	help	help	VERB
ajst-20197	23	12	in	in	ADP
ajst-20197	23	13	transmitting	transmit	VERB
ajst-20197	23	14	information	information	NOUN
ajst-20197	23	15	in	in	ADP
ajst-20197	23	16	a	a	DET
ajst-20197	23	17	particular	particular	ADJ
ajst-20197	23	18	direction	direction	NOUN
ajst-20197	23	19	.	.	PUNCT
ajst-20197	24	1	traditional	traditional	ADJ
ajst-20197	24	2	“	"	PUNCT
ajst-20197	24	3	neural	neural	ADJ
ajst-20197	24	4	”	"	PUNCT
ajst-20197	24	5	networks	network	NOUN
ajst-20197	24	6	consist	consist	VERB
ajst-20197	24	7	of	of	ADP
ajst-20197	24	8	more	more	ADJ
ajst-20197	24	9	than	than	ADP
ajst-20197	24	10	one	one	NUM
ajst-20197	24	11	hidden	hide	VERB
ajst-20197	24	12	layer	layer	NOUN
ajst-20197	24	13	,	,	PUNCT
ajst-20197	24	14	output	output	NOUN
ajst-20197	24	15	layer	layer	NOUN
ajst-20197	24	16	and	and	CCONJ
ajst-20197	24	17	input	input	NOUN
ajst-20197	24	18	layer	layer	NOUN
ajst-20197	24	19	.	.	PUNCT
ajst-20197	25	1	there	there	PRON
ajst-20197	25	2	is	be	VERB
ajst-20197	25	3	availability	availability	NOUN
ajst-20197	25	4	of	of	ADP
ajst-20197	25	5	connection	connection	NOUN
ajst-20197	25	6	within	within	ADP
ajst-20197	25	7	different	different	ADJ
ajst-20197	25	8	nodes	node	NOUN
ajst-20197	25	9	by	by	ADP
ajst-20197	25	10	having	have	VERB
ajst-20197	25	11	thresholds	threshold	NOUN
ajst-20197	25	12	and	and	CCONJ
ajst-20197	25	13	associated	associate	VERB
ajst-20197	25	14	weights	weight	NOUN
ajst-20197	25	15	.	.	PUNCT
ajst-20197	26	1	multiplied	multiply	VERB
ajst-20197	26	2	versions	version	NOUN
ajst-20197	26	3	of	of	ADP
ajst-20197	26	4	random	random	ADJ
ajst-20197	26	5	weights	weight	NOUN
ajst-20197	26	6	and	and	CCONJ
ajst-20197	26	7	inputs	input	NOUN
ajst-20197	26	8	are	be	AUX
ajst-20197	26	9	received	receive	VERB
ajst-20197	26	10	by	by	ADP
ajst-20197	26	11	neurons	neuron	NOUN
ajst-20197	26	12	.	.	PUNCT
ajst-20197	27	1	there	there	PRON
ajst-20197	27	2	has	have	AUX
ajst-20197	27	3	been	be	AUX
ajst-20197	27	4	addition	addition	NOUN
ajst-20197	27	5	of	of	ADP
ajst-20197	27	6	activated	activate	VERB
ajst-20197	27	7	functions	function	NOUN
ajst-20197	27	8	and	and	CCONJ
ajst-20197	27	9	a	a	DET
ajst-20197	27	10	“	"	PUNCT
ajst-20197	27	11	static	static	ADJ
ajst-20197	27	12	bias	bias	NOUN
ajst-20197	27	13	value	value	NOUN
ajst-20197	27	14	”	"	PUNCT
ajst-20197	27	15	that	that	PRON
ajst-20197	27	16	helps	help	VERB
ajst-20197	27	17	in	in	ADP
ajst-20197	27	18	the	the	DET
ajst-20197	27	19	formation	formation	NOUN
ajst-20197	27	20	of	of	ADP
ajst-20197	27	21	a	a	DET
ajst-20197	27	22	final	final	ADJ
ajst-20197	27	23	outcome	outcome	NOUN
ajst-20197	27	24	.	.	PUNCT
ajst-20197	28	1	the	the	DET
ajst-20197	28	2	performance	performance	NOUN
ajst-20197	28	3	of	of	ADP
ajst-20197	28	4	“	"	PUNCT
ajst-20197	28	5	traditional	traditional	ADJ
ajst-20197	28	6	neural	neural	ADJ
ajst-20197	28	7	networks	network	NOUN
ajst-20197	28	8	”	"	PUNCT
ajst-20197	28	9	are	be	AUX
ajst-20197	28	10	getting	get	VERB
ajst-20197	28	11	beyond	beyond	ADP
ajst-20197	28	12	scope	scope	NOUN
ajst-20197	28	13	and	and	CCONJ
ajst-20197	28	14	heavy	heavy	ADJ
ajst-20197	28	15	to	to	PART
ajst-20197	28	16	handle	handle	VERB
ajst-20197	28	17	some	some	DET
ajst-20197	28	18	typical	typical	ADJ
ajst-20197	28	19	application	application	NOUN
ajst-20197	28	20	scenarios	scenario	NOUN
ajst-20197	28	21	(	(	PUNCT
ajst-20197	28	22	for	for	ADP
ajst-20197	28	23	example	example	NOUN
ajst-20197	28	24	,	,	PUNCT
ajst-20197	28	25	medical	medical	ADJ
ajst-20197	28	26	diagnosis)[22	diagnosis)[22	PROPN
ajst-20197	28	27	]	]	X
ajst-20197	28	28	(	(	PUNCT
ajst-20197	28	29	salehi	salehi	X
ajst-20197	28	30	et	et	PROPN
ajst-20197	28	31	al	al	PROPN
ajst-20197	28	32	.	.	PROPN
ajst-20197	28	33	2023	2023	NUM
ajst-20197	28	34	)	)	PUNCT
ajst-20197	28	35	.	.	PUNCT
ajst-20197	29	1	there	there	PRON
ajst-20197	29	2	is	be	VERB
ajst-20197	29	3	a	a	DET
ajst-20197	29	4	limitation	limitation	NOUN
ajst-20197	29	5	within	within	ADP
ajst-20197	29	6	“	"	PUNCT
ajst-20197	29	7	traditional	traditional	ADJ
ajst-20197	29	8	neural	neural	ADJ
ajst-20197	29	9	networks	network	NOUN
ajst-20197	29	10	”	"	PUNCT
ajst-20197	29	11	in	in	ADP
ajst-20197	29	12	handling	handle	VERB
ajst-20197	29	13	sequential	sequential	ADJ
ajst-20197	29	14	data	datum	NOUN
ajst-20197	29	15	.	.	PUNCT
ajst-20197	30	1	this	this	PRON
ajst-20197	30	2	is	be	AUX
ajst-20197	30	3	because	because	SCONJ
ajst-20197	30	4	these	these	DET
ajst-20197	30	5	networks	network	NOUN
ajst-20197	30	6	help	help	VERB
ajst-20197	30	7	in	in	ADP
ajst-20197	30	8	the	the	DET
ajst-20197	30	9	independent	independent	ADJ
ajst-20197	30	10	processing	processing	NOUN
ajst-20197	30	11	of	of	ADP
ajst-20197	30	12	inputs	input	NOUN
ajst-20197	30	13	without	without	ADP
ajst-20197	30	14	having	have	VERB
ajst-20197	30	15	any	any	DET
ajst-20197	30	16	presence	presence	NOUN
ajst-20197	30	17	of	of	ADP
ajst-20197	30	18	past	past	ADJ
ajst-20197	30	19	memories	memory	NOUN
ajst-20197	30	20	[	[	X
ajst-20197	30	21	1](alzubaidi	1](alzubaidi	NUM
ajst-20197	30	22	et	et	NOUN
ajst-20197	30	23	al	al	PROPN
ajst-20197	30	24	.	.	PUNCT
ajst-20197	31	1	2021)[1	2021)[1	NUM
ajst-20197	31	2	]	]	PUNCT
ajst-20197	31	3	.	.	PUNCT
ajst-20197	32	1	as	as	ADP
ajst-20197	32	2	a	a	DET
ajst-20197	32	3	result	result	NOUN
ajst-20197	32	4	,	,	PUNCT
ajst-20197	32	5	it	it	PRON
ajst-20197	32	6	is	be	AUX
ajst-20197	32	7	unable	unable	ADJ
ajst-20197	32	8	to	to	PART
ajst-20197	32	9	capture	capture	VERB
ajst-20197	32	10	the	the	DET
ajst-20197	32	11	dependencies	dependency	NOUN
ajst-20197	32	12	and	and	CCONJ
ajst-20197	32	13	relationships	relationship	NOUN
ajst-20197	32	14	with	with	ADP
ajst-20197	32	15	respect	respect	NOUN
ajst-20197	32	16	to	to	ADP
ajst-20197	32	17	time	time	NOUN
ajst-20197	32	18	and	and	CCONJ
ajst-20197	32	19	thereby	thereby	ADV
ajst-20197	32	20	leading	lead	VERB
ajst-20197	32	21	to	to	ADP
ajst-20197	32	22	the	the	DET
ajst-20197	32	23	decline	decline	NOUN
ajst-20197	32	24	in	in	ADP
ajst-20197	32	25	the	the	DET
ajst-20197	32	26	model	model	NOUN
ajst-20197	32	27	performance	performance	NOUN
ajst-20197	32	28	.	.	PUNCT
ajst-20197	33	1	this	this	PRON
ajst-20197	33	2	has	have	AUX
ajst-20197	33	3	led	lead	VERB
ajst-20197	33	4	to	to	ADP
ajst-20197	33	5	the	the	DET
ajst-20197	33	6	generation	generation	NOUN
ajst-20197	33	7	of	of	ADP
ajst-20197	33	8	attention	attention	NOUN
ajst-20197	33	9	and	and	CCONJ
ajst-20197	33	10	memory	memory	NOUN
ajst-20197	33	11	mechanisms	mechanism	NOUN
ajst-20197	33	12	within	within	ADP
ajst-20197	33	13	deep	deep	ADJ
ajst-20197	33	14	learning	learning	NOUN
ajst-20197	33	15	that	that	PRON
ajst-20197	33	16	helps	help	VERB
ajst-20197	33	17	in	in	ADP
ajst-20197	33	18	the	the	DET
ajst-20197	33	19	sequential	sequential	ADJ
ajst-20197	33	20	handling	handling	NOUN
ajst-20197	33	21	of	of	ADP
ajst-20197	33	22	data	data	PROPN
ajst-20197	33	23	.	.	PUNCT
ajst-20197	34	1	memory	memory	NOUN
ajst-20197	34	2	mechanisms	mechanism	NOUN
ajst-20197	34	3	help	help	VERB
ajst-20197	34	4	in	in	ADP
ajst-20197	34	5	increasing	increase	VERB
ajst-20197	34	6	the	the	DET
ajst-20197	34	7	capability	capability	NOUN
ajst-20197	34	8	for	for	ADP
ajst-20197	34	9	retaining	retain	VERB
ajst-20197	34	10	data	datum	NOUN
ajst-20197	34	11	in	in	ADP
ajst-20197	34	12	models	model	NOUN
ajst-20197	34	13	that	that	PRON
ajst-20197	34	14	are	be	AUX
ajst-20197	34	15	available	available	ADJ
ajst-20197	34	16	within	within	ADP
ajst-20197	34	17	a	a	DET
ajst-20197	34	18	particular	particular	ADJ
ajst-20197	34	19	input	input	NOUN
ajst-20197	34	20	sequence[27	sequence[27	NOUN
ajst-20197	34	21	]	]	X
ajst-20197	34	22	(	(	PUNCT
ajst-20197	34	23	xi	xi	X
ajst-20197	34	24	et	et	PROPN
ajst-20197	34	25	al	al	PROPN
ajst-20197	34	26	.	.	PROPN
ajst-20197	34	27	2023	2023	NUM
ajst-20197	34	28	)	)	PUNCT
ajst-20197	34	29	.	.	PUNCT
ajst-20197	35	1	the	the	DET
ajst-20197	35	2	tasks	task	NOUN
ajst-20197	35	3	containing	contain	VERB
ajst-20197	35	4	massive	massive	ADJ
ajst-20197	35	5	datasets	dataset	NOUN
ajst-20197	35	6	are	be	AUX
ajst-20197	35	7	retrieved	retrieve	VERB
ajst-20197	35	8	and	and	CCONJ
ajst-20197	35	9	stored	store	VERB
ajst-20197	35	10	by	by	ADP
ajst-20197	35	11	memory	memory	NOUN
ajst-20197	35	12	mechanisms	mechanism	NOUN
ajst-20197	35	13	for	for	ADP
ajst-20197	35	14	understanding	understand	VERB
ajst-20197	35	15	the	the	DET
ajst-20197	35	16	tasks	task	NOUN
ajst-20197	35	17	minutely	minutely	ADV
ajst-20197	35	18	.	.	PUNCT
ajst-20197	36	1	attention	attention	NOUN
ajst-20197	36	2	mechanisms	mechanism	NOUN
ajst-20197	36	3	help	help	VERB
ajst-20197	36	4	in	in	ADP
ajst-20197	36	5	developing	develop	VERB
ajst-20197	36	6	the	the	DET
ajst-20197	36	7	model	model	NOUN
ajst-20197	36	8	capability	capability	NOUN
ajst-20197	36	9	by	by	ADP
ajst-20197	36	10	developing	develop	VERB
ajst-20197	36	11	focus	focus	NOUN
ajst-20197	36	12	over	over	ADP
ajst-20197	36	13	a	a	DET
ajst-20197	36	14	specific	specific	ADJ
ajst-20197	36	15	portion	portion	NOUN
ajst-20197	36	16	of	of	ADP
ajst-20197	36	17	a	a	DET
ajst-20197	36	18	dataset	dataset	NOUN
ajst-20197	37	1	[	[	X
ajst-20197	37	2	3](dai	3](dai	NUM
ajst-20197	37	3	et	et	NOUN
ajst-20197	37	4	al	al	PROPN
ajst-20197	37	5	.	.	PROPN
ajst-20197	37	6	2021	2021	NUM
ajst-20197	37	7	)	)	PUNCT
ajst-20197	37	8	.	.	PUNCT
ajst-20197	38	1	this	this	PRON
ajst-20197	38	2	is	be	AUX
ajst-20197	38	3	effective	effective	ADJ
ajst-20197	38	4	in	in	ADP
ajst-20197	38	5	the	the	DET
ajst-20197	38	6	complete	complete	ADJ
ajst-20197	38	7	elimination	elimination	NOUN
ajst-20197	38	8	of	of	ADP
ajst-20197	38	9	large	large	ADJ
ajst-20197	38	10	background	background	NOUN
ajst-20197	38	11	data	datum	NOUN
ajst-20197	38	12	and	and	CCONJ
ajst-20197	38	13	thereby	thereby	ADV
ajst-20197	38	14	leading	lead	VERB
ajst-20197	38	15	to	to	ADP
ajst-20197	38	16	the	the	DET
ajst-20197	38	17	decline	decline	NOUN
ajst-20197	38	18	in	in	ADP
ajst-20197	38	19	the	the	DET
ajst-20197	38	20	irrelevant	irrelevant	ADJ
ajst-20197	38	21	data	datum	NOUN
ajst-20197	38	22	within	within	ADP
ajst-20197	38	23	a	a	DET
ajst-20197	38	24	dataset	dataset	NOUN
ajst-20197	38	25	.	.	PUNCT
ajst-20197	39	1	attention	attention	NOUN
ajst-20197	39	2	mechanisms	mechanism	NOUN
ajst-20197	39	3	help	help	VERB
ajst-20197	39	4	in	in	ADP
ajst-20197	39	5	understanding	understand	VERB
ajst-20197	39	6	the	the	DET
ajst-20197	39	7	sequence	sequence	NOUN
ajst-20197	39	8	through	through	ADP
ajst-20197	39	9	the	the	DET
ajst-20197	39	10	acquisition	acquisition	NOUN
ajst-20197	39	11	of	of	ADP
ajst-20197	39	12	dependencies	dependency	NOUN
ajst-20197	39	13	available	available	ADJ
ajst-20197	39	14	within	within	ADP
ajst-20197	39	15	input	input	NOUN
ajst-20197	39	16	data	datum	NOUN
ajst-20197	39	17	weights	weight	NOUN
ajst-20197	39	18	.	.	PUNCT
ajst-20197	40	1	the	the	DET
ajst-20197	40	2	aim	aim	NOUN
ajst-20197	40	3	of	of	ADP
ajst-20197	40	4	this	this	DET
ajst-20197	40	5	paper	paper	NOUN
ajst-20197	40	6	is	be	AUX
ajst-20197	40	7	to	to	PART
ajst-20197	40	8	determine	determine	VERB
ajst-20197	40	9	the	the	DET
ajst-20197	40	10	different	different	ADJ
ajst-20197	40	11	attention	attention	NOUN
ajst-20197	40	12	and	and	CCONJ
ajst-20197	40	13	memory	memory	NOUN
ajst-20197	40	14	mechanisms	mechanism	NOUN
ajst-20197	40	15	that	that	PRON
ajst-20197	40	16	have	have	AUX
ajst-20197	40	17	helped	help	VERB
ajst-20197	40	18	in	in	ADP
ajst-20197	40	19	increasing	increase	VERB
ajst-20197	40	20	the	the	DET
ajst-20197	40	21	accuracy	accuracy	NOUN
ajst-20197	40	22	,	,	PUNCT
ajst-20197	40	23	reliability	reliability	NOUN
ajst-20197	40	24	and	and	CCONJ
ajst-20197	40	25	model	model	NOUN
ajst-20197	40	26	performance	performance	NOUN
ajst-20197	40	27	of	of	ADP
ajst-20197	40	28	“	"	PUNCT
ajst-20197	40	29	deep	deep	ADJ
ajst-20197	40	30	learning	learning	NOUN
ajst-20197	40	31	models	model	NOUN
ajst-20197	40	32	”	"	PUNCT
ajst-20197	40	33	.	.	PUNCT
ajst-20197	41	1	2.1	2.1	NUM
ajst-20197	41	2	.	.	PUNCT
ajst-20197	41	3	methods	method	NOUN
ajst-20197	41	4	linking	link	VERB
ajst-20197	41	5	with	with	ADP
ajst-20197	41	6	the	the	DET
ajst-20197	41	7	aim	aim	NOUN
ajst-20197	41	8	of	of	ADP
ajst-20197	41	9	this	this	DET
ajst-20197	41	10	research	research	NOUN
ajst-20197	41	11	,	,	PUNCT
ajst-20197	41	12	it	it	PRON
ajst-20197	41	13	is	be	AUX
ajst-20197	41	14	suggested	suggest	VERB
ajst-20197	41	15	that	that	SCONJ
ajst-20197	41	16	this	this	DET
ajst-20197	41	17	research	research	NOUN
ajst-20197	41	18	is	be	AUX
ajst-20197	41	19	accomplished	accomplish	VERB
ajst-20197	41	20	through	through	ADP
ajst-20197	41	21	the	the	DET
ajst-20197	41	22	consideration	consideration	NOUN
ajst-20197	41	23	of	of	ADP
ajst-20197	41	24	secondary	secondary	ADJ
ajst-20197	41	25	data	datum	NOUN
ajst-20197	41	26	for	for	ADP
ajst-20197	41	27	gathering	gather	VERB
ajst-20197	41	28	detailed	detailed	ADJ
ajst-20197	41	29	information	information	NOUN
ajst-20197	41	30	about	about	ADP
ajst-20197	41	31	different	different	ADJ
ajst-20197	41	32	attention	attention	NOUN
ajst-20197	41	33	and	and	CCONJ
ajst-20197	41	34	memory	memory	NOUN
ajst-20197	41	35	mechanisms	mechanism	NOUN
ajst-20197	41	36	used	use	VERB
ajst-20197	41	37	in	in	ADP
ajst-20197	41	38	deep	deep	ADJ
ajst-20197	41	39	learning	learning	NOUN
ajst-20197	41	40	.	.	PUNCT
ajst-20197	42	1	peer	peer	NOUN
ajst-20197	42	2	review	review	NOUN
ajst-20197	42	3	is	be	AUX
ajst-20197	42	4	done	do	VERB
ajst-20197	42	5	in	in	ADP
ajst-20197	42	6	this	this	DET
ajst-20197	42	7	research	research	NOUN
ajst-20197	42	8	through	through	ADP
ajst-20197	42	9	the	the	DET
ajst-20197	42	10	110	110	NUM
ajst-20197	42	11	consideration	consideration	NOUN
ajst-20197	42	12	of	of	ADP
ajst-20197	42	13	different	different	ADJ
ajst-20197	42	14	journals	journal	NOUN
ajst-20197	42	15	and	and	CCONJ
ajst-20197	42	16	articles	article	NOUN
ajst-20197	42	17	.	.	PUNCT
ajst-20197	43	1	qualitative	qualitative	ADJ
ajst-20197	43	2	research	research	NOUN
ajst-20197	43	3	is	be	AUX
ajst-20197	43	4	done	do	VERB
ajst-20197	43	5	through	through	ADP
ajst-20197	43	6	the	the	DET
ajst-20197	43	7	use	use	NOUN
ajst-20197	43	8	of	of	ADP
ajst-20197	43	9	secondary	secondary	ADJ
ajst-20197	43	10	data	datum	NOUN
ajst-20197	43	11	that	that	PRON
ajst-20197	43	12	is	be	AUX
ajst-20197	43	13	available	available	ADJ
ajst-20197	43	14	in	in	ADP
ajst-20197	43	15	different	different	ADJ
ajst-20197	43	16	journals	journal	NOUN
ajst-20197	43	17	[	[	X
ajst-20197	43	18	15](nassaji	15](nassaji	NUM
ajst-20197	43	19	2020	2020	NUM
ajst-20197	43	20	)	)	PUNCT
ajst-20197	43	21	.	.	PUNCT
ajst-20197	44	1	there	there	PRON
ajst-20197	44	2	is	be	VERB
ajst-20197	44	3	accumulation	accumulation	NOUN
ajst-20197	44	4	of	of	ADP
ajst-20197	44	5	qualitative	qualitative	ADJ
ajst-20197	44	6	information	information	NOUN
ajst-20197	44	7	based	base	VERB
ajst-20197	44	8	on	on	ADP
ajst-20197	44	9	different	different	ADJ
ajst-20197	44	10	models	model	NOUN
ajst-20197	44	11	used	use	VERB
ajst-20197	44	12	in	in	ADP
ajst-20197	44	13	attention	attention	NOUN
ajst-20197	44	14	and	and	CCONJ
ajst-20197	44	15	memory	memory	NOUN
ajst-20197	44	16	mechanisms	mechanism	NOUN
ajst-20197	44	17	within	within	ADP
ajst-20197	44	18	deep	deep	ADJ
ajst-20197	44	19	learning	learning	NOUN
ajst-20197	44	20	.	.	PUNCT
ajst-20197	45	1	this	this	DET
ajst-20197	45	2	research	research	NOUN
ajst-20197	45	3	has	have	AUX
ajst-20197	45	4	made	make	VERB
ajst-20197	45	5	a	a	DET
ajst-20197	45	6	selection	selection	NOUN
ajst-20197	45	7	of	of	ADP
ajst-20197	45	8	journals	journal	NOUN
ajst-20197	45	9	that	that	PRON
ajst-20197	45	10	are	be	AUX
ajst-20197	45	11	available	available	ADJ
ajst-20197	45	12	within	within	ADP
ajst-20197	45	13	the	the	DET
ajst-20197	45	14	last	last	ADJ
ajst-20197	45	15	6	6	NUM
ajst-20197	45	16	years	year	NOUN
ajst-20197	45	17	and	and	CCONJ
ajst-20197	45	18	have	have	AUX
ajst-20197	45	19	accessed	access	VERB
ajst-20197	45	20	them	they	PRON
ajst-20197	45	21	from	from	ADP
ajst-20197	45	22	google	google	PROPN
ajst-20197	45	23	scholar	scholar	NOUN
ajst-20197	45	24	database	database	NOUN
ajst-20197	45	25	.	.	PUNCT
ajst-20197	46	1	appropriate	appropriate	ADJ
ajst-20197	46	2	descriptions	description	NOUN
ajst-20197	46	3	are	be	AUX
ajst-20197	46	4	provided	provide	VERB
ajst-20197	46	5	on	on	ADP
ajst-20197	46	6	different	different	ADJ
ajst-20197	46	7	attention	attention	NOUN
ajst-20197	46	8	and	and	CCONJ
ajst-20197	46	9	memory	memory	NOUN
ajst-20197	46	10	mechanisms	mechanism	NOUN
ajst-20197	46	11	to	to	PART
ajst-20197	46	12	elaborate	elaborate	VERB
ajst-20197	46	13	the	the	DET
ajst-20197	46	14	contribution	contribution	NOUN
ajst-20197	46	15	made	make	VERB
ajst-20197	46	16	by	by	ADP
ajst-20197	46	17	them	they	PRON
ajst-20197	46	18	in	in	ADP
ajst-20197	46	19	deep	deep	ADJ
ajst-20197	46	20	learning	learning	NOUN
ajst-20197	46	21	models	model	NOUN
ajst-20197	46	22	.	.	PUNCT
ajst-20197	47	1	3	3	X
ajst-20197	47	2	.	.	NOUN
ajst-20197	47	3	results	result	VERB
ajst-20197	47	4	3.1	3.1	NUM
ajst-20197	47	5	.	.	PUNCT
ajst-20197	47	6	memory	memory	NOUN
ajst-20197	47	7	mechanisms	mechanism	NOUN
ajst-20197	47	8	3.1.1	3.1.1	NOUN
ajst-20197	47	9	.	.	PUNCT
ajst-20197	48	1	gated	gate	VERB
ajst-20197	48	2	recurrent	recurrent	ADJ
ajst-20197	48	3	units	unit	NOUN
ajst-20197	48	4	according	accord	VERB
ajst-20197	48	5	to	to	ADP
ajst-20197	48	6	[	[	X
ajst-20197	48	7	20]rajamani	20]rajamani	NUM
ajst-20197	48	8	et	et	PROPN
ajst-20197	48	9	al	al	PROPN
ajst-20197	48	10	.	.	PROPN
ajst-20197	48	11	2021	2021	NUM
ajst-20197	48	12	,	,	PUNCT
ajst-20197	48	13	there	there	PRON
ajst-20197	48	14	has	have	AUX
ajst-20197	48	15	been	be	AUX
ajst-20197	48	16	wide	wide	ADJ
ajst-20197	48	17	usage	usage	NOUN
ajst-20197	48	18	of	of	ADP
ajst-20197	48	19	gating	gate	VERB
ajst-20197	48	20	mechanisms	mechanism	NOUN
ajst-20197	48	21	within	within	ADP
ajst-20197	48	22	"	"	PUNCT
ajst-20197	48	23	neural	neural	ADJ
ajst-20197	48	24	network	network	NOUN
ajst-20197	48	25	models	model	NOUN
ajst-20197	48	26	"	"	PUNCT
ajst-20197	48	27	.	.	PUNCT
ajst-20197	49	1	these	these	DET
ajst-20197	49	2	mechanisms	mechanism	NOUN
ajst-20197	49	3	permit	permit	VERB
ajst-20197	49	4	gradients	gradient	NOUN
ajst-20197	49	5	in	in	ADP
ajst-20197	49	6	backpropagation	backpropagation	NOUN
ajst-20197	49	7	in	in	ADP
ajst-20197	49	8	a	a	DET
ajst-20197	49	9	much	much	ADV
ajst-20197	49	10	easier	easy	ADJ
ajst-20197	49	11	manner	manner	NOUN
ajst-20197	49	12	through	through	ADP
ajst-20197	49	13	time	time	NOUN
ajst-20197	49	14	and	and	CCONJ
ajst-20197	49	15	depth	depth	NOUN
ajst-20197	49	16	.	.	PUNCT
ajst-20197	50	1	one	one	NUM
ajst-20197	50	2	of	of	ADP
ajst-20197	50	3	the	the	DET
ajst-20197	50	4	most	most	ADV
ajst-20197	50	5	prominent	prominent	ADJ
ajst-20197	50	6	features	feature	NOUN
ajst-20197	50	7	of	of	ADP
ajst-20197	50	8	gru	gru	NOUN
ajst-20197	50	9	is	be	AUX
ajst-20197	50	10	the	the	DET
ajst-20197	50	11	presence	presence	NOUN
ajst-20197	50	12	of	of	ADP
ajst-20197	50	13	activation	activation	NOUN
ajst-20197	50	14	functions	function	NOUN
ajst-20197	50	15	.	.	PUNCT
ajst-20197	51	1	rnns	rnn	NOUN
ajst-20197	51	2	that	that	PRON
ajst-20197	51	3	include	include	VERB
ajst-20197	51	4	gru	gru	NOUN
ajst-20197	51	5	and	and	CCONJ
ajst-20197	51	6	lstm	lstm	NOUN
ajst-20197	51	7	help	help	NOUN
ajst-20197	51	8	in	in	ADP
ajst-20197	51	9	capturing	capture	VERB
ajst-20197	51	10	the	the	DET
ajst-20197	51	11	sequential	sequential	ADJ
ajst-20197	51	12	data	datum	NOUN
ajst-20197	51	13	that	that	PRON
ajst-20197	51	14	possesses	possess	VERB
ajst-20197	51	15	temporal	temporal	ADJ
ajst-20197	51	16	dynamics	dynamic	NOUN
ajst-20197	51	17	.	.	PUNCT
ajst-20197	52	1	"	"	PUNCT
ajst-20197	52	2	gated	gate	VERB
ajst-20197	52	3	recurrent	recurrent	ADJ
ajst-20197	52	4	units	unit	NOUN
ajst-20197	52	5	"	"	PUNCT
ajst-20197	52	6	are	be	AUX
ajst-20197	52	7	considered	consider	VERB
ajst-20197	52	8	to	to	PART
ajst-20197	52	9	be	be	AUX
ajst-20197	52	10	a	a	DET
ajst-20197	52	11	feature	feature	NOUN
ajst-20197	52	12	of	of	ADP
ajst-20197	52	13	rnn	rnn	NOUN
ajst-20197	52	14	that	that	PRON
ajst-20197	52	15	helps	help	VERB
ajst-20197	52	16	in	in	ADP
ajst-20197	52	17	the	the	DET
ajst-20197	52	18	implementation	implementation	NOUN
ajst-20197	52	19	of	of	ADP
ajst-20197	52	20	gating	gate	VERB
ajst-20197	52	21	mechanisms	mechanism	NOUN
ajst-20197	52	22	.	.	PUNCT
ajst-20197	53	1	this	this	DET
ajst-20197	53	2	feature	feature	NOUN
ajst-20197	53	3	helps	help	VERB
ajst-20197	53	4	in	in	ADP
ajst-20197	53	5	the	the	DET
ajst-20197	53	6	development	development	NOUN
ajst-20197	53	7	of	of	ADP
ajst-20197	53	8	control	control	NOUN
ajst-20197	53	9	over	over	ADP
ajst-20197	53	10	information	information	NOUN
ajst-20197	53	11	flow	flow	NOUN
ajst-20197	53	12	and	and	CCONJ
ajst-20197	53	13	thereby	thereby	ADV
ajst-20197	53	14	help	help	VERB
ajst-20197	53	15	in	in	ADP
ajst-20197	53	16	managing	manage	VERB
ajst-20197	53	17	cells	cell	NOUN
ajst-20197	53	18	available	available	ADJ
ajst-20197	53	19	within	within	ADP
ajst-20197	53	20	one	one	NUM
ajst-20197	53	21	neural	neural	ADJ
ajst-20197	53	22	network	network	NOUN
ajst-20197	53	23	.	.	PUNCT
ajst-20197	54	1	the	the	DET
ajst-20197	54	2	gru	gru	NOUN
ajst-20197	54	3	structure	structure	NOUN
ajst-20197	54	4	helps	help	VERB
ajst-20197	54	5	in	in	ADP
ajst-20197	54	6	the	the	DET
ajst-20197	54	7	quick	quick	ADJ
ajst-20197	54	8	acquisition	acquisition	NOUN
ajst-20197	54	9	of	of	ADP
ajst-20197	54	10	dependencies	dependency	NOUN
ajst-20197	54	11	that	that	PRON
ajst-20197	54	12	are	be	AUX
ajst-20197	54	13	available	available	ADJ
ajst-20197	54	14	within	within	ADP
ajst-20197	54	15	the	the	DET
ajst-20197	54	16	data	data	NOUN
ajst-20197	54	17	sequences	sequence	NOUN
ajst-20197	54	18	of	of	ADP
ajst-20197	54	19	a	a	DET
ajst-20197	54	20	large	large	ADJ
ajst-20197	54	21	dataset	dataset	NOUN
ajst-20197	54	22	,	,	PUNCT
ajst-20197	54	23	especially	especially	ADV
ajst-20197	54	24	videos[18	videos[18	ADP
ajst-20197	54	25	]	]	X
ajst-20197	54	26	(	(	PUNCT
ajst-20197	54	27	rafiq	rafiq	PROPN
ajst-20197	54	28	,	,	PUNCT
ajst-20197	54	29	rafiq	rafiq	PROPN
ajst-20197	54	30	and	and	CCONJ
ajst-20197	54	31	choi	choi	NOUN
ajst-20197	54	32	2023	2023	NUM
ajst-20197	54	33	)	)	PUNCT
ajst-20197	54	34	.	.	PUNCT
ajst-20197	55	1	this	this	PRON
ajst-20197	55	2	helps	help	VERB
ajst-20197	55	3	in	in	ADP
ajst-20197	55	4	ensuring	ensure	VERB
ajst-20197	55	5	the	the	DET
ajst-20197	55	6	data	datum	NOUN
ajst-20197	55	7	available	available	ADJ
ajst-20197	55	8	in	in	ADP
ajst-20197	55	9	the	the	DET
ajst-20197	55	10	previous	previous	ADJ
ajst-20197	55	11	sections	section	NOUN
ajst-20197	55	12	of	of	ADP
ajst-20197	55	13	a	a	DET
ajst-20197	55	14	data	data	NOUN
ajst-20197	55	15	sequence	sequence	NOUN
ajst-20197	55	16	.	.	PUNCT
ajst-20197	56	1	information	information	NOUN
ajst-20197	56	2	regulation	regulation	NOUN
ajst-20197	56	3	is	be	AUX
ajst-20197	56	4	accomplished	accomplish	VERB
ajst-20197	56	5	through	through	ADP
ajst-20197	56	6	the	the	DET
ajst-20197	56	7	implementation	implementation	NOUN
ajst-20197	56	8	of	of	ADP
ajst-20197	56	9	gating	gate	VERB
ajst-20197	56	10	mechanisms	mechanism	NOUN
ajst-20197	56	11	that	that	PRON
ajst-20197	56	12	helps	help	VERB
ajst-20197	56	13	in	in	ADP
ajst-20197	56	14	distorting	distort	VERB
ajst-20197	56	15	or	or	CCONJ
ajst-20197	56	16	keeping	keep	VERB
ajst-20197	56	17	information	information	NOUN
ajst-20197	56	18	at	at	ADP
ajst-20197	56	19	different	different	ADJ
ajst-20197	56	20	time	time	NOUN
ajst-20197	56	21	steps	step	NOUN
ajst-20197	56	22	.	.	PUNCT
ajst-20197	57	1	therefore	therefore	ADV
ajst-20197	57	2	,	,	PUNCT
ajst-20197	57	3	the	the	DET
ajst-20197	57	4	argument	argument	NOUN
ajst-20197	57	5	could	could	AUX
ajst-20197	57	6	be	be	AUX
ajst-20197	57	7	made	make	VERB
ajst-20197	57	8	that	that	SCONJ
ajst-20197	57	9	there	there	PRON
ajst-20197	57	10	is	be	VERB
ajst-20197	57	11	a	a	DET
ajst-20197	57	12	gradual	gradual	ADJ
ajst-20197	57	13	increase	increase	NOUN
ajst-20197	57	14	in	in	ADP
ajst-20197	57	15	the	the	DET
ajst-20197	57	16	model	model	NOUN
ajst-20197	57	17	performance	performance	NOUN
ajst-20197	57	18	of	of	ADP
ajst-20197	57	19	deep	deep	ADJ
ajst-20197	57	20	learning	learning	NOUN
ajst-20197	57	21	through	through	ADP
ajst-20197	57	22	grus	grus	NOUN
ajst-20197	57	23	as	as	SCONJ
ajst-20197	57	24	it	it	PRON
ajst-20197	57	25	helps	help	VERB
ajst-20197	57	26	in	in	ADP
ajst-20197	57	27	the	the	DET
ajst-20197	57	28	complete	complete	ADJ
ajst-20197	57	29	elimination	elimination	NOUN
ajst-20197	57	30	of	of	ADP
ajst-20197	57	31	the	the	DET
ajst-20197	57	32	"	"	PUNCT
ajst-20197	57	33	vanishing	vanish	VERB
ajst-20197	57	34	gradient	gradient	ADJ
ajst-20197	57	35	problem	problem	NOUN
ajst-20197	57	36	"	"	PUNCT
ajst-20197	58	1	[	[	X
ajst-20197	58	2	30](zulqarnain	30](zulqarnain	NUM
ajst-20197	58	3	et	et	NOUN
ajst-20197	58	4	al	al	PROPN
ajst-20197	58	5	.	.	PROPN
ajst-20197	58	6	2020	2020	NUM
ajst-20197	58	7	)	)	PUNCT
ajst-20197	58	8	.	.	PUNCT
ajst-20197	59	1	this	this	PRON
ajst-20197	59	2	is	be	AUX
ajst-20197	59	3	effective	effective	ADJ
ajst-20197	59	4	in	in	ADP
ajst-20197	59	5	increasing	increase	VERB
ajst-20197	59	6	the	the	DET
ajst-20197	59	7	speed	speed	NOUN
ajst-20197	59	8	of	of	ADP
ajst-20197	59	9	model	model	NOUN
ajst-20197	59	10	training	training	NOUN
ajst-20197	59	11	as	as	SCONJ
ajst-20197	59	12	that	that	PRON
ajst-20197	59	13	is	be	AUX
ajst-20197	59	14	a	a	DET
ajst-20197	59	15	considerable	considerable	ADJ
ajst-20197	59	16	reduction	reduction	NOUN
ajst-20197	59	17	in	in	ADP
ajst-20197	59	18	parameters	parameter	NOUN
ajst-20197	59	19	through	through	ADP
ajst-20197	59	20	the	the	DET
ajst-20197	59	21	usage	usage	NOUN
ajst-20197	59	22	of	of	ADP
ajst-20197	59	23	optimization	optimization	NOUN
ajst-20197	59	24	technique	technique	NOUN
ajst-20197	59	25	.	.	PUNCT
ajst-20197	60	1	according	accord	VERB
ajst-20197	60	2	to	to	ADP
ajst-20197	60	3	[	[	X
ajst-20197	60	4	8]jordan	8]jordan	NUM
ajst-20197	60	5	et	et	PROPN
ajst-20197	60	6	al	al	PROPN
ajst-20197	60	7	.	.	PROPN
ajst-20197	60	8	2021	2021	NUM
ajst-20197	60	9	,	,	PUNCT
ajst-20197	60	10	data	datum	NOUN
ajst-20197	60	11	is	be	AUX
ajst-20197	60	12	written	write	VERB
ajst-20197	60	13	within	within	ADP
ajst-20197	60	14	the	the	DET
ajst-20197	60	15	models	model	NOUN
ajst-20197	60	16	through	through	ADP
ajst-20197	60	17	the	the	DET
ajst-20197	60	18	implementation	implementation	NOUN
ajst-20197	60	19	of	of	ADP
ajst-20197	60	20	memory	memory	NOUN
ajst-20197	60	21	mechanisms	mechanism	NOUN
ajst-20197	60	22	.	.	PUNCT
ajst-20197	61	1	this	this	PRON
ajst-20197	61	2	is	be	AUX
ajst-20197	61	3	accomplished	accomplish	VERB
ajst-20197	61	4	through	through	ADP
ajst-20197	61	5	the	the	DET
ajst-20197	61	6	implementation	implementation	NOUN
ajst-20197	61	7	of	of	ADP
ajst-20197	61	8	"	"	PUNCT
ajst-20197	61	9	gated	gate	VERB
ajst-20197	61	10	recurrent	recurrent	ADJ
ajst-20197	61	11	units	unit	NOUN
ajst-20197	61	12	"	"	PUNCT
ajst-20197	61	13	that	that	PRON
ajst-20197	61	14	helps	help	VERB
ajst-20197	61	15	in	in	ADP
ajst-20197	61	16	the	the	DET
ajst-20197	61	17	development	development	NOUN
ajst-20197	61	18	of	of	ADP
ajst-20197	61	19	rnns	rnns	PROPN
ajst-20197	61	20	through	through	ADP
ajst-20197	61	21	the	the	DET
ajst-20197	61	22	inclusion	inclusion	NOUN
ajst-20197	61	23	of	of	ADP
ajst-20197	61	24	"	"	PUNCT
ajst-20197	61	25	specialized	specialized	ADJ
ajst-20197	61	26	memory	memory	NOUN
ajst-20197	61	27	elements	element	NOUN
ajst-20197	61	28	"	"	PUNCT
ajst-20197	62	1	[	[	X
ajst-20197	62	2	2](chandra	2](chandra	NUM
ajst-20197	62	3	et	et	NOUN
ajst-20197	62	4	al	al	PROPN
ajst-20197	62	5	.	.	PROPN
ajst-20197	62	6	2021	2021	NUM
ajst-20197	62	7	)	)	PUNCT
ajst-20197	62	8	.	.	PUNCT
ajst-20197	63	1	the	the	DET
ajst-20197	63	2	rnns	rnns	PROPN
ajst-20197	63	3	help	help	VERB
ajst-20197	63	4	in	in	ADP
ajst-20197	63	5	utilizing	utilize	VERB
ajst-20197	63	6	and	and	CCONJ
ajst-20197	63	7	capturing	capture	VERB
ajst-20197	63	8	the	the	DET
ajst-20197	63	9	sequential	sequential	ADJ
ajst-20197	63	10	structure	structure	NOUN
ajst-20197	63	11	that	that	PRON
ajst-20197	63	12	is	be	AUX
ajst-20197	63	13	available	available	ADJ
ajst-20197	63	14	within	within	ADP
ajst-20197	63	15	artificial	artificial	ADJ
ajst-20197	63	16	and	and	CCONJ
ajst-20197	63	17	natural	natural	ADJ
ajst-20197	63	18	languages	language	NOUN
ajst-20197	63	19	that	that	PRON
ajst-20197	63	20	include	include	VERB
ajst-20197	63	21	videos	video	NOUN
ajst-20197	63	22	,	,	PUNCT
ajst-20197	63	23	speech	speech	NOUN
ajst-20197	63	24	and	and	CCONJ
ajst-20197	63	25	time	time	NOUN
ajst-20197	63	26	series	series	NOUN
ajst-20197	63	27	sequences	sequence	NOUN
ajst-20197	64	1	[	[	X
ajst-20197	64	2	24](torfi	24](torfi	NUM
ajst-20197	64	3	et	et	NOUN
ajst-20197	64	4	al	al	PROPN
ajst-20197	64	5	.	.	PROPN
ajst-20197	64	6	2020	2020	NUM
ajst-20197	64	7	)	)	PUNCT
ajst-20197	64	8	.	.	PUNCT
ajst-20197	65	1	this	this	PRON
ajst-20197	65	2	is	be	AUX
ajst-20197	65	3	effective	effective	ADJ
ajst-20197	65	4	in	in	ADP
ajst-20197	65	5	capturing	capture	VERB
ajst-20197	65	6	the	the	DET
ajst-20197	65	7	information	information	NOUN
ajst-20197	65	8	flow	flow	NOUN
ajst-20197	65	9	by	by	ADP
ajst-20197	65	10	developing	develop	VERB
ajst-20197	65	11	influence	influence	NOUN
ajst-20197	65	12	over	over	ADP
ajst-20197	65	13	the	the	DET
ajst-20197	65	14	presence	presence	NOUN
ajst-20197	65	15	state	state	NOUN
ajst-20197	65	16	based	base	VERB
ajst-20197	65	17	on	on	ADP
ajst-20197	65	18	the	the	DET
ajst-20197	65	19	past	past	ADJ
ajst-20197	65	20	data	data	NOUN
ajst-20197	65	21	sequence	sequence	NOUN
ajst-20197	65	22	.	.	PUNCT
ajst-20197	66	1	park	park	NOUN
ajst-20197	66	2	et	et	PROPN
ajst-20197	66	3	al	al	PROPN
ajst-20197	66	4	.	.	PROPN
ajst-20197	66	5	2021	2021	NUM
ajst-20197	66	6	states	state	VERB
ajst-20197	66	7	that	that	SCONJ
ajst-20197	66	8	the	the	DET
ajst-20197	66	9	grus	grus	NOUN
ajst-20197	66	10	implements	implement	VERB
ajst-20197	66	11	two	two	NUM
ajst-20197	66	12	different	different	ADJ
ajst-20197	66	13	"	"	PUNCT
ajst-20197	66	14	internal	internal	ADJ
ajst-20197	66	15	gating	gate	VERB
ajst-20197	66	16	variables	variable	NOUN
ajst-20197	66	17	"	"	PUNCT
ajst-20197	66	18	that	that	PRON
ajst-20197	66	19	include	include	VERB
ajst-20197	66	20	one	one	NUM
ajst-20197	66	21	reset	reset	NOUN
ajst-20197	66	22	gate	gate	NOUN
ajst-20197	66	23	and	and	CCONJ
ajst-20197	66	24	one	one	NUM
ajst-20197	66	25	update	update	NOUN
ajst-20197	66	26	gate	gate	NOUN
ajst-20197	66	27	.	.	PUNCT
ajst-20197	67	1	a	a	DET
ajst-20197	67	2	reset	reset	ADJ
ajst-20197	67	3	gate	gate	NOUN
ajst-20197	67	4	permits	permit	VERB
ajst-20197	67	5	hidden	hide	VERB
ajst-20197	67	6	state	state	NOUN
ajst-20197	67	7	overwriting	overwriting	NOUN
ajst-20197	67	8	and	and	CCONJ
ajst-20197	67	9	thereby	thereby	ADV
ajst-20197	67	10	helps	help	VERB
ajst-20197	67	11	in	in	ADP
ajst-20197	67	12	developing	develop	VERB
ajst-20197	67	13	control	control	NOUN
ajst-20197	67	14	over	over	ADP
ajst-20197	67	15	the	the	DET
ajst-20197	67	16	interactions	interaction	NOUN
ajst-20197	67	17	developed	develop	VERB
ajst-20197	67	18	with	with	ADP
ajst-20197	67	19	an	an	DET
ajst-20197	67	20	input	input	NOUN
ajst-20197	67	21	.	.	PUNCT
ajst-20197	68	1	an	an	DET
ajst-20197	68	2	update	update	NOUN
ajst-20197	68	3	gate	gate	NOUN
ajst-20197	68	4	helps	help	VERB
ajst-20197	68	5	in	in	ADP
ajst-20197	68	6	safeguarding	safeguard	VERB
ajst-20197	68	7	the	the	DET
ajst-20197	68	8	d	d	ADJ
ajst-20197	68	9	-	-	ADJ
ajst-20197	68	10	dimensional	dimensional	ADJ
ajst-20197	68	11	"	"	PUNCT
ajst-20197	68	12	hidden	hidden	ADJ
ajst-20197	68	13	state	state	NOUN
ajst-20197	68	14	"	"	PUNCT
ajst-20197	68	15	.	.	PUNCT
ajst-20197	69	1	through	through	ADP
ajst-20197	69	2	the	the	DET
ajst-20197	69	3	implementation	implementation	NOUN
ajst-20197	69	4	of	of	ADP
ajst-20197	69	5	this	this	DET
ajst-20197	69	6	update	update	NOUN
ajst-20197	69	7	gate	gate	NOUN
ajst-20197	69	8	,	,	PUNCT
ajst-20197	69	9	there	there	PRON
ajst-20197	69	10	is	be	VERB
ajst-20197	69	11	formation	formation	NOUN
ajst-20197	69	12	of	of	ADP
ajst-20197	69	13	control	control	NOUN
ajst-20197	69	14	over	over	ADP
ajst-20197	69	15	different	different	ADJ
ajst-20197	69	16	dimensions	dimension	NOUN
ajst-20197	69	17	that	that	PRON
ajst-20197	69	18	are	be	AUX
ajst-20197	69	19	available	available	ADJ
ajst-20197	69	20	within	within	ADP
ajst-20197	69	21	a	a	DET
ajst-20197	69	22	"	"	PUNCT
ajst-20197	69	23	hidden	hidden	ADJ
ajst-20197	69	24	state	state	NOUN
ajst-20197	69	25	decay	decay	NOUN
ajst-20197	69	26	"	"	PUNCT
ajst-20197	69	27	and	and	CCONJ
ajst-20197	69	28	thereby	thereby	ADV
ajst-20197	69	29	help	help	VERB
ajst-20197	69	30	in	in	ADP
ajst-20197	69	31	the	the	DET
ajst-20197	69	32	formation	formation	NOUN
ajst-20197	69	33	of	of	ADP
ajst-20197	69	34	one	one	NUM
ajst-20197	69	35	adaptive	adaptive	ADJ
ajst-20197	69	36	"	"	PUNCT
ajst-20197	69	37	time	time	NOUN
ajst-20197	69	38	constant	constant	ADJ
ajst-20197	69	39	"	"	PUNCT
ajst-20197	69	40	for	for	ADP
ajst-20197	69	41	memory	memory	NOUN
ajst-20197	69	42	.	.	PUNCT
ajst-20197	70	1	therefore	therefore	ADV
ajst-20197	70	2	,	,	PUNCT
ajst-20197	70	3	the	the	DET
ajst-20197	70	4	argument	argument	NOUN
ajst-20197	70	5	could	could	AUX
ajst-20197	70	6	be	be	AUX
ajst-20197	70	7	made	make	VERB
ajst-20197	70	8	that	that	SCONJ
ajst-20197	70	9	there	there	PRON
ajst-20197	70	10	has	have	AUX
ajst-20197	70	11	been	be	AUX
ajst-20197	70	12	a	a	DET
ajst-20197	70	13	gradual	gradual	ADJ
ajst-20197	70	14	increase	increase	NOUN
ajst-20197	70	15	in	in	ADP
ajst-20197	70	16	the	the	DET
ajst-20197	70	17	model	model	NOUN
ajst-20197	70	18	performance	performance	NOUN
ajst-20197	70	19	through	through	ADP
ajst-20197	70	20	the	the	DET
ajst-20197	70	21	implementation	implementation	NOUN
ajst-20197	70	22	of	of	ADP
ajst-20197	70	23	differential	differential	ADJ
ajst-20197	70	24	equations	equation	NOUN
ajst-20197	70	25	in	in	ADP
ajst-20197	70	26	rnns	rnn	NOUN
ajst-20197	70	27	.	.	PUNCT
ajst-20197	71	1	this	this	PRON
ajst-20197	71	2	is	be	AUX
ajst-20197	71	3	effective	effective	ADJ
ajst-20197	71	4	in	in	ADP
ajst-20197	71	5	retrieving	retrieve	VERB
ajst-20197	71	6	and	and	CCONJ
ajst-20197	71	7	storing	store	VERB
ajst-20197	71	8	memory	memory	NOUN
ajst-20197	71	9	by	by	ADP
ajst-20197	71	10	using	use	VERB
ajst-20197	71	11	numerical	numerical	ADJ
ajst-20197	71	12	approximation	approximation	NOUN
ajst-20197	71	13	techniques	technique	NOUN
ajst-20197	71	14	in	in	ADP
ajst-20197	71	15	grus	grus	NOUN
ajst-20197	71	16	.	.	PUNCT
ajst-20197	72	1	3.2	3.2	NUM
ajst-20197	72	2	.	.	PUNCT
ajst-20197	73	1	long	long	ADJ
ajst-20197	73	2	short	short	ADJ
ajst-20197	73	3	-	-	PUNCT
ajst-20197	73	4	term	term	NOUN
ajst-20197	73	5	memory	memory	NOUN
ajst-20197	73	6	according	accord	VERB
ajst-20197	73	7	to	to	ADP
ajst-20197	73	8	[	[	PUNCT
ajst-20197	73	9	11]landi	11]landi	NUM
ajst-20197	73	10	et	et	NOUN
ajst-20197	73	11	al	al	PROPN
ajst-20197	73	12	.	.	PROPN
ajst-20197	73	13	2021	2021	NUM
ajst-20197	73	14	,	,	PUNCT
ajst-20197	73	15	there	there	PRON
ajst-20197	73	16	has	have	AUX
ajst-20197	73	17	been	be	AUX
ajst-20197	73	18	usage	usage	NOUN
ajst-20197	73	19	of	of	ADP
ajst-20197	73	20	lstms	lstms	NOUN
ajst-20197	73	21	that	that	PRON
ajst-20197	73	22	implements	implement	VERB
ajst-20197	73	23	rnns	rnn	NOUN
ajst-20197	73	24	for	for	ADP
ajst-20197	73	25	developing	develop	VERB
ajst-20197	73	26	gating	gate	VERB
ajst-20197	73	27	mechanisms	mechanism	NOUN
ajst-20197	73	28	figure	figure	VERB
ajst-20197	73	29	1	1	NUM
ajst-20197	73	30	.	.	PUNCT
ajst-20197	74	1	this	this	PRON
ajst-20197	74	2	is	be	AUX
ajst-20197	74	3	effective	effective	ADJ
ajst-20197	74	4	in	in	ADP
ajst-20197	74	5	enhancing	enhance	VERB
ajst-20197	74	6	the	the	DET
ajst-20197	74	7	model	model	NOUN
ajst-20197	74	8	performance	performance	NOUN
ajst-20197	74	9	by	by	ADP
ajst-20197	74	10	learning	learn	VERB
ajst-20197	74	11	different	different	ADJ
ajst-20197	74	12	long	long	ADJ
ajst-20197	74	13	-	-	PUNCT
ajst-20197	74	14	term	term	NOUN
ajst-20197	74	15	dependencies	dependency	NOUN
ajst-20197	74	16	by	by	ADP
ajst-20197	74	17	eliminating	eliminate	VERB
ajst-20197	74	18	vanishing	vanish	VERB
ajst-20197	74	19	and	and	CCONJ
ajst-20197	74	20	exploding	explode	VERB
ajst-20197	74	21	gradients	gradient	NOUN
ajst-20197	74	22	available	available	ADJ
ajst-20197	74	23	within	within	ADP
ajst-20197	74	24	a	a	DET
ajst-20197	74	25	dataset	dataset	NOUN
ajst-20197	74	26	.	.	PUNCT
ajst-20197	75	1	during	during	ADP
ajst-20197	75	2	a	a	DET
ajst-20197	75	3	forward	forward	ADJ
ajst-20197	75	4	pass	pass	NOUN
ajst-20197	75	5	,	,	PUNCT
ajst-20197	75	6	the	the	DET
ajst-20197	75	7	"	"	PUNCT
ajst-20197	75	8	long	long	ADJ
ajst-20197	75	9	-	-	PUNCT
ajst-20197	75	10	term	term	NOUN
ajst-20197	75	11	memory	memory	NOUN
ajst-20197	75	12	cell	cell	NOUN
ajst-20197	75	13	"	"	PUNCT
ajst-20197	75	14	is	be	AUX
ajst-20197	75	15	involved	involve	VERB
ajst-20197	75	16	in	in	ADP
ajst-20197	75	17	storing	store	VERB
ajst-20197	75	18	information	information	NOUN
ajst-20197	75	19	and	and	CCONJ
ajst-20197	75	20	thereby	thereby	ADV
ajst-20197	75	21	helps	help	VERB
ajst-20197	75	22	in	in	ADP
ajst-20197	75	23	the	the	DET
ajst-20197	75	24	accomplishment	accomplishment	NOUN
ajst-20197	75	25	of	of	ADP
ajst-20197	75	26	back	back	ADJ
ajst-20197	75	27	propagation	propagation	NOUN
ajst-20197	75	28	of	of	ADP
ajst-20197	75	29	one	one	NUM
ajst-20197	75	30	error	error	NOUN
ajst-20197	75	31	signal	signal	NOUN
ajst-20197	75	32	through	through	ADP
ajst-20197	75	33	one	one	NUM
ajst-20197	75	34	safe	safe	ADJ
ajst-20197	75	35	path	path	NOUN
ajst-20197	76	1	[	[	X
ajst-20197	76	2	19](ragab	19](ragab	NUM
ajst-20197	76	3	et	et	NOUN
ajst-20197	76	4	al	al	PROPN
ajst-20197	76	5	.	.	PROPN
ajst-20197	76	6	2020	2020	NUM
ajst-20197	76	7	)	)	PUNCT
ajst-20197	76	8	.	.	PUNCT
ajst-20197	77	1	there	there	PRON
ajst-20197	77	2	has	have	AUX
ajst-20197	77	3	been	be	AUX
ajst-20197	77	4	establishment	establishment	NOUN
ajst-20197	77	5	of	of	ADP
ajst-20197	77	6	“	"	PUNCT
ajst-20197	77	7	working	work	VERB
ajst-20197	77	8	memory	memory	NOUN
ajst-20197	77	9	connections	connection	NOUN
ajst-20197	77	10	”	"	PUNCT
ajst-20197	77	11	that	that	PRON
ajst-20197	77	12	helps	help	VERB
ajst-20197	77	13	in	in	ADP
ajst-20197	77	14	enhancing	enhance	VERB
ajst-20197	77	15	the	the	DET
ajst-20197	77	16	gates	gate	NOUN
ajst-20197	77	17	’	'	PUNCT
ajst-20197	77	18	values	value	NOUN
ajst-20197	77	19	under	under	ADP
ajst-20197	77	20	the	the	DET
ajst-20197	77	21	influence	influence	NOUN
ajst-20197	77	22	of	of	ADP
ajst-20197	77	23	memory	memory	NOUN
ajst-20197	77	24	cells	cell	NOUN
ajst-20197	77	25	through	through	ADP
ajst-20197	77	26	recurrent	recurrent	ADJ
ajst-20197	77	27	weights	weight	NOUN
ajst-20197	77	28	.	.	PUNCT
ajst-20197	78	1	a	a	DET
ajst-20197	78	2	constant	constant	ADJ
ajst-20197	78	3	“	"	PUNCT
ajst-20197	78	4	error	error	NOUN
ajst-20197	78	5	paths	path	NOUN
ajst-20197	78	6	are	be	AUX
ajst-20197	78	7	developed	develop	VERB
ajst-20197	78	8	within	within	ADP
ajst-20197	78	9	subsequent	subsequent	ADJ
ajst-20197	78	10	time	time	NOUN
ajst-20197	78	11	steps	step	NOUN
ajst-20197	78	12	”	"	PUNCT
ajst-20197	78	13	through	through	ADP
ajst-20197	78	14	the	the	DET
ajst-20197	78	15	implementation	implementation	NOUN
ajst-20197	78	16	of	of	ADP
ajst-20197	78	17	lstm	lstm	NOUN
ajst-20197	78	18	networks	network	NOUN
ajst-20197	78	19	.	.	PUNCT
ajst-20197	79	1	figure	figure	NOUN
ajst-20197	79	2	1	1	NUM
ajst-20197	79	3	.	.	PUNCT
ajst-20197	79	4	protection	protection	PROPN
ajst-20197	79	5	mechanism(source	mechanism(source	PROPN
ajst-20197	79	6	:	:	PUNCT
ajst-20197	80	1	baraldi	baraldi	PROPN
ajst-20197	80	2	et	et	PROPN
ajst-20197	80	3	al	al	PROPN
ajst-20197	80	4	.	.	PROPN
ajst-20197	80	5	2021	2021	NUM
ajst-20197	80	6	)	)	PUNCT
ajst-20197	80	7	baraldi	baraldi	PROPN
ajst-20197	80	8	et	et	PROPN
ajst-20197	80	9	al	al	PROPN
ajst-20197	80	10	.	.	PROPN
ajst-20197	80	11	2021	2021	NUM
ajst-20197	80	12	highlights	highlight	NOUN
ajst-20197	80	13	that	that	PRON
ajst-20197	80	14	a	a	DET
ajst-20197	80	15	protection	protection	NOUN
ajst-20197	80	16	mechanism	mechanism	NOUN
ajst-20197	80	17	is	be	AUX
ajst-20197	80	18	developed	develop	VERB
ajst-20197	80	19	through	through	ADP
ajst-20197	80	20	the	the	DET
ajst-20197	80	21	implementation	implementation	NOUN
ajst-20197	80	22	of	of	ADP
ajst-20197	80	23	lstm	lstm	NOUN
ajst-20197	80	24	networks	network	NOUN
ajst-20197	80	25	.	.	PUNCT
ajst-20197	81	1	this	this	PRON
ajst-20197	81	2	helps	help	VERB
ajst-20197	81	3	in	in	ADP
ajst-20197	81	4	the	the	DET
ajst-20197	81	5	development	development	NOUN
ajst-20197	81	6	of	of	ADP
ajst-20197	81	7	a	a	DET
ajst-20197	81	8	connection	connection	NOUN
ajst-20197	81	9	within	within	ADP
ajst-20197	81	10	gates	gate	NOUN
ajst-20197	81	11	and	and	CCONJ
ajst-20197	81	12	memory	memory	NOUN
ajst-20197	81	13	cells	cell	NOUN
ajst-20197	81	14	through	through	ADP
ajst-20197	81	15	the	the	DET
ajst-20197	81	16	generation	generation	NOUN
ajst-20197	81	17	of	of	ADP
ajst-20197	81	18	one	one	NUM
ajst-20197	81	19	“	"	PUNCT
ajst-20197	81	20	non	non	ADJ
ajst-20197	81	21	-	-	ADJ
ajst-20197	81	22	linear	linear	ADJ
ajst-20197	81	23	activation	activation	NOUN
ajst-20197	81	24	function	function	NOUN
ajst-20197	81	25	”	"	PUNCT
ajst-20197	81	26	.	.	PUNCT
ajst-20197	82	1	information	information	NOUN
ajst-20197	82	2	accessibility	accessibility	NOUN
ajst-20197	82	3	is	be	AUX
ajst-20197	82	4	raised	raise	VERB
ajst-20197	82	5	through	through	ADP
ajst-20197	82	6	the	the	DET
ajst-20197	82	7	consideration	consideration	NOUN
ajst-20197	82	8	of	of	ADP
ajst-20197	82	9	one	one	NUM
ajst-20197	82	10	lstm	lstm	ADJ
ajst-20197	82	11	cell	cell	NOUN
ajst-20197	82	12	that	that	PRON
ajst-20197	82	13	uses	use	VERB
ajst-20197	82	14	“	"	PUNCT
ajst-20197	82	15	working	work	VERB
ajst-20197	82	16	memory	memory	NOUN
ajst-20197	82	17	connections	connection	NOUN
ajst-20197	82	18	”	"	PUNCT
ajst-20197	82	19	for	for	ADP
ajst-20197	82	20	developing	develop	VERB
ajst-20197	82	21	a	a	DET
ajst-20197	82	22	suitable	suitable	ADJ
ajst-20197	82	23	access	access	NOUN
ajst-20197	82	24	within	within	ADP
ajst-20197	82	25	a	a	DET
ajst-20197	82	26	gate	gate	NOUN
ajst-20197	82	27	structure	structure	NOUN
ajst-20197	82	28	.	.	PUNCT
ajst-20197	83	1	the	the	DET
ajst-20197	83	2	improvements	improvement	NOUN
ajst-20197	83	3	that	that	PRON
ajst-20197	83	4	are	be	AUX
ajst-20197	83	5	made	make	VERB
ajst-20197	83	6	through	through	ADP
ajst-20197	83	7	the	the	DET
ajst-20197	83	8	inclusion	inclusion	NOUN
ajst-20197	83	9	of	of	ADP
ajst-20197	83	10	wmcs	wmcs	NOUN
ajst-20197	83	11	are	be	AUX
ajst-20197	83	12	increase	increase	NOUN
ajst-20197	83	13	in	in	ADP
ajst-20197	83	14	gate	gate	NOUN
ajst-20197	83	15	controls	control	NOUN
ajst-20197	83	16	and	and	CCONJ
ajst-20197	83	17	rise	rise	VERB
ajst-20197	83	18	in	in	ADP
ajst-20197	83	19	stability	stability	NOUN
ajst-20197	83	20	for	for	ADP
ajst-20197	83	21	training	training	NOUN
ajst-20197	83	22	function	function	NOUN
ajst-20197	83	23	.	.	PUNCT
ajst-20197	84	1	according	accord	VERB
ajst-20197	84	2	to	to	ADP
ajst-20197	84	3	le	le	X
ajst-20197	84	4	2021	2021	NUM
ajst-20197	84	5	,	,	PUNCT
ajst-20197	84	6	there	there	PRON
ajst-20197	84	7	is	be	VERB
ajst-20197	84	8	availability	availability	NOUN
ajst-20197	84	9	of	of	ADP
ajst-20197	84	10	a	a	DET
ajst-20197	84	11	mathematical	mathematical	ADJ
ajst-20197	84	12	challenge	challenge	NOUN
ajst-20197	84	13	of	of	ADP
ajst-20197	84	14	rnns	rnn	NOUN
ajst-20197	84	15	in	in	ADP
ajst-20197	84	16	reading	read	VERB
ajst-20197	84	17	long	long	ADJ
ajst-20197	84	18	sequences	sequence	NOUN
ajst-20197	84	19	.	.	PUNCT
ajst-20197	85	1	this	this	PRON
ajst-20197	85	2	is	be	AUX
ajst-20197	85	3	because	because	SCONJ
ajst-20197	85	4	the	the	DET
ajst-20197	85	5	gradients	gradient	NOUN
ajst-20197	85	6	either	either	CCONJ
ajst-20197	85	7	explode	explode	VERB
ajst-20197	85	8	or	or	CCONJ
ajst-20197	85	9	vanish	vanish	VERB
ajst-20197	85	10	while	while	SCONJ
ajst-20197	85	11	accomplishing	accomplish	VERB
ajst-20197	85	12	step	step	NOUN
ajst-20197	85	13	propagation	propagation	NOUN
ajst-20197	85	14	.	.	PUNCT
ajst-20197	86	1	figure	figure	NOUN
ajst-20197	86	2	2	2	NUM
ajst-20197	86	3	the	the	DET
ajst-20197	86	4	complexity	complexity	NOUN
ajst-20197	86	5	is	be	AUX
ajst-20197	86	6	developed	develop	VERB
ajst-20197	86	7	through	through	ADP
ajst-20197	86	8	the	the	DET
ajst-20197	86	9	allocation	allocation	NOUN
ajst-20197	86	10	of	of	ADP
ajst-20197	86	11	smaller	small	ADJ
ajst-20197	86	12	weights	weight	NOUN
ajst-20197	86	13	for	for	ADP
ajst-20197	86	14	accomplishing	accomplish	VERB
ajst-20197	86	15	long	long	ADJ
ajst-20197	86	16	-	-	PUNCT
ajst-20197	86	17	term	term	NOUN
ajst-20197	86	18	interactions	interaction	NOUN
ajst-20197	86	19	that	that	PRON
ajst-20197	86	20	are	be	AUX
ajst-20197	86	21	linked	link	VERB
ajst-20197	86	22	with	with	ADP
ajst-20197	86	23	different	different	ADJ
ajst-20197	86	24	long	long	ADJ
ajst-20197	86	25	-	-	PUNCT
ajst-20197	86	26	term	term	NOUN
ajst-20197	86	27	dependencies	dependency	NOUN
ajst-20197	86	28	.	.	PUNCT
ajst-20197	87	1	this	this	DET
ajst-20197	87	2	issue	issue	NOUN
ajst-20197	87	3	is	be	AUX
ajst-20197	87	4	resolved	resolve	VERB
ajst-20197	87	5	through	through	ADP
ajst-20197	87	6	the	the	DET
ajst-20197	87	7	inclusion	inclusion	NOUN
ajst-20197	87	8	of	of	ADP
ajst-20197	87	9	lstm	lstm	NOUN
ajst-20197	87	10	that	that	PRON
ajst-20197	87	11	helps	help	VERB
ajst-20197	87	12	in	in	ADP
ajst-20197	87	13	the	the	DET
ajst-20197	87	14	addition	addition	NOUN
ajst-20197	87	15	of	of	ADP
ajst-20197	87	16	one	one	NUM
ajst-20197	87	17	“	"	PUNCT
ajst-20197	87	18	linear	linear	ADJ
ajst-20197	87	19	self	self	NOUN
ajst-20197	87	20	-	-	PUNCT
ajst-20197	87	21	loop	loop	NOUN
ajst-20197	87	22	memory	memory	NOUN
ajst-20197	87	23	cell	cell	NOUN
ajst-20197	87	24	”	"	PUNCT
ajst-20197	87	25	for	for	ADP
ajst-20197	87	26	computing	compute	VERB
ajst-20197	87	27	one	one	NUM
ajst-20197	87	28	hidden	hide	VERB
ajst-20197	87	29	unit	unit	NOUN
ajst-20197	87	30	.	.	PUNCT
ajst-20197	88	1	figure	figure	NOUN
ajst-20197	88	2	2	2	NUM
ajst-20197	88	3	.	.	PUNCT
ajst-20197	88	4	lstm	lstm	ADJ
ajst-20197	88	5	unit	unit	NOUN
ajst-20197	89	1	[	[	X
ajst-20197	89	2	13](source	13](source	NUM
ajst-20197	89	3	:	:	PUNCT
ajst-20197	89	4	le	le	X
ajst-20197	89	5	2021	2021	NUM
ajst-20197	89	6	)	)	PUNCT
ajst-20197	89	7	the	the	DET
ajst-20197	89	8	weight	weight	NOUN
ajst-20197	89	9	associated	associate	VERB
ajst-20197	89	10	with	with	ADP
ajst-20197	89	11	this	this	DET
ajst-20197	89	12	memory	memory	NOUN
ajst-20197	89	13	cell	cell	NOUN
ajst-20197	89	14	is	be	AUX
ajst-20197	89	15	gated	gate	VERB
ajst-20197	89	16	and	and	CCONJ
ajst-20197	89	17	directly	directly	ADV
ajst-20197	89	18	depends	depend	VERB
ajst-20197	89	19	over	over	ADP
ajst-20197	89	20	an	an	DET
ajst-20197	89	21	input	input	NOUN
ajst-20197	89	22	.	.	PUNCT
ajst-20197	90	1	as	as	ADP
ajst-20197	90	2	a	a	DET
ajst-20197	90	3	result	result	NOUN
ajst-20197	90	4	,	,	PUNCT
ajst-20197	90	5	dynamic	dynamic	ADJ
ajst-20197	90	6	111	111	NUM
ajst-20197	90	7	moderation	moderation	NOUN
ajst-20197	90	8	is	be	AUX
ajst-20197	90	9	developed	develop	VERB
ajst-20197	90	10	within	within	ADP
ajst-20197	90	11	the	the	DET
ajst-20197	90	12	network	network	NOUN
ajst-20197	90	13	for	for	ADP
ajst-20197	90	14	controlling	control	VERB
ajst-20197	90	15	the	the	DET
ajst-20197	90	16	data	datum	NOUN
ajst-20197	90	17	passing	pass	VERB
ajst-20197	90	18	through	through	ADP
ajst-20197	90	19	one	one	NUM
ajst-20197	90	20	hidden	hide	VERB
ajst-20197	90	21	unit	unit	NOUN
ajst-20197	90	22	.	.	PUNCT
ajst-20197	91	1	figure	figure	VERB
ajst-20197	91	2	3	3	NUM
ajst-20197	91	3	shows	show	VERB
ajst-20197	91	4	that	that	SCONJ
ajst-20197	91	5	there	there	PRON
ajst-20197	91	6	has	have	AUX
ajst-20197	91	7	been	be	AUX
ajst-20197	91	8	inclusion	inclusion	NOUN
ajst-20197	91	9	of	of	ADP
ajst-20197	91	10	different	different	ADJ
ajst-20197	91	11	additional	additional	ADJ
ajst-20197	91	12	parameters	parameter	NOUN
ajst-20197	91	13	to	to	PART
ajst-20197	91	14	achieve	achieve	VERB
ajst-20197	91	15	vector	vector	NOUN
ajst-20197	91	16	representation	representation	NOUN
ajst-20197	91	17	.	.	PUNCT
ajst-20197	92	1	figure	figure	NOUN
ajst-20197	92	2	3	3	NUM
ajst-20197	92	3	.	.	PUNCT
ajst-20197	92	4	hidden	hide	VERB
ajst-20197	92	5	state	state	NOUN
ajst-20197	92	6	computation	computation	NOUN
ajst-20197	92	7	of	of	ADP
ajst-20197	92	8	forget	forget	PROPN
ajst-20197	92	9	gate	gate	NOUN
ajst-20197	93	1	[	[	X
ajst-20197	93	2	13](source	13](source	NUM
ajst-20197	93	3	:	:	PUNCT
ajst-20197	93	4	le	le	X
ajst-20197	93	5	2021	2021	NUM
ajst-20197	93	6	)	)	PUNCT
ajst-20197	93	7	the	the	DET
ajst-20197	93	8	“	"	PUNCT
ajst-20197	93	9	forget	forget	NOUN
ajst-20197	93	10	gate	gate	NOUN
ajst-20197	93	11	”	"	PUNCT
ajst-20197	93	12	is	be	AUX
ajst-20197	93	13	involved	involve	VERB
ajst-20197	93	14	in	in	ADP
ajst-20197	93	15	capturing	capture	VERB
ajst-20197	93	16	long	long	ADJ
ajst-20197	93	17	-	-	PUNCT
ajst-20197	93	18	term	term	NOUN
ajst-20197	93	19	dependencies	dependency	NOUN
ajst-20197	93	20	by	by	ADP
ajst-20197	93	21	allowing	allow	VERB
ajst-20197	93	22	the	the	DET
ajst-20197	93	23	network	network	NOUN
ajst-20197	93	24	to	to	PART
ajst-20197	93	25	preserve	preserve	VERB
ajst-20197	93	26	the	the	DET
ajst-20197	93	27	previous	previous	ADJ
ajst-20197	93	28	memory	memory	NOUN
ajst-20197	93	29	.	.	PUNCT
ajst-20197	94	1	as	as	ADP
ajst-20197	94	2	a	a	DET
ajst-20197	94	3	result	result	NOUN
ajst-20197	94	4	,	,	PUNCT
ajst-20197	94	5	there	there	PRON
ajst-20197	94	6	is	be	VERB
ajst-20197	94	7	formation	formation	NOUN
ajst-20197	94	8	of	of	ADP
ajst-20197	94	9	backpropagation	backpropagation	NOUN
ajst-20197	94	10	for	for	ADP
ajst-20197	94	11	one	one	NUM
ajst-20197	94	12	distant	distant	ADJ
ajst-20197	94	13	input	input	NOUN
ajst-20197	94	14	through	through	ADP
ajst-20197	94	15	the	the	DET
ajst-20197	94	16	complement	complement	NOUN
ajst-20197	94	17	elimination	elimination	NOUN
ajst-20197	94	18	of	of	ADP
ajst-20197	94	19	the	the	DET
ajst-20197	94	20	“	"	PUNCT
ajst-20197	94	21	gradient	gradient	ADJ
ajst-20197	94	22	vanishing	vanish	VERB
ajst-20197	94	23	process	process	NOUN
ajst-20197	94	24	”	"	PUNCT
ajst-20197	94	25	.	.	PUNCT
ajst-20197	95	1	3.3	3.3	NUM
ajst-20197	95	2	.	.	PUNCT
ajst-20197	95	3	memory	memory	NOUN
ajst-20197	95	4	augmented	augment	VERB
ajst-20197	95	5	neural	neural	ADJ
ajst-20197	95	6	networks	network	NOUN
ajst-20197	95	7	according	accord	VERB
ajst-20197	95	8	to	to	ADP
ajst-20197	95	9	park	park	NOUN
ajst-20197	95	10	et	et	PROPN
ajst-20197	95	11	al	al	PROPN
ajst-20197	95	12	.	.	PROPN
ajst-20197	95	13	2018	2018	NUM
ajst-20197	95	14	,	,	PUNCT
ajst-20197	95	15	manns	mann	NOUN
ajst-20197	95	16	involve	involve	VERB
ajst-20197	95	17	external	external	ADJ
ajst-20197	95	18	networks	network	NOUN
ajst-20197	95	19	like	like	ADP
ajst-20197	95	20	memory	memory	NOUN
ajst-20197	95	21	networks	network	NOUN
ajst-20197	95	22	and	and	CCONJ
ajst-20197	95	23	turing	ture	VERB
ajst-20197	95	24	machines	machine	NOUN
ajst-20197	95	25	that	that	PRON
ajst-20197	95	26	helps	help	VERB
ajst-20197	95	27	in	in	ADP
ajst-20197	95	28	increasing	increase	VERB
ajst-20197	95	29	the	the	DET
ajst-20197	95	30	performance	performance	NOUN
ajst-20197	95	31	of	of	ADP
ajst-20197	95	32	crnns	crnn	NOUN
ajst-20197	95	33	through	through	ADP
ajst-20197	95	34	the	the	DET
ajst-20197	95	35	increase	increase	NOUN
ajst-20197	95	36	in	in	ADP
ajst-20197	95	37	learning	learn	VERB
ajst-20197	95	38	capabilities	capability	NOUN
ajst-20197	95	39	for	for	ADP
ajst-20197	95	40	long	long	ADJ
ajst-20197	95	41	-	-	PUNCT
ajst-20197	95	42	term	term	NOUN
ajst-20197	95	43	dependencies	dependency	NOUN
ajst-20197	95	44	.	.	PUNCT
ajst-20197	96	1	there	there	PRON
ajst-20197	96	2	is	be	VERB
ajst-20197	96	3	availability	availability	NOUN
ajst-20197	96	4	of	of	ADP
ajst-20197	96	5	two	two	NUM
ajst-20197	96	6	components	component	NOUN
ajst-20197	96	7	in	in	ADP
ajst-20197	96	8	manns	mann	NOUN
ajst-20197	96	9	that	that	PRON
ajst-20197	96	10	help	help	VERB
ajst-20197	96	11	in	in	ADP
ajst-20197	96	12	increasing	increase	VERB
ajst-20197	96	13	the	the	DET
ajst-20197	96	14	model	model	NOUN
ajst-20197	96	15	performance	performance	NOUN
ajst-20197	96	16	.	.	PUNCT
ajst-20197	97	1	these	these	PRON
ajst-20197	97	2	include	include	VERB
ajst-20197	97	3	external	external	ADJ
ajst-20197	97	4	memory	memory	NOUN
ajst-20197	97	5	and	and	CCONJ
ajst-20197	97	6	memory	memory	NOUN
ajst-20197	97	7	controllers	controller	NOUN
ajst-20197	97	8	.	.	PUNCT
ajst-20197	98	1	manns	manns	PROPN
ajst-20197	98	2	help	help	VERB
ajst-20197	98	3	in	in	ADP
ajst-20197	98	4	writing	write	VERB
ajst-20197	98	5	and	and	CCONJ
ajst-20197	98	6	reading	read	VERB
ajst-20197	98	7	memory	memory	NOUN
ajst-20197	98	8	by	by	ADP
ajst-20197	98	9	considering	consider	VERB
ajst-20197	98	10	the	the	DET
ajst-20197	98	11	information	information	NOUN
ajst-20197	98	12	.	.	PUNCT
ajst-20197	99	1	the	the	DET
ajst-20197	99	2	memory	memory	NOUN
ajst-20197	99	3	addressing	address	VERB
ajst-20197	99	4	feature	feature	NOUN
ajst-20197	99	5	plays	play	VERB
ajst-20197	99	6	a	a	DET
ajst-20197	99	7	critical	critical	ADJ
ajst-20197	99	8	role	role	NOUN
ajst-20197	99	9	in	in	ADP
ajst-20197	99	10	interference	interference	NOUN
ajst-20197	99	11	and	and	CCONJ
ajst-20197	99	12	training	training	NOUN
ajst-20197	99	13	.	.	PUNCT
ajst-20197	100	1	content	content	NOUN
ajst-20197	100	2	-	-	PUNCT
ajst-20197	100	3	based	base	VERB
ajst-20197	100	4	addressing[17](source	addressing[17](source	NOUN
ajst-20197	100	5	:	:	PUNCT
ajst-20197	100	6	park	park	NOUN
ajst-20197	100	7	et	et	PROPN
ajst-20197	100	8	al	al	PROPN
ajst-20197	100	9	.	.	PROPN
ajst-20197	100	10	2018	2018	NUM
ajst-20197	100	11	)	)	PUNCT
ajst-20197	101	1	there	there	PRON
ajst-20197	101	2	has	have	AUX
ajst-20197	101	3	been	be	AUX
ajst-20197	101	4	implementation	implementation	NOUN
ajst-20197	101	5	of	of	ADP
ajst-20197	101	6	content	content	NOUN
ajst-20197	101	7	-	-	PUNCT
ajst-20197	101	8	based	base	VERB
ajst-20197	101	9	addressing	addressing	NOUN
ajst-20197	101	10	by	by	ADP
ajst-20197	101	11	manns	mann	NOUN
ajst-20197	101	12	that	that	PRON
ajst-20197	101	13	helps	help	VERB
ajst-20197	101	14	in	in	ADP
ajst-20197	101	15	increasing	increase	VERB
ajst-20197	101	16	the	the	DET
ajst-20197	101	17	performance	performance	NOUN
ajst-20197	101	18	of	of	ADP
ajst-20197	101	19	rnns	rnn	NOUN
ajst-20197	101	20	.	.	PUNCT
ajst-20197	102	1	this	this	PRON
ajst-20197	102	2	is	be	AUX
ajst-20197	102	3	accomplished	accomplish	VERB
ajst-20197	102	4	through	through	ADP
ajst-20197	102	5	the	the	DET
ajst-20197	102	6	consideration	consideration	NOUN
ajst-20197	102	7	of	of	ADP
ajst-20197	102	8	a	a	DET
ajst-20197	102	9	“	"	PUNCT
ajst-20197	102	10	similarity	similarity	NOUN
ajst-20197	102	11	measure	measure	NOUN
ajst-20197	102	12	”	"	PUNCT
ajst-20197	102	13	that	that	PRON
ajst-20197	102	14	is	be	AUX
ajst-20197	102	15	represented	represent	VERB
ajst-20197	102	16	as	as	ADP
ajst-20197	102	17	s	s	NOUN
ajst-20197	102	18	that	that	PRON
ajst-20197	102	19	helps	help	VERB
ajst-20197	102	20	in	in	ADP
ajst-20197	102	21	representing	represent	VERB
ajst-20197	102	22	the	the	DET
ajst-20197	102	23	cosine	cosine	NOUN
ajst-20197	102	24	similarity	similarity	NOUN
ajst-20197	102	25	within	within	ADP
ajst-20197	102	26	one	one	NUM
ajst-20197	102	27	key	key	ADJ
ajst-20197	102	28	vector	vector	NOUN
ajst-20197	102	29	and	and	CCONJ
ajst-20197	102	30	a	a	DET
ajst-20197	102	31	memory	memory	NOUN
ajst-20197	102	32	element	element	NOUN
ajst-20197	102	33	.	.	PUNCT
ajst-20197	103	1	the	the	DET
ajst-20197	103	2	normalization	normalization	NOUN
ajst-20197	103	3	is	be	AUX
ajst-20197	103	4	accomplished	accomplish	VERB
ajst-20197	103	5	through	through	ADP
ajst-20197	103	6	the	the	DET
ajst-20197	103	7	consideration	consideration	NOUN
ajst-20197	103	8	of	of	ADP
ajst-20197	103	9	one	one	NUM
ajst-20197	103	10	softmax	softmax	NOUN
ajst-20197	103	11	function	function	NOUN
ajst-20197	103	12	.	.	PUNCT
ajst-20197	104	1	due	due	ADP
ajst-20197	104	2	to	to	ADP
ajst-20197	104	3	the	the	DET
ajst-20197	104	4	availability	availability	NOUN
ajst-20197	104	5	of	of	ADP
ajst-20197	104	6	differentiable	differentiable	ADJ
ajst-20197	104	7	components	component	NOUN
ajst-20197	104	8	within	within	ADP
ajst-20197	104	9	softmax	softmax	PROPN
ajst-20197	104	10	and	and	CCONJ
ajst-20197	104	11	cosine	cosine	NOUN
ajst-20197	104	12	similarity	similarity	NOUN
ajst-20197	104	13	,	,	PUNCT
ajst-20197	104	14	the	the	DET
ajst-20197	104	15	gradient	gradient	NOUN
ajst-20197	104	16	backpropagation	backpropagation	NOUN
ajst-20197	104	17	is	be	AUX
ajst-20197	104	18	implemented	implement	VERB
ajst-20197	104	19	by	by	ADP
ajst-20197	104	20	mann	mann	PROPN
ajst-20197	104	21	for	for	ADP
ajst-20197	104	22	establishing	establish	VERB
ajst-20197	104	23	an	an	DET
ajst-20197	104	24	end	end	NOUN
ajst-20197	104	25	-	-	PUNCT
ajst-20197	104	26	to	to	ADP
ajst-20197	104	27	-	-	PUNCT
ajst-20197	104	28	end	end	NOUN
ajst-20197	104	29	trainable	trainable	ADJ
ajst-20197	104	30	feature	feature	NOUN
ajst-20197	104	31	.	.	PUNCT
ajst-20197	105	1	kim	kim	PROPN
ajst-20197	105	2	et	et	PROPN
ajst-20197	105	3	al	al	PROPN
ajst-20197	105	4	.	.	PROPN
ajst-20197	105	5	2018	2018	NUM
ajst-20197	105	6	states	state	VERB
ajst-20197	105	7	that	that	SCONJ
ajst-20197	105	8	binary	binary	NOUN
ajst-20197	105	9	and	and	CCONJ
ajst-20197	105	10	fixed	fix	VERB
ajst-20197	105	11	-	-	PUNCT
ajst-20197	105	12	point	point	NOUN
ajst-20197	105	13	quantization	quantization	NOUN
ajst-20197	105	14	is	be	AUX
ajst-20197	105	15	done	do	VERB
ajst-20197	105	16	for	for	ADP
ajst-20197	105	17	developing	develop	VERB
ajst-20197	105	18	accountability	accountability	NOUN
ajst-20197	105	19	over	over	ADP
ajst-20197	105	20	the	the	DET
ajst-20197	105	21	“	"	PUNCT
ajst-20197	105	22	quantization	quantization	NOUN
ajst-20197	105	23	error	error	NOUN
ajst-20197	105	24	”	"	PUNCT
ajst-20197	105	25	.	.	PUNCT
ajst-20197	106	1	there	there	PRON
ajst-20197	106	2	has	have	AUX
ajst-20197	106	3	been	be	AUX
ajst-20197	106	4	implementation	implementation	NOUN
ajst-20197	106	5	of	of	ADP
ajst-20197	106	6	xnornet	xnornet	NOUN
ajst-20197	106	7	and	and	CCONJ
ajst-20197	106	8	binarynet	binarynet	NOUN
ajst-20197	106	9	in	in	ADP
ajst-20197	106	10	cnn	cnn	PROPN
ajst-20197	106	11	for	for	ADP
ajst-20197	106	12	“	"	PUNCT
ajst-20197	106	13	binary	binary	ADJ
ajst-20197	106	14	quantization	quantization	NOUN
ajst-20197	106	15	”	"	PUNCT
ajst-20197	106	16	of	of	ADP
ajst-20197	106	17	activations	activation	NOUN
ajst-20197	106	18	and	and	CCONJ
ajst-20197	106	19	parameters	parameter	NOUN
ajst-20197	107	1	[	[	X
ajst-20197	107	2	25](weng	25](weng	NUM
ajst-20197	107	3	2021	2021	NUM
ajst-20197	107	4	)	)	PUNCT
ajst-20197	107	5	.	.	PUNCT
ajst-20197	108	1	this	this	PRON
ajst-20197	108	2	is	be	AUX
ajst-20197	108	3	effective	effective	ADJ
ajst-20197	108	4	in	in	ADP
ajst-20197	108	5	minimizing	minimize	VERB
ajst-20197	108	6	energy	energy	NOUN
ajst-20197	108	7	consumption	consumption	NOUN
ajst-20197	108	8	.	.	PUNCT
ajst-20197	109	1	binarynet	binarynet	NOUN
ajst-20197	109	2	has	have	AUX
ajst-20197	109	3	been	be	AUX
ajst-20197	109	4	implemented	implement	VERB
ajst-20197	109	5	over	over	ADP
ajst-20197	109	6	cifar10	cifar10	NOUN
ajst-20197	109	7	and	and	CCONJ
ajst-20197	109	8	mnist	mnist	PROPN
ajst-20197	109	9	datasets	dataset	NOUN
ajst-20197	109	10	.	.	PUNCT
ajst-20197	110	1	information	information	NOUN
ajst-20197	110	2	loss	loss	NOUN
ajst-20197	110	3	is	be	AUX
ajst-20197	110	4	compensated	compensate	VERB
ajst-20197	110	5	through	through	ADP
ajst-20197	110	6	the	the	DET
ajst-20197	110	7	use	use	NOUN
ajst-20197	110	8	of	of	ADP
ajst-20197	110	9	xnor	xnor	NOUN
ajst-20197	110	10	-	-	NOUN
ajst-20197	110	11	net	net	NOUN
ajst-20197	110	12	that	that	PRON
ajst-20197	110	13	helps	help	VERB
ajst-20197	110	14	in	in	ADP
ajst-20197	110	15	the	the	DET
ajst-20197	110	16	adjustment	adjustment	NOUN
ajst-20197	110	17	of	of	ADP
ajst-20197	110	18	layer	layer	NOUN
ajst-20197	110	19	positions	position	NOUN
ajst-20197	110	20	through	through	ADP
ajst-20197	110	21	the	the	DET
ajst-20197	110	22	consideration	consideration	NOUN
ajst-20197	110	23	of	of	ADP
ajst-20197	110	24	one	one	NUM
ajst-20197	110	25	scale	scale	NOUN
ajst-20197	110	26	factor	factor	NOUN
ajst-20197	110	27	.	.	PUNCT
ajst-20197	111	1	this	this	PRON
ajst-20197	111	2	is	be	AUX
ajst-20197	111	3	effective	effective	ADJ
ajst-20197	111	4	in	in	ADP
ajst-20197	111	5	increasing	increase	VERB
ajst-20197	111	6	the	the	DET
ajst-20197	111	7	error	error	NOUN
ajst-20197	111	8	rates	rate	NOUN
ajst-20197	111	9	by	by	ADP
ajst-20197	111	10	at	at	ADV
ajst-20197	111	11	least	least	ADV
ajst-20197	111	12	50	50	NUM
ajst-20197	111	13	%	%	NOUN
ajst-20197	111	14	to	to	PART
ajst-20197	111	15	diagnose	diagnose	VERB
ajst-20197	111	16	the	the	DET
ajst-20197	111	17	layer	layer	NOUN
ajst-20197	111	18	positions	position	NOUN
ajst-20197	111	19	available	available	ADJ
ajst-20197	111	20	within	within	ADP
ajst-20197	111	21	image	image	NOUN
ajst-20197	111	22	datasets	dataset	NOUN
ajst-20197	111	23	.	.	PUNCT
ajst-20197	112	1	3.4	3.4	NUM
ajst-20197	112	2	.	.	PUNCT
ajst-20197	112	3	attention	attention	NOUN
ajst-20197	112	4	mechanisms	mechanism	NOUN
ajst-20197	112	5	3.4.1	3.4.1	NUM
ajst-20197	112	6	.	.	PUNCT
ajst-20197	112	7	temporal	temporal	ADJ
ajst-20197	112	8	and	and	CCONJ
ajst-20197	112	9	spatial	spatial	ADJ
ajst-20197	112	10	attention	attention	NOUN
ajst-20197	112	11	mechanisms	mechanism	NOUN
ajst-20197	112	12	according	accord	VERB
ajst-20197	112	13	to	to	ADP
ajst-20197	112	14	[	[	X
ajst-20197	112	15	5]guo	5]guo	NUM
ajst-20197	112	16	et	et	NOUN
ajst-20197	112	17	al	al	PROPN
ajst-20197	112	18	.	.	PROPN
ajst-20197	112	19	2022	2022	NUM
ajst-20197	112	20	,	,	PUNCT
ajst-20197	112	21	there	there	PRON
ajst-20197	112	22	has	have	AUX
ajst-20197	112	23	been	be	AUX
ajst-20197	112	24	a	a	DET
ajst-20197	112	25	considerable	considerable	ADJ
ajst-20197	112	26	increase	increase	NOUN
ajst-20197	112	27	in	in	ADP
ajst-20197	112	28	computer	computer	NOUN
ajst-20197	112	29	vision	vision	NOUN
ajst-20197	112	30	through	through	ADP
ajst-20197	112	31	the	the	DET
ajst-20197	112	32	consideration	consideration	NOUN
ajst-20197	112	33	of	of	ADP
ajst-20197	112	34	temporal	temporal	ADJ
ajst-20197	112	35	and	and	CCONJ
ajst-20197	112	36	spatial	spatial	ADJ
ajst-20197	112	37	mechanisms	mechanism	NOUN
ajst-20197	112	38	that	that	PRON
ajst-20197	112	39	help	help	VERB
ajst-20197	112	40	in	in	ADP
ajst-20197	112	41	the	the	DET
ajst-20197	112	42	appropriate	appropriate	ADJ
ajst-20197	112	43	recognition	recognition	NOUN
ajst-20197	112	44	of	of	ADP
ajst-20197	112	45	human	human	ADJ
ajst-20197	112	46	actions	action	NOUN
ajst-20197	112	47	.	.	PUNCT
ajst-20197	113	1	attention	attention	NOUN
ajst-20197	113	2	mechanisms	mechanism	NOUN
ajst-20197	113	3	help	help	VERB
ajst-20197	113	4	in	in	ADP
ajst-20197	113	5	the	the	DET
ajst-20197	113	6	enhancement	enhancement	NOUN
ajst-20197	113	7	of	of	ADP
ajst-20197	113	8	“	"	PUNCT
ajst-20197	113	9	computer	computer	NOUN
ajst-20197	113	10	vision	vision	NOUN
ajst-20197	113	11	”	"	PUNCT
ajst-20197	113	12	by	by	ADP
ajst-20197	113	13	developing	develop	VERB
ajst-20197	113	14	imitation	imitation	NOUN
ajst-20197	113	15	over	over	ADP
ajst-20197	113	16	the	the	DET
ajst-20197	113	17	visual	visual	ADJ
ajst-20197	113	18	system	system	NOUN
ajst-20197	113	19	of	of	ADP
ajst-20197	113	20	humans	human	NOUN
ajst-20197	113	21	.	.	PUNCT
ajst-20197	114	1	these	these	DET
ajst-20197	114	2	mechanisms	mechanism	NOUN
ajst-20197	114	3	help	help	VERB
ajst-20197	114	4	in	in	ADP
ajst-20197	114	5	semantic	semantic	ADJ
ajst-20197	114	6	segmentation	segmentation	NOUN
ajst-20197	114	7	,	,	PUNCT
ajst-20197	114	8	multi	multi	ADJ
ajst-20197	114	9	-	-	ADJ
ajst-20197	114	10	video	video	ADJ
ajst-20197	114	11	understanding	understanding	NOUN
ajst-20197	114	12	,	,	PUNCT
ajst-20197	114	13	object	object	VERB
ajst-20197	114	14	detection	detection	NOUN
ajst-20197	114	15	and	and	CCONJ
ajst-20197	114	16	image	image	NOUN
ajst-20197	114	17	classification	classification	NOUN
ajst-20197	114	18	.	.	PUNCT
ajst-20197	115	1	spatial	spatial	ADJ
ajst-20197	115	2	attention	attention	NOUN
ajst-20197	115	3	helps	help	VERB
ajst-20197	115	4	in	in	ADP
ajst-20197	115	5	the	the	DET
ajst-20197	115	6	generation	generation	NOUN
ajst-20197	115	7	of	of	ADP
ajst-20197	115	8	an	an	DET
ajst-20197	115	9	attention	attention	NOUN
ajst-20197	115	10	mask	mask	NOUN
ajst-20197	115	11	over	over	ADP
ajst-20197	115	12	spatial	spatial	ADJ
ajst-20197	115	13	domains	domain	NOUN
ajst-20197	115	14	.	.	PUNCT
ajst-20197	116	1	this	this	PRON
ajst-20197	116	2	is	be	AUX
ajst-20197	116	3	effective	effective	ADJ
ajst-20197	116	4	in	in	ADP
ajst-20197	116	5	selection	selection	NOUN
ajst-20197	116	6	of	of	ADP
ajst-20197	116	7	spatial	spatial	ADJ
ajst-20197	116	8	regions	region	NOUN
ajst-20197	116	9	and	and	CCONJ
ajst-20197	116	10	positions	position	NOUN
ajst-20197	116	11	.	.	PUNCT
ajst-20197	117	1	temporal	temporal	ADJ
ajst-20197	117	2	attention	attention	NOUN
ajst-20197	117	3	helps	help	VERB
ajst-20197	117	4	in	in	ADP
ajst-20197	117	5	the	the	DET
ajst-20197	117	6	generation	generation	NOUN
ajst-20197	117	7	of	of	ADP
ajst-20197	117	8	an	an	DET
ajst-20197	117	9	attention	attention	NOUN
ajst-20197	117	10	mask	mask	NOUN
ajst-20197	117	11	within	within	ADP
ajst-20197	117	12	time	time	NOUN
ajst-20197	117	13	for	for	ADP
ajst-20197	117	14	the	the	DET
ajst-20197	117	15	appropriate	appropriate	ADJ
ajst-20197	117	16	selection	selection	NOUN
ajst-20197	117	17	of	of	ADP
ajst-20197	117	18	key	key	ADJ
ajst-20197	117	19	frames	frame	NOUN
ajst-20197	117	20	.	.	PUNCT
ajst-20197	118	1	spatial	spatial	ADJ
ajst-20197	118	2	attention	attention	NOUN
ajst-20197	118	3	helps	help	VERB
ajst-20197	118	4	in	in	ADP
ajst-20197	118	5	the	the	DET
ajst-20197	118	6	establishment	establishment	NOUN
ajst-20197	118	7	of	of	ADP
ajst-20197	118	8	one	one	NUM
ajst-20197	118	9	adaptive	adaptive	ADJ
ajst-20197	118	10	spatial	spatial	ADJ
ajst-20197	118	11	region	region	NOUN
ajst-20197	118	12	selection	selection	NOUN
ajst-20197	118	13	mechanism	mechanism	NOUN
ajst-20197	118	14	”	"	PUNCT
ajst-20197	118	15	for	for	ADP
ajst-20197	118	16	developing	develop	VERB
ajst-20197	118	17	focus	focus	NOUN
ajst-20197	118	18	on	on	ADP
ajst-20197	118	19	a	a	DET
ajst-20197	118	20	particular	particular	ADJ
ajst-20197	118	21	region	region	NOUN
ajst-20197	118	22	for	for	ADP
ajst-20197	118	23	human	human	ADJ
ajst-20197	118	24	actions	action	NOUN
ajst-20197	118	25	.	.	PUNCT
ajst-20197	119	1	genet	genet	NOUN
ajst-20197	119	2	,	,	PUNCT
ajst-20197	119	3	ram	ram	NOUN
ajst-20197	119	4	,	,	PUNCT
ajst-20197	119	5	non	non	ADJ
ajst-20197	119	6	-	-	ADJ
ajst-20197	119	7	local	local	ADJ
ajst-20197	119	8	and	and	CCONJ
ajst-20197	119	9	stn	stn	PROPN
ajst-20197	119	10	are	be	AUX
ajst-20197	119	11	the	the	DET
ajst-20197	119	12	different	different	ADJ
ajst-20197	119	13	spatial	spatial	ADJ
ajst-20197	119	14	attention	attention	NOUN
ajst-20197	119	15	techniques	technique	NOUN
ajst-20197	120	1	[	[	X
ajst-20197	120	2	12](lau	12](lau	NUM
ajst-20197	120	3	et	et	PROPN
ajst-20197	120	4	al	al	PROPN
ajst-20197	120	5	.	.	PROPN
ajst-20197	120	6	2024	2024	NUM
ajst-20197	120	7	)	)	PUNCT
ajst-20197	120	8	.	.	PUNCT
ajst-20197	121	1	relevant	relevant	ADJ
ajst-20197	121	2	regions	region	NOUN
ajst-20197	121	3	are	be	AUX
ajst-20197	121	4	explicitly	explicitly	ADV
ajst-20197	121	5	predicted	predict	VERB
ajst-20197	121	6	through	through	ADP
ajst-20197	121	7	the	the	DET
ajst-20197	121	8	implementation	implementation	NOUN
ajst-20197	121	9	of	of	ADP
ajst-20197	121	10	one	one	NUM
ajst-20197	121	11	subsequent	subsequent	ADJ
ajst-20197	121	12	network	network	NOUN
ajst-20197	121	13	by	by	ADP
ajst-20197	121	14	using	use	VERB
ajst-20197	121	15	stn	stn	PROPN
ajst-20197	121	16	.	.	PUNCT
ajst-20197	122	1	soft	soft	ADJ
ajst-20197	122	2	max	max	PROPN
ajst-20197	122	3	is	be	AUX
ajst-20197	122	4	implicitly	implicitly	ADV
ajst-20197	122	5	predicted	predict	VERB
ajst-20197	122	6	through	through	ADP
ajst-20197	122	7	the	the	DET
ajst-20197	122	8	implementation	implementation	NOUN
ajst-20197	122	9	of	of	ADP
ajst-20197	122	10	genet	genet	NOUN
ajst-20197	122	11	that	that	PRON
ajst-20197	122	12	uses	use	VERB
ajst-20197	122	13	one	one	NUM
ajst-20197	122	14	subnetwork	subnetwork	NOUN
ajst-20197	122	15	for	for	ADP
ajst-20197	122	16	choosing	choose	VERB
ajst-20197	122	17	important	important	ADJ
ajst-20197	122	18	regions	region	NOUN
ajst-20197	122	19	.	.	PUNCT
ajst-20197	123	1	visual	visual	ADJ
ajst-20197	123	2	attention	attention	NOUN
ajst-20197	123	3	is	be	AUX
ajst-20197	123	4	developed	develop	VERB
ajst-20197	123	5	through	through	ADP
ajst-20197	123	6	the	the	DET
ajst-20197	123	7	use	use	NOUN
ajst-20197	123	8	of	of	ADP
ajst-20197	123	9	reinforcement	reinforcement	NOUN
ajst-20197	123	10	learning	learning	NOUN
ajst-20197	123	11	techniques	technique	NOUN
ajst-20197	123	12	in	in	ADP
ajst-20197	123	13	ram	ram	NOUN
ajst-20197	123	14	and	and	CCONJ
ajst-20197	123	15	thereby	thereby	ADV
ajst-20197	123	16	help	help	VERB
ajst-20197	123	17	in	in	ADP
ajst-20197	123	18	developing	develop	VERB
ajst-20197	123	19	focus	focus	NOUN
ajst-20197	123	20	over	over	ADP
ajst-20197	123	21	a	a	DET
ajst-20197	123	22	particular	particular	ADJ
ajst-20197	123	23	location	location	NOUN
ajst-20197	124	1	[	[	X
ajst-20197	124	2	4](de	4](de	NUM
ajst-20197	124	3	santana	santana	PROPN
ajst-20197	124	4	correia	correia	PROPN
ajst-20197	124	5	and	and	CCONJ
ajst-20197	124	6	colombini	colombini	PROPN
ajst-20197	124	7	2022	2022	NUM
ajst-20197	124	8	)	)	PUNCT
ajst-20197	124	9	.	.	PUNCT
ajst-20197	125	1	xu	xu	PROPN
ajst-20197	125	2	et	et	PROPN
ajst-20197	126	1	al	al	PROPN
ajst-20197	126	2	.	.	PROPN
ajst-20197	126	3	2022	2022	NUM
ajst-20197	126	4	stated	state	VERB
ajst-20197	126	5	that	that	SCONJ
ajst-20197	126	6	temporal	temporal	ADJ
ajst-20197	126	7	attention	attention	NOUN
ajst-20197	126	8	helps	help	VERB
ajst-20197	126	9	in	in	ADP
ajst-20197	126	10	the	the	DET
ajst-20197	126	11	implementation	implementation	NOUN
ajst-20197	126	12	of	of	ADP
ajst-20197	126	13	a	a	DET
ajst-20197	126	14	“	"	PUNCT
ajst-20197	126	15	dynamic	dynamic	ADJ
ajst-20197	126	16	time	time	NOUN
ajst-20197	126	17	selection	selection	NOUN
ajst-20197	126	18	mechanism	mechanism	NOUN
ajst-20197	126	19	”	"	PUNCT
ajst-20197	126	20	for	for	ADP
ajst-20197	126	21	selecting	select	VERB
ajst-20197	126	22	a	a	DET
ajst-20197	126	23	particular	particular	ADJ
ajst-20197	126	24	time	time	NOUN
ajst-20197	126	25	in	in	ADP
ajst-20197	126	26	paying	pay	VERB
ajst-20197	126	27	attention	attention	NOUN
ajst-20197	126	28	.	.	PUNCT
ajst-20197	127	1	this	this	DET
ajst-20197	127	2	technique	technique	NOUN
ajst-20197	127	3	is	be	AUX
ajst-20197	127	4	effective	effective	ADJ
ajst-20197	127	5	in	in	ADP
ajst-20197	127	6	video	video	NOUN
ajst-20197	127	7	processing	processing	NOUN
ajst-20197	127	8	by	by	ADP
ajst-20197	127	9	picking	pick	VERB
ajst-20197	127	10	a	a	DET
ajst-20197	127	11	particular	particular	ADJ
ajst-20197	127	12	time	time	NOUN
ajst-20197	127	13	in	in	ADP
ajst-20197	127	14	analyzing	analyze	VERB
ajst-20197	127	15	the	the	DET
ajst-20197	127	16	information	information	NOUN
ajst-20197	127	17	flow	flow	NOUN
ajst-20197	127	18	sequence	sequence	NOUN
ajst-20197	127	19	.	.	PUNCT
ajst-20197	128	1	the	the	DET
ajst-20197	128	2	tam	tam	NOUN
ajst-20197	128	3	module	module	NOUN
ajst-20197	128	4	is	be	AUX
ajst-20197	128	5	implemented	implement	VERB
ajst-20197	128	6	for	for	ADP
ajst-20197	128	7	developing	develop	VERB
ajst-20197	128	8	temporal	temporal	ADJ
ajst-20197	128	9	attention	attention	NOUN
ajst-20197	128	10	through	through	ADP
ajst-20197	128	11	the	the	DET
ajst-20197	128	12	consideration	consideration	NOUN
ajst-20197	128	13	of	of	ADP
ajst-20197	128	14	one	one	NUM
ajst-20197	128	15	adaptive	adaptive	ADJ
ajst-20197	128	16	kernel	kernel	NOUN
ajst-20197	128	17	.	.	PUNCT
ajst-20197	129	1	this	this	PRON
ajst-20197	129	2	helps	help	VERB
ajst-20197	129	3	in	in	ADP
ajst-20197	129	4	capturing	capture	VERB
ajst-20197	129	5	the	the	DET
ajst-20197	129	6	“	"	PUNCT
ajst-20197	129	7	global	global	ADJ
ajst-20197	129	8	contextual	contextual	ADJ
ajst-20197	129	9	data	datum	NOUN
ajst-20197	129	10	”	"	PUNCT
ajst-20197	129	11	for	for	ADP
ajst-20197	129	12	minimizing	minimize	VERB
ajst-20197	129	13	time	time	NOUN
ajst-20197	129	14	complexity	complexity	NOUN
ajst-20197	129	15	.	.	PUNCT
ajst-20197	130	1	computation	computation	NOUN
ajst-20197	130	2	cost	cost	NOUN
ajst-20197	130	3	is	be	AUX
ajst-20197	130	4	reduced	reduce	VERB
ajst-20197	130	5	through	through	ADP
ajst-20197	130	6	the	the	DET
ajst-20197	130	7	consideration	consideration	NOUN
ajst-20197	130	8	of	of	ADP
ajst-20197	130	9	1d	1d	NUM
ajst-20197	130	10	convolutions	convolution	NOUN
ajst-20197	130	11	that	that	PRON
ajst-20197	130	12	include	include	VERB
ajst-20197	130	13	relu	relu	NOUN
ajst-20197	130	14	nonlinearity	nonlinearity	NOUN
ajst-20197	130	15	within	within	ADP
ajst-20197	130	16	the	the	DET
ajst-20197	130	17	“	"	PUNCT
ajst-20197	130	18	temporal	temporal	ADJ
ajst-20197	130	19	domain	domain	NOUN
ajst-20197	130	20	”	"	PUNCT
ajst-20197	131	1	[	[	X
ajst-20197	131	2	29](zhao	29](zhao	NUM
ajst-20197	131	3	et	et	NOUN
ajst-20197	131	4	al	al	PROPN
ajst-20197	131	5	.	.	PROPN
ajst-20197	131	6	2020	2020	NUM
ajst-20197	131	7	)	)	PUNCT
ajst-20197	131	8	.	.	PUNCT
ajst-20197	132	1	this	this	PRON
ajst-20197	132	2	helps	help	VERB
ajst-20197	132	3	in	in	ADP
ajst-20197	132	4	generating	generate	VERB
ajst-20197	132	5	location	location	NOUN
ajst-20197	132	6	-	-	PUNCT
ajst-20197	132	7	sensitive	sensitive	ADJ
ajst-20197	132	8	maps	map	NOUN
ajst-20197	132	9	for	for	ADP
ajst-20197	132	10	frame	frame	NOUN
ajst-20197	132	11	-	-	PUNCT
ajst-20197	132	12	wise	wise	ADJ
ajst-20197	132	13	features	feature	NOUN
ajst-20197	132	14	enhancement	enhancement	NOUN
ajst-20197	132	15	.	.	PUNCT
ajst-20197	133	1	3.5	3.5	NUM
ajst-20197	133	2	.	.	PUNCT
ajst-20197	133	3	self	self	NOUN
ajst-20197	133	4	-	-	PUNCT
ajst-20197	133	5	attention	attention	NOUN
ajst-20197	133	6	mechanisms	mechanism	NOUN
ajst-20197	133	7	according	accord	VERB
ajst-20197	133	8	to	to	ADP
ajst-20197	133	9	hafiz	hafiz	PROPN
ajst-20197	133	10	et	et	PROPN
ajst-20197	133	11	al	al	PROPN
ajst-20197	133	12	.	.	PROPN
ajst-20197	133	13	2021	2021	NUM
ajst-20197	133	14	,	,	PUNCT
ajst-20197	133	15	in	in	ADP
ajst-20197	133	16	the	the	DET
ajst-20197	133	17	current	current	ADJ
ajst-20197	133	18	scenario	scenario	NOUN
ajst-20197	133	19	,	,	PUNCT
ajst-20197	133	20	the	the	DET
ajst-20197	133	21	attention	attention	NOUN
ajst-20197	133	22	-	-	PUNCT
ajst-20197	133	23	based	base	VERB
ajst-20197	133	24	mechanisms	mechanism	NOUN
ajst-20197	133	25	act	act	VERB
ajst-20197	133	26	as	as	ADP
ajst-20197	133	27	transformers	transformer	NOUN
ajst-20197	133	28	for	for	ADP
ajst-20197	133	29	the	the	DET
ajst-20197	133	30	accomplishment	accomplishment	NOUN
ajst-20197	133	31	of	of	ADP
ajst-20197	133	32	“	"	PUNCT
ajst-20197	133	33	machine	machine	NOUN
ajst-20197	133	34	vision	vision	NOUN
ajst-20197	133	35	tasks	task	NOUN
ajst-20197	133	36	”	"	PUNCT
ajst-20197	133	37	.	.	PUNCT
ajst-20197	134	1	this	this	PRON
ajst-20197	134	2	is	be	AUX
ajst-20197	134	3	done	do	VERB
ajst-20197	134	4	through	through	ADP
ajst-20197	134	5	the	the	DET
ajst-20197	134	6	usage	usage	NOUN
ajst-20197	134	7	of	of	ADP
ajst-20197	134	8	vision	vision	NOUN
ajst-20197	134	9	transformers	transformer	NOUN
ajst-20197	134	10	that	that	PRON
ajst-20197	134	11	help	help	VERB
ajst-20197	134	12	in	in	ADP
ajst-20197	134	13	object	object	NOUN
ajst-20197	134	14	detection	detection	NOUN
ajst-20197	134	15	,	,	PUNCT
ajst-20197	134	16	image	image	NOUN
ajst-20197	134	17	recognition	recognition	NOUN
ajst-20197	134	18	and	and	CCONJ
ajst-20197	134	19	video	video	NOUN
ajst-20197	134	20	understanding	understanding	NOUN
ajst-20197	134	21	.	.	PUNCT
ajst-20197	135	1	the	the	DET
ajst-20197	135	2	implementation	implementation	NOUN
ajst-20197	135	3	of	of	ADP
ajst-20197	135	4	a	a	DET
ajst-20197	135	5	self	self	NOUN
ajst-20197	135	6	-	-	PUNCT
ajst-20197	135	7	attention	attention	NOUN
ajst-20197	135	8	technique	technique	NOUN
ajst-20197	135	9	helps	help	VERB
ajst-20197	135	10	in	in	ADP
ajst-20197	135	11	the	the	DET
ajst-20197	135	12	formation	formation	NOUN
ajst-20197	135	13	of	of	ADP
ajst-20197	135	14	one	one	NUM
ajst-20197	135	15	measurable	measurable	ADJ
ajst-20197	135	16	estimate	estimate	NOUN
ajst-20197	135	17	based	base	VERB
ajst-20197	135	18	on	on	ADP
ajst-20197	135	19	data	datum	NOUN
ajst-20197	135	20	relevance	relevance	NOUN
ajst-20197	135	21	available	available	ADJ
ajst-20197	135	22	within	within	ADP
ajst-20197	135	23	other	other	ADJ
ajst-20197	135	24	elements	element	NOUN
ajst-20197	135	25	[	[	X
ajst-20197	135	26	26](wilson	26](wilson	NUM
ajst-20197	135	27	et	et	NOUN
ajst-20197	135	28	al	al	PROPN
ajst-20197	135	29	.	.	PROPN
ajst-20197	135	30	2023	2023	NUM
ajst-20197	135	31	)	)	PUNCT
ajst-20197	135	32	.	.	PUNCT
ajst-20197	136	1	this	this	DET
ajst-20197	136	2	technique	technique	NOUN
ajst-20197	136	3	acts	act	VERB
ajst-20197	136	4	like	like	ADP
ajst-20197	136	5	transformers	transformer	NOUN
ajst-20197	136	6	for	for	ADP
ajst-20197	136	7	modeling	model	VERB
ajst-20197	136	8	dependencies	dependency	NOUN
ajst-20197	136	9	available	available	ADJ
ajst-20197	136	10	within	within	ADP
ajst-20197	136	11	a	a	DET
ajst-20197	136	12	sequence	sequence	NOUN
ajst-20197	136	13	.	.	PUNCT
ajst-20197	137	1	this	this	PRON
ajst-20197	137	2	helps	help	VERB
ajst-20197	137	3	in	in	ADP
ajst-20197	137	4	enhancing	enhance	VERB
ajst-20197	137	5	the	the	DET
ajst-20197	137	6	prediction	prediction	NOUN
ajst-20197	137	7	accuracy	accuracy	NOUN
ajst-20197	137	8	and	and	CCONJ
ajst-20197	137	9	capability	capability	NOUN
ajst-20197	137	10	within	within	ADP
ajst-20197	137	11	deep	deep	ADJ
ajst-20197	137	12	learning	learning	NOUN
ajst-20197	137	13	techniques	technique	NOUN
ajst-20197	137	14	.	.	PUNCT
ajst-20197	138	1	a	a	DET
ajst-20197	138	2	value	value	NOUN
ajst-20197	138	3	is	be	AUX
ajst-20197	138	4	assigned	assign	VERB
ajst-20197	138	5	within	within	ADP
ajst-20197	138	6	an	an	DET
ajst-20197	138	7	element	element	NOUN
ajst-20197	138	8	within	within	ADP
ajst-20197	138	9	a	a	DET
ajst-20197	138	10	particular	particular	ADJ
ajst-20197	138	11	sequence	sequence	NOUN
ajst-20197	138	12	through	through	ADP
ajst-20197	138	13	the	the	DET
ajst-20197	138	14	use	use	NOUN
ajst-20197	138	15	of	of	ADP
ajst-20197	138	16	a	a	DET
ajst-20197	138	17	“	"	PUNCT
ajst-20197	138	18	self	self	NOUN
ajst-20197	138	19	-	-	PUNCT
ajst-20197	138	20	attention	attention	NOUN
ajst-20197	138	21	model	model	NOUN
ajst-20197	138	22	layer	layer	NOUN
ajst-20197	138	23	”	"	PUNCT
ajst-20197	138	24	.	.	PUNCT
ajst-20197	139	1	this	this	PRON
ajst-20197	139	2	helps	help	VERB
ajst-20197	139	3	in	in	ADP
ajst-20197	139	4	the	the	DET
ajst-20197	139	5	global	global	ADJ
ajst-20197	139	6	combination	combination	NOUN
ajst-20197	139	7	of	of	ADP
ajst-20197	139	8	the	the	DET
ajst-20197	139	9	information	information	NOUN
ajst-20197	139	10	that	that	PRON
ajst-20197	139	11	is	be	AUX
ajst-20197	139	12	derived	derive	VERB
ajst-20197	139	13	from	from	ADP
ajst-20197	139	14	an	an	DET
ajst-20197	139	15	“	"	PUNCT
ajst-20197	139	16	input	input	NOUN
ajst-20197	139	17	sequence	sequence	NOUN
ajst-20197	139	18	”	"	PUNCT
ajst-20197	139	19	.	.	PUNCT
ajst-20197	140	1	determination	determination	NOUN
ajst-20197	140	2	of	of	ADP
ajst-20197	140	3	output	output	NOUN
ajst-20197	140	4	from	from	ADP
ajst-20197	140	5	self	self	NOUN
ajst-20197	140	6	-	-	PUNCT
ajst-20197	140	7	attention	attention	NOUN
ajst-20197	140	8	layer[6](source	layer[6](source	NOUN
ajst-20197	140	9	:	:	PUNCT
ajst-20197	140	10	hafiz	hafiz	PROPN
ajst-20197	140	11	et	et	PROPN
ajst-20197	140	12	al	al	PROPN
ajst-20197	140	13	.	.	PROPN
ajst-20197	140	14	2021	2021	NUM
ajst-20197	140	15	)	)	PUNCT
ajst-20197	140	16	self	self	NOUN
ajst-20197	140	17	-	-	PUNCT
ajst-20197	140	18	attention	attention	NOUN
ajst-20197	140	19	helps	help	VERB
ajst-20197	140	20	in	in	ADP
ajst-20197	140	21	capturing	capture	VERB
ajst-20197	140	22	dependencies	dependency	NOUN
ajst-20197	140	23	by	by	ADP
ajst-20197	140	24	transforming	transform	VERB
ajst-20197	140	25	values	value	NOUN
ajst-20197	140	26	and	and	CCONJ
ajst-20197	140	27	queries	query	VERB
ajst-20197	141	1	[	[	X
ajst-20197	141	2	28](xie	28](xie	NUM
ajst-20197	141	3	et	et	NOUN
ajst-20197	141	4	al	al	PROPN
ajst-20197	141	5	.	.	PROPN
ajst-20197	141	6	2020	2020	NUM
ajst-20197	141	7	)	)	PUNCT
ajst-20197	141	8	.	.	PUNCT
ajst-20197	142	1	the	the	DET
ajst-20197	142	2	query	query	NOUN
ajst-20197	142	3	dot	dot	NOUN
ajst-20197	142	4	product	product	NOUN
ajst-20197	142	5	that	that	PRON
ajst-20197	142	6	is	be	AUX
ajst-20197	142	7	available	available	ADJ
ajst-20197	142	8	within	within	ADP
ajst-20197	142	9	a	a	DET
ajst-20197	142	10	sequence	sequence	NOUN
ajst-20197	142	11	is	be	AUX
ajst-20197	142	12	diagnosed	diagnose	VERB
ajst-20197	142	13	through	through	ADP
ajst-20197	142	14	the	the	DET
ajst-20197	142	15	implementation	implementation	NOUN
ajst-20197	142	16	of	of	ADP
ajst-20197	142	17	self	self	NOUN
ajst-20197	142	18	-	-	PUNCT
ajst-20197	142	19	attention	attention	NOUN
ajst-20197	142	20	techniques	technique	NOUN
ajst-20197	142	21	.	.	PUNCT
ajst-20197	143	1	the	the	DET
ajst-20197	143	2	normalization	normalization	NOUN
ajst-20197	143	3	of	of	ADP
ajst-20197	143	4	this	this	DET
ajst-20197	143	5	product	product	NOUN
ajst-20197	143	6	is	be	AUX
ajst-20197	143	7	accomplished	accomplish	VERB
ajst-20197	143	8	through	through	ADP
ajst-20197	143	9	the	the	DET
ajst-20197	143	10	consideration	consideration	NOUN
ajst-20197	143	11	of	of	ADP
ajst-20197	143	12	a	a	DET
ajst-20197	143	13	softmax	softmax	NOUN
ajst-20197	143	14	function	function	NOUN
ajst-20197	144	1	[	[	X
ajst-20197	144	2	16](pan	16](pan	PROPN
ajst-20197	144	3	et	et	PROPN
ajst-20197	144	4	al	al	PROPN
ajst-20197	144	5	.	.	PROPN
ajst-20197	144	6	2022	2022	NUM
ajst-20197	144	7	)	)	PUNCT
ajst-20197	144	8	.	.	PUNCT
ajst-20197	145	1	as	as	ADP
ajst-20197	145	2	a	a	DET
ajst-20197	145	3	result	result	NOUN
ajst-20197	145	4	,	,	PUNCT
ajst-20197	145	5	there	there	PRON
ajst-20197	145	6	is	be	VERB
ajst-20197	145	7	achievement	achievement	NOUN
ajst-20197	145	8	of	of	ADP
ajst-20197	145	9	attention	attention	NOUN
ajst-20197	145	10	-	-	PUNCT
ajst-20197	145	11	map	map	NOUN
ajst-20197	145	12	scores	score	NOUN
ajst-20197	145	13	that	that	PRON
ajst-20197	145	14	helps	help	VERB
ajst-20197	145	15	in	in	ADP
ajst-20197	145	16	determining	determine	VERB
ajst-20197	145	17	the	the	DET
ajst-20197	145	18	“	"	PUNCT
ajst-20197	145	19	weighted	weighted	ADJ
ajst-20197	145	20	summation	summation	NOUN
ajst-20197	145	21	”	"	PUNCT
ajst-20197	145	22	of	of	ADP
ajst-20197	145	23	112	112	NUM
ajst-20197	145	24	all	all	DET
ajst-20197	145	25	elements	element	NOUN
ajst-20197	145	26	available	available	ADJ
ajst-20197	145	27	within	within	ADP
ajst-20197	145	28	a	a	DET
ajst-20197	145	29	sequence	sequence	NOUN
ajst-20197	145	30	.	.	PUNCT
ajst-20197	146	1	parah	parah	PROPN
ajst-20197	146	2	et	et	PROPN
ajst-20197	146	3	al	al	PROPN
ajst-20197	146	4	.	.	PROPN
ajst-20197	146	5	2021	2021	NUM
ajst-20197	146	6	stated	state	VERB
ajst-20197	146	7	that	that	SCONJ
ajst-20197	146	8	recently	recently	ADV
ajst-20197	146	9	,	,	PUNCT
ajst-20197	146	10	this	this	DET
ajst-20197	146	11	feature	feature	NOUN
ajst-20197	146	12	has	have	AUX
ajst-20197	146	13	been	be	AUX
ajst-20197	146	14	implemented	implement	VERB
ajst-20197	146	15	in	in	ADP
ajst-20197	146	16	image	image	NOUN
ajst-20197	146	17	classification	classification	NOUN
ajst-20197	146	18	and	and	CCONJ
ajst-20197	146	19	face	face	NOUN
ajst-20197	146	20	recognition	recognition	NOUN
ajst-20197	146	21	for	for	ADP
ajst-20197	146	22	analyzing	analyze	VERB
ajst-20197	146	23	the	the	DET
ajst-20197	146	24	sequence	sequence	NOUN
ajst-20197	146	25	and	and	CCONJ
ajst-20197	146	26	thereby	thereby	ADV
ajst-20197	146	27	developing	develop	VERB
ajst-20197	146	28	outputs	output	NOUN
ajst-20197	146	29	based	base	VERB
ajst-20197	146	30	on	on	ADP
ajst-20197	146	31	the	the	DET
ajst-20197	146	32	weighted	weight	VERB
ajst-20197	146	33	summation	summation	NOUN
ajst-20197	146	34	.	.	PUNCT
ajst-20197	147	1	“	"	PUNCT
ajst-20197	147	2	convolution	convolution	NOUN
ajst-20197	147	3	vision	vision	NOUN
ajst-20197	147	4	transformers	transformer	NOUN
ajst-20197	147	5	”	"	PUNCT
ajst-20197	147	6	are	be	AUX
ajst-20197	147	7	used	use	VERB
ajst-20197	147	8	for	for	ADP
ajst-20197	147	9	the	the	DET
ajst-20197	147	10	establishment	establishment	NOUN
ajst-20197	147	11	of	of	ADP
ajst-20197	147	12	one	one	NUM
ajst-20197	147	13	linear	linear	ADJ
ajst-20197	147	14	projection	projection	NOUN
ajst-20197	147	15	for	for	ADP
ajst-20197	147	16	analyzing	analyze	VERB
ajst-20197	147	17	the	the	DET
ajst-20197	147	18	sequence	sequence	NOUN
ajst-20197	147	19	and	and	CCONJ
ajst-20197	147	20	thereby	thereby	ADV
ajst-20197	147	21	increasing	increase	VERB
ajst-20197	147	22	the	the	DET
ajst-20197	147	23	prediction	prediction	NOUN
ajst-20197	147	24	accuracy	accuracy	NOUN
ajst-20197	148	1	[	[	X
ajst-20197	148	2	21](ranftl	21](ranftl	NUM
ajst-20197	148	3	,	,	PUNCT
ajst-20197	148	4	bochkovskiy	bochkovskiy	NOUN
ajst-20197	148	5	and	and	CCONJ
ajst-20197	148	6	koltun	koltun	VERB
ajst-20197	148	7	2021	2021	NUM
ajst-20197	148	8	)	)	PUNCT
ajst-20197	148	9	.	.	PUNCT
ajst-20197	149	1	according	accord	VERB
ajst-20197	149	2	to	to	ADP
ajst-20197	149	3	[	[	X
ajst-20197	149	4	7]hernández	7]hernández	NUM
ajst-20197	149	5	et	et	NOUN
ajst-20197	149	6	al	al	PROPN
ajst-20197	149	7	.	.	PROPN
ajst-20197	149	8	2021	2021	NUM
ajst-20197	149	9	,	,	PUNCT
ajst-20197	149	10	self	self	NOUN
ajst-20197	149	11	-	-	PUNCT
ajst-20197	149	12	attention	attention	NOUN
ajst-20197	149	13	helps	help	VERB
ajst-20197	149	14	in	in	ADP
ajst-20197	149	15	the	the	DET
ajst-20197	149	16	establishment	establishment	NOUN
ajst-20197	149	17	of	of	ADP
ajst-20197	149	18	relationships	relationship	NOUN
ajst-20197	149	19	within	within	ADP
ajst-20197	149	20	numerous	numerous	ADJ
ajst-20197	149	21	positions	position	NOUN
ajst-20197	149	22	available	available	ADJ
ajst-20197	149	23	within	within	ADP
ajst-20197	149	24	a	a	DET
ajst-20197	149	25	particular	particular	ADJ
ajst-20197	149	26	sequence	sequence	NOUN
ajst-20197	149	27	.	.	PUNCT
ajst-20197	150	1	sequence	sequence	NOUN
ajst-20197	150	2	transformation	transformation	NOUN
ajst-20197	150	3	is	be	AUX
ajst-20197	150	4	accomplished	accomplish	VERB
ajst-20197	150	5	for	for	ADP
ajst-20197	150	6	developing	develop	VERB
ajst-20197	150	7	computation	computation	NOUN
ajst-20197	150	8	using	use	VERB
ajst-20197	150	9	queries	query	NOUN
ajst-20197	150	10	,	,	PUNCT
ajst-20197	150	11	keys	key	NOUN
ajst-20197	150	12	and	and	CCONJ
ajst-20197	150	13	values	value	NOUN
ajst-20197	150	14	[	[	X
ajst-20197	150	15	14](lin	14](lin	NUM
ajst-20197	150	16	et	et	NOUN
ajst-20197	150	17	al	al	PROPN
ajst-20197	150	18	.	.	PROPN
ajst-20197	150	19	2020	2020	NUM
ajst-20197	150	20	)	)	PUNCT
ajst-20197	150	21	.	.	PUNCT
ajst-20197	151	1	this	this	DET
ajst-20197	151	2	technique	technique	NOUN
ajst-20197	151	3	helps	help	VERB
ajst-20197	151	4	in	in	ADP
ajst-20197	151	5	the	the	DET
ajst-20197	151	6	dynamic	dynamic	ADJ
ajst-20197	151	7	calculation	calculation	NOUN
ajst-20197	151	8	of	of	ADP
ajst-20197	151	9	weights	weight	NOUN
ajst-20197	151	10	by	by	ADP
ajst-20197	151	11	availing	avail	VERB
ajst-20197	151	12	direct	direct	ADJ
ajst-20197	151	13	transformation	transformation	NOUN
ajst-20197	151	14	within	within	ADP
ajst-20197	151	15	inputs	input	NOUN
ajst-20197	151	16	.	.	PUNCT
ajst-20197	152	1	direct	direct	ADJ
ajst-20197	152	2	relationships	relationship	NOUN
ajst-20197	152	3	are	be	AUX
ajst-20197	152	4	established	establish	VERB
ajst-20197	152	5	within	within	ADP
ajst-20197	152	6	an	an	DET
ajst-20197	152	7	output	output	NOUN
ajst-20197	152	8	vector	vector	NOUN
ajst-20197	152	9	and	and	CCONJ
ajst-20197	152	10	inputs	input	NOUN
ajst-20197	152	11	through	through	ADP
ajst-20197	152	12	the	the	DET
ajst-20197	152	13	use	use	NOUN
ajst-20197	152	14	of	of	ADP
ajst-20197	152	15	this	this	DET
ajst-20197	152	16	module	module	NOUN
ajst-20197	152	17	[	[	X
ajst-20197	152	18	10](kossen	10](kossen	NUM
ajst-20197	152	19	et	et	NOUN
ajst-20197	152	20	al	al	PROPN
ajst-20197	152	21	.	.	PROPN
ajst-20197	152	22	2021	2021	NUM
ajst-20197	152	23	)	)	PUNCT
ajst-20197	152	24	.	.	PUNCT
ajst-20197	153	1	however	however	ADV
ajst-20197	153	2	,	,	PUNCT
ajst-20197	153	3	the	the	DET
ajst-20197	153	4	local	local	ADJ
ajst-20197	153	5	interactions	interaction	NOUN
ajst-20197	153	6	are	be	AUX
ajst-20197	153	7	not	not	PART
ajst-20197	153	8	prioritized	prioritize	VERB
ajst-20197	153	9	through	through	ADP
ajst-20197	153	10	the	the	DET
ajst-20197	153	11	use	use	NOUN
ajst-20197	153	12	of	of	ADP
ajst-20197	153	13	self	self	NOUN
ajst-20197	153	14	-	-	PUNCT
ajst-20197	153	15	attention	attention	NOUN
ajst-20197	153	16	techniques	technique	NOUN
ajst-20197	153	17	.	.	PUNCT
ajst-20197	154	1	4	4	X
ajst-20197	154	2	.	.	X
ajst-20197	154	3	discussion	discussion	NOUN
ajst-20197	154	4	referring	refer	VERB
ajst-20197	154	5	to	to	ADP
ajst-20197	154	6	the	the	DET
ajst-20197	154	7	above	above	ADJ
ajst-20197	154	8	findings	finding	NOUN
ajst-20197	154	9	,	,	PUNCT
ajst-20197	154	10	it	it	PRON
ajst-20197	154	11	is	be	AUX
ajst-20197	154	12	summarized	summarize	VERB
ajst-20197	154	13	that	that	SCONJ
ajst-20197	154	14	there	there	PRON
ajst-20197	154	15	is	be	VERB
ajst-20197	154	16	accomplishment	accomplishment	NOUN
ajst-20197	154	17	of	of	ADP
ajst-20197	154	18	sequential	sequential	ADJ
ajst-20197	154	19	modeling	modeling	NOUN
ajst-20197	154	20	through	through	ADP
ajst-20197	154	21	the	the	DET
ajst-20197	154	22	consideration	consideration	NOUN
ajst-20197	154	23	of	of	ADP
ajst-20197	154	24	attention	attention	NOUN
ajst-20197	154	25	and	and	CCONJ
ajst-20197	154	26	memory	memory	NOUN
ajst-20197	154	27	mechanisms	mechanism	NOUN
ajst-20197	154	28	.	.	PUNCT
ajst-20197	155	1	through	through	ADP
ajst-20197	155	2	the	the	DET
ajst-20197	155	3	use	use	NOUN
ajst-20197	155	4	of	of	ADP
ajst-20197	155	5	these	these	DET
ajst-20197	155	6	mechanisms	mechanism	NOUN
ajst-20197	155	7	,	,	PUNCT
ajst-20197	155	8	focus	focus	NOUN
ajst-20197	155	9	is	be	AUX
ajst-20197	155	10	developed	develop	VERB
ajst-20197	155	11	over	over	ADP
ajst-20197	155	12	a	a	DET
ajst-20197	155	13	particular	particular	ADJ
ajst-20197	155	14	location	location	NOUN
ajst-20197	155	15	and	and	CCONJ
ajst-20197	155	16	time	time	NOUN
ajst-20197	155	17	within	within	ADP
ajst-20197	155	18	an	an	DET
ajst-20197	155	19	available	available	ADJ
ajst-20197	155	20	dataset	dataset	NOUN
ajst-20197	155	21	.	.	PUNCT
ajst-20197	156	1	this	this	PRON
ajst-20197	156	2	is	be	AUX
ajst-20197	156	3	effective	effective	ADJ
ajst-20197	156	4	in	in	ADP
ajst-20197	156	5	the	the	DET
ajst-20197	156	6	suitable	suitable	ADJ
ajst-20197	156	7	recognition	recognition	NOUN
ajst-20197	156	8	of	of	ADP
ajst-20197	156	9	hidden	hidden	ADJ
ajst-20197	156	10	parameters	parameter	NOUN
ajst-20197	156	11	and	and	CCONJ
ajst-20197	156	12	thereby	thereby	ADV
ajst-20197	156	13	helps	help	VERB
ajst-20197	156	14	in	in	ADP
ajst-20197	156	15	increasing	increase	VERB
ajst-20197	156	16	prediction	prediction	NOUN
ajst-20197	156	17	accuracy	accuracy	NOUN
ajst-20197	156	18	.	.	PUNCT
ajst-20197	157	1	there	there	PRON
ajst-20197	157	2	has	have	AUX
ajst-20197	157	3	been	be	AUX
ajst-20197	157	4	consideration	consideration	NOUN
ajst-20197	157	5	of	of	ADP
ajst-20197	157	6	the	the	DET
ajst-20197	157	7	state	state	NOUN
ajst-20197	157	8	-	-	PUNCT
ajst-20197	157	9	of	of	ADP
ajst-20197	157	10	-	-	PUNCT
ajst-20197	157	11	the	the	DET
ajst-20197	157	12	art	art	NOUN
ajst-20197	157	13	approach	approach	NOUN
ajst-20197	157	14	through	through	ADP
ajst-20197	157	15	the	the	DET
ajst-20197	157	16	use	use	NOUN
ajst-20197	157	17	of	of	ADP
ajst-20197	157	18	attention	attention	NOUN
ajst-20197	157	19	and	and	CCONJ
ajst-20197	157	20	memory	memory	NOUN
ajst-20197	157	21	mechanisms	mechanism	NOUN
ajst-20197	157	22	in	in	ADP
ajst-20197	157	23	deep	deep	ADJ
ajst-20197	157	24	learning	learning	NOUN
ajst-20197	157	25	.	.	PUNCT
ajst-20197	158	1	referring	refer	VERB
ajst-20197	158	2	to	to	ADP
ajst-20197	158	3	the	the	DET
ajst-20197	158	4	context	context	NOUN
ajst-20197	158	5	of	of	ADP
ajst-20197	158	6	grus	grus	NOUN
ajst-20197	158	7	,	,	PUNCT
ajst-20197	158	8	the	the	DET
ajst-20197	158	9	"	"	PUNCT
ajst-20197	158	10	vanishing	vanish	VERB
ajst-20197	158	11	gradient	gradient	ADJ
ajst-20197	158	12	problem	problem	NOUN
ajst-20197	158	13	"	"	PUNCT
ajst-20197	158	14	is	be	AUX
ajst-20197	158	15	eliminated	eliminate	VERB
ajst-20197	158	16	through	through	ADP
ajst-20197	158	17	the	the	DET
ajst-20197	158	18	usage	usage	NOUN
ajst-20197	158	19	of	of	ADP
ajst-20197	158	20	gating	gate	VERB
ajst-20197	158	21	mechanisms	mechanism	NOUN
ajst-20197	158	22	.	.	PUNCT
ajst-20197	159	1	this	this	PRON
ajst-20197	159	2	is	be	AUX
ajst-20197	159	3	done	do	VERB
ajst-20197	159	4	through	through	ADP
ajst-20197	159	5	the	the	DET
ajst-20197	159	6	effective	effective	ADJ
ajst-20197	159	7	consideration	consideration	NOUN
ajst-20197	159	8	of	of	ADP
ajst-20197	159	9	the	the	DET
ajst-20197	159	10	dependencies	dependency	NOUN
ajst-20197	159	11	that	that	PRON
ajst-20197	159	12	are	be	AUX
ajst-20197	159	13	diagnosed	diagnose	VERB
ajst-20197	159	14	within	within	ADP
ajst-20197	159	15	a	a	DET
ajst-20197	159	16	dataset	dataset	NOUN
ajst-20197	159	17	.	.	PUNCT
ajst-20197	160	1	the	the	DET
ajst-20197	160	2	usage	usage	NOUN
ajst-20197	160	3	of	of	ADP
ajst-20197	160	4	activation	activation	NOUN
ajst-20197	160	5	functions	function	NOUN
ajst-20197	160	6	in	in	ADP
ajst-20197	160	7	grus	grus	NOUN
ajst-20197	160	8	helps	help	VERB
ajst-20197	160	9	in	in	ADP
ajst-20197	160	10	increasing	increase	VERB
ajst-20197	160	11	model	model	NOUN
ajst-20197	160	12	performance	performance	NOUN
ajst-20197	160	13	by	by	ADP
ajst-20197	160	14	developing	develop	VERB
ajst-20197	160	15	in	in	ADP
ajst-20197	160	16	-	-	PUNCT
ajst-20197	160	17	depth	depth	NOUN
ajst-20197	160	18	investigation	investigation	NOUN
ajst-20197	160	19	over	over	ADP
ajst-20197	160	20	dataset	dataset	ADJ
ajst-20197	160	21	sequence	sequence	NOUN
ajst-20197	160	22	.	.	PUNCT
ajst-20197	161	1	further	far	ADV
ajst-20197	161	2	,	,	PUNCT
ajst-20197	161	3	numerical	numerical	ADJ
ajst-20197	161	4	approximation	approximation	NOUN
ajst-20197	161	5	is	be	AUX
ajst-20197	161	6	done	do	VERB
ajst-20197	161	7	through	through	ADP
ajst-20197	161	8	the	the	DET
ajst-20197	161	9	usage	usage	NOUN
ajst-20197	161	10	of	of	ADP
ajst-20197	161	11	gate	gate	NOUN
ajst-20197	161	12	controllers	controller	NOUN
ajst-20197	161	13	that	that	PRON
ajst-20197	161	14	helps	help	VERB
ajst-20197	161	15	in	in	ADP
ajst-20197	161	16	analyzing	analyze	VERB
ajst-20197	161	17	the	the	DET
ajst-20197	161	18	“	"	PUNCT
ajst-20197	161	19	hidden	hidden	ADJ
ajst-20197	161	20	state	state	NOUN
ajst-20197	161	21	decay	decay	NOUN
ajst-20197	161	22	”	"	PUNCT
ajst-20197	161	23	in	in	ADP
ajst-20197	161	24	terms	term	NOUN
ajst-20197	161	25	of	of	ADP
ajst-20197	161	26	a	a	DET
ajst-20197	161	27	time	time	NOUN
ajst-20197	161	28	constant	constant	ADJ
ajst-20197	161	29	.	.	PUNCT
ajst-20197	162	1	referring	refer	VERB
ajst-20197	162	2	to	to	ADP
ajst-20197	162	3	the	the	DET
ajst-20197	162	4	context	context	NOUN
ajst-20197	162	5	of	of	ADP
ajst-20197	162	6	lstm	lstm	PROPN
ajst-20197	162	7	,	,	PUNCT
ajst-20197	162	8	the	the	DET
ajst-20197	162	9	usage	usage	NOUN
ajst-20197	162	10	of	of	ADP
ajst-20197	162	11	“	"	PUNCT
ajst-20197	162	12	working	work	VERB
ajst-20197	162	13	memory	memory	NOUN
ajst-20197	162	14	connections	connection	NOUN
ajst-20197	162	15	”	"	PUNCT
ajst-20197	162	16	helps	help	VERB
ajst-20197	162	17	in	in	ADP
ajst-20197	162	18	the	the	DET
ajst-20197	162	19	establishment	establishment	NOUN
ajst-20197	162	20	of	of	ADP
ajst-20197	162	21	a	a	DET
ajst-20197	162	22	protection	protection	NOUN
ajst-20197	162	23	mechanism	mechanism	NOUN
ajst-20197	162	24	by	by	ADP
ajst-20197	162	25	developing	develop	VERB
ajst-20197	162	26	error	error	NOUN
ajst-20197	162	27	paths	path	NOUN
ajst-20197	162	28	based	base	VERB
ajst-20197	162	29	on	on	ADP
ajst-20197	162	30	time	time	NOUN
ajst-20197	162	31	.	.	PUNCT
ajst-20197	163	1	this	this	DET
ajst-20197	163	2	approach	approach	NOUN
ajst-20197	163	3	helps	help	VERB
ajst-20197	163	4	in	in	ADP
ajst-20197	163	5	increasing	increase	VERB
ajst-20197	163	6	the	the	DET
ajst-20197	163	7	performance	performance	NOUN
ajst-20197	163	8	of	of	ADP
ajst-20197	163	9	“	"	PUNCT
ajst-20197	163	10	deep	deep	ADJ
ajst-20197	163	11	learning	learning	NOUN
ajst-20197	163	12	models	model	NOUN
ajst-20197	163	13	”	"	PUNCT
ajst-20197	163	14	by	by	ADP
ajst-20197	163	15	establishing	establish	VERB
ajst-20197	163	16	one	one	NUM
ajst-20197	163	17	“	"	PUNCT
ajst-20197	163	18	non	non	ADJ
ajst-20197	163	19	-	-	ADJ
ajst-20197	163	20	linear	linear	ADJ
ajst-20197	163	21	activation	activation	NOUN
ajst-20197	163	22	function	function	NOUN
ajst-20197	163	23	”	"	PUNCT
ajst-20197	163	24	.	.	PUNCT
ajst-20197	164	1	lstm	lstm	NOUN
ajst-20197	164	2	helps	help	VERB
ajst-20197	164	3	in	in	ADP
ajst-20197	164	4	adding	add	VERB
ajst-20197	164	5	one	one	NUM
ajst-20197	164	6	“	"	PUNCT
ajst-20197	164	7	linear	linear	ADJ
ajst-20197	164	8	self	self	NOUN
ajst-20197	164	9	-	-	PUNCT
ajst-20197	164	10	loop	loop	NOUN
ajst-20197	164	11	memory	memory	NOUN
ajst-20197	164	12	cell	cell	NOUN
ajst-20197	164	13	”	"	PUNCT
ajst-20197	164	14	and	and	CCONJ
ajst-20197	164	15	thereby	thereby	ADV
ajst-20197	164	16	helps	help	VERB
ajst-20197	164	17	in	in	ADP
ajst-20197	164	18	the	the	DET
ajst-20197	164	19	suitable	suitable	ADJ
ajst-20197	164	20	computation	computation	NOUN
ajst-20197	164	21	of	of	ADP
ajst-20197	164	22	hidden	hidden	ADJ
ajst-20197	164	23	units	unit	NOUN
ajst-20197	164	24	.	.	PUNCT
ajst-20197	165	1	referring	refer	VERB
ajst-20197	165	2	to	to	ADP
ajst-20197	165	3	the	the	DET
ajst-20197	165	4	context	context	NOUN
ajst-20197	165	5	of	of	ADP
ajst-20197	165	6	manns	mann	NOUN
ajst-20197	165	7	,	,	PUNCT
ajst-20197	165	8	there	there	PRON
ajst-20197	165	9	has	have	AUX
ajst-20197	165	10	been	be	AUX
ajst-20197	165	11	formation	formation	NOUN
ajst-20197	165	12	of	of	ADP
ajst-20197	165	13	content	content	NOUN
ajst-20197	165	14	-	-	PUNCT
ajst-20197	165	15	based	base	VERB
ajst-20197	165	16	addressing	addressing	NOUN
ajst-20197	165	17	that	that	PRON
ajst-20197	165	18	helps	help	VERB
ajst-20197	165	19	in	in	ADP
ajst-20197	165	20	the	the	DET
ajst-20197	165	21	effective	effective	ADJ
ajst-20197	165	22	diagnosis	diagnosis	NOUN
ajst-20197	165	23	of	of	ADP
ajst-20197	165	24	similar	similar	ADJ
ajst-20197	165	25	features	feature	NOUN
ajst-20197	165	26	available	available	ADJ
ajst-20197	165	27	within	within	ADP
ajst-20197	165	28	vector	vector	NOUN
ajst-20197	165	29	and	and	CCONJ
ajst-20197	165	30	memory	memory	NOUN
ajst-20197	165	31	elements	element	NOUN
ajst-20197	165	32	.	.	PUNCT
ajst-20197	166	1	quantization	quantization	NOUN
ajst-20197	166	2	error	error	NOUN
ajst-20197	166	3	is	be	AUX
ajst-20197	166	4	reduced	reduce	VERB
ajst-20197	166	5	through	through	ADP
ajst-20197	166	6	the	the	DET
ajst-20197	166	7	use	use	NOUN
ajst-20197	166	8	of	of	ADP
ajst-20197	166	9	lstm	lstm	NOUN
ajst-20197	166	10	that	that	PRON
ajst-20197	166	11	helps	help	VERB
ajst-20197	166	12	in	in	ADP
ajst-20197	166	13	the	the	DET
ajst-20197	166	14	implementation	implementation	NOUN
ajst-20197	166	15	of	of	ADP
ajst-20197	166	16	binary	binary	NOUN
ajst-20197	166	17	and	and	CCONJ
ajst-20197	166	18	fixed	fix	VERB
ajst-20197	166	19	-	-	PUNCT
ajst-20197	166	20	point	point	NOUN
ajst-20197	166	21	quantization	quantization	NOUN
ajst-20197	166	22	.	.	PUNCT
ajst-20197	167	1	referring	refer	VERB
ajst-20197	167	2	to	to	ADP
ajst-20197	167	3	the	the	DET
ajst-20197	167	4	context	context	NOUN
ajst-20197	167	5	of	of	ADP
ajst-20197	167	6	“	"	PUNCT
ajst-20197	167	7	temporal	temporal	ADJ
ajst-20197	167	8	and	and	CCONJ
ajst-20197	167	9	spatial	spatial	ADJ
ajst-20197	167	10	mechanisms	mechanism	NOUN
ajst-20197	167	11	”	"	PUNCT
ajst-20197	167	12	,	,	PUNCT
ajst-20197	167	13	there	there	PRON
ajst-20197	167	14	has	have	AUX
ajst-20197	167	15	been	be	AUX
ajst-20197	167	16	a	a	DET
ajst-20197	167	17	considerable	considerable	ADJ
ajst-20197	167	18	enhancement	enhancement	NOUN
ajst-20197	167	19	in	in	ADP
ajst-20197	167	20	the	the	DET
ajst-20197	167	21	visual	visual	ADJ
ajst-20197	167	22	systems	system	NOUN
ajst-20197	167	23	through	through	ADP
ajst-20197	167	24	the	the	DET
ajst-20197	167	25	effective	effective	ADJ
ajst-20197	167	26	diagnosis	diagnosis	NOUN
ajst-20197	167	27	of	of	ADP
ajst-20197	167	28	spatial	spatial	ADJ
ajst-20197	167	29	regions	region	NOUN
ajst-20197	167	30	and	and	CCONJ
ajst-20197	167	31	positions	position	NOUN
ajst-20197	167	32	.	.	PUNCT
ajst-20197	168	1	these	these	DET
ajst-20197	168	2	mechanisms	mechanism	NOUN
ajst-20197	168	3	help	help	VERB
ajst-20197	168	4	in	in	ADP
ajst-20197	168	5	the	the	DET
ajst-20197	168	6	gradual	gradual	ADJ
ajst-20197	168	7	increase	increase	NOUN
ajst-20197	168	8	in	in	ADP
ajst-20197	168	9	the	the	DET
ajst-20197	168	10	model	model	NOUN
ajst-20197	168	11	performance	performance	NOUN
ajst-20197	168	12	by	by	ADP
ajst-20197	168	13	analyzing	analyze	VERB
ajst-20197	168	14	location	location	NOUN
ajst-20197	168	15	and	and	CCONJ
ajst-20197	168	16	time	time	NOUN
ajst-20197	168	17	for	for	ADP
ajst-20197	168	18	capturing	capture	VERB
ajst-20197	168	19	the	the	DET
ajst-20197	168	20	information	information	NOUN
ajst-20197	168	21	flow	flow	NOUN
ajst-20197	168	22	within	within	ADP
ajst-20197	168	23	a	a	DET
ajst-20197	168	24	data	data	NOUN
ajst-20197	168	25	sequence	sequence	NOUN
ajst-20197	168	26	.	.	PUNCT
ajst-20197	169	1	there	there	PRON
ajst-20197	169	2	is	be	VERB
ajst-20197	169	3	formation	formation	NOUN
ajst-20197	169	4	of	of	ADP
ajst-20197	169	5	location	location	NOUN
ajst-20197	169	6	-	-	PUNCT
ajst-20197	169	7	sensitive	sensitive	ADJ
ajst-20197	169	8	maps	map	NOUN
ajst-20197	169	9	that	that	PRON
ajst-20197	169	10	helps	help	VERB
ajst-20197	169	11	in	in	ADP
ajst-20197	169	12	understanding	understand	VERB
ajst-20197	169	13	the	the	DET
ajst-20197	169	14	changes	change	NOUN
ajst-20197	169	15	occurring	occur	VERB
ajst-20197	169	16	within	within	ADP
ajst-20197	169	17	human	human	ADJ
ajst-20197	169	18	actions	action	NOUN
ajst-20197	169	19	with	with	ADP
ajst-20197	169	20	respect	respect	NOUN
ajst-20197	169	21	to	to	ADP
ajst-20197	169	22	time	time	NOUN
ajst-20197	169	23	.	.	PUNCT
ajst-20197	170	1	referring	refer	VERB
ajst-20197	170	2	to	to	ADP
ajst-20197	170	3	the	the	DET
ajst-20197	170	4	context	context	NOUN
ajst-20197	170	5	of	of	ADP
ajst-20197	170	6	self	self	NOUN
ajst-20197	170	7	-	-	PUNCT
ajst-20197	170	8	attention	attention	NOUN
ajst-20197	170	9	mechanisms	mechanism	NOUN
ajst-20197	170	10	,	,	PUNCT
ajst-20197	170	11	there	there	PRON
ajst-20197	170	12	has	have	AUX
ajst-20197	170	13	been	be	AUX
ajst-20197	170	14	formation	formation	NOUN
ajst-20197	170	15	of	of	ADP
ajst-20197	170	16	weighted	weight	VERB
ajst-20197	170	17	summation	summation	NOUN
ajst-20197	170	18	based	base	VERB
ajst-20197	170	19	on	on	ADP
ajst-20197	170	20	attention	attention	NOUN
ajst-20197	170	21	-	-	PUNCT
ajst-20197	170	22	map	map	NOUN
ajst-20197	170	23	scores	score	NOUN
ajst-20197	170	24	.	.	PUNCT
ajst-20197	171	1	this	this	PRON
ajst-20197	171	2	helps	help	VERB
ajst-20197	171	3	in	in	ADP
ajst-20197	171	4	enhancing	enhance	VERB
ajst-20197	171	5	prediction	prediction	NOUN
ajst-20197	171	6	accuracy	accuracy	NOUN
ajst-20197	171	7	by	by	ADP
ajst-20197	171	8	developing	develop	VERB
ajst-20197	171	9	accountability	accountability	NOUN
ajst-20197	171	10	over	over	ADP
ajst-20197	171	11	position	position	NOUN
ajst-20197	171	12	vectors	vector	NOUN
ajst-20197	171	13	[	[	X
ajst-20197	171	14	9](kardakis	9](kardakis	NUM
ajst-20197	171	15	et	et	NOUN
ajst-20197	171	16	al	al	PROPN
ajst-20197	171	17	.	.	PROPN
ajst-20197	171	18	2021	2021	NUM
ajst-20197	171	19	)	)	PUNCT
ajst-20197	171	20	.	.	PUNCT
ajst-20197	172	1	the	the	DET
ajst-20197	172	2	softmax	softmax	NOUN
ajst-20197	172	3	function	function	NOUN
ajst-20197	172	4	helps	help	VERB
ajst-20197	172	5	in	in	ADP
ajst-20197	172	6	the	the	DET
ajst-20197	172	7	development	development	NOUN
ajst-20197	172	8	of	of	ADP
ajst-20197	172	9	measurable	measurable	ADJ
ajst-20197	172	10	outcomes	outcome	NOUN
ajst-20197	172	11	by	by	ADP
ajst-20197	172	12	establishing	establish	VERB
ajst-20197	172	13	proper	proper	ADJ
ajst-20197	172	14	relationships	relationship	NOUN
ajst-20197	172	15	within	within	ADP
ajst-20197	172	16	output	output	NOUN
ajst-20197	172	17	vectors	vector	NOUN
ajst-20197	172	18	and	and	CCONJ
ajst-20197	172	19	inputs	input	NOUN
ajst-20197	172	20	.	.	PUNCT
ajst-20197	173	1	5	5	NUM
ajst-20197	173	2	.	.	X
ajst-20197	173	3	conclusion	conclusion	NOUN
ajst-20197	173	4	to	to	PART
ajst-20197	173	5	summarize	summarize	VERB
ajst-20197	173	6	,	,	PUNCT
ajst-20197	173	7	the	the	DET
ajst-20197	173	8	evolution	evolution	NOUN
ajst-20197	173	9	of	of	ADP
ajst-20197	173	10	attention	attention	NOUN
ajst-20197	173	11	and	and	CCONJ
ajst-20197	173	12	memory	memory	NOUN
ajst-20197	173	13	mechanisms	mechanism	NOUN
ajst-20197	173	14	helped	help	VERB
ajst-20197	173	15	in	in	ADP
ajst-20197	173	16	increasing	increase	VERB
ajst-20197	173	17	prediction	prediction	NOUN
ajst-20197	173	18	accuracy	accuracy	NOUN
ajst-20197	173	19	in	in	ADP
ajst-20197	173	20	deep	deep	ADJ
ajst-20197	173	21	learning	learning	NOUN
ajst-20197	173	22	.	.	PUNCT
ajst-20197	174	1	these	these	DET
ajst-20197	174	2	mechanisms	mechanism	NOUN
ajst-20197	174	3	have	have	AUX
ajst-20197	174	4	helped	help	VERB
ajst-20197	174	5	in	in	ADP
ajst-20197	174	6	the	the	DET
ajst-20197	174	7	accomplishment	accomplishment	NOUN
ajst-20197	174	8	of	of	ADP
ajst-20197	174	9	sequential	sequential	ADJ
ajst-20197	174	10	modeling	modeling	NOUN
ajst-20197	174	11	by	by	ADP
ajst-20197	174	12	analyzing	analyze	VERB
ajst-20197	174	13	vectors	vector	NOUN
ajst-20197	174	14	to	to	PART
ajst-20197	174	15	determine	determine	VERB
ajst-20197	174	16	human	human	ADJ
ajst-20197	174	17	actions	action	NOUN
ajst-20197	174	18	.	.	PUNCT
ajst-20197	175	1	lstms	lstms	PROPN
ajst-20197	175	2	are	be	AUX
ajst-20197	175	3	considered	consider	VERB
ajst-20197	175	4	to	to	PART
ajst-20197	175	5	be	be	AUX
ajst-20197	175	6	an	an	DET
ajst-20197	175	7	innovative	innovative	ADJ
ajst-20197	175	8	feature	feature	NOUN
ajst-20197	175	9	of	of	ADP
ajst-20197	175	10	rnns	rnn	NOUN
ajst-20197	175	11	as	as	SCONJ
ajst-20197	175	12	it	it	PRON
ajst-20197	175	13	helps	help	VERB
ajst-20197	175	14	in	in	ADP
ajst-20197	175	15	development	development	NOUN
ajst-20197	175	16	of	of	ADP
ajst-20197	175	17	outcomes	outcome	NOUN
ajst-20197	175	18	based	base	VERB
ajst-20197	175	19	on	on	ADP
ajst-20197	175	20	the	the	DET
ajst-20197	175	21	changes	change	NOUN
ajst-20197	175	22	occurring	occur	VERB
ajst-20197	175	23	within	within	ADP
ajst-20197	175	24	a	a	DET
ajst-20197	175	25	dataset	dataset	NOUN
ajst-20197	175	26	at	at	ADP
ajst-20197	175	27	a	a	DET
ajst-20197	175	28	particular	particular	ADJ
ajst-20197	175	29	time	time	NOUN
ajst-20197	175	30	.	.	PUNCT
ajst-20197	176	1	time	time	NOUN
ajst-20197	176	2	and	and	CCONJ
ajst-20197	176	3	location	location	NOUN
ajst-20197	176	4	are	be	AUX
ajst-20197	176	5	the	the	DET
ajst-20197	176	6	major	major	ADJ
ajst-20197	176	7	aspects	aspect	NOUN
ajst-20197	176	8	that	that	PRON
ajst-20197	176	9	are	be	AUX
ajst-20197	176	10	considered	consider	VERB
ajst-20197	176	11	by	by	ADP
ajst-20197	176	12	lstms	lstms	NOUN
ajst-20197	176	13	and	and	CCONJ
ajst-20197	176	14	manns	mann	NOUN
ajst-20197	176	15	for	for	ADP
ajst-20197	176	16	generating	generate	VERB
ajst-20197	176	17	prediction	prediction	NOUN
ajst-20197	176	18	outcomes	outcome	NOUN
ajst-20197	176	19	.	.	PUNCT
ajst-20197	177	1	these	these	DET
ajst-20197	177	2	mechanisms	mechanism	NOUN
ajst-20197	177	3	help	help	VERB
ajst-20197	177	4	in	in	ADP
ajst-20197	177	5	the	the	DET
ajst-20197	177	6	accomplishment	accomplishment	NOUN
ajst-20197	177	7	of	of	ADP
ajst-20197	177	8	dynamic	dynamic	ADJ
ajst-20197	177	9	moderation	moderation	NOUN
ajst-20197	177	10	by	by	ADP
ajst-20197	177	11	understanding	understand	VERB
ajst-20197	177	12	the	the	DET
ajst-20197	177	13	accurate	accurate	ADJ
ajst-20197	177	14	event	event	NOUN
ajst-20197	177	15	for	for	ADP
ajst-20197	177	16	developing	develop	VERB
ajst-20197	177	17	focus	focus	NOUN
ajst-20197	177	18	.	.	PUNCT
ajst-20197	178	1	gate	gate	NOUN
ajst-20197	178	2	controllers	controller	NOUN
ajst-20197	178	3	are	be	AUX
ajst-20197	178	4	deployed	deploy	VERB
ajst-20197	178	5	by	by	ADP
ajst-20197	178	6	lstm	lstm	NOUN
ajst-20197	178	7	for	for	ADP
ajst-20197	178	8	increasing	increase	VERB
ajst-20197	178	9	control	control	NOUN
ajst-20197	178	10	over	over	ADP
ajst-20197	178	11	inputs	input	NOUN
ajst-20197	178	12	.	.	PUNCT
ajst-20197	179	1	this	this	PRON
ajst-20197	179	2	helps	help	VERB
ajst-20197	179	3	in	in	ADP
ajst-20197	179	4	the	the	DET
ajst-20197	179	5	significant	significant	ADJ
ajst-20197	179	6	reduction	reduction	NOUN
ajst-20197	179	7	of	of	ADP
ajst-20197	179	8	errors	error	NOUN
ajst-20197	179	9	in	in	ADP
ajst-20197	179	10	the	the	DET
ajst-20197	179	11	dataset	dataset	NOUN
ajst-20197	179	12	and	and	CCONJ
ajst-20197	179	13	thereby	thereby	ADV
ajst-20197	179	14	helps	help	VERB
ajst-20197	179	15	in	in	ADP
ajst-20197	179	16	increasing	increase	VERB
ajst-20197	179	17	prediction	prediction	NOUN
ajst-20197	179	18	accuracy	accuracy	NOUN
ajst-20197	179	19	.	.	PUNCT
ajst-20197	180	1	further	further	ADJ
ajst-20197	180	2	investigations	investigation	NOUN
ajst-20197	180	3	could	could	AUX
ajst-20197	180	4	be	be	AUX
ajst-20197	180	5	developed	develop	VERB
ajst-20197	180	6	in	in	ADP
ajst-20197	180	7	the	the	DET
ajst-20197	180	8	future	future	NOUN
ajst-20197	180	9	concerning	concern	VERB
ajst-20197	180	10	the	the	DET
ajst-20197	180	11	architectures	architecture	NOUN
ajst-20197	180	12	of	of	ADP
ajst-20197	180	13	each	each	DET
ajst-20197	180	14	mechanism	mechanism	NOUN
ajst-20197	180	15	.	.	PUNCT
ajst-20197	181	1	this	this	PRON
ajst-20197	181	2	will	will	AUX
ajst-20197	181	3	be	be	AUX
ajst-20197	181	4	effective	effective	ADJ
ajst-20197	181	5	to	to	PART
ajst-20197	181	6	analyze	analyze	VERB
ajst-20197	181	7	the	the	DET
ajst-20197	181	8	arithmetic	arithmetic	ADJ
ajst-20197	181	9	calculations	calculation	NOUN
ajst-20197	181	10	developed	develop	VERB
ajst-20197	181	11	by	by	ADP
ajst-20197	181	12	different	different	ADJ
ajst-20197	181	13	models	model	NOUN
ajst-20197	181	14	in	in	ADP
ajst-20197	181	15	generating	generate	VERB
ajst-20197	181	16	accurate	accurate	ADJ
ajst-20197	181	17	outcomes	outcome	NOUN
ajst-20197	181	18	.	.	PUNCT
ajst-20197	182	1	references	reference	NOUN
ajst-20197	182	2	[	[	X
ajst-20197	182	3	1	1	NUM
ajst-20197	182	4	]	]	PUNCT
ajst-20197	182	5	alzubaidi	alzubaidi	NOUN
ajst-20197	182	6	,	,	PUNCT
ajst-20197	182	7	l.	l.	PROPN
ajst-20197	182	8	,	,	PUNCT
ajst-20197	182	9	zhang	zhang	PROPN
ajst-20197	182	10	,	,	PUNCT
ajst-20197	182	11	j.	j.	PROPN
ajst-20197	182	12	,	,	PUNCT
ajst-20197	182	13	humaidi	humaidi	PROPN
ajst-20197	182	14	,	,	PUNCT
ajst-20197	182	15	a.j	a.j	PROPN
ajst-20197	182	16	.	.	PROPN
ajst-20197	182	17	,	,	PUNCT
ajst-20197	182	18	al	al	PROPN
ajst-20197	182	19	-	-	PUNCT
ajst-20197	182	20	dujaili	dujaili	PROPN
ajst-20197	182	21	,	,	PUNCT
ajst-20197	182	22	a.	a.	PROPN
ajst-20197	182	23	,	,	PUNCT
ajst-20197	182	24	duan	duan	PROPN
ajst-20197	182	25	,	,	PUNCT
ajst-20197	182	26	y.	y.	PROPN
ajst-20197	182	27	,	,	PUNCT
ajst-20197	182	28	al	al	PROPN
ajst-20197	182	29	-	-	PUNCT
ajst-20197	182	30	shamma	shamma	PROPN
ajst-20197	182	31	,	,	PUNCT
ajst-20197	182	32	o.	o.	PROPN
ajst-20197	182	33	,	,	PUNCT
ajst-20197	182	34	santamaría	santamaría	VERB
ajst-20197	182	35	,	,	PUNCT
ajst-20197	182	36	j.	j.	PROPN
ajst-20197	182	37	,	,	PUNCT
ajst-20197	182	38	fadhel	fadhel	PROPN
ajst-20197	182	39	,	,	PUNCT
ajst-20197	182	40	m.a	m.a	PROPN
ajst-20197	182	41	.	.	PROPN
ajst-20197	182	42	,	,	PUNCT
ajst-20197	182	43	al	al	PROPN
ajst-20197	182	44	-	-	PUNCT
ajst-20197	182	45	amidie	amidie	NOUN
ajst-20197	182	46	,	,	PUNCT
ajst-20197	182	47	m.	m.	NOUN
ajst-20197	182	48	and	and	CCONJ
ajst-20197	182	49	farhan	farhan	PROPN
ajst-20197	182	50	,	,	PUNCT
ajst-20197	182	51	l.	l.	PROPN
ajst-20197	182	52	,	,	PUNCT
ajst-20197	182	53	2021	2021	NUM
ajst-20197	182	54	.	.	PUNCT
ajst-20197	183	1	review	review	NOUN
ajst-20197	183	2	of	of	ADP
ajst-20197	183	3	deep	deep	ADJ
ajst-20197	183	4	learning	learning	NOUN
ajst-20197	183	5	:	:	PUNCT
ajst-20197	183	6	concepts	concept	NOUN
ajst-20197	183	7	,	,	PUNCT
ajst-20197	183	8	cnn	cnn	PROPN
ajst-20197	183	9	architectures	architecture	NOUN
ajst-20197	183	10	,	,	PUNCT
ajst-20197	183	11	challenges	challenge	NOUN
ajst-20197	183	12	,	,	PUNCT
ajst-20197	183	13	applications	application	NOUN
ajst-20197	183	14	,	,	PUNCT
ajst-20197	183	15	future	future	ADJ
ajst-20197	183	16	directions	direction	NOUN
ajst-20197	183	17	.	.	PUNCT
ajst-20197	184	1	journal	journal	NOUN
ajst-20197	184	2	of	of	ADP
ajst-20197	184	3	big	big	ADJ
ajst-20197	184	4	data	datum	NOUN
ajst-20197	184	5	,	,	PUNCT
ajst-20197	184	6	8	8	NUM
ajst-20197	184	7	,	,	PUNCT
ajst-20197	184	8	pp.1	pp.1	NOUN
ajst-20197	184	9	-	-	PUNCT
ajst-20197	184	10	74	74	NUM
ajst-20197	184	11	.	.	PUNCT
ajst-20197	185	1	[	[	X
ajst-20197	185	2	2	2	NUM
ajst-20197	185	3	]	]	X
ajst-20197	185	4	chandra	chandra	PROPN
ajst-20197	185	5	,	,	PUNCT
ajst-20197	185	6	n.	n.	PROPN
ajst-20197	185	7	,	,	PUNCT
ajst-20197	185	8	ahuja	ahuja	PROPN
ajst-20197	185	9	,	,	PUNCT
ajst-20197	185	10	l.	l.	PROPN
ajst-20197	185	11	,	,	PUNCT
ajst-20197	185	12	khatri	khatri	PROPN
ajst-20197	185	13	,	,	PUNCT
ajst-20197	185	14	s.k	s.k	PROPN
ajst-20197	185	15	.	.	PROPN
ajst-20197	185	16	and	and	CCONJ
ajst-20197	185	17	monga	monga	PROPN
ajst-20197	185	18	,	,	PUNCT
ajst-20197	185	19	h.	h.	NOUN
ajst-20197	185	20	,	,	PUNCT
ajst-20197	185	21	2021	2021	NUM
ajst-20197	185	22	.	.	PUNCT
ajst-20197	186	1	utilizing	utilize	VERB
ajst-20197	186	2	gated	gate	VERB
ajst-20197	186	3	recurrent	recurrent	ADJ
ajst-20197	186	4	units	unit	NOUN
ajst-20197	186	5	to	to	PART
ajst-20197	186	6	retain	retain	VERB
ajst-20197	186	7	long	long	ADJ
ajst-20197	186	8	term	term	NOUN
ajst-20197	186	9	dependencies	dependency	NOUN
ajst-20197	186	10	with	with	ADP
ajst-20197	186	11	recurrent	recurrent	ADJ
ajst-20197	186	12	neural	neural	ADJ
ajst-20197	186	13	network	network	NOUN
ajst-20197	186	14	in	in	ADP
ajst-20197	186	15	text	text	NOUN
ajst-20197	186	16	classification	classification	NOUN
ajst-20197	186	17	.	.	PUNCT
ajst-20197	187	1	j.	j.	PROPN
ajst-20197	187	2	inf	inf	PROPN
ajst-20197	187	3	.	.	PUNCT
ajst-20197	187	4	syst	syst	PROPN
ajst-20197	187	5	.	.	PUNCT
ajst-20197	188	1	telecommun	telecommun	PROPN
ajst-20197	188	2	,	,	PUNCT
ajst-20197	188	3	2	2	NUM
ajst-20197	188	4	,	,	PUNCT
ajst-20197	188	5	p.89	p.89	NOUN
ajst-20197	188	6	.	.	PUNCT
ajst-20197	189	1	[	[	X
ajst-20197	189	2	3	3	X
ajst-20197	189	3	]	]	X
ajst-20197	189	4	dai	dai	PROPN
ajst-20197	189	5	,	,	PUNCT
ajst-20197	189	6	x.	x.	PROPN
ajst-20197	189	7	,	,	PUNCT
ajst-20197	189	8	chen	chen	PROPN
ajst-20197	189	9	,	,	PUNCT
ajst-20197	189	10	y.	y.	PROPN
ajst-20197	189	11	,	,	PUNCT
ajst-20197	189	12	xiao	xiao	PROPN
ajst-20197	189	13	,	,	PUNCT
ajst-20197	189	14	b.	b.	PROPN
ajst-20197	189	15	,	,	PUNCT
ajst-20197	189	16	chen	chen	PROPN
ajst-20197	189	17	,	,	PUNCT
ajst-20197	189	18	d.	d.	PROPN
ajst-20197	189	19	,	,	PUNCT
ajst-20197	189	20	liu	liu	PROPN
ajst-20197	189	21	,	,	PUNCT
ajst-20197	189	22	m.	m.	NOUN
ajst-20197	189	23	,	,	PUNCT
ajst-20197	189	24	yuan	yuan	PROPN
ajst-20197	189	25	,	,	PUNCT
ajst-20197	189	26	l.	l.	PROPN
ajst-20197	189	27	and	and	CCONJ
ajst-20197	189	28	zhang	zhang	PROPN
ajst-20197	189	29	,	,	PUNCT
ajst-20197	189	30	l.	l.	PROPN
ajst-20197	189	31	,	,	PUNCT
ajst-20197	189	32	2021	2021	NUM
ajst-20197	189	33	.	.	PUNCT
ajst-20197	189	34	dynamic	dynamic	ADJ
ajst-20197	189	35	head	head	NOUN
ajst-20197	189	36	:	:	PUNCT
ajst-20197	189	37	unifying	unify	VERB
ajst-20197	189	38	object	object	NOUN
ajst-20197	189	39	detection	detection	NOUN
ajst-20197	189	40	heads	head	NOUN
ajst-20197	189	41	with	with	ADP
ajst-20197	189	42	attentions	attention	NOUN
ajst-20197	189	43	.	.	PUNCT
ajst-20197	190	1	in	in	ADP
ajst-20197	190	2	proceedings	proceeding	NOUN
ajst-20197	190	3	of	of	ADP
ajst-20197	190	4	the	the	DET
ajst-20197	190	5	ieee	ieee	NOUN
ajst-20197	190	6	/	/	SYM
ajst-20197	190	7	cvf	cvf	NOUN
ajst-20197	190	8	conference	conference	NOUN
ajst-20197	190	9	on	on	ADP
ajst-20197	190	10	computer	computer	NOUN
ajst-20197	190	11	vision	vision	NOUN
ajst-20197	190	12	and	and	CCONJ
ajst-20197	190	13	pattern	pattern	NOUN
ajst-20197	190	14	recognition	recognition	NOUN
ajst-20197	190	15	(	(	PUNCT
ajst-20197	190	16	pp	pp	ADJ
ajst-20197	190	17	.	.	PUNCT
ajst-20197	191	1	7373	7373	NUM
ajst-20197	191	2	-	-	SYM
ajst-20197	191	3	7382	7382	NUM
ajst-20197	191	4	)	)	PUNCT
ajst-20197	191	5	.	.	PUNCT
ajst-20197	192	1	[	[	X
ajst-20197	192	2	4	4	NUM
ajst-20197	192	3	]	]	X
ajst-20197	192	4	de	de	PROPN
ajst-20197	192	5	santana	santana	PROPN
ajst-20197	192	6	correia	correia	PROPN
ajst-20197	192	7	,	,	PUNCT
ajst-20197	192	8	a.	a.	NOUN
ajst-20197	192	9	and	and	CCONJ
ajst-20197	192	10	colombini	colombini	PROPN
ajst-20197	192	11	,	,	PUNCT
ajst-20197	192	12	e.l	e.l	PROPN
ajst-20197	192	13	.	.	PROPN
ajst-20197	192	14	,	,	PUNCT
ajst-20197	192	15	2022	2022	NUM
ajst-20197	192	16	.	.	PUNCT
ajst-20197	193	1	attention	attention	NOUN
ajst-20197	193	2	,	,	PUNCT
ajst-20197	193	3	please	please	INTJ
ajst-20197	193	4	!	!	PUNCT
ajst-20197	194	1	a	a	DET
ajst-20197	194	2	survey	survey	NOUN
ajst-20197	194	3	of	of	ADP
ajst-20197	194	4	neural	neural	ADJ
ajst-20197	194	5	attention	attention	NOUN
ajst-20197	194	6	models	model	NOUN
ajst-20197	194	7	in	in	ADP
ajst-20197	194	8	deep	deep	ADJ
ajst-20197	194	9	learning	learning	NOUN
ajst-20197	194	10	.	.	PUNCT
ajst-20197	195	1	artificial	artificial	ADJ
ajst-20197	195	2	intelligence	intelligence	NOUN
ajst-20197	195	3	review	review	NOUN
ajst-20197	195	4	,	,	PUNCT
ajst-20197	195	5	55(8	55(8	NUM
ajst-20197	195	6	)	)	PUNCT
ajst-20197	195	7	,	,	PUNCT
ajst-20197	195	8	pp.6037	pp.6037	PROPN
ajst-20197	195	9	-	-	PUNCT
ajst-20197	195	10	6124	6124	NUM
ajst-20197	195	11	.	.	PUNCT
ajst-20197	196	1	[	[	X
ajst-20197	196	2	5	5	NUM
ajst-20197	196	3	]	]	X
ajst-20197	196	4	guo	guo	PROPN
ajst-20197	196	5	,	,	PUNCT
ajst-20197	196	6	m.h	m.h	PROPN
ajst-20197	196	7	.	.	PROPN
ajst-20197	196	8	,	,	PUNCT
ajst-20197	196	9	xu	xu	PROPN
ajst-20197	196	10	,	,	PUNCT
ajst-20197	196	11	t.x	t.x	PROPN
ajst-20197	196	12	.	.	PROPN
ajst-20197	196	13	,	,	PUNCT
ajst-20197	196	14	liu	liu	PROPN
ajst-20197	196	15	,	,	PUNCT
ajst-20197	196	16	j.j	j.j	PROPN
ajst-20197	196	17	.	.	PROPN
ajst-20197	196	18	,	,	PUNCT
ajst-20197	196	19	liu	liu	PROPN
ajst-20197	196	20	,	,	PUNCT
ajst-20197	196	21	z.n	z.n	PROPN
ajst-20197	196	22	.	.	PROPN
ajst-20197	196	23	,	,	PUNCT
ajst-20197	196	24	jiang	jiang	PROPN
ajst-20197	196	25	,	,	PUNCT
ajst-20197	196	26	p.t	p.t	PROPN
ajst-20197	196	27	.	.	PROPN
ajst-20197	196	28	,	,	PUNCT
ajst-20197	196	29	mu	mu	PROPN
ajst-20197	196	30	,	,	PUNCT
ajst-20197	196	31	t.j	t.j	PROPN
ajst-20197	196	32	.	.	PROPN
ajst-20197	196	33	,	,	PUNCT
ajst-20197	196	34	zhang	zhang	PROPN
ajst-20197	196	35	,	,	PUNCT
ajst-20197	196	36	s.h	s.h	PROPN
ajst-20197	196	37	.	.	PROPN
ajst-20197	196	38	,	,	PUNCT
ajst-20197	196	39	martin	martin	PROPN
ajst-20197	196	40	,	,	PUNCT
ajst-20197	196	41	r.r	r.r	PROPN
ajst-20197	196	42	.	.	PROPN
ajst-20197	196	43	,	,	PUNCT
ajst-20197	196	44	cheng	cheng	PROPN
ajst-20197	196	45	,	,	PUNCT
ajst-20197	196	46	m.m	m.m	PROPN
ajst-20197	196	47	.	.	PROPN
ajst-20197	196	48	and	and	CCONJ
ajst-20197	196	49	hu	hu	PROPN
ajst-20197	196	50	,	,	PUNCT
ajst-20197	196	51	s.m	s.m	PROPN
ajst-20197	196	52	.	.	PROPN
ajst-20197	196	53	,	,	PUNCT
ajst-20197	196	54	2022	2022	NUM
ajst-20197	196	55	.	.	PUNCT
ajst-20197	197	1	attention	attention	NOUN
ajst-20197	197	2	mechanisms	mechanism	NOUN
ajst-20197	197	3	in	in	ADP
ajst-20197	197	4	computer	computer	NOUN
ajst-20197	197	5	vision	vision	NOUN
ajst-20197	197	6	:	:	PUNCT
ajst-20197	197	7	a	a	DET
ajst-20197	197	8	survey	survey	NOUN
ajst-20197	197	9	.	.	PUNCT
ajst-20197	198	1	computational	computational	ADJ
ajst-20197	198	2	visual	visual	ADJ
ajst-20197	198	3	media	medium	NOUN
ajst-20197	198	4	,	,	PUNCT
ajst-20197	198	5	8(3	8(3	NUM
ajst-20197	198	6	)	)	PUNCT
ajst-20197	198	7	,	,	PUNCT
ajst-20197	198	8	pp.331	pp.331	PROPN
ajst-20197	198	9	-	-	PUNCT
ajst-20197	198	10	368	368	NUM
ajst-20197	198	11	.	.	PUNCT
ajst-20197	199	1	[	[	X
ajst-20197	199	2	6	6	NUM
ajst-20197	199	3	]	]	X
ajst-20197	199	4	hafiz	hafiz	PROPN
ajst-20197	199	5	,	,	PUNCT
ajst-20197	199	6	a.m.	a.m.	ADV
ajst-20197	199	7	,	,	PUNCT
ajst-20197	199	8	parah	parah	PROPN
ajst-20197	199	9	,	,	PUNCT
ajst-20197	199	10	s.a	s.a	PROPN
ajst-20197	199	11	.	.	PROPN
ajst-20197	199	12	and	and	CCONJ
ajst-20197	199	13	bhat	bhat	PROPN
ajst-20197	199	14	,	,	PUNCT
ajst-20197	199	15	r.u.a	r.u.a	PROPN
ajst-20197	199	16	.	.	PROPN
ajst-20197	199	17	,	,	PUNCT
ajst-20197	199	18	2021	2021	NUM
ajst-20197	199	19	.	.	PUNCT
ajst-20197	200	1	attention	attention	NOUN
ajst-20197	200	2	mechanisms	mechanism	NOUN
ajst-20197	200	3	and	and	CCONJ
ajst-20197	200	4	deep	deep	ADJ
ajst-20197	200	5	learning	learning	NOUN
ajst-20197	200	6	for	for	ADP
ajst-20197	200	7	machine	machine	NOUN
ajst-20197	200	8	vision	vision	NOUN
ajst-20197	200	9	:	:	PUNCT
ajst-20197	200	10	a	a	DET
ajst-20197	200	11	survey	survey	NOUN
ajst-20197	200	12	of	of	ADP
ajst-20197	200	13	the	the	DET
ajst-20197	200	14	state	state	NOUN
ajst-20197	200	15	of	of	ADP
ajst-20197	200	16	the	the	DET
ajst-20197	200	17	art	art	NOUN
ajst-20197	200	18	.	.	PUNCT
ajst-20197	201	1	arxiv	arxiv	PROPN
ajst-20197	201	2	preprint	preprint	NOUN
ajst-20197	201	3	arxiv:2106.07550	arxiv:2106.07550	ADV
ajst-20197	201	4	.	.	PUNCT
ajst-20197	202	1	[	[	X
ajst-20197	202	2	7	7	NUM
ajst-20197	202	3	]	]	X
ajst-20197	202	4	hernández	hernández	PROPN
ajst-20197	202	5	,	,	PUNCT
ajst-20197	202	6	a.	a.	NOUN
ajst-20197	202	7	and	and	CCONJ
ajst-20197	202	8	amigó	amigó	NOUN
ajst-20197	202	9	,	,	PUNCT
ajst-20197	202	10	j.m	j.m	PROPN
ajst-20197	202	11	.	.	PROPN
ajst-20197	202	12	,	,	PUNCT
ajst-20197	202	13	2021	2021	NUM
ajst-20197	202	14	.	.	PUNCT
ajst-20197	203	1	attention	attention	NOUN
ajst-20197	203	2	mechanisms	mechanism	NOUN
ajst-20197	203	3	and	and	CCONJ
ajst-20197	203	4	their	their	PRON
ajst-20197	203	5	applications	application	NOUN
ajst-20197	203	6	to	to	ADP
ajst-20197	203	7	complex	complex	ADJ
ajst-20197	203	8	systems	system	NOUN
ajst-20197	203	9	.	.	PUNCT
ajst-20197	204	1	entropy	entropy	PROPN
ajst-20197	204	2	,	,	PUNCT
ajst-20197	204	3	23(3	23(3	NUM
ajst-20197	204	4	)	)	PUNCT
ajst-20197	204	5	,	,	PUNCT
ajst-20197	204	6	p.283	p.283	NOUN
ajst-20197	204	7	.	.	PUNCT
ajst-20197	205	1	[	[	X
ajst-20197	205	2	8	8	NUM
ajst-20197	205	3	]	]	X
ajst-20197	205	4	jordan	jordan	PROPN
ajst-20197	205	5	,	,	PUNCT
ajst-20197	205	6	i.d	i.d	PROPN
ajst-20197	205	7	.	.	PROPN
ajst-20197	205	8	,	,	PUNCT
ajst-20197	205	9	sokół	sokół	PROPN
ajst-20197	205	10	,	,	PUNCT
ajst-20197	205	11	p.a	p.a	PROPN
ajst-20197	205	12	.	.	PROPN
ajst-20197	205	13	and	and	CCONJ
ajst-20197	205	14	park	park	NOUN
ajst-20197	205	15	,	,	PUNCT
ajst-20197	205	16	i.m	i.m	PROPN
ajst-20197	205	17	.	.	PROPN
ajst-20197	205	18	,	,	PUNCT
ajst-20197	205	19	2021	2021	NUM
ajst-20197	205	20	.	.	PUNCT
ajst-20197	205	21	gated	gate	VERB
ajst-20197	205	22	recurrent	recurrent	ADJ
ajst-20197	205	23	units	unit	NOUN
ajst-20197	205	24	viewed	view	VERB
ajst-20197	205	25	through	through	ADP
ajst-20197	205	26	the	the	DET
ajst-20197	205	27	lens	lens	NOUN
ajst-20197	205	28	of	of	ADP
ajst-20197	205	29	continuous	continuous	ADJ
ajst-20197	205	30	time	time	NOUN
ajst-20197	205	31	dynamical	dynamical	ADJ
ajst-20197	205	32	systems	system	NOUN
ajst-20197	205	33	.	.	PUNCT
ajst-20197	206	1	frontiers	frontier	NOUN
ajst-20197	206	2	in	in	ADP
ajst-20197	206	3	computational	computational	ADJ
ajst-20197	206	4	neuroscience	neuroscience	NOUN
ajst-20197	206	5	,	,	PUNCT
ajst-20197	206	6	15	15	NUM
ajst-20197	206	7	,	,	PUNCT
ajst-20197	206	8	p.678158	p.678158	NOUN
ajst-20197	206	9	.	.	PUNCT
ajst-20197	207	1	[	[	X
ajst-20197	207	2	9	9	NUM
ajst-20197	207	3	]	]	PUNCT
ajst-20197	207	4	kardakis	kardaki	NOUN
ajst-20197	207	5	,	,	PUNCT
ajst-20197	207	6	s.	s.	PROPN
ajst-20197	207	7	,	,	PUNCT
ajst-20197	207	8	perikos	perikos	PROPN
ajst-20197	207	9	,	,	PUNCT
ajst-20197	207	10	i.	i.	PROPN
ajst-20197	207	11	,	,	PUNCT
ajst-20197	207	12	grivokostopoulou	grivokostopoulou	PROPN
ajst-20197	207	13	,	,	PUNCT
ajst-20197	207	14	f.	f.	PROPN
ajst-20197	207	15	and	and	CCONJ
ajst-20197	207	16	hatzilygeroudis	hatzilygeroudis	PROPN
ajst-20197	207	17	,	,	PUNCT
ajst-20197	207	18	i.	i.	NOUN
ajst-20197	207	19	,	,	PUNCT
ajst-20197	207	20	2021	2021	NUM
ajst-20197	207	21	.	.	PUNCT
ajst-20197	208	1	examining	examine	VERB
ajst-20197	208	2	attention	attention	NOUN
ajst-20197	208	3	mechanisms	mechanism	NOUN
ajst-20197	208	4	in	in	ADP
ajst-20197	208	5	deep	deep	ADJ
ajst-20197	208	6	learning	learning	NOUN
ajst-20197	208	7	models	model	NOUN
ajst-20197	208	8	for	for	ADP
ajst-20197	208	9	sentiment	sentiment	NOUN
ajst-20197	208	10	analysis	analysis	NOUN
ajst-20197	208	11	.	.	PUNCT
ajst-20197	209	1	applied	apply	VERB
ajst-20197	209	2	sciences	science	NOUN
ajst-20197	209	3	,	,	PUNCT
ajst-20197	209	4	11(9	11(9	PROPN
ajst-20197	209	5	)	)	PUNCT
ajst-20197	209	6	,	,	PUNCT
ajst-20197	209	7	p.3883	p.3883	NOUN
ajst-20197	209	8	.	.	PUNCT
ajst-20197	210	1	[	[	X
ajst-20197	210	2	10	10	NUM
ajst-20197	210	3	]	]	X
ajst-20197	210	4	kossen	kossen	PROPN
ajst-20197	210	5	,	,	PUNCT
ajst-20197	210	6	j.	j.	PROPN
ajst-20197	210	7	,	,	PUNCT
ajst-20197	210	8	band	band	PROPN
ajst-20197	210	9	,	,	PUNCT
ajst-20197	210	10	n.	n.	NOUN
ajst-20197	210	11	,	,	PUNCT
ajst-20197	210	12	lyle	lyle	PROPN
ajst-20197	210	13	,	,	PUNCT
ajst-20197	210	14	c.	c.	PROPN
ajst-20197	210	15	,	,	PUNCT
ajst-20197	210	16	gomez	gomez	PROPN
ajst-20197	210	17	,	,	PUNCT
ajst-20197	210	18	a.n	a.n	PROPN
ajst-20197	210	19	.	.	PROPN
ajst-20197	210	20	,	,	PUNCT
ajst-20197	210	21	rainforth	rainforth	NOUN
ajst-20197	210	22	,	,	PUNCT
ajst-20197	210	23	t.	t.	NOUN
ajst-20197	210	24	and	and	CCONJ
ajst-20197	210	25	gal	gal	PROPN
ajst-20197	210	26	,	,	PUNCT
ajst-20197	210	27	y.	y.	NOUN
ajst-20197	210	28	,	,	PUNCT
ajst-20197	210	29	2021	2021	NUM
ajst-20197	210	30	.	.	PUNCT
ajst-20197	211	1	self	self	NOUN
ajst-20197	211	2	-	-	PUNCT
ajst-20197	211	3	attention	attention	NOUN
ajst-20197	211	4	between	between	ADP
ajst-20197	211	5	datapoints	datapoint	NOUN
ajst-20197	211	6	:	:	PUNCT
ajst-20197	211	7	going	go	VERB
ajst-20197	211	8	113	113	NUM
ajst-20197	211	9	beyond	beyond	ADP
ajst-20197	211	10	individual	individual	ADJ
ajst-20197	211	11	input	input	NOUN
ajst-20197	211	12	-	-	PUNCT
ajst-20197	211	13	output	output	NOUN
ajst-20197	211	14	pairs	pair	NOUN
ajst-20197	211	15	in	in	ADP
ajst-20197	211	16	deep	deep	ADJ
ajst-20197	211	17	learning	learning	NOUN
ajst-20197	211	18	.	.	PUNCT
ajst-20197	212	1	advances	advance	NOUN
ajst-20197	212	2	in	in	ADP
ajst-20197	212	3	neural	neural	ADJ
ajst-20197	212	4	information	information	NOUN
ajst-20197	212	5	processing	processing	NOUN
ajst-20197	212	6	systems	system	NOUN
ajst-20197	212	7	,	,	PUNCT
ajst-20197	212	8	34	34	NUM
ajst-20197	212	9	,	,	PUNCT
ajst-20197	212	10	pp.28742	pp.28742	PROPN
ajst-20197	212	11	-	-	PUNCT
ajst-20197	212	12	28756	28756	NUM
ajst-20197	212	13	.	.	PUNCT
ajst-20197	213	1	[	[	X
ajst-20197	213	2	11	11	NUM
ajst-20197	213	3	]	]	PUNCT
ajst-20197	213	4	landi	landi	PROPN
ajst-20197	213	5	,	,	PUNCT
ajst-20197	213	6	f.	f.	PROPN
ajst-20197	213	7	,	,	PUNCT
ajst-20197	213	8	baraldi	baraldi	PROPN
ajst-20197	213	9	,	,	PUNCT
ajst-20197	213	10	l.	l.	PROPN
ajst-20197	213	11	,	,	PUNCT
ajst-20197	213	12	cornia	cornia	PROPN
ajst-20197	213	13	,	,	PUNCT
ajst-20197	213	14	m.	m.	NOUN
ajst-20197	213	15	and	and	CCONJ
ajst-20197	213	16	cucchiara	cucchiara	PROPN
ajst-20197	213	17	,	,	PUNCT
ajst-20197	213	18	r.	r.	PROPN
ajst-20197	213	19	,	,	PUNCT
ajst-20197	213	20	2021	2021	NUM
ajst-20197	213	21	.	.	PUNCT
ajst-20197	214	1	working	work	VERB
ajst-20197	214	2	memory	memory	NOUN
ajst-20197	214	3	connections	connection	NOUN
ajst-20197	214	4	for	for	ADP
ajst-20197	214	5	lstm	lstm	NOUN
ajst-20197	214	6	.	.	PUNCT
ajst-20197	215	1	neural	neural	ADJ
ajst-20197	215	2	networks	network	NOUN
ajst-20197	215	3	,	,	PUNCT
ajst-20197	215	4	144	144	NUM
ajst-20197	215	5	,	,	PUNCT
ajst-20197	215	6	pp.334	pp.334	NOUN
ajst-20197	215	7	-	-	PUNCT
ajst-20197	215	8	341	341	NUM
ajst-20197	215	9	.	.	PUNCT
ajst-20197	216	1	[	[	X
ajst-20197	216	2	12	12	NUM
ajst-20197	216	3	]	]	X
ajst-20197	216	4	lau	lau	PROPN
ajst-20197	216	5	,	,	PUNCT
ajst-20197	216	6	k.w	k.w	PROPN
ajst-20197	216	7	.	.	PROPN
ajst-20197	216	8	,	,	PUNCT
ajst-20197	216	9	po	po	PROPN
ajst-20197	216	10	,	,	PUNCT
ajst-20197	216	11	l.m	l.m	PROPN
ajst-20197	216	12	.	.	PROPN
ajst-20197	216	13	and	and	CCONJ
ajst-20197	216	14	rehman	rehman	PROPN
ajst-20197	216	15	,	,	PUNCT
ajst-20197	216	16	y.a.u	y.a.u	PROPN
ajst-20197	216	17	.	.	PUNCT
ajst-20197	216	18	,	,	PUNCT
ajst-20197	216	19	2024	2024	NUM
ajst-20197	216	20	.	.	PUNCT
ajst-20197	216	21	large	large	ADJ
ajst-20197	216	22	separable	separable	ADJ
ajst-20197	216	23	kernel	kernel	PROPN
ajst-20197	216	24	attention	attention	NOUN
ajst-20197	216	25	:	:	PUNCT
ajst-20197	216	26	rethinking	rethink	VERB
ajst-20197	216	27	the	the	DET
ajst-20197	216	28	large	large	ADJ
ajst-20197	216	29	kernel	kernel	NOUN
ajst-20197	216	30	attention	attention	NOUN
ajst-20197	216	31	design	design	NOUN
ajst-20197	216	32	in	in	ADP
ajst-20197	216	33	cnn	cnn	PROPN
ajst-20197	216	34	.	.	PUNCT
ajst-20197	217	1	expert	expert	NOUN
ajst-20197	217	2	systems	system	NOUN
ajst-20197	217	3	with	with	ADP
ajst-20197	217	4	applications	application	NOUN
ajst-20197	217	5	,	,	PUNCT
ajst-20197	217	6	236	236	NUM
ajst-20197	217	7	,	,	PUNCT
ajst-20197	217	8	p.121352	p.121352	ADV
ajst-20197	217	9	.	.	PUNCT
ajst-20197	218	1	[	[	X
ajst-20197	218	2	13	13	NUM
ajst-20197	218	3	]	]	SYM
ajst-20197	218	4	le	le	X
ajst-20197	218	5	,	,	PUNCT
ajst-20197	218	6	h.t	h.t	PROPN
ajst-20197	218	7	.	.	PROPN
ajst-20197	218	8	,	,	PUNCT
ajst-20197	218	9	2021	2021	NUM
ajst-20197	218	10	.	.	PUNCT
ajst-20197	218	11	memory	memory	NOUN
ajst-20197	218	12	and	and	CCONJ
ajst-20197	218	13	attention	attention	NOUN
ajst-20197	218	14	in	in	ADP
ajst-20197	218	15	deep	deep	ADJ
ajst-20197	218	16	learning	learning	NOUN
ajst-20197	218	17	,	,	PUNCT
ajst-20197	218	18	1	1	X
ajst-20197	218	19	.	.	X
ajst-20197	218	20	doi	doi	NOUN
ajst-20197	218	21	:	:	PUNCT
ajst-20197	218	22	https://doi.org/10.48550	https://doi.org/10.48550	PROPN
ajst-20197	218	23	/	/	SYM
ajst-20197	218	24	arxiv.2107.01390	arxiv.2107.01390	PROPN
ajst-20197	218	25	.	.	PUNCT
ajst-20197	219	1	[	[	X
ajst-20197	219	2	14	14	NUM
ajst-20197	219	3	]	]	X
ajst-20197	219	4	lin	lin	PROPN
ajst-20197	219	5	,	,	PUNCT
ajst-20197	219	6	z.	z.	PROPN
ajst-20197	219	7	,	,	PUNCT
ajst-20197	219	8	li	li	PROPN
ajst-20197	219	9	,	,	PUNCT
ajst-20197	219	10	m.	m.	NOUN
ajst-20197	219	11	,	,	PUNCT
ajst-20197	219	12	zheng	zheng	PROPN
ajst-20197	219	13	,	,	PUNCT
ajst-20197	219	14	z.	z.	PROPN
ajst-20197	219	15	,	,	PUNCT
ajst-20197	219	16	cheng	cheng	PROPN
ajst-20197	219	17	,	,	PUNCT
ajst-20197	219	18	y.	y.	PROPN
ajst-20197	219	19	and	and	CCONJ
ajst-20197	219	20	yuan	yuan	PROPN
ajst-20197	219	21	,	,	PUNCT
ajst-20197	219	22	c.	c.	PROPN
ajst-20197	219	23	,	,	PUNCT
ajst-20197	219	24	2020	2020	NUM
ajst-20197	219	25	,	,	PUNCT
ajst-20197	219	26	april	april	PROPN
ajst-20197	219	27	.	.	PUNCT
ajst-20197	220	1	self	self	NOUN
ajst-20197	220	2	-	-	PUNCT
ajst-20197	220	3	attention	attention	NOUN
ajst-20197	220	4	convlstm	convlstm	NOUN
ajst-20197	220	5	forspatiotemporal	forspatiotemporal	ADJ
ajst-20197	220	6	prediction	prediction	NOUN
ajst-20197	220	7	.	.	PUNCT
ajst-20197	221	1	in	in	ADP
ajst-20197	221	2	proceedings	proceeding	NOUN
ajst-20197	221	3	of	of	ADP
ajst-20197	221	4	the	the	DET
ajst-20197	221	5	aaai	aaai	PROPN
ajst-20197	221	6	conference	conference	NOUN
ajst-20197	221	7	on	on	ADP
ajst-20197	221	8	artificial	artificial	ADJ
ajst-20197	221	9	intelligence	intelligence	NOUN
ajst-20197	221	10	(	(	PUNCT
ajst-20197	221	11	vol	vol	NOUN
ajst-20197	221	12	.	.	PROPN
ajst-20197	222	1	34	34	NUM
ajst-20197	222	2	,	,	PUNCT
ajst-20197	222	3	no	no	INTJ
ajst-20197	222	4	.	.	NOUN
ajst-20197	222	5	07	07	NUM
ajst-20197	222	6	,	,	PUNCT
ajst-20197	222	7	pp	pp	ADJ
ajst-20197	222	8	.	.	PUNCT
ajst-20197	222	9	11531	11531	NUM
ajst-20197	222	10	-	-	SYM
ajst-20197	222	11	11538	11538	NUM
ajst-20197	222	12	)	)	PUNCT
ajst-20197	222	13	.	.	PUNCT
ajst-20197	223	1	[	[	X
ajst-20197	223	2	15	15	NUM
ajst-20197	223	3	]	]	X
ajst-20197	223	4	nassaji	nassaji	PROPN
ajst-20197	223	5	,	,	PUNCT
ajst-20197	223	6	h.	h.	PROPN
ajst-20197	223	7	,	,	PUNCT
ajst-20197	223	8	2020	2020	NUM
ajst-20197	223	9	.	.	PUNCT
ajst-20197	224	1	good	good	ADJ
ajst-20197	224	2	qualitative	qualitative	ADJ
ajst-20197	224	3	research	research	NOUN
ajst-20197	224	4	.	.	PUNCT
ajst-20197	225	1	language	language	NOUN
ajst-20197	225	2	teaching	teaching	NOUN
ajst-20197	225	3	research	research	NOUN
ajst-20197	225	4	,	,	PUNCT
ajst-20197	225	5	24(4	24(4	NUM
ajst-20197	225	6	)	)	PUNCT
ajst-20197	225	7	,	,	PUNCT
ajst-20197	225	8	pp.427	pp.427	PROPN
ajst-20197	225	9	-	-	PUNCT
ajst-20197	225	10	431	431	NUM
ajst-20197	225	11	.	.	PUNCT
ajst-20197	226	1	[	[	X
ajst-20197	226	2	16	16	NUM
ajst-20197	226	3	]	]	X
ajst-20197	226	4	pan	pan	PROPN
ajst-20197	226	5	,	,	PUNCT
ajst-20197	226	6	x.	x.	PROPN
ajst-20197	226	7	,	,	PUNCT
ajst-20197	226	8	ge	ge	PROPN
ajst-20197	226	9	,	,	PUNCT
ajst-20197	226	10	c.	c.	PROPN
ajst-20197	226	11	,	,	PUNCT
ajst-20197	226	12	lu	lu	PROPN
ajst-20197	226	13	,	,	PUNCT
ajst-20197	226	14	r.	r.	PROPN
ajst-20197	226	15	,	,	PUNCT
ajst-20197	226	16	song	song	NOUN
ajst-20197	226	17	,	,	PUNCT
ajst-20197	226	18	s.	s.	PROPN
ajst-20197	226	19	,	,	PUNCT
ajst-20197	226	20	chen	chen	PROPN
ajst-20197	226	21	,	,	PUNCT
ajst-20197	226	22	g.	g.	PROPN
ajst-20197	226	23	,	,	PUNCT
ajst-20197	226	24	huang	huang	PROPN
ajst-20197	226	25	,	,	PUNCT
ajst-20197	226	26	z.	z.	PROPN
ajst-20197	226	27	and	and	CCONJ
ajst-20197	226	28	huang	huang	PROPN
ajst-20197	226	29	,	,	PUNCT
ajst-20197	226	30	g.	g.	PROPN
ajst-20197	226	31	,	,	PUNCT
ajst-20197	226	32	2022	2022	NUM
ajst-20197	226	33	.	.	PUNCT
ajst-20197	227	1	on	on	ADP
ajst-20197	227	2	the	the	DET
ajst-20197	227	3	integration	integration	NOUN
ajst-20197	227	4	of	of	ADP
ajst-20197	227	5	self	self	NOUN
ajst-20197	227	6	-	-	PUNCT
ajst-20197	227	7	attention	attention	NOUN
ajst-20197	227	8	and	and	CCONJ
ajst-20197	227	9	convolution	convolution	NOUN
ajst-20197	227	10	.	.	PUNCT
ajst-20197	228	1	in	in	ADP
ajst-20197	228	2	proceedings	proceeding	NOUN
ajst-20197	228	3	of	of	ADP
ajst-20197	228	4	the	the	DET
ajst-20197	228	5	ieee	ieee	NOUN
ajst-20197	228	6	/	/	SYM
ajst-20197	228	7	cvf	cvf	NOUN
ajst-20197	228	8	conference	conference	NOUN
ajst-20197	228	9	on	on	ADP
ajst-20197	228	10	computer	computer	NOUN
ajst-20197	228	11	vision	vision	NOUN
ajst-20197	228	12	and	and	CCONJ
ajst-20197	228	13	pattern	pattern	NOUN
ajst-20197	228	14	recognition	recognition	NOUN
ajst-20197	228	15	(	(	PUNCT
ajst-20197	228	16	pp	pp	PROPN
ajst-20197	228	17	.	.	PUNCT
ajst-20197	228	18	815	815	NUM
ajst-20197	228	19	-	-	NUM
ajst-20197	228	20	825	825	NUM
ajst-20197	228	21	)	)	PUNCT
ajst-20197	228	22	.	.	PUNCT
ajst-20197	229	1	[	[	X
ajst-20197	229	2	17	17	NUM
ajst-20197	229	3	]	]	PUNCT
ajst-20197	229	4	park	park	NOUN
ajst-20197	229	5	,	,	PUNCT
ajst-20197	229	6	s.	s.	PROPN
ajst-20197	229	7	,	,	PUNCT
ajst-20197	229	8	kim	kim	PROPN
ajst-20197	229	9	,	,	PUNCT
ajst-20197	229	10	s.	s.	PROPN
ajst-20197	229	11	,	,	PUNCT
ajst-20197	229	12	lee	lee	PROPN
ajst-20197	229	13	,	,	PUNCT
ajst-20197	229	14	s.	s.	PROPN
ajst-20197	229	15	,	,	PUNCT
ajst-20197	229	16	bae	bae	PROPN
ajst-20197	229	17	,	,	PUNCT
ajst-20197	229	18	h.	h.	PROPN
ajst-20197	229	19	and	and	CCONJ
ajst-20197	229	20	yoon	yoon	PROPN
ajst-20197	229	21	,	,	PUNCT
ajst-20197	229	22	s.	s.	PROPN
ajst-20197	229	23	,	,	PUNCT
ajst-20197	229	24	2018	2018	NUM
ajst-20197	229	25	,	,	PUNCT
ajst-20197	229	26	april	april	PROPN
ajst-20197	229	27	.	.	PUNCT
ajst-20197	230	1	quantized	quantize	VERB
ajst-20197	230	2	memory	memory	NOUN
ajst-20197	230	3	-	-	PUNCT
ajst-20197	230	4	augmented	augment	VERB
ajst-20197	230	5	neural	neural	ADJ
ajst-20197	230	6	networks	network	NOUN
ajst-20197	230	7	.	.	PUNCT
ajst-20197	231	1	in	in	ADP
ajst-20197	231	2	proceedings	proceeding	NOUN
ajst-20197	231	3	of	of	ADP
ajst-20197	231	4	the	the	DET
ajst-20197	231	5	aaai	aaai	PROPN
ajst-20197	231	6	conference	conference	NOUN
ajst-20197	231	7	on	on	ADP
ajst-20197	231	8	artificial	artificial	ADJ
ajst-20197	231	9	intelligence	intelligence	NOUN
ajst-20197	231	10	(	(	PUNCT
ajst-20197	231	11	vol	vol	NOUN
ajst-20197	231	12	.	.	PROPN
ajst-20197	231	13	32	32	NUM
ajst-20197	231	14	,	,	PUNCT
ajst-20197	231	15	no	no	INTJ
ajst-20197	231	16	.	.	NOUN
ajst-20197	231	17	1	1	NUM
ajst-20197	231	18	)	)	PUNCT
ajst-20197	231	19	.	.	PUNCT
ajst-20197	232	1	[	[	X
ajst-20197	232	2	18	18	NUM
ajst-20197	232	3	]	]	X
ajst-20197	232	4	rafiq	rafiq	PROPN
ajst-20197	232	5	,	,	PUNCT
ajst-20197	232	6	g.	g.	PROPN
ajst-20197	232	7	,	,	PUNCT
ajst-20197	232	8	rafiq	rafiq	PROPN
ajst-20197	232	9	,	,	PUNCT
ajst-20197	232	10	m.	m.	NOUN
ajst-20197	232	11	and	and	CCONJ
ajst-20197	232	12	choi	choi	NOUN
ajst-20197	232	13	,	,	PUNCT
ajst-20197	232	14	g.s	g.s	PROPN
ajst-20197	232	15	.	.	PROPN
ajst-20197	232	16	,	,	PUNCT
ajst-20197	232	17	2023	2023	NUM
ajst-20197	232	18	.	.	PUNCT
ajst-20197	232	19	video	video	NOUN
ajst-20197	232	20	description	description	NOUN
ajst-20197	232	21	:	:	PUNCT
ajst-20197	232	22	a	a	DET
ajst-20197	232	23	comprehensive	comprehensive	ADJ
ajst-20197	232	24	survey	survey	NOUN
ajst-20197	232	25	of	of	ADP
ajst-20197	232	26	deep	deep	ADJ
ajst-20197	232	27	learning	learning	NOUN
ajst-20197	232	28	approaches	approach	NOUN
ajst-20197	232	29	.	.	PUNCT
ajst-20197	233	1	artificial	artificial	ADJ
ajst-20197	233	2	intelligence	intelligence	NOUN
ajst-20197	233	3	review	review	NOUN
ajst-20197	233	4	,	,	PUNCT
ajst-20197	233	5	pp.1	pp.1	NOUN
ajst-20197	233	6	-	-	PUNCT
ajst-20197	233	7	80	80	NUM
ajst-20197	233	8	.	.	PUNCT
ajst-20197	234	1	[	[	X
ajst-20197	234	2	19	19	NUM
ajst-20197	234	3	]	]	X
ajst-20197	234	4	ragab	ragab	NOUN
ajst-20197	234	5	,	,	PUNCT
ajst-20197	234	6	m.	m.	NOUN
ajst-20197	234	7	,	,	PUNCT
ajst-20197	234	8	chen	chen	PROPN
ajst-20197	234	9	,	,	PUNCT
ajst-20197	234	10	z.	z.	PROPN
ajst-20197	234	11	,	,	PUNCT
ajst-20197	234	12	wu	wu	PROPN
ajst-20197	234	13	,	,	PUNCT
ajst-20197	234	14	m.	m.	NOUN
ajst-20197	234	15	,	,	PUNCT
ajst-20197	234	16	kwoh	kwoh	PROPN
ajst-20197	234	17	,	,	PUNCT
ajst-20197	234	18	c.k	c.k	PROPN
ajst-20197	234	19	.	.	PROPN
ajst-20197	234	20	,	,	PUNCT
ajst-20197	234	21	yan	yan	PROPN
ajst-20197	234	22	,	,	PUNCT
ajst-20197	234	23	r.	r.	PROPN
ajst-20197	234	24	and	and	CCONJ
ajst-20197	234	25	li	li	PROPN
ajst-20197	234	26	,	,	PUNCT
ajst-20197	234	27	x.	x.	NOUN
ajst-20197	234	28	,	,	PUNCT
ajst-20197	234	29	2020	2020	NUM
ajst-20197	234	30	.	.	PUNCT
ajst-20197	235	1	attention	attention	NOUN
ajst-20197	235	2	sequence	sequence	NOUN
ajst-20197	235	3	to	to	ADP
ajst-20197	235	4	sequence	sequence	NOUN
ajst-20197	235	5	model	model	NOUN
ajst-20197	235	6	for	for	ADP
ajst-20197	235	7	machine	machine	NOUN
ajst-20197	235	8	remaining	remain	VERB
ajst-20197	235	9	useful	useful	ADJ
ajst-20197	235	10	life	life	NOUN
ajst-20197	235	11	prediction	prediction	NOUN
ajst-20197	235	12	.	.	PUNCT
ajst-20197	236	1	arxiv	arxiv	PROPN
ajst-20197	236	2	preprint	preprint	NOUN
ajst-20197	236	3	arxiv:2007.09868	arxiv:2007.09868	NOUN
ajst-20197	236	4	.	.	PUNCT
ajst-20197	237	1	[	[	X
ajst-20197	237	2	20	20	NUM
ajst-20197	237	3	]	]	PUNCT
ajst-20197	237	4	rajamani	rajamani	NOUN
ajst-20197	237	5	,	,	PUNCT
ajst-20197	237	6	s.t	s.t	PROPN
ajst-20197	237	7	.	.	PROPN
ajst-20197	237	8	,	,	PUNCT
ajst-20197	237	9	rajamani	rajamani	PROPN
ajst-20197	237	10	,	,	PUNCT
ajst-20197	237	11	k.t	k.t	PROPN
ajst-20197	237	12	.	.	PROPN
ajst-20197	237	13	,	,	PUNCT
ajst-20197	237	14	mallol	mallol	NOUN
ajst-20197	237	15	-	-	PUNCT
ajst-20197	237	16	ragolta	ragolta	NOUN
ajst-20197	237	17	,	,	PUNCT
ajst-20197	237	18	a.	a.	NOUN
ajst-20197	237	19	,	,	PUNCT
ajst-20197	237	20	liu	liu	PROPN
ajst-20197	237	21	,	,	PUNCT
ajst-20197	237	22	s.	s.	PROPN
ajst-20197	237	23	and	and	CCONJ
ajst-20197	237	24	schuller	schuller	PROPN
ajst-20197	237	25	,	,	PUNCT
ajst-20197	237	26	b.	b.	PROPN
ajst-20197	237	27	,	,	PUNCT
ajst-20197	237	28	2021	2021	NUM
ajst-20197	237	29	,	,	PUNCT
ajst-20197	237	30	june	june	PROPN
ajst-20197	237	31	.	.	PUNCT
ajst-20197	238	1	a	a	DET
ajst-20197	238	2	novel	novel	ADJ
ajst-20197	238	3	attention	attention	NOUN
ajst-20197	238	4	-	-	PUNCT
ajst-20197	238	5	based	base	VERB
ajst-20197	238	6	gated	gate	VERB
ajst-20197	238	7	recurrent	recurrent	ADJ
ajst-20197	238	8	unit	unit	NOUN
ajst-20197	238	9	and	and	CCONJ
ajst-20197	238	10	its	its	PRON
ajst-20197	238	11	efficacy	efficacy	NOUN
ajst-20197	238	12	in	in	ADP
ajst-20197	238	13	speech	speech	NOUN
ajst-20197	238	14	emotion	emotion	NOUN
ajst-20197	238	15	recognition	recognition	NOUN
ajst-20197	238	16	.	.	PUNCT
ajst-20197	239	1	in	in	ADP
ajst-20197	239	2	icassp	icassp	NOUN
ajst-20197	239	3	2021	2021	NUM
ajst-20197	239	4	-	-	SYM
ajst-20197	239	5	2021	2021	NUM
ajst-20197	239	6	ieee	ieee	NOUN
ajst-20197	239	7	international	international	ADJ
ajst-20197	239	8	conference	conference	NOUN
ajst-20197	239	9	on	on	ADP
ajst-20197	239	10	acoustics	acoustic	NOUN
ajst-20197	239	11	,	,	PUNCT
ajst-20197	239	12	speech	speech	NOUN
ajst-20197	239	13	and	and	CCONJ
ajst-20197	239	14	signal	signal	NOUN
ajst-20197	239	15	processing	processing	NOUN
ajst-20197	239	16	(	(	PUNCT
ajst-20197	239	17	icassp	icassp	PROPN
ajst-20197	239	18	)	)	PUNCT
ajst-20197	239	19	(	(	PUNCT
ajst-20197	239	20	pp	pp	ADP
ajst-20197	239	21	.	.	PUNCT
ajst-20197	239	22	62946298	62946298	NUM
ajst-20197	239	23	)	)	PUNCT
ajst-20197	239	24	.	.	PUNCT
ajst-20197	240	1	ieee	ieee	NOUN
ajst-20197	240	2	.	.	PUNCT
ajst-20197	241	1	[	[	X
ajst-20197	241	2	21	21	NUM
ajst-20197	241	3	]	]	X
ajst-20197	241	4	ranftl	ranftl	PROPN
ajst-20197	241	5	,	,	PUNCT
ajst-20197	241	6	r.	r.	PROPN
ajst-20197	241	7	,	,	PUNCT
ajst-20197	241	8	bochkovskiy	bochkovskiy	PROPN
ajst-20197	241	9	,	,	PUNCT
ajst-20197	241	10	a.	a.	NOUN
ajst-20197	241	11	and	and	CCONJ
ajst-20197	241	12	koltun	koltun	PROPN
ajst-20197	241	13	,	,	PUNCT
ajst-20197	241	14	v.	v.	ADV
ajst-20197	241	15	,	,	PUNCT
ajst-20197	241	16	2021	2021	NUM
ajst-20197	241	17	.	.	PUNCT
ajst-20197	242	1	vision	vision	NOUN
ajst-20197	242	2	transformers	transformer	NOUN
ajst-20197	242	3	for	for	ADP
ajst-20197	242	4	dense	dense	ADJ
ajst-20197	242	5	prediction	prediction	NOUN
ajst-20197	242	6	.	.	PUNCT
ajst-20197	243	1	in	in	ADP
ajst-20197	243	2	proceedings	proceeding	NOUN
ajst-20197	243	3	of	of	ADP
ajst-20197	243	4	the	the	DET
ajst-20197	243	5	ieee	ieee	NOUN
ajst-20197	243	6	/	/	SYM
ajst-20197	243	7	cvf	cvf	NOUN
ajst-20197	243	8	international	international	ADJ
ajst-20197	243	9	conference	conference	NOUN
ajst-20197	243	10	on	on	ADP
ajst-20197	243	11	computer	computer	NOUN
ajst-20197	243	12	vision	vision	NOUN
ajst-20197	243	13	(	(	PUNCT
ajst-20197	243	14	pp	pp	ADJ
ajst-20197	243	15	.	.	PUNCT
ajst-20197	243	16	12179	12179	NUM
ajst-20197	243	17	-	-	SYM
ajst-20197	243	18	12188	12188	NUM
ajst-20197	243	19	)	)	PUNCT
ajst-20197	243	20	.	.	PUNCT
ajst-20197	244	1	[	[	X
ajst-20197	244	2	22	22	NUM
ajst-20197	244	3	]	]	X
ajst-20197	244	4	salehi	salehi	X
ajst-20197	244	5	,	,	PUNCT
ajst-20197	244	6	a.w	a.w	PROPN
ajst-20197	244	7	.	.	PROPN
ajst-20197	244	8	,	,	PUNCT
ajst-20197	244	9	khan	khan	PROPN
ajst-20197	244	10	,	,	PUNCT
ajst-20197	244	11	s.	s.	PROPN
ajst-20197	244	12	,	,	PUNCT
ajst-20197	244	13	gupta	gupta	PROPN
ajst-20197	244	14	,	,	PUNCT
ajst-20197	244	15	g.	g.	PROPN
ajst-20197	244	16	,	,	PUNCT
ajst-20197	244	17	alabduallah	alabduallah	PROPN
ajst-20197	244	18	,	,	PUNCT
ajst-20197	244	19	b.i	b.i	PROPN
ajst-20197	244	20	.	.	PROPN
ajst-20197	244	21	,	,	PUNCT
ajst-20197	244	22	almjally	almjally	ADV
ajst-20197	244	23	,	,	PUNCT
ajst-20197	244	24	a.	a.	NOUN
ajst-20197	244	25	,	,	PUNCT
ajst-20197	244	26	alsolai	alsolai	PROPN
ajst-20197	244	27	,	,	PUNCT
ajst-20197	244	28	h.	h.	PROPN
ajst-20197	244	29	,	,	PUNCT
ajst-20197	244	30	siddiqui	siddiqui	NOUN
ajst-20197	244	31	,	,	PUNCT
ajst-20197	244	32	t.	t.	PROPN
ajst-20197	244	33	and	and	CCONJ
ajst-20197	244	34	mellit	mellit	VERB
ajst-20197	244	35	,	,	PUNCT
ajst-20197	244	36	a.	a.	NOUN
ajst-20197	244	37	,	,	PUNCT
ajst-20197	244	38	2023	2023	NUM
ajst-20197	244	39	.	.	PUNCT
ajst-20197	245	1	a	a	DET
ajst-20197	245	2	study	study	NOUN
ajst-20197	245	3	of	of	ADP
ajst-20197	245	4	cnn	cnn	PROPN
ajst-20197	245	5	and	and	CCONJ
ajst-20197	245	6	transfer	transfer	VERB
ajst-20197	245	7	learning	learning	NOUN
ajst-20197	245	8	in	in	ADP
ajst-20197	245	9	medical	medical	ADJ
ajst-20197	245	10	imaging	imaging	NOUN
ajst-20197	245	11	:	:	PUNCT
ajst-20197	245	12	advantages	advantage	NOUN
ajst-20197	245	13	,	,	PUNCT
ajst-20197	245	14	challenges	challenge	NOUN
ajst-20197	245	15	,	,	PUNCT
ajst-20197	245	16	future	future	ADJ
ajst-20197	245	17	scope	scope	NOUN
ajst-20197	245	18	.	.	PUNCT
ajst-20197	246	1	sustainability	sustainability	NOUN
ajst-20197	246	2	,	,	PUNCT
ajst-20197	246	3	15(7	15(7	NUM
ajst-20197	246	4	)	)	PUNCT
ajst-20197	246	5	,	,	PUNCT
ajst-20197	246	6	p.5930	p.5930	PROPN
ajst-20197	246	7	.	.	PUNCT
ajst-20197	247	1	[	[	X
ajst-20197	247	2	23	23	NUM
ajst-20197	247	3	]	]	X
ajst-20197	247	4	smys	smys	PROPN
ajst-20197	247	5	,	,	PUNCT
ajst-20197	247	6	s.	s.	PROPN
ajst-20197	247	7	,	,	PUNCT
ajst-20197	247	8	chen	chen	PROPN
ajst-20197	247	9	,	,	PUNCT
ajst-20197	247	10	j.i.z	j.i.z	PROPN
ajst-20197	247	11	.	.	PUNCT
ajst-20197	247	12	and	and	CCONJ
ajst-20197	247	13	shakya	shakya	PROPN
ajst-20197	247	14	,	,	PUNCT
ajst-20197	247	15	s.	s.	PROPN
ajst-20197	247	16	,	,	PUNCT
ajst-20197	247	17	2020	2020	NUM
ajst-20197	247	18	.	.	PUNCT
ajst-20197	248	1	survey	survey	NOUN
ajst-20197	248	2	on	on	ADP
ajst-20197	248	3	neural	neural	ADJ
ajst-20197	248	4	network	network	NOUN
ajst-20197	248	5	architectures	architecture	NOUN
ajst-20197	248	6	with	with	ADP
ajst-20197	248	7	deep	deep	ADJ
ajst-20197	248	8	learning	learning	NOUN
ajst-20197	248	9	.	.	PUNCT
ajst-20197	249	1	journal	journal	NOUN
ajst-20197	249	2	of	of	ADP
ajst-20197	249	3	soft	soft	ADJ
ajst-20197	249	4	computing	computing	NOUN
ajst-20197	249	5	paradigm	paradigm	NOUN
ajst-20197	249	6	(	(	PUNCT
ajst-20197	249	7	jscp	jscp	PROPN
ajst-20197	249	8	)	)	PUNCT
ajst-20197	249	9	,	,	PUNCT
ajst-20197	249	10	2(03	2(03	NOUN
ajst-20197	249	11	)	)	PUNCT
ajst-20197	249	12	,	,	PUNCT
ajst-20197	249	13	pp.186	pp.186	PROPN
ajst-20197	249	14	-	-	SYM
ajst-20197	249	15	194	194	NUM
ajst-20197	249	16	.	.	PUNCT
ajst-20197	250	1	[	[	X
ajst-20197	250	2	24	24	NUM
ajst-20197	250	3	]	]	PUNCT
ajst-20197	250	4	torfi	torfi	PROPN
ajst-20197	250	5	,	,	PUNCT
ajst-20197	250	6	a.	a.	PROPN
ajst-20197	250	7	,	,	PUNCT
ajst-20197	250	8	shirvani	shirvani	PROPN
ajst-20197	250	9	,	,	PUNCT
ajst-20197	250	10	r.a	r.a	PROPN
ajst-20197	250	11	.	.	PROPN
ajst-20197	250	12	,	,	PUNCT
ajst-20197	250	13	keneshloo	keneshloo	PROPN
ajst-20197	250	14	,	,	PUNCT
ajst-20197	250	15	y.	y.	PROPN
ajst-20197	250	16	,	,	PUNCT
ajst-20197	250	17	tavaf	tavaf	INTJ
ajst-20197	250	18	,	,	PUNCT
ajst-20197	250	19	n.	n.	PROPN
ajst-20197	250	20	and	and	CCONJ
ajst-20197	250	21	fox	fox	PROPN
ajst-20197	250	22	,	,	PUNCT
ajst-20197	250	23	e.a	e.a	PROPN
ajst-20197	250	24	.	.	PROPN
ajst-20197	250	25	,	,	PUNCT
ajst-20197	250	26	2020	2020	NUM
ajst-20197	250	27	.	.	PUNCT
ajst-20197	250	28	natural	natural	ADJ
ajst-20197	250	29	language	language	NOUN
ajst-20197	250	30	processing	processing	NOUN
ajst-20197	250	31	advancements	advancement	NOUN
ajst-20197	250	32	by	by	ADP
ajst-20197	250	33	deep	deep	ADJ
ajst-20197	250	34	learning	learning	NOUN
ajst-20197	250	35	:	:	PUNCT
ajst-20197	250	36	a	a	DET
ajst-20197	250	37	survey	survey	NOUN
ajst-20197	250	38	.	.	PUNCT
ajst-20197	251	1	arxiv	arxiv	PROPN
ajst-20197	251	2	preprint	preprint	VERB
ajst-20197	251	3	arxiv:2003.01200	arxiv:2003.01200	PROPN
ajst-20197	251	4	.	.	PUNCT
ajst-20197	252	1	[	[	X
ajst-20197	252	2	25	25	NUM
ajst-20197	252	3	]	]	X
ajst-20197	252	4	weng	weng	PROPN
ajst-20197	252	5	,	,	PUNCT
ajst-20197	252	6	o.	o.	NOUN
ajst-20197	252	7	,	,	PUNCT
ajst-20197	252	8	2021	2021	NUM
ajst-20197	252	9	.	.	PUNCT
ajst-20197	253	1	neural	neural	ADJ
ajst-20197	253	2	network	network	NOUN
ajst-20197	253	3	quantization	quantization	NOUN
ajst-20197	253	4	for	for	ADP
ajst-20197	253	5	efficient	efficient	ADJ
ajst-20197	253	6	inference	inference	NOUN
ajst-20197	253	7	:	:	PUNCT
ajst-20197	253	8	a	a	DET
ajst-20197	253	9	survey	survey	NOUN
ajst-20197	253	10	.	.	PUNCT
ajst-20197	254	1	arxiv	arxiv	PROPN
ajst-20197	254	2	preprint	preprint	NOUN
ajst-20197	254	3	arxiv:2112.06126	arxiv:2112.06126	NOUN
ajst-20197	254	4	.	.	PUNCT
ajst-20197	255	1	[	[	X
ajst-20197	255	2	26	26	NUM
ajst-20197	255	3	]	]	X
ajst-20197	255	4	wilson	wilson	PROPN
ajst-20197	255	5	,	,	PUNCT
ajst-20197	255	6	m.	m.	NOUN
ajst-20197	255	7	,	,	PUNCT
ajst-20197	255	8	wellington	wellington	PROPN
ajst-20197	255	9	,	,	PUNCT
ajst-20197	255	10	b.	b.	PROPN
ajst-20197	255	11	,	,	PUNCT
ajst-20197	255	12	merrick	merrick	PROPN
ajst-20197	255	13	,	,	PUNCT
ajst-20197	255	14	a.	a.	NOUN
ajst-20197	255	15	and	and	CCONJ
ajst-20197	255	16	huxley	huxley	PROPN
ajst-20197	255	17	,	,	PUNCT
ajst-20197	255	18	i.	i.	NOUN
ajst-20197	255	19	,	,	PUNCT
ajst-20197	255	20	2023	2023	NUM
ajst-20197	255	21	.	.	PUNCT
ajst-20197	256	1	a	a	DET
ajst-20197	256	2	recommendation	recommendation	NOUN
ajst-20197	256	3	model	model	NOUN
ajst-20197	256	4	based	base	VERB
ajst-20197	256	5	on	on	ADP
ajst-20197	256	6	deep	deep	ADJ
ajst-20197	256	7	feature	feature	NOUN
ajst-20197	256	8	representation	representation	NOUN
ajst-20197	256	9	and	and	CCONJ
ajst-20197	256	10	multi	multi	ADJ
ajst-20197	256	11	-	-	ADJ
ajst-20197	256	12	head	head	ADJ
ajst-20197	256	13	self	self	NOUN
ajst-20197	256	14	-	-	PUNCT
ajst-20197	256	15	attention	attention	NOUN
ajst-20197	256	16	mechanism	mechanism	NOUN
ajst-20197	256	17	.	.	PUNCT
ajst-20197	257	1	[	[	X
ajst-20197	257	2	27	27	NUM
ajst-20197	257	3	]	]	SYM
ajst-20197	257	4	xi	xi	PROPN
ajst-20197	257	5	,	,	PUNCT
ajst-20197	257	6	z.	z.	PROPN
ajst-20197	257	7	,	,	PUNCT
ajst-20197	257	8	chen	chen	PROPN
ajst-20197	257	9	,	,	PUNCT
ajst-20197	257	10	w.	w.	PROPN
ajst-20197	257	11	,	,	PUNCT
ajst-20197	257	12	guo	guo	PROPN
ajst-20197	257	13	,	,	PUNCT
ajst-20197	257	14	x.	x.	PROPN
ajst-20197	257	15	,	,	PUNCT
ajst-20197	257	16	he	he	PRON
ajst-20197	257	17	,	,	PUNCT
ajst-20197	257	18	w.	w.	PROPN
ajst-20197	257	19	,	,	PUNCT
ajst-20197	257	20	ding	ding	NOUN
ajst-20197	257	21	,	,	PUNCT
ajst-20197	257	22	y.	y.	PROPN
ajst-20197	257	23	,	,	PUNCT
ajst-20197	257	24	hong	hong	PROPN
ajst-20197	257	25	,	,	PUNCT
ajst-20197	257	26	b.	b.	PROPN
ajst-20197	257	27	,	,	PUNCT
ajst-20197	257	28	zhang	zhang	PROPN
ajst-20197	257	29	,	,	PUNCT
ajst-20197	257	30	m.	m.	NOUN
ajst-20197	257	31	,	,	PUNCT
ajst-20197	257	32	wang	wang	PROPN
ajst-20197	257	33	,	,	PUNCT
ajst-20197	257	34	j.	j.	PROPN
ajst-20197	257	35	,	,	PUNCT
ajst-20197	257	36	jin	jin	PROPN
ajst-20197	257	37	,	,	PUNCT
ajst-20197	257	38	s.	s.	PROPN
ajst-20197	257	39	,	,	PUNCT
ajst-20197	257	40	zhou	zhou	PROPN
ajst-20197	257	41	,	,	PUNCT
ajst-20197	257	42	e.	e.	PROPN
ajst-20197	257	43	and	and	CCONJ
ajst-20197	257	44	zheng	zheng	PROPN
ajst-20197	257	45	,	,	PUNCT
ajst-20197	257	46	r.	r.	PROPN
ajst-20197	257	47	,	,	PUNCT
ajst-20197	257	48	2023	2023	NUM
ajst-20197	257	49	.	.	PUNCT
ajst-20197	258	1	the	the	DET
ajst-20197	258	2	rise	rise	NOUN
ajst-20197	258	3	and	and	CCONJ
ajst-20197	258	4	potential	potential	NOUN
ajst-20197	258	5	of	of	ADP
ajst-20197	258	6	large	large	ADJ
ajst-20197	258	7	language	language	NOUN
ajst-20197	258	8	model	model	NOUN
ajst-20197	258	9	based	base	VERB
ajst-20197	258	10	agents	agent	NOUN
ajst-20197	258	11	:	:	PUNCT
ajst-20197	258	12	a	a	DET
ajst-20197	258	13	survey	survey	NOUN
ajst-20197	258	14	.	.	PUNCT
ajst-20197	259	1	arxiv	arxiv	PROPN
ajst-20197	259	2	preprint	preprint	NOUN
ajst-20197	259	3	arxiv:2309.07864	arxiv:2309.07864	PROPN
ajst-20197	259	4	.	.	PUNCT
ajst-20197	260	1	[	[	X
ajst-20197	260	2	28	28	NUM
ajst-20197	260	3	]	]	X
ajst-20197	260	4	xie	xie	PROPN
ajst-20197	260	5	,	,	PUNCT
ajst-20197	260	6	y.	y.	PROPN
ajst-20197	260	7	,	,	PUNCT
ajst-20197	260	8	zhou	zhou	PROPN
ajst-20197	260	9	,	,	PUNCT
ajst-20197	260	10	t.	t.	PROPN
ajst-20197	260	11	,	,	PUNCT
ajst-20197	260	12	mao	mao	PROPN
ajst-20197	260	13	,	,	PUNCT
ajst-20197	260	14	y.	y.	PROPN
ajst-20197	260	15	and	and	CCONJ
ajst-20197	260	16	chen	chen	PROPN
ajst-20197	260	17	,	,	PUNCT
ajst-20197	260	18	w.	w.	PROPN
ajst-20197	260	19	,	,	PUNCT
ajst-20197	260	20	2020	2020	NUM
ajst-20197	260	21	.	.	PUNCT
ajst-20197	261	1	conditional	conditional	ADJ
ajst-20197	261	2	self	self	NOUN
ajst-20197	261	3	-	-	PUNCT
ajst-20197	261	4	attention	attention	NOUN
ajst-20197	261	5	for	for	ADP
ajst-20197	261	6	query	query	NOUN
ajst-20197	261	7	-	-	PUNCT
ajst-20197	261	8	based	base	VERB
ajst-20197	261	9	summarization	summarization	NOUN
ajst-20197	261	10	.	.	PUNCT
ajst-20197	262	1	arxiv	arxiv	PROPN
ajst-20197	262	2	preprint	preprint	NOUN
ajst-20197	262	3	arxiv:2002.07338	arxiv:2002.07338	NOUN
ajst-20197	262	4	.	.	PUNCT
ajst-20197	263	1	[	[	X
ajst-20197	263	2	29	29	NUM
ajst-20197	263	3	]	]	X
ajst-20197	263	4	zhao	zhao	X
ajst-20197	263	5	,	,	PUNCT
ajst-20197	263	6	y.	y.	PROPN
ajst-20197	263	7	,	,	PUNCT
ajst-20197	263	8	wang	wang	PROPN
ajst-20197	263	9	,	,	PUNCT
ajst-20197	263	10	d.	d.	PROPN
ajst-20197	263	11	,	,	PUNCT
ajst-20197	263	12	xu	xu	PROPN
ajst-20197	263	13	,	,	PUNCT
ajst-20197	263	14	b.	b.	PROPN
ajst-20197	263	15	and	and	CCONJ
ajst-20197	263	16	zhang	zhang	PROPN
ajst-20197	263	17	,	,	PUNCT
ajst-20197	263	18	t.	t.	PROPN
ajst-20197	263	19	,	,	PUNCT
ajst-20197	263	20	2020	2020	NUM
ajst-20197	263	21	.	.	PUNCT
ajst-20197	264	1	monaural	monaural	ADJ
ajst-20197	264	2	speech	speech	NOUN
ajst-20197	264	3	dereverberation	dereverberation	NOUN
ajst-20197	264	4	using	use	VERB
ajst-20197	264	5	temporal	temporal	ADJ
ajst-20197	264	6	convolutional	convolutional	ADJ
ajst-20197	264	7	networks	network	NOUN
ajst-20197	264	8	with	with	ADP
ajst-20197	264	9	self	self	NOUN
ajst-20197	264	10	attention	attention	NOUN
ajst-20197	264	11	.	.	PUNCT
ajst-20197	265	1	ieee	ieee	PROPN
ajst-20197	265	2	/	/	SYM
ajst-20197	265	3	acm	acm	PROPN
ajst-20197	265	4	transactions	transaction	NOUN
ajst-20197	265	5	on	on	ADP
ajst-20197	265	6	audio	audio	NOUN
ajst-20197	265	7	,	,	PUNCT
ajst-20197	265	8	speech	speech	NOUN
ajst-20197	265	9	,	,	PUNCT
ajst-20197	265	10	and	and	CCONJ
ajst-20197	265	11	language	language	NOUN
ajst-20197	265	12	processing	processing	NOUN
ajst-20197	265	13	,	,	PUNCT
ajst-20197	265	14	28	28	NUM
ajst-20197	265	15	,	,	PUNCT
ajst-20197	265	16	pp.1598	pp.1598	PROPN
ajst-20197	265	17	-	-	PUNCT
ajst-20197	265	18	1607	1607	NUM
ajst-20197	265	19	.	.	PUNCT
ajst-20197	266	1	[	[	X
ajst-20197	266	2	30	30	NUM
ajst-20197	266	3	]	]	X
ajst-20197	266	4	zulqarnain	zulqarnain	NOUN
ajst-20197	266	5	,	,	PUNCT
ajst-20197	266	6	m.	m.	NOUN
ajst-20197	266	7	,	,	PUNCT
ajst-20197	266	8	abd	abd	PROPN
ajst-20197	266	9	ishak	ishak	PROPN
ajst-20197	266	10	,	,	PUNCT
ajst-20197	266	11	s.	s.	PROPN
ajst-20197	266	12	,	,	PUNCT
ajst-20197	266	13	ghazali	ghazali	PROPN
ajst-20197	266	14	,	,	PUNCT
ajst-20197	266	15	r.	r.	PROPN
ajst-20197	266	16	,	,	PUNCT
ajst-20197	266	17	nawi	nawi	PROPN
ajst-20197	266	18	,	,	PUNCT
ajst-20197	266	19	n.m	n.m	PROPN
ajst-20197	266	20	.	.	PROPN
ajst-20197	266	21	,	,	PUNCT
ajst-20197	266	22	aamir	aamir	PROPN
ajst-20197	266	23	,	,	PUNCT
ajst-20197	266	24	m.	m.	NOUN
ajst-20197	266	25	and	and	CCONJ
ajst-20197	266	26	hassim	hassim	NOUN
ajst-20197	266	27	,	,	PUNCT
ajst-20197	266	28	y.m.m	y.m.m	PROPN
ajst-20197	266	29	.	.	PROPN
ajst-20197	266	30	,	,	PUNCT
ajst-20197	266	31	2020	2020	NUM
ajst-20197	266	32	.	.	PUNCT
ajst-20197	267	1	an	an	DET
ajst-20197	267	2	improved	improved	ADJ
ajst-20197	267	3	deep	deep	ADJ
ajst-20197	267	4	learning	learning	NOUN
ajst-20197	267	5	approach	approach	NOUN
ajst-20197	267	6	based	base	VERB
ajst-20197	267	7	on	on	ADP
ajst-20197	267	8	variant	variant	ADJ
ajst-20197	267	9	two	two	NUM
ajst-20197	267	10	-	-	PUNCT
ajst-20197	267	11	state	state	NOUN
ajst-20197	267	12	gated	gate	VERB
ajst-20197	267	13	recurrent	recurrent	ADJ
ajst-20197	267	14	unit	unit	NOUN
ajst-20197	267	15	and	and	CCONJ
ajst-20197	267	16	word	word	NOUN
ajst-20197	267	17	embeddings	embedding	NOUN
ajst-20197	267	18	for	for	ADP
ajst-20197	267	19	sentiment	sentiment	NOUN
ajst-20197	267	20	classification	classification	NOUN
ajst-20197	267	21	.	.	PUNCT
ajst-20197	268	1	international	international	ADJ
ajst-20197	268	2	journal	journal	NOUN
ajst-20197	268	3	of	of	ADP
ajst-20197	268	4	advanced	advanced	ADJ
ajst-20197	268	5	computer	computer	NOUN
ajst-20197	268	6	science	science	NOUN
ajst-20197	268	7	and	and	CCONJ
ajst-20197	268	8	applications	application	NOUN
ajst-20197	268	9	,	,	PUNCT
ajst-20197	268	10	11(1	11(1	NUM
ajst-20197	268	11	)	)	PUNCT
ajst-20197	268	12	.	.	PUNCT