id	sid	tid	token	lemma	pos
fcis-6025	1	1	frontiers	frontier	NOUN
fcis-6025	1	2	in	in	ADP
fcis-6025	1	3	computing	computing	NOUN
fcis-6025	1	4	and	and	CCONJ
fcis-6025	1	5	intelligent	intelligent	ADJ
fcis-6025	1	6	systems	system	NOUN
fcis-6025	1	7	issn	issn	VERB
fcis-6025	1	8	:	:	PUNCT
fcis-6025	1	9	2832	2832	NUM
fcis-6025	1	10	-	-	SYM
fcis-6025	1	11	6024	6024	NUM
fcis-6025	1	12	|	|	NOUN
fcis-6025	1	13	vol	vol	NOUN
fcis-6025	1	14	.	.	PROPN
fcis-6025	2	1	3	3	NUM
fcis-6025	2	2	,	,	PUNCT
fcis-6025	2	3	no	no	INTJ
fcis-6025	2	4	.	.	NOUN
fcis-6025	2	5	1	1	NUM
fcis-6025	2	6	,	,	PUNCT
fcis-6025	2	7	2023	2023	NUM
fcis-6025	2	8	60	60	NUM
fcis-6025	2	9	research	research	NOUN
fcis-6025	2	10	on	on	ADP
fcis-6025	2	11	speech	speech	NOUN
fcis-6025	2	12	emotion	emotion	NOUN
fcis-6025	2	13	recognition	recognition	NOUN
fcis-6025	2	14	analysis	analysis	NOUN
fcis-6025	2	15	based	base	VERB
fcis-6025	2	16	on	on	ADP
fcis-6025	2	17	deep	deep	ADJ
fcis-6025	2	18	learning	learn	VERB
fcis-6025	2	19	ailiang	ailiang	PROPN
fcis-6025	2	20	yin	yin	PROPN
fcis-6025	2	21	*	*	PROPN
fcis-6025	2	22	,	,	PUNCT
fcis-6025	2	23	chunhao	chunhao	PROPN
fcis-6025	2	24	li	li	PROPN
fcis-6025	2	25	school	school	PROPN
fcis-6025	2	26	of	of	ADP
fcis-6025	2	27	electronic	electronic	ADJ
fcis-6025	2	28	information	information	NOUN
fcis-6025	2	29	,	,	PUNCT
fcis-6025	2	30	southwest	southwest	PROPN
fcis-6025	2	31	minzu	minzu	PROPN
fcis-6025	2	32	university	university	PROPN
fcis-6025	2	33	,	,	PUNCT
fcis-6025	2	34	chengdu	chengdu	PROPN
fcis-6025	2	35	,	,	PUNCT
fcis-6025	2	36	china	china	PROPN
fcis-6025	2	37	*	*	PUNCT
fcis-6025	2	38	corresponding	correspond	VERB
fcis-6025	2	39	author	author	NOUN
fcis-6025	2	40	:	:	PUNCT
fcis-6025	2	41	ailiang	ailiang	PROPN
fcis-6025	2	42	yin	yin	PROPN
fcis-6025	2	43	abstract	abstract	NOUN
fcis-6025	2	44	:	:	PUNCT
fcis-6025	3	1	this	this	DET
fcis-6025	3	2	paper	paper	NOUN
fcis-6025	3	3	combines	combine	VERB
fcis-6025	3	4	two	two	NUM
fcis-6025	3	5	aspects	aspect	NOUN
fcis-6025	3	6	of	of	ADP
fcis-6025	3	7	feature	feature	NOUN
fcis-6025	3	8	selection	selection	NOUN
fcis-6025	3	9	and	and	CCONJ
fcis-6025	3	10	building	build	VERB
fcis-6025	3	11	deep	deep	ADJ
fcis-6025	3	12	neural	neural	ADJ
fcis-6025	3	13	networks	network	NOUN
fcis-6025	3	14	to	to	PART
fcis-6025	3	15	carry	carry	VERB
fcis-6025	3	16	out	out	ADP
fcis-6025	3	17	targeted	target	VERB
fcis-6025	3	18	research	research	NOUN
fcis-6025	3	19	to	to	PART
fcis-6025	3	20	improve	improve	VERB
fcis-6025	3	21	recognition	recognition	NOUN
fcis-6025	3	22	accuracy	accuracy	NOUN
fcis-6025	3	23	.	.	PUNCT
fcis-6025	4	1	firstly	firstly	ADV
fcis-6025	4	2	,	,	PUNCT
fcis-6025	4	3	speech	speech	NOUN
fcis-6025	4	4	preprocessing	preprocessing	NOUN
fcis-6025	4	5	techniques	technique	NOUN
fcis-6025	4	6	are	be	AUX
fcis-6025	4	7	introduced	introduce	VERB
fcis-6025	4	8	to	to	PART
fcis-6025	4	9	extract	extract	VERB
fcis-6025	4	10	the	the	DET
fcis-6025	4	11	speech	speech	NOUN
fcis-6025	4	12	spectrogram	spectrogram	NOUN
fcis-6025	4	13	and	and	CCONJ
fcis-6025	4	14	lay	lie	VERB
fcis-6025	4	15	the	the	DET
fcis-6025	4	16	foundation	foundation	NOUN
fcis-6025	4	17	for	for	ADP
fcis-6025	4	18	building	build	VERB
fcis-6025	4	19	the	the	DET
fcis-6025	4	20	speech	speech	NOUN
fcis-6025	4	21	emotion	emotion	NOUN
fcis-6025	4	22	recognition	recognition	NOUN
fcis-6025	4	23	network	network	NOUN
fcis-6025	4	24	model	model	NOUN
fcis-6025	4	25	study	study	PROPN
fcis-6025	4	26	.	.	PUNCT
fcis-6025	5	1	the	the	DET
fcis-6025	5	2	focus	focus	NOUN
fcis-6025	5	3	is	be	AUX
fcis-6025	5	4	on	on	ADP
fcis-6025	5	5	building	build	VERB
fcis-6025	5	6	a	a	DET
fcis-6025	5	7	speech	speech	NOUN
fcis-6025	5	8	emotion	emotion	NOUN
fcis-6025	5	9	recognition	recognition	NOUN
fcis-6025	5	10	network	network	NOUN
fcis-6025	5	11	model	model	NOUN
fcis-6025	5	12	based	base	VERB
fcis-6025	5	13	on	on	ADP
fcis-6025	5	14	residual	residual	ADJ
fcis-6025	5	15	network	network	NOUN
fcis-6025	5	16	improvement	improvement	NOUN
fcis-6025	5	17	and	and	CCONJ
fcis-6025	5	18	comparing	compare	VERB
fcis-6025	5	19	experiments	experiment	NOUN
fcis-6025	5	20	with	with	ADP
fcis-6025	5	21	alexnet	alexnet	ADJ
fcis-6025	5	22	model	model	NOUN
fcis-6025	5	23	network	network	NOUN
fcis-6025	5	24	and	and	CCONJ
fcis-6025	5	25	resnet-18	resnet-18	PROPN
fcis-6025	5	26	network	network	NOUN
fcis-6025	5	27	model	model	NOUN
fcis-6025	5	28	.	.	PUNCT
fcis-6025	6	1	keywords	keyword	NOUN
fcis-6025	6	2	:	:	PUNCT
fcis-6025	6	3	speech	speech	NOUN
fcis-6025	6	4	spectrogram	spectrogram	NOUN
fcis-6025	6	5	;	;	PUNCT
fcis-6025	6	6	residual	residual	ADJ
fcis-6025	6	7	network	network	NOUN
fcis-6025	6	8	;	;	PUNCT
fcis-6025	6	9	speech	speech	NOUN
fcis-6025	6	10	emotion	emotion	NOUN
fcis-6025	6	11	recognition	recognition	NOUN
fcis-6025	6	12	.	.	PUNCT
fcis-6025	7	1	1	1	X
fcis-6025	7	2	.	.	X
fcis-6025	7	3	introduction	introduction	NOUN
fcis-6025	7	4	in	in	ADP
fcis-6025	7	5	addition	addition	NOUN
fcis-6025	7	6	to	to	ADP
fcis-6025	7	7	the	the	DET
fcis-6025	7	8	basic	basic	ADJ
fcis-6025	7	9	semantic	semantic	ADJ
fcis-6025	7	10	information	information	NOUN
fcis-6025	7	11	,	,	PUNCT
fcis-6025	7	12	speech	speech	NOUN
fcis-6025	7	13	signals	signal	NOUN
fcis-6025	7	14	also	also	ADV
fcis-6025	7	15	contain	contain	VERB
fcis-6025	7	16	a	a	DET
fcis-6025	7	17	variety	variety	NOUN
fcis-6025	7	18	of	of	ADP
fcis-6025	7	19	complex	complex	ADJ
fcis-6025	7	20	human	human	ADJ
fcis-6025	7	21	emotions	emotion	NOUN
fcis-6025	7	22	,	,	PUNCT
fcis-6025	7	23	which	which	PRON
fcis-6025	7	24	play	play	VERB
fcis-6025	7	25	a	a	DET
fcis-6025	7	26	very	very	ADV
fcis-6025	7	27	important	important	ADJ
fcis-6025	7	28	role	role	NOUN
fcis-6025	7	29	in	in	ADP
fcis-6025	7	30	human	human	ADJ
fcis-6025	7	31	communication	communication	NOUN
fcis-6025	7	32	and	and	CCONJ
fcis-6025	7	33	life	life	NOUN
fcis-6025	7	34	.	.	PUNCT
fcis-6025	8	1	the	the	DET
fcis-6025	8	2	importance	importance	NOUN
fcis-6025	8	3	of	of	ADP
fcis-6025	8	4	speech	speech	NOUN
fcis-6025	8	5	emotion	emotion	NOUN
fcis-6025	8	6	recognition	recognition	NOUN
fcis-6025	8	7	in	in	ADP
fcis-6025	8	8	the	the	DET
fcis-6025	8	9	field	field	NOUN
fcis-6025	8	10	of	of	ADP
fcis-6025	8	11	artificial	artificial	ADJ
fcis-6025	8	12	intelligence	intelligence	NOUN
fcis-6025	8	13	is	be	AUX
fcis-6025	8	14	self	self	NOUN
fcis-6025	8	15	-	-	PUNCT
fcis-6025	8	16	evident	evident	ADJ
fcis-6025	8	17	.	.	PUNCT
fcis-6025	9	1	with	with	ADP
fcis-6025	9	2	the	the	DET
fcis-6025	9	3	development	development	NOUN
fcis-6025	9	4	of	of	ADP
fcis-6025	9	5	science	science	NOUN
fcis-6025	9	6	and	and	CCONJ
fcis-6025	9	7	technology	technology	NOUN
fcis-6025	9	8	,	,	PUNCT
fcis-6025	9	9	artificial	artificial	ADJ
fcis-6025	9	10	intelligence	intelligence	NOUN
fcis-6025	9	11	is	be	AUX
fcis-6025	9	12	becoming	become	VERB
fcis-6025	9	13	more	more	ADV
fcis-6025	9	14	and	and	CCONJ
fcis-6025	9	15	more	more	ADV
fcis-6025	9	16	mature	mature	ADJ
fcis-6025	9	17	,	,	PUNCT
fcis-6025	9	18	but	but	CCONJ
fcis-6025	9	19	the	the	DET
fcis-6025	9	20	development	development	NOUN
fcis-6025	9	21	of	of	ADP
fcis-6025	9	22	self	self	NOUN
fcis-6025	9	23	-	-	PUNCT
fcis-6025	9	24	awareness	awareness	NOUN
fcis-6025	9	25	and	and	CCONJ
fcis-6025	9	26	self	self	NOUN
fcis-6025	9	27	-	-	PUNCT
fcis-6025	9	28	awareness	awareness	NOUN
fcis-6025	9	29	of	of	ADP
fcis-6025	9	30	artificial	artificial	ADJ
fcis-6025	9	31	intelligence	intelligence	NOUN
fcis-6025	9	32	is	be	AUX
fcis-6025	9	33	slow	slow	ADJ
fcis-6025	9	34	,	,	PUNCT
fcis-6025	9	35	which	which	PRON
fcis-6025	9	36	hinders	hinder	VERB
fcis-6025	9	37	the	the	DET
fcis-6025	9	38	natural	natural	ADJ
fcis-6025	9	39	interaction	interaction	NOUN
fcis-6025	9	40	process	process	NOUN
fcis-6025	9	41	between	between	ADP
fcis-6025	9	42	human	human	NOUN
fcis-6025	9	43	and	and	CCONJ
fcis-6025	9	44	machine	machine	NOUN
fcis-6025	9	45	.	.	PUNCT
fcis-6025	10	1	one	one	NUM
fcis-6025	10	2	important	important	ADJ
fcis-6025	10	3	way	way	NOUN
fcis-6025	10	4	to	to	PART
fcis-6025	10	5	solve	solve	VERB
fcis-6025	10	6	the	the	DET
fcis-6025	10	7	selfawareness	selfawareness	NOUN
fcis-6025	10	8	and	and	CCONJ
fcis-6025	10	9	self	self	NOUN
fcis-6025	10	10	-	-	PUNCT
fcis-6025	10	11	awareness	awareness	NOUN
fcis-6025	10	12	of	of	ADP
fcis-6025	10	13	ai	ai	NOUN
fcis-6025	10	14	is	be	AUX
fcis-6025	10	15	to	to	PART
fcis-6025	10	16	make	make	VERB
fcis-6025	10	17	the	the	DET
fcis-6025	10	18	machine	machine	NOUN
fcis-6025	10	19	understand	understand	VERB
fcis-6025	10	20	human	human	NOUN
fcis-6025	10	21	's	's	PART
fcis-6025	10	22	emotional	emotional	ADJ
fcis-6025	10	23	state	state	NOUN
fcis-6025	10	24	,	,	PUNCT
fcis-6025	10	25	so	so	SCONJ
fcis-6025	10	26	that	that	SCONJ
fcis-6025	10	27	the	the	DET
fcis-6025	10	28	intelligent	intelligent	ADJ
fcis-6025	10	29	machine	machine	NOUN
fcis-6025	10	30	has	have	VERB
fcis-6025	10	31	the	the	DET
fcis-6025	10	32	ability	ability	NOUN
fcis-6025	10	33	of	of	ADP
fcis-6025	10	34	emotion.[1][2	emotion.[1][2	PROPN
fcis-6025	10	35	]	]	PUNCT
fcis-6025	10	36	this	this	PRON
fcis-6025	10	37	is	be	AUX
fcis-6025	10	38	an	an	DET
fcis-6025	10	39	important	important	ADJ
fcis-6025	10	40	way	way	NOUN
fcis-6025	10	41	to	to	PART
fcis-6025	10	42	address	address	VERB
fcis-6025	10	43	the	the	DET
fcis-6025	10	44	self	self	NOUN
fcis-6025	10	45	-	-	PUNCT
fcis-6025	10	46	awareness	awareness	NOUN
fcis-6025	10	47	and	and	CCONJ
fcis-6025	10	48	self	self	NOUN
fcis-6025	10	49	-	-	PUNCT
fcis-6025	10	50	awareness	awareness	NOUN
fcis-6025	10	51	of	of	ADP
fcis-6025	10	52	ai	ai	NOUN
fcis-6025	10	53	is	be	AUX
fcis-6025	10	54	to	to	PART
fcis-6025	10	55	enable	enable	VERB
fcis-6025	10	56	machines	machine	NOUN
fcis-6025	10	57	to	to	PART
fcis-6025	10	58	understand	understand	VERB
fcis-6025	10	59	human	human	ADJ
fcis-6025	10	60	emotional	emotional	ADJ
fcis-6025	10	61	states	state	NOUN
fcis-6025	10	62	,	,	PUNCT
fcis-6025	10	63	so	so	SCONJ
fcis-6025	10	64	that	that	SCONJ
fcis-6025	10	65	intelligent	intelligent	ADJ
fcis-6025	10	66	machines	machine	NOUN
fcis-6025	10	67	have	have	VERB
fcis-6025	10	68	emotional	emotional	ADJ
fcis-6025	10	69	capabilities	capability	NOUN
fcis-6025	10	70	.	.	PUNCT
fcis-6025	11	1	in	in	ADP
fcis-6025	11	2	recent	recent	ADJ
fcis-6025	11	3	years	year	NOUN
fcis-6025	11	4	,	,	PUNCT
fcis-6025	11	5	deep	deep	ADJ
fcis-6025	11	6	learning	learning	NOUN
fcis-6025	11	7	algorithms	algorithm	NOUN
fcis-6025	11	8	have	have	AUX
fcis-6025	11	9	achieved	achieve	VERB
fcis-6025	11	10	excellent	excellent	ADJ
fcis-6025	11	11	results	result	NOUN
fcis-6025	11	12	in	in	ADP
fcis-6025	11	13	the	the	DET
fcis-6025	11	14	field	field	NOUN
fcis-6025	11	15	of	of	ADP
fcis-6025	11	16	natural	natural	ADJ
fcis-6025	11	17	language	language	NOUN
fcis-6025	11	18	,	,	PUNCT
fcis-6025	11	19	and	and	CCONJ
fcis-6025	11	20	the	the	DET
fcis-6025	11	21	classic	classic	ADJ
fcis-6025	11	22	result	result	NOUN
fcis-6025	11	23	is	be	AUX
fcis-6025	11	24	the	the	DET
fcis-6025	11	25	recurrent	recurrent	ADJ
fcis-6025	11	26	neural	neural	ADJ
fcis-6025	11	27	network	network	NOUN
fcis-6025	11	28	(	(	PUNCT
fcis-6025	11	29	rnn).[3	rnn).[3	PROPN
fcis-6025	11	30	]	]	X
fcis-6025	11	31	rnns	rnn	NOUN
fcis-6025	11	32	have	have	AUX
fcis-6025	11	33	been	be	AUX
fcis-6025	11	34	shown	show	VERB
fcis-6025	11	35	to	to	PART
fcis-6025	11	36	be	be	AUX
fcis-6025	11	37	effective	effective	ADJ
fcis-6025	11	38	in	in	ADP
fcis-6025	11	39	solving	solve	VERB
fcis-6025	11	40	serialization	serialization	NOUN
fcis-6025	11	41	problems	problem	NOUN
fcis-6025	11	42	,	,	PUNCT
fcis-6025	11	43	making	make	VERB
fcis-6025	11	44	use	use	NOUN
fcis-6025	11	45	of	of	ADP
fcis-6025	11	46	contextual	contextual	ADJ
fcis-6025	11	47	information	information	NOUN
fcis-6025	11	48	,	,	PUNCT
fcis-6025	11	49	but	but	CCONJ
fcis-6025	11	50	they	they	PRON
fcis-6025	11	51	suffer	suffer	VERB
fcis-6025	11	52	from	from	ADP
fcis-6025	11	53	gradient	gradient	ADJ
fcis-6025	11	54	explosion	explosion	NOUN
fcis-6025	11	55	and	and	CCONJ
fcis-6025	11	56	vanishing	vanish	VERB
fcis-6025	11	57	problems	problem	NOUN
fcis-6025	11	58	in	in	ADP
fcis-6025	11	59	the	the	DET
fcis-6025	11	60	solution	solution	NOUN
fcis-6025	11	61	process	process	NOUN
fcis-6025	11	62	,	,	PUNCT
fcis-6025	11	63	and	and	CCONJ
fcis-6025	11	64	are	be	AUX
fcis-6025	11	65	ineffective	ineffective	ADJ
fcis-6025	11	66	in	in	ADP
fcis-6025	11	67	processing	process	VERB
fcis-6025	11	68	long	long	ADJ
fcis-6025	11	69	texts	text	NOUN
fcis-6025	11	70	.	.	PUNCT
fcis-6025	12	1	however	however	ADV
fcis-6025	12	2	,	,	PUNCT
fcis-6025	12	3	rnns	rnn	NOUN
fcis-6025	12	4	suffer	suffer	VERB
fcis-6025	12	5	from	from	ADP
fcis-6025	12	6	the	the	DET
fcis-6025	12	7	gradient	gradient	NOUN
fcis-6025	12	8	explosion	explosion	NOUN
fcis-6025	12	9	and	and	CCONJ
fcis-6025	12	10	disappearance	disappearance	NOUN
fcis-6025	12	11	problems	problem	NOUN
fcis-6025	12	12	during	during	ADP
fcis-6025	12	13	the	the	DET
fcis-6025	12	14	solution	solution	NOUN
fcis-6025	12	15	process	process	NOUN
fcis-6025	12	16	,	,	PUNCT
fcis-6025	12	17	and	and	CCONJ
fcis-6025	12	18	are	be	AUX
fcis-6025	12	19	not	not	PART
fcis-6025	12	20	effective	effective	ADJ
fcis-6025	12	21	for	for	ADP
fcis-6025	12	22	long	long	ADJ
fcis-6025	12	23	text	text	NOUN
fcis-6025	12	24	.	.	PUNCT
fcis-6025	13	1	the	the	DET
fcis-6025	13	2	problem	problem	NOUN
fcis-6025	13	3	has	have	AUX
fcis-6025	13	4	been	be	AUX
fcis-6025	13	5	effectively	effectively	ADV
fcis-6025	13	6	solved	solve	VERB
fcis-6025	13	7	by	by	ADP
fcis-6025	13	8	the	the	DET
fcis-6025	13	9	proposed	propose	VERB
fcis-6025	13	10	long	long	ADJ
fcis-6025	13	11	-	-	PUNCT
fcis-6025	13	12	short	short	ADJ
fcis-6025	13	13	term	term	NOUN
fcis-6025	13	14	memory	memory	NOUN
fcis-6025	13	15	neural	neural	ADJ
fcis-6025	13	16	network	network	NOUN
fcis-6025	13	17	(	(	PUNCT
fcis-6025	13	18	lstm	lstm	PROPN
fcis-6025	13	19	)	)	PUNCT
fcis-6025	13	20	,	,	PUNCT
fcis-6025	13	21	which	which	PRON
fcis-6025	13	22	is	be	AUX
fcis-6025	13	23	suitable	suitable	ADJ
fcis-6025	13	24	for	for	ADP
fcis-6025	13	25	solving	solve	VERB
fcis-6025	13	26	long	long	ADJ
fcis-6025	13	27	sequential	sequential	ADJ
fcis-6025	13	28	problems	problem	NOUN
fcis-6025	13	29	by	by	ADP
fcis-6025	13	30	adding	add	VERB
fcis-6025	13	31	three	three	NUM
fcis-6025	13	32	"	"	PUNCT
fcis-6025	13	33	gates	gate	NOUN
fcis-6025	13	34	"	"	PUNCT
fcis-6025	13	35	in	in	ADP
fcis-6025	13	36	the	the	DET
fcis-6025	13	37	hidden	hide	VERB
fcis-6025	13	38	layer	layer	NOUN
fcis-6025	13	39	to	to	PART
fcis-6025	13	40	control	control	VERB
fcis-6025	13	41	cell	cell	NOUN
fcis-6025	13	42	states.[4][5	states.[4][5	NOUN
fcis-6025	13	43	]	]	PUNCT
fcis-6025	13	44	it	it	PRON
fcis-6025	13	45	is	be	AUX
fcis-6025	13	46	suitable	suitable	ADJ
fcis-6025	13	47	for	for	ADP
fcis-6025	13	48	solving	solve	VERB
fcis-6025	13	49	long	long	ADJ
fcis-6025	13	50	sequence	sequence	NOUN
fcis-6025	13	51	problems	problem	NOUN
fcis-6025	13	52	.	.	PUNCT
fcis-6025	14	1	speech	speech	NOUN
fcis-6025	14	2	emotion	emotion	NOUN
fcis-6025	14	3	recognition	recognition	NOUN
fcis-6025	14	4	,	,	PUNCT
fcis-6025	14	5	as	as	ADP
fcis-6025	14	6	a	a	DET
fcis-6025	14	7	crucial	crucial	ADJ
fcis-6025	14	8	technology	technology	NOUN
fcis-6025	14	9	to	to	PART
fcis-6025	14	10	realize	realize	VERB
fcis-6025	14	11	human	human	ADJ
fcis-6025	14	12	-	-	PUNCT
fcis-6025	14	13	computer	computer	NOUN
fcis-6025	14	14	interaction	interaction	NOUN
fcis-6025	14	15	,	,	PUNCT
fcis-6025	14	16	has	have	VERB
fcis-6025	14	17	a	a	DET
fcis-6025	14	18	wide	wide	ADJ
fcis-6025	14	19	range	range	NOUN
fcis-6025	14	20	of	of	ADP
fcis-6025	14	21	applications	application	NOUN
fcis-6025	14	22	in	in	ADP
fcis-6025	14	23	many	many	ADJ
fcis-6025	14	24	fields	field	NOUN
fcis-6025	14	25	with	with	ADP
fcis-6025	14	26	natural	natural	ADJ
fcis-6025	14	27	human	human	ADJ
fcis-6025	14	28	-	-	PUNCT
fcis-6025	14	29	computer	computer	NOUN
fcis-6025	14	30	interaction	interaction	NOUN
fcis-6025	14	31	needs	need	NOUN
fcis-6025	14	32	.	.	PUNCT
fcis-6025	15	1	for	for	ADP
fcis-6025	15	2	example	example	NOUN
fcis-6025	15	3	,	,	PUNCT
fcis-6025	15	4	we	we	PRON
fcis-6025	15	5	can	can	AUX
fcis-6025	15	6	track	track	VERB
fcis-6025	15	7	the	the	DET
fcis-6025	15	8	emotional	emotional	ADJ
fcis-6025	15	9	changes	change	NOUN
fcis-6025	15	10	of	of	ADP
fcis-6025	15	11	depression	depression	NOUN
fcis-6025	15	12	patients	patient	NOUN
fcis-6025	15	13	to	to	PART
fcis-6025	15	14	understand	understand	VERB
fcis-6025	15	15	the	the	DET
fcis-6025	15	16	specific	specific	ADJ
fcis-6025	15	17	condition	condition	NOUN
fcis-6025	15	18	of	of	ADP
fcis-6025	15	19	patients	patient	NOUN
fcis-6025	15	20	from	from	ADP
fcis-6025	15	21	these	these	DET
fcis-6025	15	22	emotional	emotional	ADJ
fcis-6025	15	23	changes	change	NOUN
fcis-6025	15	24	,	,	PUNCT
fcis-6025	15	25	so	so	SCONJ
fcis-6025	15	26	as	as	SCONJ
fcis-6025	15	27	to	to	PART
fcis-6025	15	28	make	make	VERB
fcis-6025	15	29	corresponding	corresponding	ADJ
fcis-6025	15	30	diagnosis	diagnosis	NOUN
fcis-6025	15	31	and	and	CCONJ
fcis-6025	15	32	treatment	treatment	NOUN
fcis-6025	15	33	;	;	PUNCT
fcis-6025	15	34	we	we	PRON
fcis-6025	15	35	can	can	AUX
fcis-6025	15	36	monitor	monitor	VERB
fcis-6025	15	37	the	the	DET
fcis-6025	15	38	emotional	emotional	ADJ
fcis-6025	15	39	changes	change	NOUN
fcis-6025	15	40	of	of	ADP
fcis-6025	15	41	each	each	DET
fcis-6025	15	42	student	student	NOUN
fcis-6025	15	43	in	in	ADP
fcis-6025	15	44	real	real	ADJ
fcis-6025	15	45	time	time	NOUN
fcis-6025	15	46	in	in	ADP
fcis-6025	15	47	the	the	DET
fcis-6025	15	48	teaching	teaching	NOUN
fcis-6025	15	49	classroom	classroom	NOUN
fcis-6025	15	50	to	to	PART
fcis-6025	15	51	find	find	VERB
fcis-6025	15	52	out	out	ADP
fcis-6025	15	53	the	the	DET
fcis-6025	15	54	interest	interest	NOUN
fcis-6025	15	55	of	of	ADP
fcis-6025	15	56	each	each	DET
fcis-6025	15	57	student	student	NOUN
fcis-6025	15	58	in	in	ADP
fcis-6025	15	59	the	the	DET
fcis-6025	15	60	lecture	lecture	NOUN
fcis-6025	15	61	content	content	NOUN
fcis-6025	15	62	from	from	ADP
fcis-6025	15	63	these	these	DET
fcis-6025	15	64	emotional	emotional	ADJ
fcis-6025	15	65	changes	change	NOUN
fcis-6025	15	66	,	,	PUNCT
fcis-6025	15	67	so	so	SCONJ
fcis-6025	15	68	as	as	SCONJ
fcis-6025	15	69	to	to	PART
fcis-6025	15	70	make	make	VERB
fcis-6025	15	71	timely	timely	ADJ
fcis-6025	15	72	adjustments	adjustment	NOUN
fcis-6025	15	73	to	to	ADP
fcis-6025	15	74	the	the	DET
fcis-6025	15	75	lecture	lecture	NOUN
fcis-6025	15	76	content	content	NOUN
fcis-6025	15	77	and	and	CCONJ
fcis-6025	15	78	lecture	lecture	VERB
fcis-6025	15	79	style.[6	style.[6	NOUN
fcis-6025	15	80	]	]	PUNCT
fcis-6025	15	81	the	the	DET
fcis-6025	15	82	class	class	NOUN
fcis-6025	15	83	is	be	AUX
fcis-6025	15	84	monitored	monitor	VERB
fcis-6025	15	85	in	in	ADP
fcis-6025	15	86	real	real	ADJ
fcis-6025	15	87	time	time	NOUN
fcis-6025	15	88	to	to	PART
fcis-6025	15	89	detect	detect	VERB
fcis-6025	15	90	each	each	DET
fcis-6025	15	91	student	student	NOUN
fcis-6025	15	92	's	's	PART
fcis-6025	15	93	interest	interest	NOUN
fcis-6025	15	94	in	in	ADP
fcis-6025	15	95	the	the	DET
fcis-6025	15	96	lecture	lecture	NOUN
fcis-6025	15	97	content	content	NOUN
fcis-6025	15	98	,	,	PUNCT
fcis-6025	15	99	so	so	SCONJ
fcis-6025	15	100	that	that	SCONJ
fcis-6025	15	101	adjustments	adjustment	NOUN
fcis-6025	15	102	can	can	AUX
fcis-6025	15	103	be	be	AUX
fcis-6025	15	104	made	make	VERB
fcis-6025	15	105	to	to	ADP
fcis-6025	15	106	the	the	DET
fcis-6025	15	107	lecture	lecture	NOUN
fcis-6025	15	108	content	content	NOUN
fcis-6025	15	109	and	and	CCONJ
fcis-6025	15	110	lecture	lecture	NOUN
fcis-6025	15	111	style	style	NOUN
fcis-6025	15	112	.	.	PUNCT
fcis-6025	16	1	2	2	X
fcis-6025	16	2	.	.	X
fcis-6025	16	3	related	relate	VERB
fcis-6025	16	4	work	work	NOUN
fcis-6025	16	5	this	this	DET
fcis-6025	16	6	paper	paper	NOUN
fcis-6025	16	7	adopts	adopt	VERB
fcis-6025	16	8	the	the	DET
fcis-6025	16	9	features	feature	NOUN
fcis-6025	16	10	of	of	ADP
fcis-6025	16	11	the	the	DET
fcis-6025	16	12	speech	speech	NOUN
fcis-6025	16	13	spectrum	spectrum	NOUN
fcis-6025	16	14	map	map	NOUN
fcis-6025	16	15	as	as	ADP
fcis-6025	16	16	the	the	DET
fcis-6025	16	17	input	input	NOUN
fcis-6025	16	18	data	datum	NOUN
fcis-6025	16	19	of	of	ADP
fcis-6025	16	20	the	the	DET
fcis-6025	16	21	network	network	NOUN
fcis-6025	16	22	structure	structure	NOUN
fcis-6025	16	23	,	,	PUNCT
fcis-6025	16	24	which	which	PRON
fcis-6025	16	25	solves	solve	VERB
fcis-6025	16	26	the	the	DET
fcis-6025	16	27	problem	problem	NOUN
fcis-6025	16	28	that	that	SCONJ
fcis-6025	16	29	the	the	DET
fcis-6025	16	30	data	data	NOUN
fcis-6025	16	31	structure	structure	NOUN
fcis-6025	16	32	is	be	AUX
fcis-6025	16	33	not	not	PART
fcis-6025	16	34	similar	similar	ADJ
fcis-6025	16	35	to	to	ADP
fcis-6025	16	36	the	the	DET
fcis-6025	16	37	twodimensional	twodimensional	ADJ
fcis-6025	16	38	structure	structure	NOUN
fcis-6025	16	39	of	of	ADP
fcis-6025	16	40	image	image	NOUN
fcis-6025	16	41	data	datum	NOUN
fcis-6025	16	42	.	.	PUNCT
fcis-6025	17	1	the	the	DET
fcis-6025	17	2	traditional	traditional	ADJ
fcis-6025	17	3	convolutional	convolutional	ADJ
fcis-6025	17	4	neural	neural	ADJ
fcis-6025	17	5	network	network	NOUN
fcis-6025	17	6	has	have	VERB
fcis-6025	17	7	serious	serious	ADJ
fcis-6025	17	8	data	datum	NOUN
fcis-6025	17	9	loss	loss	NOUN
fcis-6025	17	10	problems	problem	NOUN
fcis-6025	17	11	in	in	ADP
fcis-6025	17	12	information	information	NOUN
fcis-6025	17	13	transfer	transfer	NOUN
fcis-6025	17	14	,	,	PUNCT
fcis-6025	17	15	and	and	CCONJ
fcis-6025	17	16	the	the	DET
fcis-6025	17	17	gradient	gradient	NOUN
fcis-6025	17	18	disappears	disappear	VERB
fcis-6025	17	19	or	or	CCONJ
fcis-6025	17	20	explodes	explode	NOUN
fcis-6025	17	21	as	as	SCONJ
fcis-6025	17	22	the	the	DET
fcis-6025	17	23	number	number	NOUN
fcis-6025	17	24	of	of	ADP
fcis-6025	17	25	layers	layer	NOUN
fcis-6025	17	26	of	of	ADP
fcis-6025	17	27	the	the	DET
fcis-6025	17	28	network	network	NOUN
fcis-6025	17	29	deepens.resnet	deepens.resnet	NOUN
fcis-6025	17	30	network	network	NOUN
fcis-6025	17	31	can	can	AUX
fcis-6025	17	32	effectively	effectively	ADV
fcis-6025	17	33	solve	solve	VERB
fcis-6025	17	34	the	the	DET
fcis-6025	17	35	above	above	ADJ
fcis-6025	17	36	problems	problem	NOUN
fcis-6025	17	37	,	,	PUNCT
fcis-6025	17	38	and	and	CCONJ
fcis-6025	17	39	this	this	DET
fcis-6025	17	40	paper	paper	NOUN
fcis-6025	17	41	will	will	AUX
fcis-6025	17	42	use	use	VERB
fcis-6025	17	43	resnet	resnet	NOUN
fcis-6025	17	44	network	network	NOUN
fcis-6025	17	45	as	as	ADP
fcis-6025	17	46	the	the	DET
fcis-6025	17	47	backbone	backbone	NOUN
fcis-6025	17	48	network	network	NOUN
fcis-6025	17	49	structure	structure	NOUN
fcis-6025	17	50	.	.	PUNCT
fcis-6025	18	1	2.1	2.1	NUM
fcis-6025	18	2	.	.	PUNCT
fcis-6025	18	3	speech	speech	NOUN
fcis-6025	18	4	pre	pre	ADJ
fcis-6025	18	5	-	-	ADJ
fcis-6025	18	6	processing	processing	ADJ
fcis-6025	18	7	technology	technology	NOUN
fcis-6025	18	8	emotion	emotion	NOUN
fcis-6025	18	9	feature	feature	NOUN
fcis-6025	18	10	extraction	extraction	NOUN
fcis-6025	18	11	is	be	AUX
fcis-6025	18	12	important	important	ADJ
fcis-6025	18	13	in	in	ADP
fcis-6025	18	14	speech	speech	NOUN
fcis-6025	18	15	emotion	emotion	NOUN
fcis-6025	18	16	recognition	recognition	NOUN
fcis-6025	18	17	.	.	PUNCT
fcis-6025	19	1	currently	currently	ADV
fcis-6025	19	2	,	,	PUNCT
fcis-6025	19	3	there	there	PRON
fcis-6025	19	4	are	be	VERB
fcis-6025	19	5	many	many	ADJ
fcis-6025	19	6	effective	effective	ADJ
fcis-6025	19	7	emotion	emotion	NOUN
fcis-6025	19	8	features	feature	NOUN
fcis-6025	19	9	commonly	commonly	ADV
fcis-6025	19	10	used	use	VERB
fcis-6025	19	11	in	in	ADP
fcis-6025	19	12	the	the	DET
fcis-6025	19	13	field	field	NOUN
fcis-6025	19	14	of	of	ADP
fcis-6025	19	15	speech	speech	NOUN
fcis-6025	19	16	emotion	emotion	NOUN
fcis-6025	19	17	feature	feature	NOUN
fcis-6025	19	18	research	research	NOUN
fcis-6025	19	19	,	,	PUNCT
fcis-6025	19	20	and	and	CCONJ
fcis-6025	19	21	they	they	PRON
fcis-6025	19	22	express	express	VERB
fcis-6025	19	23	the	the	DET
fcis-6025	19	24	emotions	emotion	NOUN
fcis-6025	19	25	embedded	embed	VERB
fcis-6025	19	26	in	in	ADP
fcis-6025	19	27	speech	speech	NOUN
fcis-6025	19	28	at	at	ADP
fcis-6025	19	29	different	different	ADJ
fcis-6025	19	30	levels	level	NOUN
fcis-6025	19	31	respectively	respectively	ADV
fcis-6025	19	32	.	.	PUNCT
fcis-6025	20	1	in	in	ADP
fcis-6025	20	2	this	this	DET
fcis-6025	20	3	thesis	thesis	NOUN
fcis-6025	20	4	,	,	PUNCT
fcis-6025	20	5	the	the	DET
fcis-6025	20	6	extracted	extract	VERB
fcis-6025	20	7	speech	speech	NOUN
fcis-6025	20	8	spectrograms	spectrogram	NOUN
fcis-6025	20	9	are	be	AUX
fcis-6025	20	10	used	use	VERB
fcis-6025	20	11	as	as	ADP
fcis-6025	20	12	the	the	DET
fcis-6025	20	13	dataset	dataset	NOUN
fcis-6025	20	14	for	for	ADP
fcis-6025	20	15	model	model	NOUN
fcis-6025	20	16	training	training	NOUN
fcis-6025	20	17	.	.	PUNCT
fcis-6025	21	1	the	the	DET
fcis-6025	21	2	speech	speech	NOUN
fcis-6025	21	3	spectrogram	spectrogram	NOUN
fcis-6025	21	4	represents	represent	VERB
fcis-6025	21	5	the	the	DET
fcis-6025	21	6	speech	speech	NOUN
fcis-6025	21	7	signal	signal	NOUN
fcis-6025	21	8	in	in	ADP
fcis-6025	21	9	visual	visual	ADJ
fcis-6025	21	10	form	form	NOUN
fcis-6025	21	11	and	and	CCONJ
fcis-6025	21	12	contains	contain	VERB
fcis-6025	21	13	many	many	ADJ
fcis-6025	21	14	essential	essential	ADJ
fcis-6025	21	15	information	information	NOUN
fcis-6025	21	16	of	of	ADP
fcis-6025	21	17	the	the	DET
fcis-6025	21	18	speech	speech	NOUN
fcis-6025	21	19	signal	signal	NOUN
fcis-6025	21	20	,	,	PUNCT
fcis-6025	21	21	such	such	ADJ
fcis-6025	21	22	as	as	ADP
fcis-6025	21	23	:	:	PUNCT
fcis-6025	21	24	fundamental	fundamental	ADJ
fcis-6025	21	25	period	period	NOUN
fcis-6025	21	26	,	,	PUNCT
fcis-6025	21	27	amplitude	amplitude	NOUN
fcis-6025	21	28	,	,	PUNCT
fcis-6025	21	29	etc	etc	X
fcis-6025	21	30	.	.	X
fcis-6025	22	1	in	in	ADP
fcis-6025	22	2	a	a	DET
fcis-6025	22	3	speech	speech	NOUN
fcis-6025	22	4	spectrogram	spectrogram	NOUN
fcis-6025	22	5	,	,	PUNCT
fcis-6025	22	6	a	a	DET
fcis-6025	22	7	two	two	NUM
fcis-6025	22	8	-	-	PUNCT
fcis-6025	22	9	dimensional	dimensional	ADJ
fcis-6025	22	10	plane	plane	NOUN
fcis-6025	22	11	is	be	AUX
fcis-6025	22	12	used	use	VERB
fcis-6025	22	13	to	to	PART
fcis-6025	22	14	represent	represent	VERB
fcis-6025	22	15	three	three	NUM
fcis-6025	22	16	-	-	PUNCT
fcis-6025	22	17	dimensional	dimensional	ADJ
fcis-6025	22	18	information	information	NOUN
fcis-6025	22	19	,	,	PUNCT
fcis-6025	22	20	the	the	DET
fcis-6025	22	21	horizontal	horizontal	ADJ
fcis-6025	22	22	axis	axis	NOUN
fcis-6025	22	23	shows	show	VERB
fcis-6025	22	24	the	the	DET
fcis-6025	22	25	time	time	NOUN
fcis-6025	22	26	and	and	CCONJ
fcis-6025	22	27	the	the	DET
fcis-6025	22	28	vertical	vertical	ADJ
fcis-6025	22	29	axis	axis	NOUN
fcis-6025	22	30	shows	show	VERB
fcis-6025	22	31	the	the	DET
fcis-6025	22	32	frequency	frequency	NOUN
fcis-6025	22	33	,	,	PUNCT
fcis-6025	22	34	and	and	CCONJ
fcis-6025	22	35	the	the	DET
fcis-6025	22	36	intensity	intensity	NOUN
fcis-6025	22	37	of	of	ADP
fcis-6025	22	38	a	a	DET
fcis-6025	22	39	frequency	frequency	NOUN
fcis-6025	22	40	at	at	ADP
fcis-6025	22	41	a	a	DET
fcis-6025	22	42	certain	certain	ADJ
fcis-6025	22	43	moment	moment	NOUN
fcis-6025	22	44	can	can	AUX
fcis-6025	22	45	be	be	AUX
fcis-6025	22	46	indicated	indicate	VERB
fcis-6025	22	47	by	by	ADP
fcis-6025	22	48	the	the	DET
fcis-6025	22	49	color	color	NOUN
fcis-6025	22	50	shade	shade	NOUN
fcis-6025	22	51	or	or	CCONJ
fcis-6025	22	52	the	the	DET
fcis-6025	22	53	gray	gray	ADJ
fcis-6025	22	54	scale	scale	NOUN
fcis-6025	22	55	of	of	ADP
fcis-6025	22	56	the	the	DET
fcis-6025	22	57	point	point	NOUN
fcis-6025	22	58	,	,	PUNCT
fcis-6025	22	59	and	and	CCONJ
fcis-6025	22	60	the	the	DET
fcis-6025	22	61	energy	energy	NOUN
fcis-6025	22	62	value	value	NOUN
fcis-6025	22	63	can	can	AUX
fcis-6025	22	64	also	also	ADV
fcis-6025	22	65	be	be	AUX
fcis-6025	22	66	indicated	indicate	VERB
fcis-6025	22	67	by	by	ADP
fcis-6025	22	68	the	the	DET
fcis-6025	22	69	color	color	NOUN
fcis-6025	22	70	shade	shade	NOUN
fcis-6025	22	71	,	,	PUNCT
fcis-6025	22	72	the	the	DET
fcis-6025	22	73	darker	dark	ADJ
fcis-6025	22	74	the	the	DET
fcis-6025	22	75	color	color	NOUN
fcis-6025	22	76	of	of	ADP
fcis-6025	22	77	a	a	DET
fcis-6025	22	78	point	point	NOUN
fcis-6025	22	79	means	mean	VERB
fcis-6025	22	80	the	the	PRON
fcis-6025	22	81	stronger	strong	ADJ
fcis-6025	22	82	the	the	DET
fcis-6025	22	83	energy	energy	NOUN
fcis-6025	22	84	at	at	ADP
fcis-6025	22	85	that	that	DET
fcis-6025	22	86	place.[7][8	place.[7][8	NOUN
fcis-6025	22	87	]	]	PUNCT
fcis-6025	22	88	the	the	DET
fcis-6025	22	89	darker	dark	ADJ
fcis-6025	22	90	the	the	DET
fcis-6025	22	91	color	color	NOUN
fcis-6025	22	92	of	of	ADP
fcis-6025	22	93	a	a	DET
fcis-6025	22	94	point	point	NOUN
fcis-6025	22	95	,	,	PUNCT
fcis-6025	22	96	the	the	PRON
fcis-6025	22	97	stronger	strong	ADJ
fcis-6025	22	98	the	the	DET
fcis-6025	22	99	energy	energy	NOUN
fcis-6025	22	100	at	at	ADP
fcis-6025	22	101	that	that	DET
fcis-6025	22	102	point	point	NOUN
fcis-6025	22	103	.	.	PUNCT
fcis-6025	23	1	the	the	DET
fcis-6025	23	2	specific	specific	ADJ
fcis-6025	23	3	speech	speech	NOUN
fcis-6025	23	4	spectrum	spectrum	NOUN
fcis-6025	23	5	extraction	extraction	NOUN
fcis-6025	23	6	technique	technique	NOUN
fcis-6025	23	7	is	be	AUX
fcis-6025	23	8	shown	show	VERB
fcis-6025	23	9	in	in	ADP
fcis-6025	23	10	figure	figure	NOUN
fcis-6025	23	11	1	1	NUM
fcis-6025	23	12	.	.	PUNCT
fcis-6025	24	1	the	the	DET
fcis-6025	24	2	speech	speech	NOUN
fcis-6025	24	3	spectrogram	spectrogram	NOUN
fcis-6025	24	4	is	be	AUX
fcis-6025	24	5	obtained	obtain	VERB
fcis-6025	24	6	by	by	ADP
fcis-6025	24	7	fourier	fourier	ADJ
fcis-6025	24	8	analysis	analysis	NOUN
fcis-6025	24	9	of	of	ADP
fcis-6025	24	10	the	the	DET
fcis-6025	24	11	speech	speech	NOUN
fcis-6025	24	12	signal	signal	NOUN
fcis-6025	24	13	.	.	PUNCT
fcis-6025	25	1	first	first	ADV
fcis-6025	25	2	,	,	PUNCT
fcis-6025	25	3	the	the	DET
fcis-6025	25	4	discrete	discrete	ADJ
fcis-6025	25	5	speech	speech	NOUN
fcis-6025	25	6	signal	signal	NOUN
fcis-6025	25	7	𝑥	𝑥	X
fcis-6025	25	8	(	(	PUNCT
fcis-6025	25	9	𝑛	𝑛	NOUN
fcis-6025	25	10	)	)	PUNCT
fcis-6025	25	11	is	be	AUX
fcis-6025	25	12	represented	represent	VERB
fcis-6025	25	13	as	as	ADP
fcis-6025	25	14	𝑥(𝑚	𝑥(𝑚	NOUN
fcis-6025	25	15	)	)	PUNCT
fcis-6025	25	16	,	,	PUNCT
fcis-6025	25	17	n	n	NOUN
fcis-6025	25	18	=	=	SYM
fcis-6025	25	19	0	0	NUM
fcis-6025	25	20	,	,	PUNCT
fcis-6025	25	21	1	1	NUM
fcis-6025	25	22	,	,	PUNCT
fcis-6025	25	23	n-1	n-1	NOUN
fcis-6025	25	24	after	after	ADP
fcis-6025	25	25	framing	frame	VERB
fcis-6025	25	26	,	,	PUNCT
fcis-6025	25	27	𝑚	𝑚	PROPN
fcis-6025	25	28	is	be	AUX
fcis-6025	25	29	the	the	DET
fcis-6025	25	30	sample	sample	NOUN
fcis-6025	25	31	point	point	NOUN
fcis-6025	25	32	number	number	NOUN
fcis-6025	25	33	within	within	ADP
fcis-6025	25	34	a	a	DET
fcis-6025	25	35	frame	frame	NOUN
fcis-6025	25	36	,	,	PUNCT
fcis-6025	25	37	𝑛	𝑛	PROPN
fcis-6025	25	38	and	and	CCONJ
fcis-6025	25	39	n	n	CCONJ
fcis-6025	25	40	denote	denote	VERB
fcis-6025	25	41	the	the	DET
fcis-6025	25	42	frame	frame	NOUN
fcis-6025	25	43	number	number	NOUN
fcis-6025	25	44	and	and	CCONJ
fcis-6025	25	45	frame	frame	NOUN
fcis-6025	25	46	length	length	NOUN
fcis-6025	25	47	,	,	PUNCT
fcis-6025	25	48	respectively	respectively	ADV
fcis-6025	25	49	.	.	PUNCT
fcis-6025	26	1	the	the	DET
fcis-6025	26	2	short	short	ADJ
fcis-6025	26	3	-	-	PUNCT
fcis-6025	26	4	time	time	NOUN
fcis-6025	26	5	61	61	NUM
fcis-6025	26	6	fourier	fourier	NOUN
fcis-6025	26	7	transform	transform	NOUN
fcis-6025	26	8	of	of	ADP
fcis-6025	26	9	the	the	DET
fcis-6025	26	10	signal	signal	NOUN
fcis-6025	26	11	𝑥	𝑥	X
fcis-6025	26	12	(	(	PUNCT
fcis-6025	26	13	𝑛	𝑛	NOUN
fcis-6025	26	14	)	)	PUNCT
fcis-6025	26	15	is	be	AUX
fcis-6025	26	16	shown	show	VERB
fcis-6025	26	17	in	in	ADP
fcis-6025	26	18	the	the	DET
fcis-6025	26	19	following	follow	VERB
fcis-6025	26	20	equation	equation	NOUN
fcis-6025	26	21	.	.	PUNCT
fcis-6025	27	1	figure	figure	NOUN
fcis-6025	27	2	1	1	NUM
fcis-6025	27	3	.	.	PUNCT
fcis-6025	28	1	the	the	DET
fcis-6025	28	2	process	process	NOUN
fcis-6025	28	3	of	of	ADP
fcis-6025	28	4	extracting	extract	VERB
fcis-6025	28	5	the	the	DET
fcis-6025	28	6	language	language	NOUN
fcis-6025	28	7	spectrum	spectrum	NOUN
fcis-6025	28	8	map	map	NOUN
fcis-6025	28	9	(	(	PUNCT
fcis-6025	28	10	)	)	PUNCT
fcis-6025	28	11	(	(	PUNCT
fcis-6025	28	12	)	)	PUNCT
fcis-6025	28	13	(	(	PUNCT
fcis-6025	28	14	)	)	PUNCT
fcis-6025	28	15	jw	jw	PROPN
fcis-6025	28	16	jwm	jwm	PROPN
fcis-6025	28	17	n	n	ADV
fcis-6025	28	18	m	m	VERB
fcis-6025	28	19	x	x	X
fcis-6025	28	20	e	e	NOUN
fcis-6025	28	21	x	x	X
fcis-6025	28	22	m	m	VERB
fcis-6025	28	23	w	w	VERB
fcis-6025	28	24	n	n	ADV
fcis-6025	28	25	m	m	PROPN
fcis-6025	28	26	e	e	NOUN
fcis-6025	28	27			NOUN
fcis-6025	28	28			VERB
fcis-6025	28	29	−	−	ADP
fcis-6025	28	30	=	=	SYM
fcis-6025	28	31	−	−	NOUN
fcis-6025	28	32	=	=	SYM
fcis-6025	28	33			NUM
fcis-6025	28	34	−	−	PROPN
fcis-6025	28	35			SYM
fcis-6025	28	36	(	(	PUNCT
fcis-6025	28	37	1	1	NUM
fcis-6025	28	38	)	)	PUNCT
fcis-6025	28	39	w(𝑛	w(𝑛	NOUN
fcis-6025	28	40	)	)	PUNCT
fcis-6025	28	41	is	be	AUX
fcis-6025	28	42	the	the	DET
fcis-6025	28	43	window	window	NOUN
fcis-6025	28	44	function	function	NOUN
fcis-6025	28	45	,	,	PUNCT
fcis-6025	28	46	and	and	CCONJ
fcis-6025	28	47	the	the	DET
fcis-6025	28	48	discrete	discrete	ADJ
fcis-6025	28	49	time	time	NOUN
fcis-6025	28	50	domain	domain	NOUN
fcis-6025	28	51	fourier	fourier	NOUN
fcis-6025	28	52	transform	transform	VERB
fcis-6025	28	53	and	and	CCONJ
fcis-6025	28	54	discrete	discrete	ADJ
fcis-6025	28	55	fourier	fourier	NOUN
fcis-6025	28	56	transform	transform	NOUN
fcis-6025	28	57	of	of	ADP
fcis-6025	28	58	x(𝑛	x(𝑛	PRON
fcis-6025	28	59	)	)	PUNCT
fcis-6025	28	60	are	be	AUX
fcis-6025	28	61	as	as	SCONJ
fcis-6025	28	62	follows	follow	VERB
fcis-6025	28	63	.	.	PUNCT
fcis-6025	29	1	(	(	PUNCT
fcis-6025	29	2	)	)	PUNCT
fcis-6025	29	3	1	1	NUM
fcis-6025	29	4	0	0	NUM
fcis-6025	29	5	21	21	NUM
fcis-6025	29	6	0	0	NUM
fcis-6025	29	7	x	x	SYM
fcis-6025	29	8	n	n	CCONJ
fcis-6025	29	9	,	,	PUNCT
fcis-6025	29	10	(	(	PUNCT
fcis-6025	29	11	)	)	PUNCT
fcis-6025	29	12	x(n	x(n	NOUN
fcis-6025	29	13	,	,	PUNCT
fcis-6025	29	14	k	k	NOUN
fcis-6025	29	15	)	)	PUNCT
fcis-6025	29	16	(	(	PUNCT
fcis-6025	29	17	)	)	PUNCT
fcis-6025	29	18	n	n	X
fcis-6025	29	19	jw	jw	PROPN
fcis-6025	29	20	jwm	jwm	PROPN
fcis-6025	29	21	n	n	PROPN
fcis-6025	29	22	m	m	PROPN
fcis-6025	29	23	kmn	kmn	PROPN
fcis-6025	29	24	j	j	PROPN
fcis-6025	29	25	n	n	CCONJ
fcis-6025	29	26	n	n	NOUN
fcis-6025	29	27	m	m	VERB
fcis-6025	29	28	e	e	NOUN
fcis-6025	29	29	x	x	X
fcis-6025	29	30	m	m	VERB
fcis-6025	29	31	e	e	NOUN
fcis-6025	29	32	x	x	X
fcis-6025	29	33	m	m	PROPN
fcis-6025	29	34	e	e	NOUN
fcis-6025	29	35			NOUN
fcis-6025	29	36	−	−	NOUN
fcis-6025	29	37	−	−	PROPN
fcis-6025	29	38	=	=	SYM
fcis-6025	29	39	−	−	NOUN
fcis-6025	29	40	−	−	PROPN
fcis-6025	29	41	=	=	SYM
fcis-6025	29	42	=	=	SYM
fcis-6025	29	43	=	=	SYM
fcis-6025	29	44			X
fcis-6025	29	45			X
fcis-6025	29	46	(	(	PUNCT
fcis-6025	29	47	2	2	X
fcis-6025	29	48	)	)	PUNCT
fcis-6025	29	49	where	where	SCONJ
fcis-6025	29	50	0	0	NUM
fcis-6025	29	51	≤	≤	NUM
fcis-6025	29	52	k	k	X
fcis-6025	29	53	≤	≤	NUM
fcis-6025	29	54	n	n	DET
fcis-6025	29	55	1	1	NUM
fcis-6025	29	56	,	,	PUNCT
fcis-6025	29	57	x(n	x(n	NOUN
fcis-6025	29	58	,	,	PUNCT
fcis-6025	29	59	k	k	NOUN
fcis-6025	29	60	)	)	PUNCT
fcis-6025	29	61	denotes	denote	VERB
fcis-6025	29	62	the	the	DET
fcis-6025	29	63	short	short	ADJ
fcis-6025	29	64	-	-	PUNCT
fcis-6025	29	65	time	time	NOUN
fcis-6025	29	66	amplitude	amplitude	NOUN
fcis-6025	29	67	spectrum	spectrum	NOUN
fcis-6025	29	68	estimate	estimate	NOUN
fcis-6025	29	69	of	of	ADP
fcis-6025	29	70	x(𝑛	x(𝑛	PROPN
fcis-6025	29	71	)	)	PUNCT
fcis-6025	29	72	,	,	PUNCT
fcis-6025	29	73	and	and	CCONJ
fcis-6025	29	74	the	the	DET
fcis-6025	29	75	spectral	spectral	ADJ
fcis-6025	29	76	energy	energy	NOUN
fcis-6025	29	77	density	density	NOUN
fcis-6025	29	78	function	function	NOUN
fcis-6025	29	79	p(𝑛,k	p(𝑛,k	ADV
fcis-6025	29	80	)	)	PUNCT
fcis-6025	29	81	at	at	ADP
fcis-6025	29	82	𝑚	𝑚	PROPN
fcis-6025	29	83	can	can	AUX
fcis-6025	29	84	be	be	AUX
fcis-6025	29	85	expressed	express	VERB
fcis-6025	29	86	as	as	ADP
fcis-6025	29	87	:	:	PUNCT
fcis-6025	29	88	𝑃(𝑛	𝑃(𝑛	PROPN
fcis-6025	29	89	,	,	PUNCT
fcis-6025	29	90	𝑘	𝑘	NOUN
fcis-6025	29	91	)	)	PUNCT
fcis-6025	29	92	=	=	SYM
fcis-6025	29	93	|𝑋(𝑛	|𝑋(𝑛	PROPN
fcis-6025	29	94	,	,	PUNCT
fcis-6025	29	95	𝑘)|2	𝑘)|2	X
fcis-6025	29	96	=	=	SYM
fcis-6025	29	97	(	(	PUNCT
fcis-6025	29	98	𝑥(𝑛	𝑥(𝑛	PROPN
fcis-6025	29	99	,	,	PUNCT
fcis-6025	29	100	𝑘	𝑘	NOUN
fcis-6025	29	101	)	)	PUNCT
fcis-6025	29	102	)	)	PUNCT
fcis-6025	29	103	×	×	NOUN
fcis-6025	29	104	(	(	PUNCT
fcis-6025	29	105	𝑐𝑜𝑛𝑗(𝑋(𝑛	𝑐𝑜𝑛𝑗(𝑋(𝑛	NOUN
fcis-6025	29	106	,	,	PUNCT
fcis-6025	29	107	𝑘	𝑘	NOUN
fcis-6025	29	108	)	)	PUNCT
fcis-6025	29	109	)	)	PUNCT
fcis-6025	29	110	)	)	PUNCT
fcis-6025	30	1	(	(	PUNCT
fcis-6025	30	2	3	3	X
fcis-6025	30	3	)	)	PUNCT
fcis-6025	30	4	the	the	DET
fcis-6025	30	5	horizontal	horizontal	ADJ
fcis-6025	30	6	coordinate	coordinate	NOUN
fcis-6025	30	7	is	be	AUX
fcis-6025	30	8	𝑛	𝑛	PRON
fcis-6025	30	9	and	and	CCONJ
fcis-6025	30	10	the	the	DET
fcis-6025	30	11	vertical	vertical	ADJ
fcis-6025	30	12	coordinate	coordinate	NOUN
fcis-6025	30	13	is	be	AUX
fcis-6025	30	14	k.	k.	PROPN
fcis-6025	30	15	the	the	DET
fcis-6025	30	16	values	value	NOUN
fcis-6025	30	17	of	of	ADP
fcis-6025	30	18	p	p	NOUN
fcis-6025	30	19	(	(	PUNCT
fcis-6025	30	20	𝑛	𝑛	PROPN
fcis-6025	30	21	,	,	PUNCT
fcis-6025	30	22	k	k	NOUN
fcis-6025	30	23	)	)	PUNCT
fcis-6025	30	24	are	be	AUX
fcis-6025	30	25	represented	represent	VERB
fcis-6025	30	26	in	in	ADP
fcis-6025	30	27	color	color	NOUN
fcis-6025	30	28	or	or	CCONJ
fcis-6025	30	29	gray	gray	ADJ
fcis-6025	30	30	,	,	PUNCT
fcis-6025	30	31	and	and	CCONJ
fcis-6025	30	32	finally	finally	ADV
fcis-6025	30	33	the	the	DET
fcis-6025	30	34	speech	speech	NOUN
fcis-6025	30	35	spectrum	spectrum	NOUN
fcis-6025	30	36	diagram	diagram	NOUN
fcis-6025	30	37	is	be	AUX
fcis-6025	30	38	obtained	obtain	VERB
fcis-6025	30	39	.	.	PUNCT
fcis-6025	31	1	the	the	DET
fcis-6025	31	2	speech	speech	NOUN
fcis-6025	31	3	spectrogram	spectrogram	NOUN
fcis-6025	31	4	is	be	AUX
fcis-6025	31	5	generally	generally	ADV
fcis-6025	31	6	a	a	DET
fcis-6025	31	7	gray	gray	ADJ
fcis-6025	31	8	or	or	CCONJ
fcis-6025	31	9	colored	colored	ADJ
fcis-6025	31	10	twodimensional	twodimensional	ADJ
fcis-6025	31	11	image	image	NOUN
fcis-6025	31	12	,	,	PUNCT
fcis-6025	31	13	and	and	CCONJ
fcis-6025	31	14	each	each	DET
fcis-6025	31	15	point	point	NOUN
fcis-6025	31	16	contains	contain	VERB
fcis-6025	31	17	the	the	DET
fcis-6025	31	18	corresponding	corresponding	ADJ
fcis-6025	31	19	time	time	NOUN
fcis-6025	31	20	,	,	PUNCT
fcis-6025	31	21	frequency	frequency	NOUN
fcis-6025	31	22	,	,	PUNCT
fcis-6025	31	23	energy	energy	NOUN
fcis-6025	31	24	and	and	CCONJ
fcis-6025	31	25	other	other	ADJ
fcis-6025	31	26	information	information	NOUN
fcis-6025	31	27	of	of	ADP
fcis-6025	31	28	the	the	DET
fcis-6025	31	29	speech	speech	NOUN
fcis-6025	31	30	signal	signal	NOUN
fcis-6025	31	31	.	.	PUNCT
fcis-6025	32	1	the	the	DET
fcis-6025	32	2	speech	speech	NOUN
fcis-6025	32	3	spectrogram	spectrogram	NOUN
fcis-6025	32	4	of	of	ADP
fcis-6025	32	5	a	a	DET
fcis-6025	32	6	speech	speech	NOUN
fcis-6025	32	7	segment	segment	NOUN
fcis-6025	32	8	of	of	ADP
fcis-6025	32	9	anger	anger	NOUN
fcis-6025	32	10	class	class	NOUN
fcis-6025	32	11	in	in	ADP
fcis-6025	32	12	casia	casia	NOUN
fcis-6025	32	13	chinese	chinese	ADJ
fcis-6025	32	14	emotion	emotion	NOUN
fcis-6025	32	15	dataset	dataset	NOUN
fcis-6025	32	16	is	be	AUX
fcis-6025	32	17	shown	show	VERB
fcis-6025	32	18	in	in	ADP
fcis-6025	32	19	figure	figure	NOUN
fcis-6025	32	20	2	2	NUM
fcis-6025	32	21	.	.	PUNCT
fcis-6025	32	22	figure	figure	NOUN
fcis-6025	32	23	2	2	NUM
fcis-6025	32	24	.	.	PUNCT
fcis-6025	33	1	the	the	DET
fcis-6025	33	2	emotional	emotional	ADJ
fcis-6025	33	3	spectrum	spectrum	NOUN
fcis-6025	33	4	of	of	ADP
fcis-6025	33	5	the	the	DET
fcis-6025	33	6	anger	anger	NOUN
fcis-6025	33	7	category	category	NOUN
fcis-6025	33	8	in	in	ADP
fcis-6025	33	9	casia	casia	NOUN
fcis-6025	33	10	chinese	chinese	ADJ
fcis-6025	33	11	emotional	emotional	ADJ
fcis-6025	33	12	database	database	NOUN
fcis-6025	33	13	from	from	ADP
fcis-6025	33	14	figure	figure	NOUN
fcis-6025	33	15	2	2	NUM
fcis-6025	33	16	,	,	PUNCT
fcis-6025	33	17	we	we	PRON
fcis-6025	33	18	can	can	AUX
fcis-6025	33	19	clearly	clearly	ADV
fcis-6025	33	20	see	see	VERB
fcis-6025	33	21	many	many	ADJ
fcis-6025	33	22	horizontal	horizontal	ADJ
fcis-6025	33	23	stripes	stripe	NOUN
fcis-6025	33	24	,	,	PUNCT
fcis-6025	33	25	i.e.	i.e.	X
fcis-6025	33	26	,	,	PUNCT
fcis-6025	33	27	sound	sound	ADJ
fcis-6025	33	28	patterns	pattern	NOUN
fcis-6025	33	29	,	,	PUNCT
fcis-6025	33	30	which	which	PRON
fcis-6025	33	31	are	be	AUX
fcis-6025	33	32	formed	form	VERB
fcis-6025	33	33	by	by	ADP
fcis-6025	33	34	the	the	DET
fcis-6025	33	35	aggregation	aggregation	NOUN
fcis-6025	33	36	of	of	ADP
fcis-6025	33	37	dark	dark	ADJ
fcis-6025	33	38	-	-	PUNCT
fcis-6025	33	39	colored	color	VERB
fcis-6025	33	40	pixels	pixel	NOUN
fcis-6025	33	41	,	,	PUNCT
fcis-6025	33	42	i.e.	i.e.	X
fcis-6025	33	43	,	,	PUNCT
fcis-6025	33	44	by	by	ADP
fcis-6025	33	45	the	the	DET
fcis-6025	33	46	extension	extension	NOUN
fcis-6025	33	47	of	of	ADP
fcis-6025	33	48	the	the	DET
fcis-6025	33	49	duration	duration	NOUN
fcis-6025	33	50	points	point	NOUN
fcis-6025	33	51	of	of	ADP
fcis-6025	33	52	time	time	NOUN
fcis-6025	33	53	,	,	PUNCT
fcis-6025	33	54	indicating	indicate	VERB
fcis-6025	33	55	that	that	SCONJ
fcis-6025	33	56	the	the	DET
fcis-6025	33	57	frequency	frequency	NOUN
fcis-6025	33	58	component	component	NOUN
fcis-6025	33	59	at	at	ADP
fcis-6025	33	60	that	that	DET
fcis-6025	33	61	moment	moment	NOUN
fcis-6025	33	62	has	have	VERB
fcis-6025	33	63	a	a	DET
fcis-6025	33	64	large	large	ADJ
fcis-6025	33	65	proportion	proportion	NOUN
fcis-6025	33	66	and	and	CCONJ
fcis-6025	33	67	is	be	AUX
fcis-6025	33	68	easily	easily	ADV
fcis-6025	33	69	perceived	perceive	VERB
fcis-6025	33	70	.	.	PUNCT
fcis-6025	34	1	the	the	PRON
fcis-6025	34	2	greater	great	ADJ
fcis-6025	34	3	the	the	DET
fcis-6025	34	4	proportion	proportion	NOUN
fcis-6025	34	5	of	of	ADP
fcis-6025	34	6	stripes	stripe	NOUN
fcis-6025	34	7	represents	represent	VERB
fcis-6025	34	8	the	the	DET
fcis-6025	34	9	greater	great	ADJ
fcis-6025	34	10	the	the	DET
fcis-6025	34	11	proportion	proportion	NOUN
fcis-6025	34	12	of	of	ADP
fcis-6025	34	13	effective	effective	ADJ
fcis-6025	34	14	information	information	NOUN
fcis-6025	34	15	in	in	ADP
fcis-6025	34	16	the	the	DET
fcis-6025	34	17	whole	whole	ADJ
fcis-6025	34	18	speech	speech	NOUN
fcis-6025	34	19	signal	signal	NOUN
fcis-6025	34	20	,	,	PUNCT
fcis-6025	34	21	and	and	CCONJ
fcis-6025	34	22	the	the	DET
fcis-6025	34	23	more	more	ADV
fcis-6025	34	24	obvious	obvious	ADJ
fcis-6025	34	25	the	the	DET
fcis-6025	34	26	influence	influence	NOUN
fcis-6025	34	27	on	on	ADP
fcis-6025	34	28	the	the	DET
fcis-6025	34	29	speech	speech	NOUN
fcis-6025	34	30	emotion	emotion	NOUN
fcis-6025	34	31	recognition	recognition	NOUN
fcis-6025	34	32	rate	rate	NOUN
fcis-6025	34	33	.	.	PUNCT
fcis-6025	35	1	there	there	PRON
fcis-6025	35	2	is	be	VERB
fcis-6025	35	3	a	a	DET
fcis-6025	35	4	lot	lot	NOUN
fcis-6025	35	5	of	of	ADP
fcis-6025	35	6	information	information	NOUN
fcis-6025	35	7	that	that	PRON
fcis-6025	35	8	can	can	AUX
fcis-6025	35	9	be	be	AUX
fcis-6025	35	10	obtained	obtain	VERB
fcis-6025	35	11	from	from	ADP
fcis-6025	35	12	the	the	DET
fcis-6025	35	13	speech	speech	NOUN
fcis-6025	35	14	spectrogram	spectrogram	NOUN
fcis-6025	35	15	,	,	PUNCT
fcis-6025	35	16	such	such	ADJ
fcis-6025	35	17	as	as	ADP
fcis-6025	35	18	the	the	DET
fcis-6025	35	19	change	change	NOUN
fcis-6025	35	20	of	of	ADP
fcis-6025	35	21	sound	sound	ADJ
fcis-6025	35	22	intensity	intensity	NOUN
fcis-6025	35	23	in	in	ADP
fcis-6025	35	24	a	a	DET
fcis-6025	35	25	certain	certain	ADJ
fcis-6025	35	26	period	period	NOUN
fcis-6025	35	27	of	of	ADP
fcis-6025	35	28	time	time	NOUN
fcis-6025	35	29	,	,	PUNCT
fcis-6025	35	30	the	the	DET
fcis-6025	35	31	frequency	frequency	NOUN
fcis-6025	35	32	distribution	distribution	NOUN
fcis-6025	35	33	and	and	CCONJ
fcis-6025	35	34	sound	sound	NOUN
fcis-6025	35	35	intensity	intensity	NOUN
fcis-6025	35	36	strength	strength	NOUN
fcis-6025	35	37	of	of	ADP
fcis-6025	35	38	the	the	DET
fcis-6025	35	39	whole	whole	ADJ
fcis-6025	35	40	speech	speech	NOUN
fcis-6025	35	41	,	,	PUNCT
fcis-6025	35	42	etc	etc	X
fcis-6025	35	43	.	.	X
fcis-6025	35	44	for	for	ADP
fcis-6025	35	45	different	different	ADJ
fcis-6025	35	46	emotions	emotion	NOUN
fcis-6025	35	47	of	of	ADP
fcis-6025	35	48	speech	speech	NOUN
fcis-6025	35	49	,	,	PUNCT
fcis-6025	35	50	there	there	PRON
fcis-6025	35	51	is	be	VERB
fcis-6025	35	52	a	a	DET
fcis-6025	35	53	certain	certain	ADJ
fcis-6025	35	54	degree	degree	NOUN
fcis-6025	35	55	of	of	ADP
fcis-6025	35	56	difference	difference	NOUN
fcis-6025	35	57	in	in	ADP
fcis-6025	35	58	the	the	DET
fcis-6025	35	59	speech	speech	NOUN
fcis-6025	35	60	spectrogram	spectrogram	NOUN
fcis-6025	35	61	,	,	PUNCT
fcis-6025	35	62	and	and	CCONJ
fcis-6025	35	63	through	through	ADP
fcis-6025	35	64	these	these	DET
fcis-6025	35	65	differences	difference	NOUN
fcis-6025	35	66	,	,	PUNCT
fcis-6025	35	67	different	different	ADJ
fcis-6025	35	68	emotions	emotion	NOUN
fcis-6025	35	69	can	can	AUX
fcis-6025	35	70	be	be	AUX
fcis-6025	35	71	classified[9][10	classified[9][10	NUM
fcis-6025	35	72	]	]	PUNCT
fcis-6025	35	73	.	.	PUNCT
fcis-6025	36	1	2.2	2.2	NUM
fcis-6025	36	2	.	.	PUNCT
fcis-6025	37	1	att	att	PROPN
fcis-6025	37	2	-	-	PUNCT
fcis-6025	37	3	res	re	NOUN
fcis-6025	37	4	speech	speech	NOUN
fcis-6025	37	5	emotion	emotion	NOUN
fcis-6025	37	6	recognition	recognition	NOUN
fcis-6025	37	7	network	network	NOUN
fcis-6025	37	8	model	model	NOUN
fcis-6025	37	9	construction	construction	NOUN
fcis-6025	37	10	based	base	VERB
fcis-6025	37	11	on	on	ADP
fcis-6025	37	12	residual	residual	ADJ
fcis-6025	37	13	network	network	NOUN
fcis-6025	37	14	improvement	improvement	NOUN
fcis-6025	37	15	in	in	ADP
fcis-6025	37	16	this	this	DET
fcis-6025	37	17	paper	paper	NOUN
fcis-6025	37	18	,	,	PUNCT
fcis-6025	37	19	the	the	DET
fcis-6025	37	20	language	language	NOUN
fcis-6025	37	21	spectrum	spectrum	NOUN
fcis-6025	37	22	map	map	NOUN
fcis-6025	37	23	is	be	AUX
fcis-6025	37	24	selected	select	VERB
fcis-6025	37	25	as	as	ADP
fcis-6025	37	26	the	the	DET
fcis-6025	37	27	primary	primary	ADJ
fcis-6025	37	28	feature	feature	NOUN
fcis-6025	37	29	for	for	ADP
fcis-6025	37	30	network	network	NOUN
fcis-6025	37	31	input	input	NOUN
fcis-6025	37	32	.	.	PUNCT
fcis-6025	38	1	compared	compare	VERB
fcis-6025	38	2	with	with	ADP
fcis-6025	38	3	the	the	DET
fcis-6025	38	4	combination	combination	NOUN
fcis-6025	38	5	of	of	ADP
fcis-6025	38	6	artificial	artificial	ADJ
fcis-6025	38	7	features	feature	NOUN
fcis-6025	38	8	,	,	PUNCT
fcis-6025	38	9	the	the	DET
fcis-6025	38	10	language	language	NOUN
fcis-6025	38	11	spectrogram	spectrogram	NOUN
fcis-6025	38	12	features	feature	NOUN
fcis-6025	38	13	bring	bring	VERB
fcis-6025	38	14	more	more	ADV
fcis-6025	38	15	comprehensive	comprehensive	ADJ
fcis-6025	38	16	and	and	CCONJ
fcis-6025	38	17	rich	rich	ADJ
fcis-6025	38	18	information	information	NOUN
fcis-6025	38	19	,	,	PUNCT
fcis-6025	38	20	which	which	PRON
fcis-6025	38	21	can	can	AUX
fcis-6025	38	22	be	be	AUX
fcis-6025	38	23	further	far	ADV
fcis-6025	38	24	extracted	extract	VERB
fcis-6025	38	25	by	by	ADP
fcis-6025	38	26	deep	deep	ADJ
fcis-6025	38	27	neural	neural	ADJ
fcis-6025	38	28	network	network	NOUN
fcis-6025	38	29	for	for	ADP
fcis-6025	38	30	deep	deep	ADJ
fcis-6025	38	31	features	feature	NOUN
fcis-6025	38	32	.	.	PUNCT
fcis-6025	39	1	at	at	ADP
fcis-6025	39	2	the	the	DET
fcis-6025	39	3	same	same	ADJ
fcis-6025	39	4	time	time	NOUN
fcis-6025	39	5	,	,	PUNCT
fcis-6025	39	6	the	the	DET
fcis-6025	39	7	spectrogram	spectrogram	NOUN
fcis-6025	39	8	features	feature	NOUN
fcis-6025	39	9	bring	bring	VERB
fcis-6025	39	10	redundant	redundant	ADJ
fcis-6025	39	11	information	information	NOUN
fcis-6025	39	12	with	with	ADP
fcis-6025	39	13	emotionally	emotionally	ADV
fcis-6025	39	14	irrelevant	irrelevant	ADJ
fcis-6025	39	15	information	information	NOUN
fcis-6025	39	16	due	due	ADP
fcis-6025	39	17	to	to	ADP
fcis-6025	39	18	the	the	DET
fcis-6025	39	19	comprehensive	comprehensive	ADJ
fcis-6025	39	20	speech	speech	NOUN
fcis-6025	39	21	information	information	NOUN
fcis-6025	39	22	they	they	PRON
fcis-6025	39	23	contain	contain	VERB
fcis-6025	39	24	,	,	PUNCT
fcis-6025	39	25	which	which	PRON
fcis-6025	39	26	brings	bring	VERB
fcis-6025	39	27	impact	impact	NOUN
fcis-6025	39	28	on	on	ADP
fcis-6025	39	29	the	the	DET
fcis-6025	39	30	final	final	ADJ
fcis-6025	39	31	recognition	recognition	NOUN
fcis-6025	39	32	accuracy	accuracy	NOUN
fcis-6025	39	33	.	.	PUNCT
fcis-6025	40	1	the	the	DET
fcis-6025	40	2	attention	attention	NOUN
fcis-6025	40	3	mechanism	mechanism	NOUN
fcis-6025	40	4	simulates	simulate	VERB
fcis-6025	40	5	human	human	ADJ
fcis-6025	40	6	brain	brain	NOUN
fcis-6025	40	7	vision	vision	NOUN
fcis-6025	40	8	,	,	PUNCT
fcis-6025	40	9	learns	learn	VERB
fcis-6025	40	10	the	the	DET
fcis-6025	40	11	weight	weight	NOUN
fcis-6025	40	12	distribution	distribution	NOUN
fcis-6025	40	13	from	from	ADP
fcis-6025	40	14	the	the	DET
fcis-6025	40	15	features	feature	NOUN
fcis-6025	40	16	and	and	CCONJ
fcis-6025	40	17	imposes	impose	VERB
fcis-6025	40	18	it	it	PRON
fcis-6025	40	19	on	on	ADP
fcis-6025	40	20	top	top	NOUN
fcis-6025	40	21	of	of	ADP
fcis-6025	40	22	the	the	DET
fcis-6025	40	23	original	original	ADJ
fcis-6025	40	24	features	feature	NOUN
fcis-6025	40	25	,	,	PUNCT
fcis-6025	40	26	focuses	focus	VERB
fcis-6025	40	27	on	on	ADP
fcis-6025	40	28	the	the	DET
fcis-6025	40	29	effective	effective	ADJ
fcis-6025	40	30	features	feature	NOUN
fcis-6025	40	31	and	and	CCONJ
fcis-6025	40	32	weakens	weaken	VERB
fcis-6025	40	33	the	the	DET
fcis-6025	40	34	invalid	invalid	ADJ
fcis-6025	40	35	features	feature	NOUN
fcis-6025	40	36	,	,	PUNCT
fcis-6025	40	37	which	which	PRON
fcis-6025	40	38	can	can	AUX
fcis-6025	40	39	effectively	effectively	ADV
fcis-6025	40	40	improve	improve	VERB
fcis-6025	40	41	the	the	DET
fcis-6025	40	42	feature	feature	NOUN
fcis-6025	40	43	extraction	extraction	NOUN
fcis-6025	40	44	efficiency	efficiency	NOUN
fcis-6025	40	45	.	.	PUNCT
fcis-6025	41	1	the	the	DET
fcis-6025	41	2	structure	structure	NOUN
fcis-6025	41	3	of	of	ADP
fcis-6025	41	4	the	the	DET
fcis-6025	41	5	spatial	spatial	ADJ
fcis-6025	41	6	visual	visual	ADJ
fcis-6025	41	7	attention	attention	NOUN
fcis-6025	41	8	mechanism	mechanism	NOUN
fcis-6025	41	9	is	be	AUX
fcis-6025	41	10	shown	show	VERB
fcis-6025	41	11	in	in	ADP
fcis-6025	41	12	figure	figure	NOUN
fcis-6025	41	13	3	3	NUM
fcis-6025	41	14	.	.	PUNCT
fcis-6025	42	1	the	the	DET
fcis-6025	42	2	spatial	spatial	ADJ
fcis-6025	42	3	attention	attention	NOUN
fcis-6025	42	4	module	module	NOUN
fcis-6025	42	5	first	first	ADV
fcis-6025	42	6	performs	perform	VERB
fcis-6025	42	7	the	the	DET
fcis-6025	42	8	input	input	NOUN
fcis-6025	42	9	feature	feature	NOUN
fcis-6025	42	10	map	map	NOUN
fcis-6025	42	11	by	by	ADP
fcis-6025	42	12	channel𝐹	channel𝐹	PROPN
fcis-6025	42	13	maximum	maximum	ADJ
fcis-6025	42	14	pooling	pooling	NOUN
fcis-6025	42	15	and	and	CCONJ
fcis-6025	42	16	average	average	ADJ
fcis-6025	42	17	pooling	pool	VERB
fcis-6025	42	18	operations	operation	NOUN
fcis-6025	42	19	are	be	AUX
fcis-6025	42	20	performed	perform	VERB
fcis-6025	42	21	.	.	PUNCT
fcis-6025	43	1	maximum	maximum	ADJ
fcis-6025	43	2	pooling	pooling	NOUN
fcis-6025	43	3	is	be	AUX
fcis-6025	43	4	to	to	PART
fcis-6025	43	5	keep	keep	VERB
fcis-6025	43	6	the	the	DET
fcis-6025	43	7	maximum	maximum	ADJ
fcis-6025	43	8	value	value	NOUN
fcis-6025	43	9	for	for	ADP
fcis-6025	43	10	the	the	DET
fcis-6025	43	11	pooling	pooling	NOUN
fcis-6025	43	12	window	window	NOUN
fcis-6025	43	13	.	.	PUNCT
fcis-6025	44	1	the	the	DET
fcis-6025	44	2	average	average	ADJ
fcis-6025	44	3	pooling	pooling	NOUN
fcis-6025	44	4	is	be	AUX
fcis-6025	44	5	to	to	PART
fcis-6025	44	6	keep	keep	VERB
fcis-6025	44	7	the	the	DET
fcis-6025	44	8	overall	overall	ADJ
fcis-6025	44	9	data	datum	NOUN
fcis-6025	44	10	features	feature	NOUN
fcis-6025	44	11	for	for	ADP
fcis-6025	44	12	the	the	DET
fcis-6025	44	13	pooling	pooling	NOUN
fcis-6025	44	14	window	window	NOUN
fcis-6025	44	15	and	and	CCONJ
fcis-6025	44	16	calculate	calculate	VERB
fcis-6025	44	17	the	the	DET
fcis-6025	44	18	average	average	ADJ
fcis-6025	44	19	value	value	NOUN
fcis-6025	44	20	.	.	PUNCT
fcis-6025	45	1	the	the	DET
fcis-6025	45	2	two	two	NUM
fcis-6025	45	3	pooling	pooling	NOUN
fcis-6025	45	4	results	result	NOUN
fcis-6025	45	5	are	be	AUX
fcis-6025	45	6	concatenated	concatenate	VERB
fcis-6025	45	7	by	by	ADP
fcis-6025	45	8	channel	channel	NOUN
fcis-6025	45	9	to	to	PART
fcis-6025	45	10	represent	represent	VERB
fcis-6025	45	11	the	the	DET
fcis-6025	45	12	input	input	NOUN
fcis-6025	45	13	holistically	holistically	ADV
fcis-6025	45	14	,	,	PUNCT
fcis-6025	45	15	and	and	CCONJ
fcis-6025	45	16	then	then	ADV
fcis-6025	45	17	a	a	DET
fcis-6025	45	18	1×	1×	NUM
fcis-6025	45	19	1	1	NUM
fcis-6025	45	20	convolution	convolution	NOUN
fcis-6025	45	21	layer	layer	NOUN
fcis-6025	45	22	for	for	ADP
fcis-6025	45	23	feature	feature	NOUN
fcis-6025	45	24	extraction	extraction	NOUN
fcis-6025	45	25	by	by	ADP
fcis-6025	45	26	convolution	convolution	NOUN
fcis-6025	45	27	,	,	PUNCT
fcis-6025	45	28	followed	follow	VERB
fcis-6025	45	29	by	by	ADP
fcis-6025	45	30	using	use	VERB
fcis-6025	45	31	the	the	DET
fcis-6025	45	32	𝑠𝑜𝑓𝑡𝑚𝑎𝑥	𝑠𝑜𝑓𝑡𝑚𝑎𝑥	ADJ
fcis-6025	45	33	activation	activation	NOUN
fcis-6025	45	34	function	function	NOUN
fcis-6025	45	35	to	to	PART
fcis-6025	45	36	convert	convert	VERB
fcis-6025	45	37	the	the	DET
fcis-6025	45	38	obtained	obtain	VERB
fcis-6025	45	39	single	single	ADJ
fcis-6025	45	40	-	-	PUNCT
fcis-6025	45	41	layer	layer	NOUN
fcis-6025	45	42	feature	feature	NOUN
fcis-6025	45	43	maps	map	NOUN
fcis-6025	45	44	into	into	ADP
fcis-6025	45	45	probability	probability	NOUN
fcis-6025	45	46	distributions	distribution	NOUN
fcis-6025	45	47	to	to	PART
fcis-6025	45	48	obtain	obtain	VERB
fcis-6025	45	49	the	the	DET
fcis-6025	45	50	attention	attention	NOUN
fcis-6025	45	51	mechanism	mechanism	NOUN
fcis-6025	45	52	scores	score	NOUN
fcis-6025	45	53	for	for	ADP
fcis-6025	45	54	each	each	DET
fcis-6025	45	55	element	element	NOUN
fcis-6025	45	56	of	of	ADP
fcis-6025	45	57	the	the	DET
fcis-6025	45	58	input	input	NOUN
fcis-6025	45	59	features	feature	NOUN
fcis-6025	45	60	.	.	PUNCT
fcis-6025	46	1	the	the	PRON
fcis-6025	46	2	larger	large	ADJ
fcis-6025	46	3	the	the	DET
fcis-6025	46	4	attention	attention	NOUN
fcis-6025	46	5	score	score	NOUN
fcis-6025	46	6	,	,	PUNCT
fcis-6025	46	7	the	the	PRON
fcis-6025	46	8	more	more	ADJ
fcis-6025	46	9	the	the	DET
fcis-6025	46	10	information	information	NOUN
fcis-6025	46	11	of	of	ADP
fcis-6025	46	12	the	the	DET
fcis-6025	46	13	corresponding	corresponding	ADJ
fcis-6025	46	14	element	element	NOUN
fcis-6025	46	15	is	be	AUX
fcis-6025	46	16	worthy	worthy	ADJ
fcis-6025	46	17	of	of	ADP
fcis-6025	46	18	attention	attention	NOUN
fcis-6025	46	19	in	in	ADP
fcis-6025	46	20	identifying	identify	VERB
fcis-6025	46	21	the	the	DET
fcis-6025	46	22	sentiment	sentiment	NOUN
fcis-6025	46	23	category	category	NOUN
fcis-6025	46	24	,	,	PUNCT
fcis-6025	46	25	and	and	CCONJ
fcis-6025	46	26	conversely	conversely	ADV
fcis-6025	46	27	,	,	PUNCT
fcis-6025	46	28	the	the	PRON
fcis-6025	46	29	smaller	small	ADJ
fcis-6025	46	30	the	the	DET
fcis-6025	46	31	attention	attention	NOUN
fcis-6025	46	32	score	score	NOUN
fcis-6025	46	33	,	,	PUNCT
fcis-6025	46	34	the	the	DET
fcis-6025	46	35	less	less	ADV
fcis-6025	46	36	impactful	impactful	ADJ
fcis-6025	46	37	and	and	CCONJ
fcis-6025	46	38	less	less	ADV
fcis-6025	46	39	valuable	valuable	ADJ
fcis-6025	46	40	the	the	DET
fcis-6025	46	41	information	information	NOUN
fcis-6025	46	42	of	of	ADP
fcis-6025	46	43	the	the	DET
fcis-6025	46	44	corresponding	corresponding	ADJ
fcis-6025	46	45	element	element	NOUN
fcis-6025	46	46	is	be	AUX
fcis-6025	46	47	in	in	ADP
fcis-6025	46	48	identifying	identify	VERB
fcis-6025	46	49	the	the	DET
fcis-6025	46	50	sentiment	sentiment	NOUN
fcis-6025	46	51	category	category	NOUN
fcis-6025	46	52	,	,	PUNCT
fcis-6025	46	53	and	and	CCONJ
fcis-6025	46	54	can	can	AUX
fcis-6025	46	55	be	be	AUX
fcis-6025	46	56	ignored	ignore	VERB
fcis-6025	46	57	.	.	PUNCT
fcis-6025	47	1	finally	finally	ADV
fcis-6025	47	2	,	,	PUNCT
fcis-6025	47	3	the	the	DET
fcis-6025	47	4	attention	attention	NOUN
fcis-6025	47	5	score	score	NOUN
fcis-6025	47	6	matrix	matrix	NOUN
fcis-6025	47	7	and	and	CCONJ
fcis-6025	47	8	the	the	DET
fcis-6025	47	9	input	input	NOUN
fcis-6025	47	10	values𝐹	values𝐹	PROPN
fcis-6025	47	11	a	a	DET
fcis-6025	47	12	dot	dot	NOUN
fcis-6025	47	13	product	product	NOUN
fcis-6025	47	14	operation	operation	NOUN
fcis-6025	47	15	is	be	AUX
fcis-6025	47	16	performed	perform	VERB
fcis-6025	47	17	to	to	PART
fcis-6025	47	18	adjust	adjust	VERB
fcis-6025	47	19	the	the	DET
fcis-6025	47	20	model	model	NOUN
fcis-6025	47	21	to	to	ADP
fcis-6025	47	22	the	the	DET
fcis-6025	47	23	input𝐹	input𝐹	PROPN
fcis-6025	47	24	importance	importance	NOUN
fcis-6025	47	25	of	of	ADP
fcis-6025	47	26	different	different	ADJ
fcis-6025	47	27	spatial	spatial	ADJ
fcis-6025	47	28	locations	location	NOUN
fcis-6025	47	29	.	.	PUNCT
fcis-6025	48	1	feature	feature	NOUN
fcis-6025	48	2	f	f	PROPN
fcis-6025	49	1	[	[	X
fcis-6025	49	2	maxpool	maxpool	NOUN
fcis-6025	49	3	,	,	PUNCT
fcis-6025	49	4	avgpool	avgpool	PROPN
fcis-6025	49	5	]	]	PUNCT
fcis-6025	49	6	conv	conv	ADJ
fcis-6025	49	7	layer	layer	NOUN
fcis-6025	49	8	spatial	spatial	ADJ
fcis-6025	49	9	attention	attention	NOUN
fcis-6025	49	10	refined	refine	VERB
fcis-6025	49	11	feature	feature	NOUN
fcis-6025	49	12	f	f	PROPN
fcis-6025	49	13	figure	figure	NOUN
fcis-6025	49	14	3	3	NUM
fcis-6025	49	15	.	.	PUNCT
fcis-6025	49	16	spatial	spatial	ADJ
fcis-6025	49	17	attention	attention	NOUN
fcis-6025	49	18	mechanism	mechanism	NOUN
fcis-6025	49	19	in	in	ADP
fcis-6025	49	20	this	this	DET
fcis-6025	49	21	paper	paper	NOUN
fcis-6025	49	22	,	,	PUNCT
fcis-6025	49	23	we	we	PRON
fcis-6025	49	24	combine	combine	VERB
fcis-6025	49	25	the	the	DET
fcis-6025	49	26	residual	residual	ADJ
fcis-6025	49	27	network	network	NOUN
fcis-6025	49	28	and	and	CCONJ
fcis-6025	49	29	introduce	introduce	VERB
fcis-6025	49	30	the	the	DET
fcis-6025	49	31	spatial	spatial	ADJ
fcis-6025	49	32	attention	attention	NOUN
fcis-6025	49	33	mechanism	mechanism	NOUN
fcis-6025	49	34	in	in	ADP
fcis-6025	49	35	the	the	DET
fcis-6025	49	36	residual	residual	ADJ
fcis-6025	49	37	block	block	NOUN
fcis-6025	49	38	to	to	PART
fcis-6025	49	39	build	build	VERB
fcis-6025	49	40	the	the	DET
fcis-6025	49	41	res	re	NOUN
fcis-6025	49	42	-	-	PUNCT
fcis-6025	49	43	a	a	DET
fcis-6025	49	44	module	module	NOUN
fcis-6025	49	45	,	,	PUNCT
fcis-6025	49	46	aiming	aim	VERB
fcis-6025	49	47	at	at	ADP
fcis-6025	49	48	the	the	DET
fcis-6025	49	49	training	training	NOUN
fcis-6025	49	50	process	process	NOUN
fcis-6025	49	51	when	when	SCONJ
fcis-6025	49	52	the	the	DET
fcis-6025	49	53	network	network	NOUN
fcis-6025	49	54	pays	pay	VERB
fcis-6025	49	55	more	more	ADJ
fcis-6025	49	56	attention	attention	NOUN
fcis-6025	49	57	to	to	ADP
fcis-6025	49	58	the	the	DET
fcis-6025	49	59	sentiment	sentiment	NOUN
fcis-6025	49	60	features	feature	VERB
fcis-6025	49	61	and	and	CCONJ
fcis-6025	49	62	ignores	ignore	VERB
fcis-6025	49	63	the	the	DET
fcis-6025	49	64	redundant	redundant	ADJ
fcis-6025	49	65	information	information	NOUN
fcis-6025	49	66	.	.	PUNCT
fcis-6025	50	1	the	the	DET
fcis-6025	50	2	structure	structure	NOUN
fcis-6025	50	3	of	of	ADP
fcis-6025	50	4	the	the	DET
fcis-6025	50	5	res	re	NOUN
fcis-6025	50	6	-	-	PUNCT
fcis-6025	50	7	a	a	DET
fcis-6025	50	8	module	module	NOUN
fcis-6025	50	9	is	be	AUX
fcis-6025	50	10	shown	show	VERB
fcis-6025	50	11	in	in	ADP
fcis-6025	50	12	figure	figure	NOUN
fcis-6025	50	13	4	4	NUM
fcis-6025	50	14	.	.	NOUN
fcis-6025	50	15	62	62	NUM
fcis-6025	50	16	bn	bn	NOUN
fcis-6025	50	17	,	,	PUNCT
fcis-6025	50	18	relu	relu	NOUN
fcis-6025	50	19	,	,	PUNCT
fcis-6025	50	20	conv	conv	PROPN
fcis-6025	50	21	bn	bn	PROPN
fcis-6025	50	22	,	,	PUNCT
fcis-6025	50	23	conv	conv	PROPN
fcis-6025	50	24	max	max	PROPN
fcis-6025	50	25	pooling	pool	VERB
fcis-6025	50	26	concat	concat	PROPN
fcis-6025	50	27	conv	conv	PROPN
fcis-6025	50	28	feature	feature	PROPN
fcis-6025	50	29	f	f	PROPN
fcis-6025	50	30	refined	refined	ADJ
fcis-6025	50	31	feature	feature	NOUN
fcis-6025	50	32	f	f	PROPN
fcis-6025	50	33	relu	relu	NOUN
fcis-6025	50	34	average	average	ADJ
fcis-6025	50	35	pooling	pool	VERB
fcis-6025	50	36	figure	figure	NOUN
fcis-6025	50	37	4	4	NUM
fcis-6025	50	38	.	.	PUNCT
fcis-6025	51	1	res	re	NOUN
fcis-6025	51	2	-	-	PUNCT
fcis-6025	51	3	a	a	DET
fcis-6025	51	4	module	module	NOUN
fcis-6025	51	5	in	in	ADP
fcis-6025	51	6	this	this	DET
fcis-6025	51	7	paper	paper	NOUN
fcis-6025	51	8	,	,	PUNCT
fcis-6025	51	9	based	base	VERB
fcis-6025	51	10	on	on	ADP
fcis-6025	51	11	the	the	DET
fcis-6025	51	12	res	re	NOUN
fcis-6025	51	13	-	-	PUNCT
fcis-6025	51	14	a	a	DET
fcis-6025	51	15	module	module	NOUN
fcis-6025	51	16	,	,	PUNCT
fcis-6025	51	17	we	we	PRON
fcis-6025	51	18	design	design	VERB
fcis-6025	51	19	and	and	CCONJ
fcis-6025	51	20	build	build	VERB
fcis-6025	51	21	the	the	DET
fcis-6025	51	22	att	att	PROPN
fcis-6025	51	23	-	-	PUNCT
fcis-6025	51	24	res	re	NOUN
fcis-6025	51	25	network	network	NOUN
fcis-6025	51	26	structure	structure	NOUN
fcis-6025	51	27	with	with	ADP
fcis-6025	51	28	15	15	NUM
fcis-6025	51	29	layers	layer	NOUN
fcis-6025	51	30	,	,	PUNCT
fcis-6025	51	31	as	as	SCONJ
fcis-6025	51	32	shown	show	VERB
fcis-6025	51	33	in	in	ADP
fcis-6025	51	34	figure	figure	NOUN
fcis-6025	51	35	5	5	NUM
fcis-6025	51	36	.	.	PUNCT
fcis-6025	52	1	in	in	ADP
fcis-6025	52	2	the	the	DET
fcis-6025	52	3	neural	neural	ADJ
fcis-6025	52	4	network	network	NOUN
fcis-6025	52	5	model	model	NOUN
fcis-6025	52	6	,	,	PUNCT
fcis-6025	52	7	it	it	PRON
fcis-6025	52	8	is	be	AUX
fcis-6025	52	9	experimentally	experimentally	ADV
fcis-6025	52	10	proven	prove	VERB
fcis-6025	52	11	that	that	SCONJ
fcis-6025	52	12	a	a	DET
fcis-6025	52	13	smaller	small	ADJ
fcis-6025	52	14	convolutional	convolutional	ADJ
fcis-6025	52	15	kernel	kernel	NOUN
fcis-6025	52	16	can	can	AUX
fcis-6025	52	17	be	be	AUX
fcis-6025	52	18	used	use	VERB
fcis-6025	52	19	instead	instead	ADV
fcis-6025	52	20	of	of	ADP
fcis-6025	52	21	a	a	DET
fcis-6025	52	22	larger	large	ADJ
fcis-6025	52	23	convolutional	convolutional	ADJ
fcis-6025	52	24	kernel	kernel	NOUN
fcis-6025	52	25	to	to	PART
fcis-6025	52	26	obtain	obtain	VERB
fcis-6025	52	27	better	well	ADJ
fcis-6025	52	28	training	training	NOUN
fcis-6025	52	29	results	result	NOUN
fcis-6025	52	30	,	,	PUNCT
fcis-6025	52	31	and	and	CCONJ
fcis-6025	52	32	inception	inception	NOUN
fcis-6025	52	33	network	network	NOUN
fcis-6025	52	34	structure	structure	NOUN
fcis-6025	52	35	proposes	propose	VERB
fcis-6025	52	36	to	to	PART
fcis-6025	52	37	use	use	VERB
fcis-6025	52	38	1×	1×	NUM
fcis-6025	52	39	1	1	NUM
fcis-6025	52	40	and	and	CCONJ
fcis-6025	52	41	3×	3×	NUM
fcis-6025	52	42	3	3	NUM
fcis-6025	52	43	convolutional	convolutional	ADJ
fcis-6025	52	44	kernels	kernel	NOUN
fcis-6025	52	45	can	can	AUX
fcis-6025	52	46	be	be	AUX
fcis-6025	52	47	used	use	VERB
fcis-6025	52	48	instead	instead	ADV
fcis-6025	52	49	of	of	ADP
fcis-6025	52	50	5×	5×	NOUN
fcis-6025	52	51	in	in	ADP
fcis-6025	52	52	the	the	DET
fcis-6025	52	53	inception	inception	ADJ
fcis-6025	52	54	network	network	NOUN
fcis-6025	52	55	structure	structure	NOUN
fcis-6025	52	56	,	,	PUNCT
fcis-6025	52	57	it	it	PRON
fcis-6025	52	58	is	be	AUX
fcis-6025	52	59	proposed	propose	VERB
fcis-6025	52	60	that	that	SCONJ
fcis-6025	52	61	1	1	NUM
fcis-6025	52	62	1	1	NUM
fcis-6025	52	63	and	and	CCONJ
fcis-6025	52	64	3	3	NUM
fcis-6025	52	65	3	3	NUM
fcis-6025	52	66	convolutional	convolutional	ADJ
fcis-6025	52	67	kernels	kernel	NOUN
fcis-6025	52	68	can	can	AUX
fcis-6025	52	69	replace	replace	VERB
fcis-6025	52	70	5	5	NUM
fcis-6025	52	71	convolutional	convolutional	ADJ
fcis-6025	52	72	kernels	kernel	NOUN
fcis-6025	52	73	and	and	CCONJ
fcis-6025	52	74	obtain	obtain	VERB
fcis-6025	52	75	better	well	ADJ
fcis-6025	52	76	results	result	NOUN
fcis-6025	52	77	.	.	PUNCT
fcis-6025	53	1	in	in	ADP
fcis-6025	53	2	the	the	DET
fcis-6025	53	3	development	development	NOUN
fcis-6025	53	4	of	of	ADP
fcis-6025	53	5	neural	neural	ADJ
fcis-6025	53	6	networks	network	NOUN
fcis-6025	53	7	,	,	PUNCT
fcis-6025	53	8	3	3	NUM
fcis-6025	53	9	×	×	NOUN
fcis-6025	53	10	3	3	NUM
fcis-6025	53	11	convolutional	convolutional	ADJ
fcis-6025	53	12	kernel	kernel	NOUN
fcis-6025	53	13	size	size	NOUN
fcis-6025	53	14	is	be	AUX
fcis-6025	53	15	the	the	DET
fcis-6025	53	16	most	most	ADV
fcis-6025	53	17	commonly	commonly	ADV
fcis-6025	53	18	used	use	VERB
fcis-6025	53	19	convolutional	convolutional	ADJ
fcis-6025	53	20	kernel	kernel	NOUN
fcis-6025	53	21	in	in	ADP
fcis-6025	53	22	the	the	DET
fcis-6025	53	23	development	development	NOUN
fcis-6025	53	24	of	of	ADP
fcis-6025	53	25	neural	neural	ADJ
fcis-6025	53	26	networks	network	NOUN
fcis-6025	53	27	,	,	PUNCT
fcis-6025	53	28	and	and	CCONJ
fcis-6025	53	29	it	it	PRON
fcis-6025	53	30	appears	appear	VERB
fcis-6025	53	31	many	many	ADJ
fcis-6025	53	32	times	time	NOUN
fcis-6025	53	33	in	in	ADP
fcis-6025	53	34	classical	classical	ADJ
fcis-6025	53	35	networks	network	NOUN
fcis-6025	53	36	such	such	ADJ
fcis-6025	53	37	as	as	ADP
fcis-6025	53	38	transformer	transformer	NOUN
fcis-6025	53	39	,	,	PUNCT
fcis-6025	53	40	vggnet	vggnet	NOUN
fcis-6025	53	41	,	,	PUNCT
fcis-6025	53	42	inception	inception	NOUN
fcis-6025	53	43	,	,	PUNCT
fcis-6025	53	44	etc	etc	X
fcis-6025	53	45	.	.	X
fcis-6025	54	1	the	the	DET
fcis-6025	54	2	convolutional	convolutional	ADJ
fcis-6025	54	3	kernel	kernel	NOUN
fcis-6025	54	4	size	size	NOUN
fcis-6025	54	5	of	of	ADP
fcis-6025	54	6	3	3	NUM
fcis-6025	54	7	is	be	AUX
fcis-6025	54	8	also	also	ADV
fcis-6025	54	9	chosen	choose	VERB
fcis-6025	54	10	for	for	ADP
fcis-6025	54	11	att	att	PROPN
fcis-6025	54	12	-	-	PUNCT
fcis-6025	54	13	res	res	PROPN
fcis-6025	54	14	network.×	network.×	PROPN
fcis-6025	54	15	for	for	ADP
fcis-6025	54	16	network	network	NOUN
fcis-6025	54	17	models	model	NOUN
fcis-6025	54	18	such	such	ADJ
fcis-6025	54	19	as	as	ADP
fcis-6025	54	20	resnet	resnet	NOUN
fcis-6025	54	21	for	for	ADP
fcis-6025	54	22	image	image	NOUN
fcis-6025	54	23	processing	processing	NOUN
fcis-6025	54	24	,	,	PUNCT
fcis-6025	54	25	the	the	DET
fcis-6025	54	26	data	datum	NOUN
fcis-6025	54	27	input	input	NOUN
fcis-6025	54	28	to	to	ADP
fcis-6025	54	29	the	the	DET
fcis-6025	54	30	network	network	NOUN
fcis-6025	54	31	appears	appear	VERB
fcis-6025	54	32	to	to	PART
fcis-6025	54	33	pass	pass	VERB
fcis-6025	54	34	through	through	ADP
fcis-6025	54	35	64	64	NUM
fcis-6025	54	36	or	or	CCONJ
fcis-6025	54	37	more	more	ADV
fcis-6025	54	38	convolutional	convolutional	ADJ
fcis-6025	54	39	kernels	kernel	NOUN
fcis-6025	54	40	,	,	PUNCT
fcis-6025	54	41	and	and	CCONJ
fcis-6025	54	42	the	the	DET
fcis-6025	54	43	convolutional	convolutional	ADJ
fcis-6025	54	44	kernel	kernel	NOUN
fcis-6025	54	45	size	size	NOUN
fcis-6025	54	46	is	be	AUX
fcis-6025	54	47	larger	large	ADJ
fcis-6025	54	48	,	,	PUNCT
fcis-6025	54	49	such	such	ADJ
fcis-6025	54	50	as	as	ADP
fcis-6025	54	51	7×	7×	NUM
fcis-6025	54	52	7	7	NUM
fcis-6025	54	53	or	or	CCONJ
fcis-6025	54	54	5×	5×	NUM
fcis-6025	54	55	5	5	NUM
fcis-6025	54	56	convolutional	convolutional	ADJ
fcis-6025	54	57	kernels	kernel	NOUN
fcis-6025	54	58	to	to	PART
fcis-6025	54	59	speed	speed	VERB
fcis-6025	54	60	up	up	ADP
fcis-6025	54	61	the	the	DET
fcis-6025	54	62	training	training	NOUN
fcis-6025	54	63	efficiency	efficiency	NOUN
fcis-6025	54	64	.	.	PUNCT
fcis-6025	55	1	the	the	DET
fcis-6025	55	2	size	size	NOUN
fcis-6025	55	3	of	of	ADP
fcis-6025	55	4	the	the	DET
fcis-6025	55	5	speech	speech	NOUN
fcis-6025	55	6	spectrogram	spectrogram	NOUN
fcis-6025	55	7	data	datum	NOUN
fcis-6025	55	8	varies	vary	VERB
fcis-6025	55	9	with	with	ADP
fcis-6025	55	10	the	the	DET
fcis-6025	55	11	length	length	NOUN
fcis-6025	55	12	of	of	ADP
fcis-6025	55	13	the	the	DET
fcis-6025	55	14	speech	speech	NOUN
fcis-6025	55	15	,	,	PUNCT
fcis-6025	55	16	so	so	SCONJ
fcis-6025	55	17	in	in	ADP
fcis-6025	55	18	order	order	NOUN
fcis-6025	55	19	to	to	PART
fcis-6025	55	20	standardize	standardize	VERB
fcis-6025	55	21	the	the	DET
fcis-6025	55	22	size	size	NOUN
fcis-6025	55	23	,	,	PUNCT
fcis-6025	55	24	the	the	DET
fcis-6025	55	25	size	size	NOUN
fcis-6025	55	26	is	be	AUX
fcis-6025	55	27	modified	modify	VERB
fcis-6025	55	28	to	to	ADP
fcis-6025	55	29	128	128	NUM
fcis-6025	55	30	after	after	ADP
fcis-6025	55	31	extracting	extract	VERB
fcis-6025	55	32	the	the	DET
fcis-6025	55	33	spectrogram	spectrogram	NOUN
fcis-6025	55	34	for	for	ADP
fcis-6025	55	35	each	each	DET
fcis-6025	55	36	speech	speech	NOUN
fcis-6025	55	37	signal	signal	NOUN
fcis-6025	55	38	using	use	VERB
fcis-6025	55	39	the	the	DET
fcis-6025	55	40	cv2.resize	cv2.resize	NOUN
fcis-6025	55	41	function.×	function.×	PROPN
fcis-6025	55	42	128	128	NUM
fcis-6025	55	43	and	and	CCONJ
fcis-6025	55	44	then	then	ADV
fcis-6025	55	45	input	input	VERB
fcis-6025	55	46	to	to	ADP
fcis-6025	55	47	the	the	DET
fcis-6025	55	48	network	network	NOUN
fcis-6025	55	49	for	for	ADP
fcis-6025	55	50	training	training	NOUN
fcis-6025	55	51	.	.	PUNCT
fcis-6025	56	1	the	the	DET
fcis-6025	56	2	spectrogram	spectrogram	NOUN
fcis-6025	56	3	size	size	NOUN
fcis-6025	56	4	is	be	AUX
fcis-6025	56	5	small	small	ADJ
fcis-6025	56	6	compared	compare	VERB
fcis-6025	56	7	with	with	ADP
fcis-6025	56	8	the	the	DET
fcis-6025	56	9	image	image	NOUN
fcis-6025	56	10	size	size	NOUN
fcis-6025	56	11	,	,	PUNCT
fcis-6025	56	12	so	so	CCONJ
fcis-6025	56	13	the	the	DET
fcis-6025	56	14	input	input	NOUN
fcis-6025	56	15	layer	layer	NOUN
fcis-6025	56	16	of	of	ADP
fcis-6025	56	17	the	the	DET
fcis-6025	56	18	att	att	PROPN
fcis-6025	56	19	-	-	PUNCT
fcis-6025	56	20	res	re	NOUN
fcis-6025	56	21	network	network	NOUN
fcis-6025	56	22	is	be	AUX
fcis-6025	56	23	chosen	choose	VERB
fcis-6025	56	24	to	to	PART
fcis-6025	56	25	have	have	VERB
fcis-6025	56	26	32	32	NUM
fcis-6025	56	27	3×	3×	NUM
fcis-6025	56	28	3	3	NUM
fcis-6025	56	29	convolutional	convolutional	ADJ
fcis-6025	56	30	kernels	kernel	NOUN
fcis-6025	56	31	.	.	PUNCT
fcis-6025	57	1	the	the	DET
fcis-6025	57	2	speech	speech	NOUN
fcis-6025	57	3	sequences	sequence	NOUN
fcis-6025	57	4	are	be	AUX
fcis-6025	57	5	temporal	temporal	ADJ
fcis-6025	57	6	in	in	ADP
fcis-6025	57	7	nature	nature	NOUN
fcis-6025	57	8	,	,	PUNCT
fcis-6025	57	9	and	and	CCONJ
fcis-6025	57	10	the	the	DET
fcis-6025	57	11	speech	speech	NOUN
fcis-6025	57	12	spectrogram	spectrogram	NOUN
fcis-6025	57	13	is	be	AUX
fcis-6025	57	14	also	also	ADV
fcis-6025	57	15	stitched	stitch	VERB
fcis-6025	57	16	according	accord	VERB
fcis-6025	57	17	to	to	ADP
fcis-6025	57	18	the	the	DET
fcis-6025	57	19	time	time	NOUN
fcis-6025	57	20	axis	axis	NOUN
fcis-6025	57	21	of	of	ADP
fcis-6025	57	22	the	the	DET
fcis-6025	57	23	speech	speech	NOUN
fcis-6025	57	24	frame	frame	NOUN
fcis-6025	57	25	data	datum	NOUN
fcis-6025	57	26	.	.	PUNCT
fcis-6025	58	1	therefore	therefore	ADV
fcis-6025	58	2	,	,	PUNCT
fcis-6025	58	3	the	the	DET
fcis-6025	58	4	convolution	convolution	NOUN
fcis-6025	58	5	step	step	NOUN
fcis-6025	58	6	in	in	ADP
fcis-6025	58	7	the	the	DET
fcis-6025	58	8	input	input	NOUN
fcis-6025	58	9	layer	layer	NOUN
fcis-6025	58	10	of	of	ADP
fcis-6025	58	11	the	the	DET
fcis-6025	58	12	att	att	PROPN
fcis-6025	58	13	-	-	PUNCT
fcis-6025	58	14	res	re	NOUN
fcis-6025	58	15	network	network	NOUN
fcis-6025	58	16	is	be	AUX
fcis-6025	58	17	set	set	VERB
fcis-6025	58	18	to	to	ADP
fcis-6025	58	19	1	1	NUM
fcis-6025	58	20	to	to	PART
fcis-6025	58	21	extract	extract	VERB
fcis-6025	58	22	the	the	DET
fcis-6025	58	23	continuous	continuous	ADJ
fcis-6025	58	24	shallow	shallow	ADJ
fcis-6025	58	25	links	link	NOUN
fcis-6025	58	26	.	.	PUNCT
fcis-6025	59	1	subsequent	subsequent	ADJ
fcis-6025	59	2	data	datum	NOUN
fcis-6025	59	3	pass	pass	VERB
fcis-6025	59	4	through	through	ADP
fcis-6025	59	5	a	a	DET
fcis-6025	59	6	normalization	normalization	NOUN
fcis-6025	59	7	layer	layer	NOUN
fcis-6025	59	8	,	,	PUNCT
fcis-6025	59	9	a	a	DET
fcis-6025	59	10	maximum	maximum	ADJ
fcis-6025	59	11	pooling	pool	VERB
fcis-6025	59	12	layer	layer	NOUN
fcis-6025	59	13	,	,	PUNCT
fcis-6025	59	14	a	a	DET
fcis-6025	59	15	constant	constant	ADJ
fcis-6025	59	16	link	link	NOUN
fcis-6025	59	17	block	block	NOUN
fcis-6025	59	18	,	,	PUNCT
fcis-6025	59	19	and	and	CCONJ
fcis-6025	59	20	again	again	ADV
fcis-6025	59	21	through	through	ADP
fcis-6025	59	22	data	datum	NOUN
fcis-6025	59	23	passing	pass	VERB
fcis-6025	59	24	through	through	ADP
fcis-6025	59	25	two	two	NUM
fcis-6025	59	26	block	block	NOUN
fcis-6025	59	27	-	-	PUNCT
fcis-6025	59	28	a	a	DET
fcis-6025	59	29	blocks	block	NOUN
fcis-6025	59	30	,	,	PUNCT
fcis-6025	59	31	each	each	DET
fcis-6025	59	32	block	block	NOUN
fcis-6025	59	33	-	-	PUNCT
fcis-6025	59	34	a	a	DET
fcis-6025	59	35	consisting	consisting	NOUN
fcis-6025	59	36	of	of	ADP
fcis-6025	59	37	a	a	DET
fcis-6025	59	38	convolutional	convolutional	ADJ
fcis-6025	59	39	layer	layer	NOUN
fcis-6025	59	40	,	,	PUNCT
fcis-6025	59	41	a	a	DET
fcis-6025	59	42	normalization	normalization	NOUN
fcis-6025	59	43	layer	layer	NOUN
fcis-6025	59	44	,	,	PUNCT
fcis-6025	59	45	a	a	DET
fcis-6025	59	46	maximum	maximum	ADJ
fcis-6025	59	47	pooling	pool	VERB
fcis-6025	59	48	layer	layer	NOUN
fcis-6025	59	49	,	,	PUNCT
fcis-6025	59	50	and	and	CCONJ
fcis-6025	59	51	a	a	DET
fcis-6025	59	52	res	re	NOUN
fcis-6025	59	53	-	-	PUNCT
fcis-6025	59	54	a	a	DET
fcis-6025	59	55	block	block	NOUN
fcis-6025	59	56	.	.	PUNCT
fcis-6025	60	1	one	one	NUM
fcis-6025	60	2	of	of	ADP
fcis-6025	60	3	the	the	DET
fcis-6025	60	4	convolutional	convolutional	ADJ
fcis-6025	60	5	layers	layer	NOUN
fcis-6025	60	6	contains	contain	VERB
fcis-6025	60	7	64	64	NUM
fcis-6025	60	8	or	or	CCONJ
fcis-6025	60	9	128	128	NUM
fcis-6025	60	10	3×	3×	NUM
fcis-6025	60	11	the	the	DET
fcis-6025	60	12	final	final	ADJ
fcis-6025	60	13	features	feature	NOUN
fcis-6025	60	14	are	be	AUX
fcis-6025	60	15	identified	identify	VERB
fcis-6025	60	16	by	by	ADP
fcis-6025	60	17	a	a	DET
fcis-6025	60	18	fully	fully	ADV
fcis-6025	60	19	connected	connect	VERB
fcis-6025	60	20	layer	layer	NOUN
fcis-6025	60	21	and	and	CCONJ
fcis-6025	60	22	a	a	DET
fcis-6025	60	23	softmax	softmax	NOUN
fcis-6025	60	24	classifier	classifier	NOUN
fcis-6025	60	25	for	for	ADP
fcis-6025	60	26	sentiment	sentiment	NOUN
fcis-6025	60	27	classification	classification	NOUN
fcis-6025	60	28	.	.	PUNCT
fcis-6025	61	1	the	the	DET
fcis-6025	61	2	network	network	NOUN
fcis-6025	61	3	layer	layer	NOUN
fcis-6025	61	4	activation	activation	NOUN
fcis-6025	61	5	functions	function	NOUN
fcis-6025	61	6	all	all	PRON
fcis-6025	61	7	use	use	VERB
fcis-6025	61	8	the	the	DET
fcis-6025	61	9	relu	relu	NOUN
fcis-6025	61	10	function	function	NOUN
fcis-6025	61	11	.	.	PUNCT
fcis-6025	62	1	conv，3x3，1	conv，3x3，1	NUM
fcis-6025	63	1	bn	bn	INTJ
fcis-6025	63	2	max	max	PROPN
fcis-6025	63	3	pooling	pool	VERB
fcis-6025	63	4	conv，3x3，2	conv，3x3，2	PROPN
fcis-6025	63	5	bn	bn	NUM
fcis-6025	64	1	conv，3x3，2	conv，3x3，2	NUM
fcis-6025	65	1	x2	x2	NUM
fcis-6025	66	1	bn	bn	INTJ
fcis-6025	66	2	max	max	PROPN
fcis-6025	66	3	pooling	pool	VERB
fcis-6025	66	4	conv，3x3，2	conv，3x3，2	PROPN
fcis-6025	66	5	bn	bn	CCONJ
fcis-6025	67	1	x2	x2	PROPN
fcis-6025	67	2	max	max	PROPN
fcis-6025	67	3	pooling	pool	VERB
fcis-6025	67	4	average	average	ADJ
fcis-6025	67	5	pooling	pool	VERB
fcis-6025	67	6	concat	concat	NOUN
fcis-6025	67	7	conv，3x3，1	conv，3x3，1	NUM
fcis-6025	67	8	dense	dense	ADJ
fcis-6025	67	9	global	global	ADJ
fcis-6025	67	10	average	average	ADJ
fcis-6025	67	11	pooling	pool	VERB
fcis-6025	67	12	dense	dense	ADJ
fcis-6025	67	13	softmax	softmax	NOUN
fcis-6025	67	14	output	output	NOUN
fcis-6025	67	15	x2	x2	PROPN
fcis-6025	67	16	figure	figure	NOUN
fcis-6025	67	17	5	5	NUM
fcis-6025	67	18	.	.	PUNCT
fcis-6025	67	19	att	att	PROPN
fcis-6025	67	20	-	-	PUNCT
fcis-6025	67	21	res	re	NOUN
fcis-6025	67	22	network	network	NOUN
fcis-6025	67	23	structure	structure	NOUN
fcis-6025	67	24	3	3	NUM
fcis-6025	67	25	.	.	PUNCT
fcis-6025	67	26	experimental	experimental	ADJ
fcis-6025	67	27	setup	setup	NOUN
fcis-6025	67	28	,	,	PUNCT
fcis-6025	67	29	results	result	NOUN
fcis-6025	67	30	and	and	CCONJ
fcis-6025	67	31	analysis	analysis	NOUN
fcis-6025	67	32	this	this	DET
fcis-6025	67	33	paper	paper	NOUN
fcis-6025	67	34	selects	select	NOUN
fcis-6025	67	35	casia	casia	VERB
fcis-6025	67	36	chinese	chinese	ADJ
fcis-6025	67	37	sentiment	sentiment	NOUN
fcis-6025	67	38	dataset	dataset	VERB
fcis-6025	67	39	recorded	record	VERB
fcis-6025	67	40	by	by	ADP
fcis-6025	67	41	the	the	DET
fcis-6025	67	42	institute	institute	PROPN
fcis-6025	67	43	of	of	ADP
fcis-6025	67	44	automation	automation	NOUN
fcis-6025	67	45	of	of	ADP
fcis-6025	67	46	chinese	chinese	PROPN
fcis-6025	67	47	academy	academy	PROPN
fcis-6025	67	48	of	of	ADP
fcis-6025	67	49	sciences	science	NOUN
fcis-6025	67	50	,	,	PUNCT
fcis-6025	67	51	and	and	CCONJ
fcis-6025	67	52	conducts	conduct	VERB
fcis-6025	67	53	training	training	NOUN
fcis-6025	67	54	and	and	CCONJ
fcis-6025	67	55	testing	testing	NOUN
fcis-6025	67	56	on	on	ADP
fcis-6025	67	57	this	this	DET
fcis-6025	67	58	dataset	dataset	NOUN
fcis-6025	67	59	.	.	PUNCT
fcis-6025	68	1	the	the	DET
fcis-6025	68	2	training	training	NOUN
fcis-6025	68	3	framework	framework	NOUN
fcis-6025	68	4	adopts	adopt	VERB
fcis-6025	68	5	nvidia	nvidia	PROPN
fcis-6025	68	6	gtx2060	gtx2060	NOUN
fcis-6025	68	7	graphics	graphic	NOUN
fcis-6025	68	8	card	card	NOUN
fcis-6025	68	9	,	,	PUNCT
fcis-6025	68	10	deep	deep	ADJ
fcis-6025	68	11	learning	learn	VERB
fcis-6025	68	12	pytorch	pytorch	NOUN
fcis-6025	68	13	environment	environment	NOUN
fcis-6025	68	14	,	,	PUNCT
fcis-6025	68	15	which	which	PRON
fcis-6025	68	16	supports	support	VERB
fcis-6025	68	17	python	python	NOUN
fcis-6025	68	18	,	,	PUNCT
fcis-6025	68	19	deep	deep	ADJ
fcis-6025	68	20	learning	learning	NOUN
fcis-6025	68	21	model	model	NOUN
fcis-6025	68	22	training	training	NOUN
fcis-6025	68	23	,	,	PUNCT
fcis-6025	68	24	its	its	PRON
fcis-6025	68	25	package	package	NOUN
fcis-6025	68	26	of	of	ADP
fcis-6025	68	27	libraries	library	NOUN
fcis-6025	68	28	is	be	AUX
fcis-6025	68	29	complete	complete	ADJ
fcis-6025	68	30	and	and	CCONJ
fcis-6025	68	31	complete	complete	ADJ
fcis-6025	68	32	,	,	PUNCT
fcis-6025	68	33	easy	easy	ADJ
fcis-6025	68	34	to	to	PART
fcis-6025	68	35	quickly	quickly	ADV
fcis-6025	68	36	build	build	VERB
fcis-6025	68	37	models	model	NOUN
fcis-6025	68	38	,	,	PUNCT
fcis-6025	68	39	support	support	VERB
fcis-6025	68	40	gpu	gpu	NOUN
fcis-6025	68	41	acceleration	acceleration	NOUN
fcis-6025	68	42	this	this	DET
fcis-6025	68	43	environment	environment	NOUN
fcis-6025	68	44	supports	support	VERB
fcis-6025	68	45	python	python	NOUN
fcis-6025	68	46	,	,	PUNCT
fcis-6025	68	47	deep	deep	ADJ
fcis-6025	68	48	learning	learning	NOUN
fcis-6025	68	49	model	model	NOUN
fcis-6025	68	50	training	training	NOUN
fcis-6025	68	51	,	,	PUNCT
fcis-6025	68	52	is	be	AUX
fcis-6025	68	53	complete	complete	ADJ
fcis-6025	68	54	,	,	PUNCT
fcis-6025	68	55	easy	easy	ADJ
fcis-6025	68	56	to	to	PART
fcis-6025	68	57	build	build	VERB
fcis-6025	68	58	models	model	NOUN
fcis-6025	68	59	quickly	quickly	ADV
fcis-6025	68	60	,	,	PUNCT
fcis-6025	68	61	supports	support	VERB
fcis-6025	68	62	gpu	gpu	NOUN
fcis-6025	68	63	acceleration	acceleration	NOUN
fcis-6025	68	64	,	,	PUNCT
fcis-6025	68	65	and	and	CCONJ
fcis-6025	68	66	enables	enable	VERB
fcis-6025	68	67	fast	fast	ADJ
fcis-6025	68	68	scientific	scientific	ADJ
fcis-6025	68	69	computing	computing	NOUN
fcis-6025	68	70	,	,	PUNCT
fcis-6025	68	71	which	which	PRON
fcis-6025	68	72	provides	provide	VERB
fcis-6025	68	73	the	the	DET
fcis-6025	68	74	conditions	condition	NOUN
fcis-6025	68	75	for	for	ADP
fcis-6025	68	76	smooth	smooth	ADJ
fcis-6025	68	77	experiments	experiment	NOUN
fcis-6025	68	78	.	.	PUNCT
fcis-6025	69	1	3.1	3.1	NUM
fcis-6025	69	2	.	.	PUNCT
fcis-6025	69	3	experimental	experimental	ADJ
fcis-6025	69	4	setup	setup	NOUN
fcis-6025	69	5	in	in	ADP
fcis-6025	69	6	this	this	DET
fcis-6025	69	7	paper	paper	NOUN
fcis-6025	69	8	,	,	PUNCT
fcis-6025	69	9	the	the	DET
fcis-6025	69	10	casia	casia	NOUN
fcis-6025	69	11	dataset	dataset	NOUN
fcis-6025	69	12	is	be	AUX
fcis-6025	69	13	used	use	VERB
fcis-6025	69	14	for	for	ADP
fcis-6025	69	15	the	the	DET
fcis-6025	69	16	experiments	experiment	NOUN
fcis-6025	69	17	.	.	PUNCT
fcis-6025	70	1	the	the	DET
fcis-6025	70	2	experiments	experiment	NOUN
fcis-6025	70	3	are	be	AUX
fcis-6025	70	4	mainly	mainly	ADV
fcis-6025	70	5	set	set	VERB
fcis-6025	70	6	up	up	ADP
fcis-6025	70	7	as	as	ADP
fcis-6025	70	8	a	a	DET
fcis-6025	70	9	set	set	NOUN
fcis-6025	70	10	of	of	ADP
fcis-6025	70	11	ablation	ablation	NOUN
fcis-6025	70	12	experiments	experiment	NOUN
fcis-6025	70	13	to	to	PART
fcis-6025	70	14	verify	verify	VERB
fcis-6025	70	15	the	the	DET
fcis-6025	70	16	effectiveness	effectiveness	NOUN
fcis-6025	70	17	of	of	ADP
fcis-6025	70	18	the	the	DET
fcis-6025	70	19	proposed	propose	VERB
fcis-6025	70	20	module	module	NOUN
fcis-6025	70	21	and	and	CCONJ
fcis-6025	70	22	a	a	DET
fcis-6025	70	23	set	set	NOUN
fcis-6025	70	24	of	of	ADP
fcis-6025	70	25	comparison	comparison	NOUN
fcis-6025	70	26	experiments	experiment	NOUN
fcis-6025	70	27	to	to	PART
fcis-6025	70	28	verify	verify	VERB
fcis-6025	70	29	the	the	DET
fcis-6025	70	30	effectiveness	effectiveness	NOUN
fcis-6025	70	31	of	of	ADP
fcis-6025	70	32	the	the	DET
fcis-6025	70	33	proposed	propose	VERB
fcis-6025	70	34	network	network	NOUN
fcis-6025	70	35	model	model	NOUN
fcis-6025	70	36	.	.	PUNCT
fcis-6025	71	1	block	block	NOUN
fcis-6025	71	2	-	-	PUNCT
fcis-6025	71	3	a	a	NOUN
fcis-6025	71	4	is	be	AUX
fcis-6025	71	5	composed	compose	VERB
fcis-6025	71	6	of	of	ADP
fcis-6025	71	7	one	one	NUM
fcis-6025	71	8	layer	layer	NOUN
fcis-6025	71	9	of	of	ADP
fcis-6025	71	10	convolution	convolution	NOUN
fcis-6025	71	11	,	,	PUNCT
fcis-6025	71	12	one	one	NUM
fcis-6025	71	13	layer	layer	NOUN
fcis-6025	71	14	of	of	ADP
fcis-6025	71	15	normalization	normalization	NOUN
fcis-6025	71	16	,	,	PUNCT
fcis-6025	71	17	one	one	NUM
fcis-6025	71	18	layer	layer	NOUN
fcis-6025	71	19	of	of	ADP
fcis-6025	71	20	maximum	maximum	ADJ
fcis-6025	71	21	pooling	pooling	NOUN
fcis-6025	71	22	and	and	CCONJ
fcis-6025	71	23	one	one	NUM
fcis-6025	71	24	module	module	NOUN
fcis-6025	71	25	of	of	ADP
fcis-6025	71	26	res	re	NOUN
fcis-6025	71	27	-	-	PUNCT
fcis-6025	71	28	a.	a.	NOUN
fcis-6025	71	29	if	if	SCONJ
fcis-6025	71	30	block	block	NOUN
fcis-6025	71	31	-	-	PUNCT
fcis-6025	71	32	a	a	NOUN
fcis-6025	71	33	is	be	AUX
fcis-6025	71	34	in	in	ADP
fcis-6025	71	35	the	the	DET
fcis-6025	71	36	input	input	NOUN
fcis-6025	71	37	layer	layer	NOUN
fcis-6025	71	38	of	of	ADP
fcis-6025	71	39	the	the	DET
fcis-6025	71	40	network	network	NOUN
fcis-6025	71	41	,	,	PUNCT
fcis-6025	71	42	the	the	DET
fcis-6025	71	43	step	step	NOUN
fcis-6025	71	44	size	size	NOUN
fcis-6025	71	45	of	of	ADP
fcis-6025	71	46	the	the	DET
fcis-6025	71	47	convolution	convolution	NOUN
fcis-6025	71	48	layer	layer	NOUN
fcis-6025	71	49	is	be	AUX
fcis-6025	71	50	1	1	NUM
fcis-6025	71	51	,	,	PUNCT
fcis-6025	71	52	otherwise	otherwise	ADV
fcis-6025	71	53	the	the	DET
fcis-6025	71	54	step	step	NOUN
fcis-6025	71	55	size	size	NOUN
fcis-6025	71	56	is	be	AUX
fcis-6025	71	57	set	set	VERB
fcis-6025	71	58	to	to	ADP
fcis-6025	71	59	2	2	NUM
fcis-6025	71	60	to	to	PART
fcis-6025	71	61	speed	speed	VERB
fcis-6025	71	62	up	up	ADP
fcis-6025	71	63	the	the	DET
fcis-6025	71	64	training	training	NOUN
fcis-6025	71	65	.	.	PUNCT
fcis-6025	72	1	conv	conv	PROPN
fcis-6025	72	2	bn	bn	PROPN
fcis-6025	72	3	max	max	PROPN
fcis-6025	72	4	pooling	pool	VERB
fcis-6025	72	5	conv	conv	PROPN
fcis-6025	72	6	bn	bn	PROPN
fcis-6025	72	7	max	max	PROPN
fcis-6025	72	8	pooling	pooling	NOUN
fcis-6025	72	9	(	(	PUNCT
fcis-6025	72	10	a	a	NOUN
fcis-6025	72	11	)	)	PUNCT
fcis-6025	72	12	block	block	NOUN
fcis-6025	72	13	(	(	PUNCT
fcis-6025	72	14	b	b	NOUN
fcis-6025	72	15	)	)	PUNCT
fcis-6025	72	16	block	block	NOUN
fcis-6025	72	17	-	-	PUNCT
fcis-6025	72	18	a	a	DET
fcis-6025	72	19	identity	identity	NOUN
fcis-6025	72	20	block	block	NOUN
fcis-6025	72	21	res	re	NOUN
fcis-6025	72	22	-	-	PUNCT
fcis-6025	72	23	a	a	DET
fcis-6025	72	24	figure	figure	NOUN
fcis-6025	72	25	6	6	NUM
fcis-6025	72	26	.	.	PUNCT
fcis-6025	72	27	block	block	NOUN
fcis-6025	72	28	and	and	CCONJ
fcis-6025	72	29	block	block	NOUN
fcis-6025	72	30	-	-	PUNCT
fcis-6025	72	31	a	a	NOUN
fcis-6025	72	32	in	in	ADP
fcis-6025	72	33	order	order	NOUN
fcis-6025	72	34	to	to	PART
fcis-6025	72	35	better	well	ADV
fcis-6025	72	36	analyze	analyze	VERB
fcis-6025	72	37	the	the	DET
fcis-6025	72	38	impact	impact	NOUN
fcis-6025	72	39	of	of	ADP
fcis-6025	72	40	res	re	NOUN
fcis-6025	72	41	-	-	PUNCT
fcis-6025	72	42	a	a	DET
fcis-6025	72	43	module	module	NOUN
fcis-6025	72	44	on	on	ADP
fcis-6025	72	45	the	the	DET
fcis-6025	72	46	performance	performance	NOUN
fcis-6025	72	47	of	of	ADP
fcis-6025	72	48	the	the	DET
fcis-6025	72	49	overall	overall	ADJ
fcis-6025	72	50	network	network	NOUN
fcis-6025	72	51	model	model	NOUN
fcis-6025	72	52	,	,	PUNCT
fcis-6025	72	53	this	this	DET
fcis-6025	72	54	paper	paper	NOUN
fcis-6025	72	55	uses	use	VERB
fcis-6025	72	56	the	the	DET
fcis-6025	72	57	att	att	PROPN
fcis-6025	72	58	-	-	PUNCT
fcis-6025	72	59	res0	res0	PROPN
fcis-6025	72	60	network	network	NOUN
fcis-6025	72	61	constructed	construct	VERB
fcis-6025	72	62	by	by	ADP
fcis-6025	72	63	three	three	NUM
fcis-6025	72	64	blocks	block	NOUN
fcis-6025	72	65	as	as	SCONJ
fcis-6025	72	66	the	the	DET
fcis-6025	72	67	63	63	NUM
fcis-6025	72	68	backbone	backbone	NOUN
fcis-6025	72	69	network	network	NOUN
fcis-6025	72	70	to	to	PART
fcis-6025	72	71	extract	extract	VERB
fcis-6025	72	72	features	feature	NOUN
fcis-6025	72	73	,	,	PUNCT
fcis-6025	72	74	and	and	CCONJ
fcis-6025	72	75	gradually	gradually	ADV
fcis-6025	72	76	replaces	replace	VERB
fcis-6025	72	77	blocks	block	NOUN
fcis-6025	72	78	with	with	ADP
fcis-6025	72	79	block	block	NOUN
fcis-6025	72	80	-	-	PUNCT
fcis-6025	72	81	a	a	NOUN
fcis-6025	72	82	to	to	PART
fcis-6025	72	83	constitute	constitute	VERB
fcis-6025	72	84	four	four	NUM
fcis-6025	72	85	network	network	NOUN
fcis-6025	72	86	structures	structure	NOUN
fcis-6025	72	87	,	,	PUNCT
fcis-6025	72	88	att	att	PROPN
fcis-6025	72	89	-	-	PUNCT
fcis-6025	72	90	res1	res1	PROPN
fcis-6025	72	91	,	,	PUNCT
fcis-6025	72	92	att	att	PROPN
fcis-6025	72	93	-	-	PUNCT
fcis-6025	72	94	res2	res2	PROPN
fcis-6025	72	95	,	,	PUNCT
fcis-6025	72	96	att	att	PROPN
fcis-6025	72	97	-	-	PUNCT
fcis-6025	72	98	res	res	PROPN
fcis-6025	72	99	and	and	CCONJ
fcis-6025	72	100	att	att	PROPN
fcis-6025	72	101	-	-	PUNCT
fcis-6025	72	102	res3	res3	PROPN
fcis-6025	72	103	,	,	PUNCT
fcis-6025	72	104	and	and	CCONJ
fcis-6025	72	105	the	the	DET
fcis-6025	72	106	five	five	NUM
fcis-6025	72	107	network	network	NOUN
fcis-6025	72	108	structures	structure	NOUN
fcis-6025	72	109	are	be	AUX
fcis-6025	72	110	shown	show	VERB
fcis-6025	72	111	in	in	ADP
fcis-6025	72	112	figure	figure	NOUN
fcis-6025	72	113	7	7	NUM
fcis-6025	72	114	shows	show	NOUN
fcis-6025	72	115	.	.	PUNCT
fcis-6025	73	1	block	block	NOUN
fcis-6025	73	2	block	block	NOUN
fcis-6025	73	3	block	block	NOUN
fcis-6025	73	4	block	block	NOUN
fcis-6025	73	5	block	block	NOUN
fcis-6025	73	6	block	block	NOUN
fcis-6025	73	7	-	-	PUNCT
fcis-6025	73	8	a	a	DET
fcis-6025	73	9	block	block	NOUN
fcis-6025	73	10	-	-	PUNCT
fcis-6025	73	11	a	a	DET
fcis-6025	73	12	block	block	NOUN
fcis-6025	73	13	-	-	PUNCT
fcis-6025	73	14	a	a	DET
fcis-6025	73	15	block	block	NOUN
fcis-6025	73	16	block	block	NOUN
fcis-6025	73	17	-	-	PUNCT
fcis-6025	73	18	a	a	DET
fcis-6025	73	19	block	block	NOUN
fcis-6025	73	20	block	block	NOUN
fcis-6025	73	21	-	-	PUNCT
fcis-6025	73	22	a	a	DET
fcis-6025	73	23	block	block	NOUN
fcis-6025	73	24	-	-	PUNCT
fcis-6025	73	25	a	a	DET
fcis-6025	73	26	block	block	NOUN
fcis-6025	73	27	-	-	PUNCT
fcis-6025	73	28	a	a	DET
fcis-6025	73	29	block	block	NOUN
fcis-6025	73	30	-	-	PUNCT
fcis-6025	73	31	a	a	NOUN
fcis-6025	73	32	(	(	PUNCT
fcis-6025	73	33	a	a	PRON
fcis-6025	73	34	)	)	PUNCT
fcis-6025	73	35	att	att	PROPN
fcis-6025	73	36	-	-	PUNCT
fcis-6025	73	37	res0	res0	PROPN
fcis-6025	73	38	(	(	PUNCT
fcis-6025	73	39	b	b	NOUN
fcis-6025	73	40	)	)	PUNCT
fcis-6025	73	41	att	att	PROPN
fcis-6025	73	42	-	-	PROPN
fcis-6025	73	43	res1	res1	PROPN
fcis-6025	73	44	(	(	PUNCT
fcis-6025	73	45	c	c	NOUN
fcis-6025	73	46	)	)	PUNCT
fcis-6025	73	47	att	att	PROPN
fcis-6025	73	48	-	-	PUNCT
fcis-6025	73	49	res2	res2	PROPN
fcis-6025	73	50	(	(	PUNCT
fcis-6025	73	51	d	d	NOUN
fcis-6025	73	52	)	)	PUNCT
fcis-6025	73	53	att	att	PROPN
fcis-6025	73	54	-	-	PUNCT
fcis-6025	73	55	res	res	PROPN
fcis-6025	73	56	(	(	PUNCT
fcis-6025	73	57	c	c	NOUN
fcis-6025	73	58	)	)	PUNCT
fcis-6025	73	59	att	att	PROPN
fcis-6025	73	60	-	-	PUNCT
fcis-6025	73	61	res3	res3	PROPN
fcis-6025	73	62	figure	figure	NOUN
fcis-6025	73	63	7	7	NUM
fcis-6025	73	64	.	.	PUNCT
fcis-6025	73	65	att	att	PROPN
fcis-6025	73	66	-	-	PUNCT
fcis-6025	73	67	res0	res0	PROPN
fcis-6025	73	68	,	,	PUNCT
fcis-6025	73	69	att	att	PROPN
fcis-6025	73	70	-	-	PUNCT
fcis-6025	73	71	res1	res1	PROPN
fcis-6025	73	72	,	,	PUNCT
fcis-6025	73	73	att	att	PROPN
fcis-6025	73	74	-	-	PUNCT
fcis-6025	73	75	res2	res2	PROPN
fcis-6025	73	76	,	,	PUNCT
fcis-6025	73	77	att	att	PROPN
fcis-6025	73	78	-	-	PUNCT
fcis-6025	73	79	res	res	PROPN
fcis-6025	73	80	and	and	CCONJ
fcis-6025	73	81	att	att	PROPN
fcis-6025	73	82	-	-	PUNCT
fcis-6025	73	83	res3	res3	PROPN
fcis-6025	73	84	network	network	NOUN
fcis-6025	73	85	structure	structure	NOUN
fcis-6025	73	86	to	to	PART
fcis-6025	73	87	discuss	discuss	VERB
fcis-6025	73	88	the	the	DET
fcis-6025	73	89	effects	effect	NOUN
fcis-6025	73	90	of	of	ADP
fcis-6025	73	91	the	the	DET
fcis-6025	73	92	two	two	NUM
fcis-6025	73	93	approaches	approach	NOUN
fcis-6025	73	94	,	,	PUNCT
fcis-6025	73	95	res	re	NOUN
fcis-6025	73	96	-	-	PUNCT
fcis-6025	73	97	a	a	DET
fcis-6025	73	98	module	module	NOUN
fcis-6025	73	99	and	and	CCONJ
fcis-6025	73	100	adding	add	VERB
fcis-6025	73	101	spatial	spatial	ADJ
fcis-6025	73	102	attention	attention	NOUN
fcis-6025	73	103	mechanism	mechanism	NOUN
fcis-6025	73	104	directly	directly	ADV
fcis-6025	73	105	after	after	ADP
fcis-6025	73	106	feature	feature	NOUN
fcis-6025	73	107	extraction	extraction	NOUN
fcis-6025	73	108	,	,	PUNCT
fcis-6025	73	109	on	on	ADP
fcis-6025	73	110	the	the	DET
fcis-6025	73	111	network	network	NOUN
fcis-6025	73	112	model	model	NOUN
fcis-6025	73	113	,	,	PUNCT
fcis-6025	73	114	this	this	DET
fcis-6025	73	115	paper	paper	NOUN
fcis-6025	73	116	constructs	construct	VERB
fcis-6025	73	117	an	an	DET
fcis-6025	73	118	attres0	attres0	PROPN
fcis-6025	73	119	-	-	PUNCT
fcis-6025	73	120	att	att	PROPN
fcis-6025	73	121	network	network	NOUN
fcis-6025	73	122	based	base	VERB
fcis-6025	73	123	on	on	ADP
fcis-6025	73	124	the	the	DET
fcis-6025	73	125	att	att	PROPN
fcis-6025	73	126	-	-	PUNCT
fcis-6025	73	127	res0	res0	PROPN
fcis-6025	73	128	network	network	NOUN
fcis-6025	73	129	,	,	PUNCT
fcis-6025	73	130	aiming	aim	VERB
fcis-6025	73	131	to	to	PART
fcis-6025	73	132	add	add	VERB
fcis-6025	73	133	the	the	DET
fcis-6025	73	134	spatial	spatial	ADJ
fcis-6025	73	135	attention	attention	NOUN
fcis-6025	73	136	mechanism	mechanism	NOUN
fcis-6025	73	137	before	before	SCONJ
fcis-6025	73	138	the	the	DET
fcis-6025	73	139	global	global	ADJ
fcis-6025	73	140	averaging	averaging	NOUN
fcis-6025	73	141	layer	layer	NOUN
fcis-6025	73	142	to	to	PART
fcis-6025	73	143	form	form	VERB
fcis-6025	73	144	a	a	DET
fcis-6025	73	145	comparison	comparison	NOUN
fcis-6025	73	146	experiment	experiment	NOUN
fcis-6025	73	147	with	with	ADP
fcis-6025	73	148	the	the	DET
fcis-6025	73	149	attres1	attres1	PROPN
fcis-6025	73	150	network	network	NOUN
fcis-6025	73	151	model	model	PROPN
fcis-6025	73	152	.	.	PUNCT
fcis-6025	74	1	att	att	PROPN
fcis-6025	74	2	-	-	PUNCT
fcis-6025	74	3	res0	res0	PROPN
fcis-6025	74	4	-	-	PUNCT
fcis-6025	74	5	att	att	PROPN
fcis-6025	74	6	network	network	NOUN
fcis-6025	74	7	is	be	AUX
fcis-6025	74	8	shown	show	VERB
fcis-6025	74	9	in	in	ADP
fcis-6025	74	10	figure	figure	NOUN
fcis-6025	74	11	33	33	NUM
fcis-6025	74	12	shows	show	NOUN
fcis-6025	74	13	.	.	PUNCT
fcis-6025	75	1	in	in	ADP
fcis-6025	75	2	this	this	DET
fcis-6025	75	3	paper	paper	NOUN
fcis-6025	75	4	,	,	PUNCT
fcis-6025	75	5	five	five	NUM
fcis-6025	75	6	network	network	NOUN
fcis-6025	75	7	structures	structure	NOUN
fcis-6025	75	8	are	be	AUX
fcis-6025	75	9	ablated	ablate	VERB
fcis-6025	75	10	with	with	ADP
fcis-6025	75	11	the	the	DET
fcis-6025	75	12	baseline	baseline	NOUN
fcis-6025	75	13	network	network	NOUN
fcis-6025	75	14	to	to	PART
fcis-6025	75	15	analyze	analyze	VERB
fcis-6025	75	16	the	the	DET
fcis-6025	75	17	effect	effect	NOUN
fcis-6025	75	18	of	of	ADP
fcis-6025	75	19	block	block	NOUN
fcis-6025	75	20	-	-	PUNCT
fcis-6025	75	21	a	a	DET
fcis-6025	75	22	modules	module	NOUN
fcis-6025	75	23	on	on	ADP
fcis-6025	75	24	the	the	DET
fcis-6025	75	25	accuracy	accuracy	NOUN
fcis-6025	75	26	of	of	ADP
fcis-6025	75	27	emotion	emotion	NOUN
fcis-6025	75	28	recognition	recognition	NOUN
fcis-6025	75	29	and	and	CCONJ
fcis-6025	75	30	the	the	DET
fcis-6025	75	31	effect	effect	NOUN
fcis-6025	75	32	of	of	ADP
fcis-6025	75	33	the	the	DET
fcis-6025	75	34	number	number	NOUN
fcis-6025	75	35	and	and	CCONJ
fcis-6025	75	36	position	position	NOUN
fcis-6025	75	37	of	of	ADP
fcis-6025	75	38	block	block	NOUN
fcis-6025	75	39	-	-	PUNCT
fcis-6025	75	40	a	a	DET
fcis-6025	75	41	modules	module	NOUN
fcis-6025	75	42	on	on	ADP
fcis-6025	75	43	the	the	DET
fcis-6025	75	44	accuracy	accuracy	NOUN
fcis-6025	75	45	of	of	ADP
fcis-6025	75	46	emotion	emotion	NOUN
fcis-6025	75	47	recognition	recognition	NOUN
fcis-6025	75	48	.	.	PUNCT
fcis-6025	76	1	conv，3x3，1	conv，3x3，1	NUM
fcis-6025	77	1	bn	bn	INTJ
fcis-6025	77	2	max	max	PROPN
fcis-6025	77	3	pooling	pool	VERB
fcis-6025	77	4	conv，3x3，2	conv，3x3，2	PROPN
fcis-6025	77	5	bn	bn	CCONJ
fcis-6025	78	1	x2	x2	PROPN
fcis-6025	78	2	max	max	PROPN
fcis-6025	78	3	pooling	pool	VERB
fcis-6025	78	4	average	average	ADJ
fcis-6025	78	5	pooling	pool	VERB
fcis-6025	78	6	concat	concat	NOUN
fcis-6025	78	7	conv，3x3，1	conv，3x3，1	NUM
fcis-6025	78	8	x3	x3	ADJ
fcis-6025	78	9	block	block	VERB
fcis-6025	78	10	dense	dense	ADJ
fcis-6025	78	11	global	global	ADJ
fcis-6025	78	12	average	average	ADJ
fcis-6025	78	13	pooling	pool	VERB
fcis-6025	78	14	dense	dense	ADJ
fcis-6025	78	15	softmax	softmax	NOUN
fcis-6025	78	16	output	output	NOUN
fcis-6025	78	17	input	input	NOUN
fcis-6025	78	18	figure	figure	NOUN
fcis-6025	78	19	8	8	NUM
fcis-6025	78	20	.	.	PUNCT
fcis-6025	79	1	att	att	PROPN
fcis-6025	79	2	-	-	PUNCT
fcis-6025	79	3	res0	res0	PROPN
fcis-6025	79	4	-	-	PUNCT
fcis-6025	79	5	att	att	PROPN
fcis-6025	79	6	network	network	NOUN
fcis-6025	79	7	structure	structure	NOUN
fcis-6025	79	8	in	in	ADP
fcis-6025	79	9	order	order	NOUN
fcis-6025	79	10	to	to	PART
fcis-6025	79	11	better	well	ADV
fcis-6025	79	12	analyze	analyze	VERB
fcis-6025	79	13	the	the	DET
fcis-6025	79	14	effectiveness	effectiveness	NOUN
fcis-6025	79	15	of	of	ADP
fcis-6025	79	16	the	the	DET
fcis-6025	79	17	network	network	NOUN
fcis-6025	79	18	model	model	NOUN
fcis-6025	79	19	on	on	ADP
fcis-6025	79	20	the	the	DET
fcis-6025	79	21	speech	speech	NOUN
fcis-6025	79	22	emotion	emotion	NOUN
fcis-6025	79	23	recognition	recognition	NOUN
fcis-6025	79	24	task	task	NOUN
fcis-6025	79	25	,	,	PUNCT
fcis-6025	79	26	this	this	DET
fcis-6025	79	27	paper	paper	NOUN
fcis-6025	79	28	further	far	ADV
fcis-6025	79	29	analyzes	analyze	VERB
fcis-6025	79	30	the	the	DET
fcis-6025	79	31	effectiveness	effectiveness	NOUN
fcis-6025	79	32	of	of	ADP
fcis-6025	79	33	att	att	PROPN
fcis-6025	79	34	-	-	PUNCT
fcis-6025	79	35	res	re	NOUN
fcis-6025	79	36	network	network	NOUN
fcis-6025	79	37	for	for	ADP
fcis-6025	79	38	small	small	ADJ
fcis-6025	79	39	sample	sample	NOUN
fcis-6025	79	40	english	english	ADJ
fcis-6025	79	41	speech	speech	NOUN
fcis-6025	79	42	emotion	emotion	NOUN
fcis-6025	79	43	recognition	recognition	NOUN
fcis-6025	79	44	by	by	ADP
fcis-6025	79	45	comparing	compare	VERB
fcis-6025	79	46	the	the	DET
fcis-6025	79	47	experimental	experimental	ADJ
fcis-6025	79	48	results	result	NOUN
fcis-6025	79	49	with	with	ADP
fcis-6025	79	50	resnet-18	resnet-18	PROPN
fcis-6025	79	51	,	,	PUNCT
fcis-6025	79	52	svm	svm	ADJ
fcis-6025	79	53	,	,	PUNCT
fcis-6025	79	54	alexnet	alexnet	ADJ
fcis-6025	79	55	and	and	CCONJ
fcis-6025	79	56	other	other	ADJ
fcis-6025	79	57	network	network	NOUN
fcis-6025	79	58	models	model	NOUN
fcis-6025	79	59	.	.	PUNCT
fcis-6025	80	1	in	in	ADP
fcis-6025	80	2	the	the	DET
fcis-6025	80	3	experiments	experiment	NOUN
fcis-6025	80	4	,	,	PUNCT
fcis-6025	80	5	the	the	DET
fcis-6025	80	6	data	datum	NOUN
fcis-6025	80	7	set	set	VERB
fcis-6025	80	8	is	be	AUX
fcis-6025	80	9	randomly	randomly	ADV
fcis-6025	80	10	divided	divide	VERB
fcis-6025	80	11	according	accord	VERB
fcis-6025	80	12	to	to	ADP
fcis-6025	80	13	the	the	DET
fcis-6025	80	14	ratio	ratio	NOUN
fcis-6025	80	15	of	of	ADP
fcis-6025	80	16	80	80	NUM
fcis-6025	80	17	%	%	NOUN
fcis-6025	80	18	being	be	AUX
fcis-6025	80	19	the	the	DET
fcis-6025	80	20	training	training	NOUN
fcis-6025	80	21	set	set	VERB
fcis-6025	80	22	and	and	CCONJ
fcis-6025	80	23	20	20	NUM
fcis-6025	80	24	%	%	NOUN
fcis-6025	80	25	being	be	AUX
fcis-6025	80	26	the	the	DET
fcis-6025	80	27	test	test	NOUN
fcis-6025	80	28	set	set	VERB
fcis-6025	80	29	.	.	PUNCT
fcis-6025	81	1	the	the	DET
fcis-6025	81	2	adam	adam	PROPN
fcis-6025	81	3	optimizer	optimizer	NOUN
fcis-6025	81	4	was	be	AUX
fcis-6025	81	5	chosen	choose	VERB
fcis-6025	81	6	to	to	PART
fcis-6025	81	7	continuously	continuously	ADV
fcis-6025	81	8	optimize	optimize	VERB
fcis-6025	81	9	the	the	DET
fcis-6025	81	10	learning	learning	NOUN
fcis-6025	81	11	network	network	NOUN
fcis-6025	81	12	parameters	parameter	NOUN
fcis-6025	81	13	,	,	PUNCT
fcis-6025	81	14	and	and	CCONJ
fcis-6025	81	15	a	a	DET
fcis-6025	81	16	total	total	NOUN
fcis-6025	81	17	of	of	ADP
fcis-6025	81	18	25	25	NUM
fcis-6025	81	19	iterations	iteration	NOUN
fcis-6025	81	20	were	be	AUX
fcis-6025	81	21	trained	train	VERB
fcis-6025	81	22	.	.	PUNCT
fcis-6025	82	1	3.2	3.2	NUM
fcis-6025	82	2	.	.	PUNCT
fcis-6025	82	3	results	result	NOUN
fcis-6025	82	4	and	and	CCONJ
fcis-6025	82	5	analysis	analysis	NOUN
fcis-6025	82	6	the	the	DET
fcis-6025	82	7	test	test	NOUN
fcis-6025	82	8	set	set	VERB
fcis-6025	82	9	accuracies	accuracy	NOUN
fcis-6025	82	10	for	for	ADP
fcis-6025	82	11	att	att	PROPN
fcis-6025	82	12	-	-	PUNCT
fcis-6025	82	13	res0	res0	PROPN
fcis-6025	82	14	,	,	PUNCT
fcis-6025	82	15	att	att	PROPN
fcis-6025	82	16	-	-	PUNCT
fcis-6025	82	17	res0+att	res0+att	PROPN
fcis-6025	82	18	,	,	PUNCT
fcis-6025	82	19	att	att	PROPN
fcis-6025	82	20	-	-	PUNCT
fcis-6025	82	21	res1	res1	PROPN
fcis-6025	82	22	,	,	PUNCT
fcis-6025	82	23	att	att	PROPN
fcis-6025	82	24	-	-	PUNCT
fcis-6025	82	25	res2	res2	PROPN
fcis-6025	82	26	,	,	PUNCT
fcis-6025	82	27	att	att	PROPN
fcis-6025	82	28	-	-	PUNCT
fcis-6025	82	29	res	res	PROPN
fcis-6025	82	30	and	and	CCONJ
fcis-6025	82	31	att	att	PROPN
fcis-6025	82	32	-	-	PUNCT
fcis-6025	82	33	res3	res3	PROPN
fcis-6025	82	34	are	be	AUX
fcis-6025	82	35	shown	show	VERB
fcis-6025	82	36	in	in	ADP
fcis-6025	82	37	table	table	NOUN
fcis-6025	82	38	1	1	NUM
fcis-6025	82	39	.	.	PUNCT
fcis-6025	82	40	comparing	compare	VERB
fcis-6025	82	41	the	the	DET
fcis-6025	82	42	experimental	experimental	ADJ
fcis-6025	82	43	results	result	NOUN
fcis-6025	82	44	in	in	ADP
fcis-6025	82	45	table	table	NOUN
fcis-6025	82	46	1	1	NUM
fcis-6025	82	47	,	,	PUNCT
fcis-6025	82	48	we	we	PRON
fcis-6025	82	49	can	can	AUX
fcis-6025	82	50	find	find	VERB
fcis-6025	82	51	that	that	SCONJ
fcis-6025	82	52	the	the	DET
fcis-6025	82	53	improvement	improvement	NOUN
fcis-6025	82	54	effect	effect	NOUN
fcis-6025	82	55	of	of	ADP
fcis-6025	82	56	the	the	DET
fcis-6025	82	57	att	att	PROPN
fcis-6025	82	58	-	-	PUNCT
fcis-6025	82	59	res0+att	res0+att	PROPN
fcis-6025	82	60	model	model	NOUN
fcis-6025	82	61	on	on	ADP
fcis-6025	82	62	recognition	recognition	NOUN
fcis-6025	82	63	accuracy	accuracy	NOUN
fcis-6025	82	64	is	be	AUX
fcis-6025	82	65	significantly	significantly	ADV
fcis-6025	82	66	lower	low	ADJ
fcis-6025	82	67	than	than	ADP
fcis-6025	82	68	that	that	PRON
fcis-6025	82	69	of	of	ADP
fcis-6025	82	70	the	the	DET
fcis-6025	82	71	attres1	attres1	PROPN
fcis-6025	82	72	network	network	NOUN
fcis-6025	82	73	model	model	NOUN
fcis-6025	82	74	,	,	PUNCT
fcis-6025	82	75	indicating	indicate	VERB
fcis-6025	82	76	that	that	SCONJ
fcis-6025	82	77	the	the	DET
fcis-6025	82	78	res	re	NOUN
fcis-6025	82	79	-	-	PUNCT
fcis-6025	82	80	a	a	DET
fcis-6025	82	81	layer	layer	NOUN
fcis-6025	82	82	is	be	AUX
fcis-6025	82	83	better	well	ADJ
fcis-6025	82	84	than	than	ADP
fcis-6025	82	85	the	the	DET
fcis-6025	82	86	direct	direct	ADJ
fcis-6025	82	87	addition	addition	NOUN
fcis-6025	82	88	of	of	ADP
fcis-6025	82	89	the	the	DET
fcis-6025	82	90	attention	attention	NOUN
fcis-6025	82	91	mechanism	mechanism	NOUN
fcis-6025	82	92	layer	layer	NOUN
fcis-6025	82	93	in	in	ADP
fcis-6025	82	94	extracting	extract	VERB
fcis-6025	82	95	effective	effective	ADJ
fcis-6025	82	96	features	feature	NOUN
fcis-6025	82	97	,	,	PUNCT
fcis-6025	82	98	which	which	PRON
fcis-6025	82	99	verifies	verify	VERB
fcis-6025	82	100	the	the	DET
fcis-6025	82	101	effectiveness	effectiveness	NOUN
fcis-6025	82	102	of	of	ADP
fcis-6025	82	103	the	the	DET
fcis-6025	82	104	res	re	NOUN
fcis-6025	82	105	-	-	PUNCT
fcis-6025	82	106	a	a	DET
fcis-6025	82	107	layer	layer	NOUN
fcis-6025	82	108	.	.	PUNCT
fcis-6025	83	1	the	the	DET
fcis-6025	83	2	recognition	recognition	NOUN
fcis-6025	83	3	rates	rate	NOUN
fcis-6025	83	4	of	of	ADP
fcis-6025	83	5	the	the	DET
fcis-6025	83	6	network	network	NOUN
fcis-6025	83	7	models	model	NOUN
fcis-6025	83	8	with	with	ADP
fcis-6025	83	9	the	the	DET
fcis-6025	83	10	block	block	NOUN
fcis-6025	83	11	-	-	PUNCT
fcis-6025	83	12	a	a	DET
fcis-6025	83	13	module	module	NOUN
fcis-6025	83	14	added	add	VERB
fcis-6025	83	15	are	be	AUX
fcis-6025	83	16	all	all	ADV
fcis-6025	83	17	better	well	ADJ
fcis-6025	83	18	than	than	ADP
fcis-6025	83	19	those	those	PRON
fcis-6025	83	20	of	of	ADP
fcis-6025	83	21	att	att	PROPN
fcis-6025	83	22	-	-	PUNCT
fcis-6025	83	23	res0	res0	PROPN
fcis-6025	83	24	composed	compose	VERB
fcis-6025	83	25	of	of	ADP
fcis-6025	83	26	the	the	DET
fcis-6025	83	27	block	block	NOUN
fcis-6025	83	28	module	module	NOUN
fcis-6025	83	29	only	only	ADV
fcis-6025	83	30	.	.	PUNCT
fcis-6025	84	1	the	the	DET
fcis-6025	84	2	biggest	big	ADJ
fcis-6025	84	3	difference	difference	NOUN
fcis-6025	84	4	between	between	ADP
fcis-6025	84	5	the	the	DET
fcis-6025	84	6	block	block	NOUN
fcis-6025	84	7	-	-	PUNCT
fcis-6025	84	8	a	a	DET
fcis-6025	84	9	module	module	NOUN
fcis-6025	84	10	and	and	CCONJ
fcis-6025	84	11	the	the	DET
fcis-6025	84	12	block	block	NOUN
fcis-6025	84	13	module	module	NOUN
fcis-6025	84	14	is	be	AUX
fcis-6025	84	15	that	that	SCONJ
fcis-6025	84	16	the	the	DET
fcis-6025	84	17	res	re	NOUN
fcis-6025	84	18	-	-	PUNCT
fcis-6025	84	19	a	a	DET
fcis-6025	84	20	layer	layer	NOUN
fcis-6025	84	21	proposed	propose	VERB
fcis-6025	84	22	in	in	ADP
fcis-6025	84	23	this	this	DET
fcis-6025	84	24	paper	paper	NOUN
fcis-6025	84	25	replaces	replace	VERB
fcis-6025	84	26	the	the	DET
fcis-6025	84	27	constant	constant	ADJ
fcis-6025	84	28	residual	residual	ADJ
fcis-6025	84	29	layer	layer	NOUN
fcis-6025	84	30	in	in	ADP
fcis-6025	84	31	the	the	DET
fcis-6025	84	32	residual	residual	ADJ
fcis-6025	84	33	network	network	NOUN
fcis-6025	84	34	,	,	PUNCT
fcis-6025	84	35	which	which	PRON
fcis-6025	84	36	effectively	effectively	ADV
fcis-6025	84	37	improves	improve	VERB
fcis-6025	84	38	the	the	DET
fcis-6025	84	39	network	network	NOUN
fcis-6025	84	40	sentiment	sentiment	NOUN
fcis-6025	84	41	recognition	recognition	NOUN
fcis-6025	84	42	accuracy	accuracy	NOUN
fcis-6025	84	43	by	by	ADP
fcis-6025	84	44	entering	enter	VERB
fcis-6025	84	45	spatial	spatial	ADJ
fcis-6025	84	46	attention	attention	NOUN
fcis-6025	84	47	.	.	PUNCT
fcis-6025	85	1	in	in	ADP
fcis-6025	85	2	the	the	DET
fcis-6025	85	3	four	four	NUM
fcis-6025	85	4	network	network	NOUN
fcis-6025	85	5	models	model	NOUN
fcis-6025	85	6	composed	compose	VERB
fcis-6025	85	7	of	of	ADP
fcis-6025	85	8	a	a	DET
fcis-6025	85	9	mixture	mixture	NOUN
fcis-6025	85	10	of	of	ADP
fcis-6025	85	11	block	block	NOUN
fcis-6025	85	12	-	-	PUNCT
fcis-6025	85	13	a	a	DET
fcis-6025	85	14	modules	module	NOUN
fcis-6025	85	15	and	and	CCONJ
fcis-6025	85	16	block	block	NOUN
fcis-6025	85	17	modules	module	NOUN
fcis-6025	85	18	,	,	PUNCT
fcis-6025	85	19	the	the	DET
fcis-6025	85	20	recognition	recognition	NOUN
fcis-6025	85	21	accuracy	accuracy	NOUN
fcis-6025	85	22	increases	increase	NOUN
fcis-6025	85	23	as	as	ADP
fcis-6025	85	24	the	the	DET
fcis-6025	85	25	number	number	NOUN
fcis-6025	85	26	of	of	ADP
fcis-6025	85	27	block	block	NOUN
fcis-6025	85	28	-	-	PUNCT
fcis-6025	85	29	a	a	PRON
fcis-6025	85	30	modules	module	NOUN
fcis-6025	85	31	increases	increase	NOUN
fcis-6025	85	32	.	.	PUNCT
fcis-6025	86	1	however	however	ADV
fcis-6025	86	2	,	,	PUNCT
fcis-6025	86	3	the	the	DET
fcis-6025	86	4	recognition	recognition	NOUN
fcis-6025	86	5	accuracy	accuracy	NOUN
fcis-6025	86	6	decreases	decrease	VERB
fcis-6025	86	7	when	when	SCONJ
fcis-6025	86	8	the	the	DET
fcis-6025	86	9	number	number	NOUN
fcis-6025	86	10	of	of	ADP
fcis-6025	86	11	block	block	NOUN
fcis-6025	86	12	-	-	PUNCT
fcis-6025	86	13	a	a	DET
fcis-6025	86	14	modules	module	NOUN
fcis-6025	86	15	is	be	AUX
fcis-6025	86	16	3	3	NUM
fcis-6025	86	17	.	.	PUNCT
fcis-6025	87	1	the	the	DET
fcis-6025	87	2	casia	casia	NOUN
fcis-6025	87	3	data	datum	NOUN
fcis-6025	87	4	are	be	AUX
fcis-6025	87	5	small	small	ADJ
fcis-6025	87	6	sample	sample	NOUN
fcis-6025	87	7	data	datum	NOUN
fcis-6025	87	8	,	,	PUNCT
fcis-6025	87	9	and	and	CCONJ
fcis-6025	87	10	the	the	DET
fcis-6025	87	11	excessive	excessive	ADJ
fcis-6025	87	12	introduction	introduction	NOUN
fcis-6025	87	13	of	of	ADP
fcis-6025	87	14	attention	attention	NOUN
fcis-6025	87	15	mechanism	mechanism	NOUN
fcis-6025	87	16	tends	tend	VERB
fcis-6025	87	17	to	to	PART
fcis-6025	87	18	make	make	VERB
fcis-6025	87	19	the	the	DET
fcis-6025	87	20	network	network	NOUN
fcis-6025	87	21	in	in	ADP
fcis-6025	87	22	an	an	DET
fcis-6025	87	23	overfitting	overfitte	VERB
fcis-6025	87	24	state	state	NOUN
fcis-6025	87	25	,	,	PUNCT
fcis-6025	87	26	which	which	PRON
fcis-6025	87	27	reduces	reduce	VERB
fcis-6025	87	28	the	the	DET
fcis-6025	87	29	model	model	NOUN
fcis-6025	87	30	effect	effect	NOUN
fcis-6025	87	31	.	.	PUNCT
fcis-6025	88	1	in	in	ADP
fcis-6025	88	2	the	the	DET
fcis-6025	88	3	network	network	NOUN
fcis-6025	88	4	consisting	consist	VERB
fcis-6025	88	5	of	of	ADP
fcis-6025	88	6	two	two	NUM
fcis-6025	88	7	layers	layer	NOUN
fcis-6025	88	8	of	of	ADP
fcis-6025	88	9	block	block	NOUN
fcis-6025	88	10	-	-	PUNCT
fcis-6025	88	11	a	a	DET
fcis-6025	88	12	modules	module	NOUN
fcis-6025	88	13	and	and	CCONJ
fcis-6025	88	14	one	one	NUM
fcis-6025	88	15	layer	layer	NOUN
fcis-6025	88	16	of	of	ADP
fcis-6025	88	17	block	block	NOUN
fcis-6025	88	18	modules	module	NOUN
fcis-6025	88	19	,	,	PUNCT
fcis-6025	88	20	we	we	PRON
fcis-6025	88	21	explored	explore	VERB
fcis-6025	88	22	the	the	DET
fcis-6025	88	23	effects	effect	NOUN
fcis-6025	88	24	of	of	ADP
fcis-6025	88	25	introducing	introduce	VERB
fcis-6025	88	26	res	re	NOUN
fcis-6025	88	27	-	-	PUNCT
fcis-6025	88	28	a	a	DET
fcis-6025	88	29	layers	layer	NOUN
fcis-6025	88	30	at	at	ADP
fcis-6025	88	31	different	different	ADJ
fcis-6025	88	32	locations	location	NOUN
fcis-6025	88	33	.	.	PUNCT
fcis-6025	89	1	this	this	PRON
fcis-6025	89	2	is	be	AUX
fcis-6025	89	3	because	because	SCONJ
fcis-6025	89	4	the	the	DET
fcis-6025	89	5	number	number	NOUN
fcis-6025	89	6	of	of	ADP
fcis-6025	89	7	deeper	deep	ADJ
fcis-6025	89	8	features	feature	NOUN
fcis-6025	89	9	extracted	extract	VERB
fcis-6025	89	10	when	when	SCONJ
fcis-6025	89	11	the	the	DET
fcis-6025	89	12	network	network	NOUN
fcis-6025	89	13	is	be	AUX
fcis-6025	89	14	deeper	deep	ADJ
fcis-6025	89	15	has	have	VERB
fcis-6025	89	16	more	more	ADJ
fcis-6025	89	17	semantic	semantic	ADJ
fcis-6025	89	18	information	information	NOUN
fcis-6025	89	19	,	,	PUNCT
fcis-6025	89	20	and	and	CCONJ
fcis-6025	89	21	then	then	ADV
fcis-6025	89	22	adding	add	VERB
fcis-6025	89	23	the	the	DET
fcis-6025	89	24	attention	attention	NOUN
fcis-6025	89	25	mechanism	mechanism	NOUN
fcis-6025	89	26	can	can	AUX
fcis-6025	89	27	further	far	ADV
fcis-6025	89	28	extract	extract	VERB
fcis-6025	89	29	the	the	DET
fcis-6025	89	30	emotional	emotional	ADJ
fcis-6025	89	31	features	feature	NOUN
fcis-6025	89	32	.	.	PUNCT
fcis-6025	90	1	table	table	NOUN
fcis-6025	90	2	.1	.1	NUM
fcis-6025	90	3	ablation	ablation	NOUN
fcis-6025	90	4	experiments	experiment	NOUN
fcis-6025	90	5	experimental	experimental	ADJ
fcis-6025	90	6	group	group	NOUN
fcis-6025	90	7	dataset	dataset	NOUN
fcis-6025	90	8	methods	method	NOUN
fcis-6025	90	9	accuracy	accuracy	NOUN
fcis-6025	90	10	1	1	NUM
fcis-6025	90	11	casia	casia	NOUN
fcis-6025	90	12	att	att	PROPN
fcis-6025	90	13	-	-	PUNCT
fcis-6025	90	14	res0	res0	PROPN
fcis-6025	90	15	57.2	57.2	NUM
fcis-6025	90	16	%	%	NOUN
fcis-6025	90	17	2	2	NUM
fcis-6025	90	18	attres0+att	attres0+att	PROPN
fcis-6025	90	19	58.7	58.7	NUM
fcis-6025	90	20	%	%	NOUN
fcis-6025	90	21	3	3	NUM
fcis-6025	90	22	att	att	PROPN
fcis-6025	90	23	-	-	PROPN
fcis-6025	90	24	res1	res1	PROPN
fcis-6025	90	25	64.4	64.4	NUM
fcis-6025	90	26	%	%	NOUN
fcis-6025	90	27	4	4	NUM
fcis-6025	90	28	att	att	PROPN
fcis-6025	90	29	-	-	PUNCT
fcis-6025	90	30	res2	res2	PROPN
fcis-6025	90	31	67.3	67.3	NUM
fcis-6025	90	32	%	%	NOUN
fcis-6025	90	33	5	5	NUM
fcis-6025	90	34	att	att	PROPN
fcis-6025	90	35	-	-	PUNCT
fcis-6025	90	36	res	re	VERB
fcis-6025	90	37	69.8	69.8	NUM
fcis-6025	90	38	%	%	NOUN
fcis-6025	90	39	6	6	NUM
fcis-6025	90	40	att	att	PROPN
fcis-6025	90	41	-	-	PUNCT
fcis-6025	90	42	res3	res3	PROPN
fcis-6025	90	43	66.1	66.1	NUM
fcis-6025	90	44	%	%	NOUN
fcis-6025	90	45	finally	finally	ADV
fcis-6025	90	46	,	,	PUNCT
fcis-6025	90	47	to	to	PART
fcis-6025	90	48	verify	verify	VERB
fcis-6025	90	49	the	the	DET
fcis-6025	90	50	effectiveness	effectiveness	NOUN
fcis-6025	90	51	of	of	ADP
fcis-6025	90	52	the	the	DET
fcis-6025	90	53	att	att	PROPN
fcis-6025	90	54	-	-	PUNCT
fcis-6025	90	55	res	re	NOUN
fcis-6025	90	56	model	model	NOUN
fcis-6025	90	57	extracted	extract	VERB
fcis-6025	90	58	in	in	ADP
fcis-6025	90	59	this	this	DET
fcis-6025	90	60	paper	paper	NOUN
fcis-6025	90	61	for	for	ADP
fcis-6025	90	62	speech	speech	NOUN
fcis-6025	90	63	emotion	emotion	NOUN
fcis-6025	90	64	recognition	recognition	NOUN
fcis-6025	90	65	,	,	PUNCT
fcis-6025	90	66	this	this	DET
fcis-6025	90	67	subsection	subsection	NOUN
fcis-6025	90	68	compares	compare	VERB
fcis-6025	90	69	five	five	NUM
fcis-6025	90	70	commonly	commonly	ADV
fcis-6025	90	71	used	use	VERB
fcis-6025	90	72	network	network	NOUN
fcis-6025	90	73	models	model	NOUN
fcis-6025	90	74	for	for	ADP
fcis-6025	90	75	the	the	DET
fcis-6025	90	76	casia	casia	NOUN
fcis-6025	90	77	dataset	dataset	NOUN
fcis-6025	90	78	,	,	PUNCT
fcis-6025	90	79	where	where	SCONJ
fcis-6025	90	80	experimental	experimental	ADJ
fcis-6025	90	81	groups	group	NOUN
fcis-6025	90	82	4	4	NUM
fcis-6025	90	83	,	,	PUNCT
fcis-6025	90	84	5	5	NUM
fcis-6025	90	85	,	,	PUNCT
fcis-6025	90	86	and	and	CCONJ
fcis-6025	90	87	6	6	NUM
fcis-6025	90	88	all	all	PRON
fcis-6025	90	89	use	use	VERB
fcis-6025	90	90	the	the	DET
fcis-6025	90	91	speech	speech	NOUN
fcis-6025	90	92	spectrogram	spectrogram	NOUN
fcis-6025	90	93	as	as	ADP
fcis-6025	90	94	the	the	DET
fcis-6025	90	95	input	input	NOUN
fcis-6025	90	96	to	to	ADP
fcis-6025	90	97	the	the	DET
fcis-6025	90	98	network	network	NOUN
fcis-6025	90	99	.	.	PUNCT
fcis-6025	91	1	the	the	DET
fcis-6025	91	2	experimental	experimental	ADJ
fcis-6025	91	3	results	result	NOUN
fcis-6025	91	4	are	be	AUX
fcis-6025	91	5	shown	show	VERB
fcis-6025	91	6	in	in	ADP
fcis-6025	91	7	table	table	NOUN
fcis-6025	91	8	2	2	NUM
fcis-6025	91	9	.	.	PUNCT
fcis-6025	91	10	table	table	NOUN
fcis-6025	91	11	2	2	NUM
fcis-6025	91	12	.	.	PUNCT
fcis-6025	91	13	speech	speech	NOUN
fcis-6025	91	14	emotion	emotion	NOUN
fcis-6025	91	15	recognition	recognition	NOUN
fcis-6025	91	16	on	on	ADP
fcis-6025	91	17	casia	casia	NOUN
fcis-6025	91	18	dataset	dataset	VERB
fcis-6025	91	19	experimental	experimental	ADJ
fcis-6025	91	20	group	group	NOUN
fcis-6025	91	21	dataset	dataset	NOUN
fcis-6025	91	22	methods	method	NOUN
fcis-6025	91	23	accuracy	accuracy	NOUN
fcis-6025	91	24	1	1	NUM
fcis-6025	91	25	casia	casia	NOUN
fcis-6025	91	26	cnn	cnn	PROPN
fcis-6025	91	27	46.9	46.9	NUM
fcis-6025	91	28	%	%	NOUN
fcis-6025	91	29	2	2	NUM
fcis-6025	91	30	lstm+cnn	lstm+cnn	PROPN
fcis-6025	91	31	49.2	49.2	NUM
fcis-6025	91	32	%	%	NOUN
fcis-6025	91	33	3	3	NUM
fcis-6025	91	34	svm	svm	NOUN
fcis-6025	91	35	50.2	50.2	NUM
fcis-6025	91	36	%	%	NOUN
fcis-6025	91	37	4	4	NUM
fcis-6025	91	38	alexnet	alexnet	NOUN
fcis-6025	91	39	51.7	51.7	NUM
fcis-6025	91	40	%	%	NOUN
fcis-6025	91	41	5	5	NUM
fcis-6025	91	42	resnet-18	resnet-18	PROPN
fcis-6025	91	43	50.3	50.3	NUM
fcis-6025	91	44	%	%	NOUN
fcis-6025	91	45	6	6	NUM
fcis-6025	91	46	att	att	PROPN
fcis-6025	91	47	-	-	PUNCT
fcis-6025	91	48	res	re	VERB
fcis-6025	91	49	69.8	69.8	NUM
fcis-6025	91	50	%	%	NOUN
fcis-6025	91	51	compared	compare	VERB
fcis-6025	91	52	with	with	ADP
fcis-6025	91	53	the	the	DET
fcis-6025	91	54	resnet-18	resnet-18	PROPN
fcis-6025	91	55	network	network	NOUN
fcis-6025	91	56	structure	structure	NOUN
fcis-6025	91	57	,	,	PUNCT
fcis-6025	91	58	the	the	DET
fcis-6025	91	59	recognition	recognition	NOUN
fcis-6025	91	60	accuracy	accuracy	NOUN
fcis-6025	91	61	of	of	ADP
fcis-6025	91	62	the	the	DET
fcis-6025	91	63	improved	improved	ADJ
fcis-6025	91	64	att	att	PROPN
fcis-6025	91	65	-	-	PUNCT
fcis-6025	91	66	res	re	NOUN
fcis-6025	91	67	network	network	NOUN
fcis-6025	91	68	based	base	VERB
fcis-6025	91	69	on	on	ADP
fcis-6025	91	70	the	the	DET
fcis-6025	91	71	residual	residual	ADJ
fcis-6025	91	72	network	network	NOUN
fcis-6025	91	73	is	be	AUX
fcis-6025	91	74	higher	high	ADJ
fcis-6025	91	75	because	because	SCONJ
fcis-6025	91	76	the	the	DET
fcis-6025	91	77	incorporated	incorporate	VERB
fcis-6025	91	78	spatial	spatial	ADJ
fcis-6025	91	79	attention	attention	NOUN
fcis-6025	91	80	mechanism	mechanism	NOUN
fcis-6025	91	81	can	can	AUX
fcis-6025	91	82	autonomously	autonomously	ADV
fcis-6025	91	83	learn	learn	VERB
fcis-6025	91	84	the	the	DET
fcis-6025	91	85	distribution	distribution	NOUN
fcis-6025	91	86	of	of	ADP
fcis-6025	91	87	emotion	emotion	NOUN
fcis-6025	91	88	features	feature	NOUN
fcis-6025	91	89	and	and	CCONJ
fcis-6025	91	90	weight	weight	VERB
fcis-6025	91	91	them	they	PRON
fcis-6025	91	92	by	by	ADP
fcis-6025	91	93	backpropagation	backpropagation	NOUN
fcis-6025	91	94	,	,	PUNCT
fcis-6025	91	95	thus	thus	ADV
fcis-6025	91	96	achieving	achieve	VERB
fcis-6025	91	97	the	the	DET
fcis-6025	91	98	effect	effect	NOUN
fcis-6025	91	99	of	of	ADP
fcis-6025	91	100	reducing	reduce	VERB
fcis-6025	91	101	the	the	DET
fcis-6025	91	102	interference	interference	NOUN
fcis-6025	91	103	of	of	ADP
fcis-6025	91	104	redundant	redundant	ADJ
fcis-6025	91	105	information	information	NOUN
fcis-6025	91	106	on	on	ADP
fcis-6025	91	107	the	the	DET
fcis-6025	91	108	recognition	recognition	NOUN
fcis-6025	91	109	accuracy	accuracy	NOUN
fcis-6025	91	110	.	.	PUNCT
fcis-6025	92	1	finally	finally	ADV
fcis-6025	92	2	,	,	PUNCT
fcis-6025	92	3	comparing	compare	VERB
fcis-6025	92	4	all	all	DET
fcis-6025	92	5	the	the	DET
fcis-6025	92	6	experimental	experimental	ADJ
fcis-6025	92	7	results	result	NOUN
fcis-6025	92	8	in	in	ADP
fcis-6025	92	9	table	table	NOUN
fcis-6025	92	10	3	3	NUM
fcis-6025	92	11	-	-	SYM
fcis-6025	92	12	2	2	NUM
fcis-6025	92	13	,	,	PUNCT
fcis-6025	92	14	the	the	DET
fcis-6025	92	15	improved	improved	ADJ
fcis-6025	92	16	att	att	PROPN
fcis-6025	92	17	-	-	PUNCT
fcis-6025	92	18	res	re	NOUN
fcis-6025	92	19	network	network	NOUN
fcis-6025	92	20	of	of	ADP
fcis-6025	92	21	the	the	DET
fcis-6025	92	22	model	model	NOUN
fcis-6025	92	23	in	in	ADP
fcis-6025	92	24	this	this	DET
fcis-6025	92	25	paper	paper	NOUN
fcis-6025	92	26	can	can	AUX
fcis-6025	92	27	improve	improve	VERB
fcis-6025	92	28	the	the	DET
fcis-6025	92	29	accuracy	accuracy	NOUN
fcis-6025	92	30	of	of	ADP
fcis-6025	92	31	the	the	DET
fcis-6025	92	32	network	network	NOUN
fcis-6025	92	33	model	model	NOUN
fcis-6025	92	34	recognition	recognition	NOUN
fcis-6025	92	35	in	in	ADP
fcis-6025	92	36	the	the	DET
fcis-6025	92	37	comparison	comparison	NOUN
fcis-6025	92	38	experiments	experiment	NOUN
fcis-6025	92	39	.	.	PUNCT
fcis-6025	93	1	att	att	PROPN
fcis-6025	93	2	-	-	PUNCT
fcis-6025	93	3	res	re	NOUN
fcis-6025	93	4	network	network	NOUN
fcis-6025	93	5	64	64	NUM
fcis-6025	93	6	model	model	NOUN
fcis-6025	93	7	improves	improve	VERB
fcis-6025	93	8	the	the	DET
fcis-6025	93	9	speech	speech	NOUN
fcis-6025	93	10	emotion	emotion	NOUN
fcis-6025	93	11	recognition	recognition	NOUN
fcis-6025	93	12	rate	rate	NOUN
fcis-6025	93	13	by	by	ADP
fcis-6025	93	14	22.9	22.9	NUM
fcis-6025	93	15	%	%	NOUN
fcis-6025	93	16	compared	compare	VERB
fcis-6025	93	17	with	with	ADP
fcis-6025	93	18	the	the	DET
fcis-6025	93	19	cnn	cnn	PROPN
fcis-6025	93	20	network	network	NOUN
fcis-6025	93	21	model	model	NOUN
fcis-6025	93	22	and	and	CCONJ
fcis-6025	93	23	19.5	19.5	NUM
fcis-6025	93	24	%	%	NOUN
fcis-6025	93	25	compared	compare	VERB
fcis-6025	93	26	with	with	ADP
fcis-6025	93	27	the	the	DET
fcis-6025	93	28	resnet-18	resnet-18	PROPN
fcis-6025	93	29	network	network	NOUN
fcis-6025	93	30	,	,	PUNCT
fcis-6025	93	31	which	which	PRON
fcis-6025	93	32	verifies	verify	VERB
fcis-6025	93	33	the	the	DET
fcis-6025	93	34	model	model	NOUN
fcis-6025	93	35	's	's	PART
fcis-6025	93	36	effectiveness	effectiveness	NOUN
fcis-6025	93	37	.	.	PUNCT
fcis-6025	94	1	4	4	X
fcis-6025	94	2	.	.	X
fcis-6025	94	3	summary	summary	NOUN
fcis-6025	94	4	in	in	ADP
fcis-6025	94	5	this	this	DET
fcis-6025	94	6	paper	paper	NOUN
fcis-6025	94	7	,	,	PUNCT
fcis-6025	94	8	we	we	PRON
fcis-6025	94	9	introduce	introduce	VERB
fcis-6025	94	10	attention	attention	NOUN
fcis-6025	94	11	mechanism	mechanism	NOUN
fcis-6025	94	12	into	into	ADP
fcis-6025	94	13	constant	constant	ADJ
fcis-6025	94	14	residual	residual	ADJ
fcis-6025	94	15	block	block	NOUN
fcis-6025	94	16	,	,	PUNCT
fcis-6025	94	17	propose	propose	VERB
fcis-6025	94	18	a	a	DET
fcis-6025	94	19	res	re	NOUN
fcis-6025	94	20	-	-	PUNCT
fcis-6025	94	21	a	a	DET
fcis-6025	94	22	module	module	NOUN
fcis-6025	94	23	based	base	VERB
fcis-6025	94	24	on	on	ADP
fcis-6025	94	25	constant	constant	ADJ
fcis-6025	94	26	residual	residual	ADJ
fcis-6025	94	27	block	block	NOUN
fcis-6025	94	28	,	,	PUNCT
fcis-6025	94	29	design	design	VERB
fcis-6025	94	30	six	six	NUM
fcis-6025	94	31	network	network	NOUN
fcis-6025	94	32	models	model	NOUN
fcis-6025	94	33	of	of	ADP
fcis-6025	94	34	attres0	attres0	PROPN
fcis-6025	94	35	,	,	PUNCT
fcis-6025	94	36	att	att	PROPN
fcis-6025	94	37	-	-	PUNCT
fcis-6025	94	38	res0+att	res0+att	PROPN
fcis-6025	94	39	,	,	PUNCT
fcis-6025	94	40	att	att	PROPN
fcis-6025	94	41	-	-	PUNCT
fcis-6025	94	42	res1	res1	PROPN
fcis-6025	94	43	,	,	PUNCT
fcis-6025	94	44	att	att	PROPN
fcis-6025	94	45	-	-	PUNCT
fcis-6025	94	46	res2	res2	PROPN
fcis-6025	94	47	,	,	PUNCT
fcis-6025	94	48	att	att	PROPN
fcis-6025	94	49	-	-	PUNCT
fcis-6025	94	50	res	res	PROPN
fcis-6025	94	51	and	and	CCONJ
fcis-6025	94	52	att	att	PROPN
fcis-6025	94	53	-	-	PUNCT
fcis-6025	94	54	res3	res3	PROPN
fcis-6025	94	55	for	for	ADP
fcis-6025	94	56	this	this	DET
fcis-6025	94	57	module	module	NOUN
fcis-6025	94	58	,	,	PUNCT
fcis-6025	94	59	verify	verify	VERB
fcis-6025	94	60	the	the	DET
fcis-6025	94	61	effectiveness	effectiveness	NOUN
fcis-6025	94	62	of	of	ADP
fcis-6025	94	63	this	this	DET
fcis-6025	94	64	module	module	NOUN
fcis-6025	94	65	with	with	ADP
fcis-6025	94	66	ablation	ablation	NOUN
fcis-6025	94	67	experiments	experiment	NOUN
fcis-6025	94	68	,	,	PUNCT
fcis-6025	94	69	and	and	CCONJ
fcis-6025	94	70	the	the	DET
fcis-6025	94	71	experimental	experimental	ADJ
fcis-6025	94	72	results	result	NOUN
fcis-6025	94	73	prove	prove	VERB
fcis-6025	94	74	that	that	SCONJ
fcis-6025	94	75	the	the	DET
fcis-6025	94	76	res	re	NOUN
fcis-6025	94	77	-a	-a	PUNCT
fcis-6025	94	78	module	module	NOUN
fcis-6025	94	79	can	can	AUX
fcis-6025	94	80	effectively	effectively	ADV
fcis-6025	94	81	improve	improve	VERB
fcis-6025	94	82	the	the	DET
fcis-6025	94	83	speech	speech	NOUN
fcis-6025	94	84	emotion	emotion	NOUN
fcis-6025	94	85	recognition	recognition	NOUN
fcis-6025	94	86	rate	rate	NOUN
fcis-6025	94	87	.	.	PUNCT
fcis-6025	95	1	finally	finally	ADV
fcis-6025	95	2	,	,	PUNCT
fcis-6025	95	3	this	this	DET
fcis-6025	95	4	paper	paper	NOUN
fcis-6025	95	5	proposes	propose	VERB
fcis-6025	95	6	an	an	DET
fcis-6025	95	7	att	att	PROPN
fcis-6025	95	8	-	-	PUNCT
fcis-6025	95	9	res	re	NOUN
fcis-6025	95	10	speech	speech	NOUN
fcis-6025	95	11	emotion	emotion	NOUN
fcis-6025	95	12	recognition	recognition	NOUN
fcis-6025	95	13	network	network	NOUN
fcis-6025	95	14	structure	structure	NOUN
fcis-6025	95	15	based	base	VERB
fcis-6025	95	16	on	on	ADP
fcis-6025	95	17	residual	residual	ADJ
fcis-6025	95	18	block	block	NOUN
fcis-6025	95	19	improvement	improvement	NOUN
fcis-6025	95	20	based	base	VERB
fcis-6025	95	21	on	on	ADP
fcis-6025	95	22	the	the	DET
fcis-6025	95	23	res	re	NOUN
fcis-6025	95	24	-	-	PUNCT
fcis-6025	95	25	a	a	DET
fcis-6025	95	26	module	module	NOUN
fcis-6025	95	27	,	,	PUNCT
fcis-6025	95	28	and	and	CCONJ
fcis-6025	95	29	trains	train	VERB
fcis-6025	95	30	the	the	DET
fcis-6025	95	31	model	model	NOUN
fcis-6025	95	32	on	on	ADP
fcis-6025	95	33	casia	casia	NOUN
fcis-6025	95	34	dataset	dataset	NOUN
fcis-6025	95	35	.	.	PUNCT
fcis-6025	96	1	by	by	ADP
fcis-6025	96	2	comparing	compare	VERB
fcis-6025	96	3	with	with	ADP
fcis-6025	96	4	four	four	NUM
fcis-6025	96	5	advanced	advanced	ADJ
fcis-6025	96	6	network	network	NOUN
fcis-6025	96	7	models	model	NOUN
fcis-6025	96	8	and	and	CCONJ
fcis-6025	96	9	a	a	DET
fcis-6025	96	10	traditional	traditional	ADJ
fcis-6025	96	11	classifier	classifier	NOUN
fcis-6025	96	12	on	on	ADP
fcis-6025	96	13	the	the	DET
fcis-6025	96	14	same	same	ADJ
fcis-6025	96	15	dataset	dataset	NOUN
fcis-6025	96	16	,	,	PUNCT
fcis-6025	96	17	the	the	DET
fcis-6025	96	18	results	result	NOUN
fcis-6025	96	19	demonstrate	demonstrate	VERB
fcis-6025	96	20	that	that	SCONJ
fcis-6025	96	21	the	the	DET
fcis-6025	96	22	att	att	PROPN
fcis-6025	96	23	-	-	PUNCT
fcis-6025	96	24	res	re	NOUN
fcis-6025	96	25	network	network	NOUN
fcis-6025	96	26	model	model	NOUN
fcis-6025	96	27	improves	improve	VERB
fcis-6025	96	28	speech	speech	NOUN
fcis-6025	96	29	emotion	emotion	NOUN
fcis-6025	96	30	recognition	recognition	NOUN
fcis-6025	96	31	to	to	ADP
fcis-6025	96	32	a	a	DET
fcis-6025	96	33	certain	certain	ADJ
fcis-6025	96	34	extent	extent	NOUN
fcis-6025	96	35	and	and	CCONJ
fcis-6025	96	36	verifies	verifie	NOUN
fcis-6025	96	37	the	the	DET
fcis-6025	96	38	feasibility	feasibility	NOUN
fcis-6025	96	39	of	of	ADP
fcis-6025	96	40	the	the	DET
fcis-6025	96	41	model	model	NOUN
fcis-6025	96	42	.	.	PUNCT
fcis-6025	97	1	acknowledgments	acknowledgment	NOUN
fcis-6025	97	2	this	this	DET
fcis-6025	97	3	work	work	NOUN
fcis-6025	97	4	was	be	AUX
fcis-6025	97	5	supported	support	VERB
fcis-6025	97	6	by	by	ADP
fcis-6025	97	7	the	the	DET
fcis-6025	97	8	southwest	southwest	PROPN
fcis-6025	97	9	minzu	minzu	PROPN
fcis-6025	97	10	university	university	PROPN
fcis-6025	97	11	for	for	ADP
fcis-6025	97	12	nationalities	nationality	NOUN
fcis-6025	97	13	graduate	graduate	NOUN
fcis-6025	97	14	student	student	NOUN
fcis-6025	97	15	innovative	innovative	ADJ
fcis-6025	97	16	project	project	NOUN
fcis-6025	97	17	no	no	INTJ
fcis-6025	97	18	.	.	PUNCT
fcis-6025	98	1	(	(	PUNCT
fcis-6025	98	2	yb2022753	yb2022753	PROPN
fcis-6025	98	3	)	)	PUNCT
fcis-6025	98	4	fund	fund	NOUN
fcis-6025	98	5	.	.	PUNCT
fcis-6025	99	1	references	reference	NOUN
fcis-6025	99	2	[	[	X
fcis-6025	99	3	1	1	X
fcis-6025	99	4	]	]	X
fcis-6025	99	5	hong	hong	PROPN
fcis-6025	99	6	zhaogin	zhaogin	PROPN
fcis-6025	99	7	,	,	PUNCT
fcis-6025	99	8	wei	wei	PROPN
fcis-6025	99	9	chenyang	chenyang	PROPN
fcis-6025	99	10	,	,	PUNCT
fcis-6025	99	11	zhuang	zhuang	PROPN
fcis-6025	99	12	yuan	yuan	PROPN
fcis-6025	99	13	,	,	PUNCT
fcis-6025	99	14	wang	wang	PROPN
fcis-6025	99	15	ying	ying	PROPN
fcis-6025	99	16	,	,	PUNCT
fcis-6025	99	17	wang	wang	PROPN
fcis-6025	99	18	yiting	yiting	PROPN
fcis-6025	99	19	,	,	PUNCT
fcis-6025	99	20	zhao	zhao	PROPN
fcis-6025	99	21	li	li	PROPN
fcis-6025	99	22	.	.	PROPN
fcis-6025	99	23	speech	speech	NOUN
fcis-6025	99	24	emotion	emotion	NOUN
fcis-6025	99	25	recognition	recognition	NOUN
fcis-6025	99	26	and	and	CCONJ
fcis-6025	99	27	personality	personality	NOUN
fcis-6025	99	28	analysis	analysis	NOUN
fcis-6025	99	29	based	base	VERB
fcis-6025	99	30	on	on	ADP
fcis-6025	99	31	deep	deep	ADJ
fcis-6025	99	32	neural	neural	ADJ
fcis-6025	99	33	network[j	network[j	NOUN
fcis-6025	99	34	]	]	PUNCT
fcis-6025	99	35	.	.	PUNCT
fcis-6025	100	1	information	information	NOUN
fcis-6025	100	2	technology	technology	NOUN
fcis-6025	100	3	research,2020,46(01):48	research,2020,46(01):48	PROPN
fcis-6025	100	4	-	-	PUNCT
fcis-6025	100	5	53	53	NUM
fcis-6025	100	6	.	.	PUNCT
fcis-6025	101	1	[	[	X
fcis-6025	101	2	2	2	NUM
fcis-6025	101	3	]	]	SYM
fcis-6025	101	4	tao	tao	PROPN
fcis-6025	101	5	,	,	PUNCT
fcis-6025	101	6	h.w	h.w	PROPN
fcis-6025	101	7	.	.	PROPN
fcis-6025	101	8	,	,	PUNCT
fcis-6025	101	9	cha	cha	PROPN
fcis-6025	101	10	,	,	PUNCT
fcis-6025	101	11	c.	c.	PROPN
fcis-6025	101	12	,	,	PUNCT
fcis-6025	101	13	liang	liang	PROPN
fcis-6025	101	14	,	,	PUNCT
fcis-6025	101	15	r.y	r.y	PROPN
fcis-6025	101	16	.	.	PROPN
fcis-6025	101	17	,	,	PUNCT
fcis-6025	101	18	et	et	PROPN
fcis-6025	101	19	al	al	PROPN
fcis-6025	101	20	.	.	PUNCT
fcis-6025	102	1	a	a	DET
fcis-6025	102	2	speech	speech	NOUN
fcis-6025	102	3	spectrogram	spectrogram	NOUN
fcis-6025	102	4	feature	feature	NOUN
fcis-6025	102	5	extraction	extraction	NOUN
fcis-6025	102	6	algorithm	algorithm	NOUN
fcis-6025	102	7	for	for	ADP
fcis-6025	102	8	speech	speech	NOUN
fcis-6025	102	9	emotion	emotion	NOUN
fcis-6025	102	10	recognition	recognition	NOUN
fcis-6025	102	11	.	.	PUNCT
fcis-6025	103	1	journal	journal	PROPN
fcis-6025	103	2	of	of	ADP
fcis-6025	103	3	southeast	southeast	PROPN
fcis-6025	103	4	university	university	PROPN
fcis-6025	103	5	(	(	PUNCT
fcis-6025	103	6	natural	natural	ADJ
fcis-6025	103	7	science	science	NOUN
fcis-6025	103	8	edition),2015,45(5):817	edition),2015,45(5):817	PROPN
fcis-6025	103	9	-	-	PUNCT
fcis-6025	103	10	827	827	NUM
fcis-6025	103	11	.	.	PUNCT
fcis-6025	104	1	[	[	X
fcis-6025	104	2	3	3	X
fcis-6025	104	3	]	]	X
fcis-6025	104	4	irsoy	irsoy	ADJ
fcis-6025	104	5	o	o	NOUN
fcis-6025	104	6	,	,	PUNCT
fcis-6025	104	7	cardie	cardie	PROPN
fcis-6025	104	8	c.	c.	PROPN
fcis-6025	104	9	deep	deep	ADJ
fcis-6025	104	10	recursive	recursive	ADJ
fcis-6025	104	11	neural	neural	ADJ
fcis-6025	104	12	networks	network	NOUN
fcis-6025	104	13	for	for	ADP
fcis-6025	104	14	compositionality	compositionality	NOUN
fcis-6025	104	15	in	in	ADP
fcis-6025	104	16	language[j	language[j	PROPN
fcis-6025	104	17	]	]	PUNCT
fcis-6025	104	18	.	.	PUNCT
fcis-6025	105	1	advances	advance	NOUN
fcis-6025	105	2	in	in	ADP
fcis-6025	105	3	neural	neural	ADJ
fcis-6025	105	4	information	information	NOUN
fcis-6025	105	5	processing	processing	NOUN
fcis-6025	105	6	systems	system	NOUN
fcis-6025	105	7	,	,	PUNCT
fcis-6025	105	8	2014(3).2096	2014(3).2096	PROPN
fcis-6025	105	9	-	-	PUNCT
fcis-6025	105	10	2104	2104	NUM
fcis-6025	105	11	.	.	PUNCT
fcis-6025	106	1	[	[	X
fcis-6025	106	2	4	4	X
fcis-6025	106	3	]	]	X
fcis-6025	106	4	shah	shah	NOUN
fcis-6025	106	5	a	a	X
fcis-6025	106	6	,	,	PUNCT
fcis-6025	106	7	bhowmik	bhowmik	ADJ
fcis-6025	106	8	t.	t.	NOUN
fcis-6025	106	9	a	a	DET
fcis-6025	106	10	comparative	comparative	ADJ
fcis-6025	106	11	study	study	NOUN
fcis-6025	106	12	on	on	ADP
fcis-6025	106	13	mfcc	mfcc	NOUN
fcis-6025	106	14	and	and	CCONJ
fcis-6025	106	15	fundamental	fundamental	ADJ
fcis-6025	106	16	frequency	frequency	NOUN
fcis-6025	106	17	based	base	VERB
fcis-6025	106	18	speech	speech	NOUN
fcis-6025	106	19	emotion	emotion	NOUN
fcis-6025	106	20	classification[c]//international	classification[c]//international	PROPN
fcis-6025	106	21	conference	conference	NOUN
fcis-6025	106	22	on	on	ADP
fcis-6025	106	23	distributed	distribute	VERB
fcis-6025	106	24	computing	computing	NOUN
fcis-6025	106	25	and	and	CCONJ
fcis-6025	106	26	internet	internet	NOUN
fcis-6025	106	27	technology	technology	NOUN
fcis-6025	106	28	.	.	PUNCT
fcis-6025	107	1	springer	springer	NOUN
fcis-6025	107	2	,	,	PUNCT
fcis-6025	107	3	cham	cham	NOUN
fcis-6025	107	4	,	,	PUNCT
fcis-6025	107	5	2022	2022	NUM
fcis-6025	107	6	:	:	PUNCT
fcis-6025	107	7	173	173	NUM
fcis-6025	107	8	-	-	SYM
fcis-6025	107	9	184	184	NUM
fcis-6025	107	10	.	.	PUNCT
fcis-6025	108	1	[	[	X
fcis-6025	108	2	5	5	X
fcis-6025	108	3	]	]	X
fcis-6025	108	4	han	han	PROPN
fcis-6025	108	5	wj	wj	PROPN
fcis-6025	108	6	,	,	PUNCT
fcis-6025	108	7	li	li	PROPN
fcis-6025	108	8	hf	hf	PROPN
fcis-6025	108	9	,	,	PUNCT
fcis-6025	108	10	ruan	ruan	PROPN
fcis-6025	108	11	hb	hb	PROPN
fcis-6025	108	12	,	,	PUNCT
fcis-6025	108	13	et	et	PROPN
fcis-6025	108	14	al	al	PROPN
fcis-6025	108	15	.	.	PUNCT
fcis-6025	109	1	a	a	DET
fcis-6025	109	2	review	review	NOUN
fcis-6025	109	3	of	of	ADP
fcis-6025	109	4	research	research	NOUN
fcis-6025	109	5	advances	advance	NOUN
fcis-6025	109	6	in	in	ADP
fcis-6025	109	7	speech	speech	NOUN
fcis-6025	109	8	emotion	emotion	NOUN
fcis-6025	109	9	recognition[j	recognition[j	NOUN
fcis-6025	109	10	]	]	PUNCT
fcis-6025	109	11	.	.	PUNCT
fcis-6025	110	1	journal	journal	PROPN
fcis-6025	110	2	of	of	ADP
fcis-6025	110	3	software,2019,25(1):37	software,2019,25(1):37	NOUN
fcis-6025	110	4	-	-	SYM
fcis-6025	110	5	50	50	NUM
fcis-6025	110	6	.	.	PUNCT
fcis-6025	111	1	[	[	X
fcis-6025	111	2	6	6	NUM
fcis-6025	111	3	]	]	PUNCT
fcis-6025	111	4	qiu	qiu	PROPN
fcis-6025	111	5	xue	xue	PROPN
fcis-6025	111	6	.	.	PUNCT
fcis-6025	112	1	research	research	NOUN
fcis-6025	112	2	on	on	ADP
fcis-6025	112	3	multimodal	multimodal	ADJ
fcis-6025	112	4	fusion	fusion	NOUN
fcis-6025	112	5	for	for	ADP
fcis-6025	112	6	student	student	NOUN
fcis-6025	112	7	emotion	emotion	NOUN
fcis-6025	112	8	recognition	recognition	NOUN
fcis-6025	112	9	and	and	CCONJ
fcis-6025	112	10	state	state	NOUN
fcis-6025	112	11	analysis[d	analysis[d	PROPN
fcis-6025	112	12	]	]	PUNCT
fcis-6025	112	13	.	.	PUNCT
fcis-6025	113	1	supervisor	supervisor	NOUN
fcis-6025	113	2	:	:	PUNCT
fcis-6025	113	3	li	li	PROPN
fcis-6025	113	4	mingyong	mingyong	PROPN
fcis-6025	113	5	.	.	PUNCT
fcis-6025	114	1	chongqing	chongqe	VERB
fcis-6025	114	2	normal	normal	ADJ
fcis-6025	114	3	university,2021	university,2021	NOUN
fcis-6025	114	4	.	.	PUNCT
fcis-6025	115	1	[	[	X
fcis-6025	115	2	7	7	X
fcis-6025	115	3	]	]	X
fcis-6025	115	4	gao	gao	PROPN
fcis-6025	115	5	l	l	PROPN
fcis-6025	115	6	,	,	PUNCT
fcis-6025	115	7	qi	qi	PROPN
fcis-6025	115	8	l	l	PROPN
fcis-6025	115	9	,	,	PUNCT
fcis-6025	115	10	chen	chen	PROPN
fcis-6025	115	11	e	e	PROPN
fcis-6025	115	12	,	,	PUNCT
fcis-6025	115	13	et	et	PROPN
fcis-6025	115	14	al	al	PROPN
fcis-6025	115	15	.	.	PUNCT
fcis-6025	115	16	discriminative	discriminative	VERB
fcis-6025	115	17	multiple	multiple	ADJ
fcis-6025	115	18	canonical	canonical	ADJ
fcis-6025	115	19	correlation	correlation	NOUN
fcis-6025	115	20	analysis	analysis	NOUN
fcis-6025	115	21	for	for	ADP
fcis-6025	115	22	information	information	NOUN
fcis-6025	115	23	fusion[j	fusion[j	NOUN
fcis-6025	115	24	]	]	PUNCT
fcis-6025	115	25	.	.	PUNCT
fcis-6025	116	1	ieee	ieee	NOUN
fcis-6025	116	2	transactions	transaction	NOUN
fcis-6025	116	3	on	on	ADP
fcis-6025	116	4	image	image	NOUN
fcis-6025	116	5	processing	processing	NOUN
fcis-6025	116	6	,	,	PUNCT
fcis-6025	116	7	2017	2017	NUM
fcis-6025	116	8	,	,	PUNCT
fcis-6025	116	9	27(4	27(4	NUM
fcis-6025	116	10	):	):	PUNCT
fcis-6025	116	11	1951	1951	NUM
fcis-6025	116	12	-	-	SYM
fcis-6025	116	13	1965	1965	NUM
fcis-6025	116	14	.	.	PUNCT
fcis-6025	117	1	[	[	X
fcis-6025	117	2	8	8	NUM
fcis-6025	117	3	]	]	X
fcis-6025	117	4	new	new	PROPN
fcis-6025	117	5	t	t	PROPN
fcis-6025	117	6	l	l	PROPN
fcis-6025	117	7	,	,	PUNCT
fcis-6025	117	8	foo	foo	PROPN
fcis-6025	117	9	s	s	PART
fcis-6025	117	10	w	w	PROPN
fcis-6025	117	11	,	,	PUNCT
fcis-6025	117	12	de	de	PROPN
fcis-6025	117	13	silva	silva	PROPN
fcis-6025	117	14	lc	lc	PROPN
fcis-6025	117	15	.	.	PROPN
fcis-6025	117	16	speech	speech	NOUN
fcis-6025	117	17	emotion	emotion	NOUN
fcis-6025	117	18	recognition	recognition	NOUN
fcis-6025	117	19	using	use	VERB
fcis-6025	117	20	hidden	hide	VERB
fcis-6025	117	21	markov	markov	NOUN
fcis-6025	117	22	models[j	models[j	PROPN
fcis-6025	117	23	]	]	PUNCT
fcis-6025	117	24	.	.	PUNCT
fcis-6025	118	1	speech	speech	NOUN
fcis-6025	118	2	communication	communication	NOUN
fcis-6025	118	3	,	,	PUNCT
fcis-6025	118	4	2003	2003	NUM
fcis-6025	118	5	,	,	PUNCT
fcis-6025	118	6	41(4	41(4	NUM
fcis-6025	118	7	):	):	PUNCT
fcis-6025	118	8	603	603	NUM
fcis-6025	118	9	-	-	SYM
fcis-6025	118	10	623	623	NUM
fcis-6025	118	11	.	.	PUNCT
fcis-6025	119	1	[	[	X
fcis-6025	119	2	9	9	NUM
fcis-6025	119	3	]	]	X
fcis-6025	119	4	liu	liu	PROPN
fcis-6025	119	5	,	,	PUNCT
fcis-6025	119	6	jingjing	jingjing	PROPN
fcis-6025	119	7	,	,	PUNCT
fcis-6025	119	8	wu	wu	PROPN
fcis-6025	119	9	,	,	PUNCT
fcis-6025	119	10	xiaofeng	xiaofeng	PROPN
fcis-6025	119	11	.	.	PUNCT
fcis-6025	119	12	multimodal	multimodal	ADJ
fcis-6025	119	13	sentiment	sentiment	NOUN
fcis-6025	119	14	recognition	recognition	NOUN
fcis-6025	119	15	and	and	CCONJ
fcis-6025	119	16	spatial	spatial	ADJ
fcis-6025	119	17	annotation	annotation	NOUN
fcis-6025	119	18	based	base	VERB
fcis-6025	119	19	on	on	ADP
fcis-6025	119	20	long	long	ADJ
fcis-6025	119	21	and	and	CCONJ
fcis-6025	119	22	short	short	ADJ
fcis-6025	119	23	term	term	NOUN
fcis-6025	119	24	memory	memory	NOUN
fcis-6025	119	25	networks	network	NOUN
fcis-6025	119	26	[	[	X
fcis-6025	119	27	j	j	X
fcis-6025	119	28	]	]	X
fcis-6025	119	29	.	.	PUNCT
fcis-6025	120	1	journal	journal	PROPN
fcis-6025	120	2	of	of	ADP
fcis-6025	120	3	fudan	fudan	PROPN
fcis-6025	120	4	(	(	PUNCT
fcis-6025	120	5	natural	natural	ADJ
fcis-6025	120	6	science	science	NOUN
fcis-6025	120	7	edition),2020,59(05):565	edition),2020,59(05):565	NOUN
fcis-6025	120	8	-	-	PUNCT
fcis-6025	120	9	574	574	NUM
fcis-6025	120	10	.	.	PUNCT
fcis-6025	121	1	[	[	X
fcis-6025	121	2	10	10	NUM
fcis-6025	121	3	]	]	X
fcis-6025	121	4	chen	chen	PROPN
fcis-6025	121	5	jia	jia	PROPN
fcis-6025	121	6	.	.	PROPN
fcis-6025	121	7	speech	speech	PROPN
fcis-6025	121	8	emotion	emotion	NOUN
fcis-6025	121	9	recognition	recognition	NOUN
fcis-6025	121	10	based	base	VERB
fcis-6025	121	11	on	on	ADP
fcis-6025	121	12	deep	deep	ADJ
fcis-6025	121	13	learning	learning	NOUN
fcis-6025	122	1	[	[	X
fcis-6025	122	2	d	d	X
fcis-6025	122	3	]	]	X
fcis-6025	122	4	.	.	PUNCT
fcis-6025	123	1	nanjing	nanjing	PROPN
fcis-6025	123	2	university	university	PROPN
fcis-6025	123	3	of	of	ADP
fcis-6025	123	4	posts	post	NOUN
fcis-6025	123	5	and	and	CCONJ
fcis-6025	123	6	telecommunications	telecommunication	NOUN
fcis-6025	123	7	,	,	PUNCT
fcis-6025	123	8	2019	2019	NUM
fcis-6025	123	9	.	.	PUNCT
fcis-6025	123	10	000586	000586	NUM
fcis-6025	123	11	.	.	PUNCT