id	sid	tid	token	lemma	pos
fcis-3178	1	1	frontiers	frontier	NOUN
fcis-3178	1	2	in	in	ADP
fcis-3178	1	3	computing	computing	NOUN
fcis-3178	1	4	and	and	CCONJ
fcis-3178	1	5	intelligent	intelligent	ADJ
fcis-3178	1	6	systems	system	NOUN
fcis-3178	1	7	issn	issn	VERB
fcis-3178	1	8	:	:	PUNCT
fcis-3178	1	9	2832	2832	NUM
fcis-3178	1	10	-	-	SYM
fcis-3178	1	11	6024	6024	NUM
fcis-3178	1	12	|	|	NOUN
fcis-3178	1	13	vol	vol	NOUN
fcis-3178	1	14	.	.	PROPN
fcis-3178	2	1	2	2	NUM
fcis-3178	2	2	,	,	PUNCT
fcis-3178	2	3	no	no	INTJ
fcis-3178	2	4	.	.	NOUN
fcis-3178	2	5	1	1	NUM
fcis-3178	2	6	,	,	PUNCT
fcis-3178	2	7	2022	2022	NUM
fcis-3178	2	8	116	116	NUM
fcis-3178	2	9	a	a	DET
fcis-3178	2	10	facial	facial	ADJ
fcis-3178	2	11	expression	expression	NOUN
fcis-3178	2	12	recognition	recognition	NOUN
fcis-3178	2	13	method	method	NOUN
fcis-3178	2	14	based	base	VERB
fcis-3178	2	15	on	on	ADP
fcis-3178	2	16	convolutional	convolutional	ADJ
fcis-3178	2	17	neural	neural	ADJ
fcis-3178	2	18	network	network	NOUN
fcis-3178	2	19	hongbin	hongbin	PROPN
fcis-3178	2	20	huang	huang	PROPN
fcis-3178	2	21	*	*	PROPN
fcis-3178	2	22	school	school	PROPN
fcis-3178	2	23	of	of	ADP
fcis-3178	2	24	computer	computer	NOUN
fcis-3178	2	25	engineering	engineering	NOUN
fcis-3178	2	26	,	,	PUNCT
fcis-3178	2	27	jimei	jimei	PROPN
fcis-3178	2	28	university	university	PROPN
fcis-3178	2	29	,	,	PUNCT
fcis-3178	2	30	xiamen	xiamen	PROPN
fcis-3178	2	31	,	,	PUNCT
fcis-3178	2	32	fujian,361021	fujian,361021	NOUN
fcis-3178	2	33	,	,	PUNCT
fcis-3178	2	34	china	china	PROPN
fcis-3178	2	35	*	*	PUNCT
fcis-3178	2	36	corresponding	correspond	VERB
fcis-3178	2	37	author	author	NOUN
fcis-3178	2	38	:	:	PUNCT
fcis-3178	2	39	email	email	NOUN
fcis-3178	2	40	:	:	PUNCT
fcis-3178	2	41	15980839993@163.com	15980839993@163.com	X
fcis-3178	2	42	abstract	abstract	NOUN
fcis-3178	2	43	:	:	PUNCT
fcis-3178	2	44	this	this	DET
fcis-3178	2	45	project	project	NOUN
fcis-3178	2	46	is	be	AUX
fcis-3178	2	47	the	the	DET
fcis-3178	2	48	implementation	implementation	NOUN
fcis-3178	2	49	method	method	NOUN
fcis-3178	2	50	of	of	ADP
fcis-3178	2	51	a	a	DET
fcis-3178	2	52	simple	simple	ADJ
fcis-3178	2	53	static	static	ADJ
fcis-3178	2	54	facial	facial	ADJ
fcis-3178	2	55	expression	expression	NOUN
fcis-3178	2	56	recognition	recognition	NOUN
fcis-3178	2	57	(	(	PUNCT
fcis-3178	2	58	fer	fer	PROPN
fcis-3178	2	59	)	)	PUNCT
fcis-3178	2	60	project	project	NOUN
fcis-3178	2	61	.	.	PUNCT
fcis-3178	3	1	under	under	ADP
fcis-3178	3	2	the	the	DET
fcis-3178	3	3	environment	environment	NOUN
fcis-3178	3	4	of	of	ADP
fcis-3178	3	5	python3	python3	PROPN
fcis-3178	3	6	,	,	PUNCT
fcis-3178	3	7	the	the	DET
fcis-3178	3	8	deep	deep	ADJ
fcis-3178	3	9	learning	learning	NOUN
fcis-3178	3	10	model	model	NOUN
fcis-3178	3	11	is	be	AUX
fcis-3178	3	12	used	use	VERB
fcis-3178	3	13	to	to	PART
fcis-3178	3	14	compare	compare	VERB
fcis-3178	3	15	with	with	ADP
fcis-3178	3	16	the	the	DET
fcis-3178	3	17	traditional	traditional	ADJ
fcis-3178	3	18	model	model	NOUN
fcis-3178	3	19	.	.	PUNCT
fcis-3178	4	1	finally	finally	ADV
fcis-3178	4	2	,	,	PUNCT
fcis-3178	4	3	cnn	cnn	PROPN
fcis-3178	4	4	(	(	PUNCT
fcis-3178	4	5	convolutional	convolutional	ADJ
fcis-3178	4	6	neural	neural	ADJ
fcis-3178	4	7	networks	network	NOUN
fcis-3178	4	8	)	)	PUNCT
fcis-3178	4	9	is	be	AUX
fcis-3178	4	10	actually	actually	ADV
fcis-3178	4	11	used	use	VERB
fcis-3178	4	12	to	to	PART
fcis-3178	4	13	construct	construct	VERB
fcis-3178	4	14	the	the	DET
fcis-3178	4	15	entire	entire	ADJ
fcis-3178	4	16	system	system	NOUN
fcis-3178	4	17	,	,	PUNCT
fcis-3178	4	18	and	and	CCONJ
fcis-3178	4	19	model	model	NOUN
fcis-3178	4	20	evaluation	evaluation	NOUN
fcis-3178	4	21	is	be	AUX
fcis-3178	4	22	carried	carry	VERB
fcis-3178	4	23	out	out	ADP
fcis-3178	4	24	on	on	ADP
fcis-3178	4	25	the	the	DET
fcis-3178	4	26	three	three	NUM
fcis-3178	4	27	fer	fer	PROPN
fcis-3178	4	28	datasets	dataset	NOUN
fcis-3178	4	29	fer2013	fer2013	ADJ
fcis-3178	4	30	,	,	PUNCT
fcis-3178	4	31	jaffe	jaffe	PROPN
fcis-3178	4	32	and	and	CCONJ
fcis-3178	4	33	ck+	ck+	PROPN
fcis-3178	4	34	.	.	PUNCT
fcis-3178	5	1	the	the	DET
fcis-3178	5	2	project	project	NOUN
fcis-3178	5	3	's	's	PART
fcis-3178	5	4	main	main	ADJ
fcis-3178	5	5	functions	function	NOUN
fcis-3178	5	6	include	include	VERB
fcis-3178	5	7	"	"	PUNCT
fcis-3178	5	8	get	get	VERB
fcis-3178	5	9	a	a	DET
fcis-3178	5	10	picture	picture	NOUN
fcis-3178	5	11	of	of	ADP
fcis-3178	5	12	a	a	DET
fcis-3178	5	13	person	person	NOUN
fcis-3178	5	14	's	's	PART
fcis-3178	5	15	face	face	NOUN
fcis-3178	5	16	"	"	PUNCT
fcis-3178	5	17	and	and	CCONJ
fcis-3178	5	18	"	"	PUNCT
fcis-3178	5	19	recognize	recognize	VERB
fcis-3178	5	20	expressions	expression	NOUN
fcis-3178	5	21	"	"	PUNCT
fcis-3178	5	22	.	.	PUNCT
fcis-3178	6	1	the	the	DET
fcis-3178	6	2	principles	principle	NOUN
fcis-3178	6	3	of	of	ADP
fcis-3178	6	4	the	the	DET
fcis-3178	6	5	above	above	ADJ
fcis-3178	6	6	functions	function	NOUN
fcis-3178	6	7	include	include	VERB
fcis-3178	6	8	the	the	DET
fcis-3178	6	9	following	following	NOUN
fcis-3178	6	10	:	:	PUNCT
fcis-3178	6	11	the	the	DET
fcis-3178	6	12	establishment	establishment	NOUN
fcis-3178	6	13	of	of	ADP
fcis-3178	6	14	neural	neural	ADJ
fcis-3178	6	15	network	network	NOUN
fcis-3178	6	16	structure	structure	NOUN
fcis-3178	6	17	,	,	PUNCT
fcis-3178	6	18	the	the	DET
fcis-3178	6	19	acquisition	acquisition	NOUN
fcis-3178	6	20	of	of	ADP
fcis-3178	6	21	data	datum	NOUN
fcis-3178	6	22	sets	set	NOUN
fcis-3178	6	23	,	,	PUNCT
fcis-3178	6	24	the	the	DET
fcis-3178	6	25	model	model	NOUN
fcis-3178	6	26	training	training	NOUN
fcis-3178	6	27	based	base	VERB
fcis-3178	6	28	on	on	ADP
fcis-3178	6	29	data	data	NOUN
fcis-3178	6	30	sets	set	NOUN
fcis-3178	6	31	,	,	PUNCT
fcis-3178	6	32	the	the	DET
fcis-3178	6	33	use	use	NOUN
fcis-3178	6	34	of	of	ADP
fcis-3178	6	35	opencv	opencv	PROPN
fcis-3178	6	36	to	to	PART
fcis-3178	6	37	obtain	obtain	VERB
fcis-3178	6	38	the	the	DET
fcis-3178	6	39	face	face	NOUN
fcis-3178	6	40	,	,	PUNCT
fcis-3178	6	41	and	and	CCONJ
fcis-3178	6	42	the	the	DET
fcis-3178	6	43	use	use	NOUN
fcis-3178	6	44	of	of	ADP
fcis-3178	6	45	the	the	DET
fcis-3178	6	46	model	model	NOUN
fcis-3178	6	47	to	to	PART
fcis-3178	6	48	recognize	recognize	VERB
fcis-3178	6	49	the	the	DET
fcis-3178	6	50	expression	expression	NOUN
fcis-3178	6	51	.	.	PUNCT
fcis-3178	7	1	the	the	DET
fcis-3178	7	2	experimental	experimental	ADJ
fcis-3178	7	3	results	result	NOUN
fcis-3178	7	4	reproduce	reproduce	VERB
fcis-3178	7	5	the	the	DET
fcis-3178	7	6	simple	simple	ADJ
fcis-3178	7	7	deep	deep	ADJ
fcis-3178	7	8	learning	learning	NOUN
fcis-3178	7	9	model	model	NOUN
fcis-3178	7	10	to	to	PART
fcis-3178	7	11	realize	realize	VERB
fcis-3178	7	12	fer	fer	PROPN
fcis-3178	7	13	and	and	CCONJ
fcis-3178	7	14	verify	verify	VERB
fcis-3178	7	15	the	the	DET
fcis-3178	7	16	different	different	ADJ
fcis-3178	7	17	effects	effect	NOUN
fcis-3178	7	18	of	of	ADP
fcis-3178	7	19	different	different	ADJ
fcis-3178	7	20	data	datum	NOUN
fcis-3178	7	21	sets	set	NOUN
fcis-3178	7	22	on	on	ADP
fcis-3178	7	23	the	the	DET
fcis-3178	7	24	results	result	NOUN
fcis-3178	7	25	.	.	PUNCT
fcis-3178	8	1	face	face	NOUN
fcis-3178	8	2	recognition	recognition	NOUN
fcis-3178	8	3	has	have	AUX
fcis-3178	8	4	been	be	AUX
fcis-3178	8	5	widely	widely	ADV
fcis-3178	8	6	used	use	VERB
fcis-3178	8	7	in	in	ADP
fcis-3178	8	8	all	all	DET
fcis-3178	8	9	aspects	aspect	NOUN
fcis-3178	8	10	of	of	ADP
fcis-3178	8	11	life	life	NOUN
fcis-3178	8	12	,	,	PUNCT
fcis-3178	8	13	but	but	CCONJ
fcis-3178	8	14	different	different	ADJ
fcis-3178	8	15	purposes	purpose	NOUN
fcis-3178	8	16	need	need	VERB
fcis-3178	8	17	different	different	ADJ
fcis-3178	8	18	models	model	NOUN
fcis-3178	8	19	and	and	CCONJ
fcis-3178	8	20	data	datum	NOUN
fcis-3178	8	21	collection	collection	NOUN
fcis-3178	8	22	methods	method	NOUN
fcis-3178	8	23	,	,	PUNCT
fcis-3178	8	24	and	and	CCONJ
fcis-3178	8	25	the	the	DET
fcis-3178	8	26	laboratory	laboratory	NOUN
fcis-3178	8	27	collection	collection	NOUN
fcis-3178	8	28	cost	cost	NOUN
fcis-3178	8	29	is	be	AUX
fcis-3178	8	30	high	high	ADJ
fcis-3178	8	31	,	,	PUNCT
fcis-3178	8	32	and	and	CCONJ
fcis-3178	8	33	the	the	DET
fcis-3178	8	34	amount	amount	NOUN
fcis-3178	8	35	of	of	ADP
fcis-3178	8	36	data	datum	NOUN
fcis-3178	8	37	is	be	AUX
fcis-3178	8	38	limited	limited	ADJ
fcis-3178	8	39	,	,	PUNCT
fcis-3178	8	40	so	so	SCONJ
fcis-3178	8	41	this	this	DET
fcis-3178	8	42	project	project	NOUN
fcis-3178	8	43	also	also	ADV
fcis-3178	8	44	discusses	discuss	VERB
fcis-3178	8	45	the	the	DET
fcis-3178	8	46	data	datum	NOUN
fcis-3178	8	47	set	set	VERB
fcis-3178	8	48	selection	selection	NOUN
fcis-3178	8	49	methods	method	NOUN
fcis-3178	8	50	under	under	ADP
fcis-3178	8	51	different	different	ADJ
fcis-3178	8	52	purposes	purpose	NOUN
fcis-3178	8	53	.	.	PUNCT
fcis-3178	9	1	keywords	keyword	NOUN
fcis-3178	9	2	:	:	PUNCT
fcis-3178	10	1	cnn	cnn	PROPN
fcis-3178	10	2	;	;	PUNCT
fcis-3178	10	3	fer	fer	PROPN
fcis-3178	10	4	;	;	PUNCT
fcis-3178	10	5	jaffe	jaffe	PROPN
fcis-3178	10	6	;	;	PUNCT
fcis-3178	10	7	ck+	ck+	NOUN
fcis-3178	10	8	;	;	PUNCT
fcis-3178	10	9	fer2013	fer2013	ADJ
fcis-3178	10	10	.	.	PUNCT
fcis-3178	11	1	1	1	X
fcis-3178	11	2	.	.	X
fcis-3178	11	3	introduction	introduction	NOUN
fcis-3178	11	4	1.1	1.1	NUM
fcis-3178	11	5	.	.	PUNCT
fcis-3178	12	1	facial	facial	ADJ
fcis-3178	12	2	expression	expression	NOUN
fcis-3178	12	3	recognition	recognition	NOUN
fcis-3178	12	4	facial	facial	ADJ
fcis-3178	12	5	expression	expression	NOUN
fcis-3178	12	6	recognition	recognition	NOUN
fcis-3178	12	7	(	(	PUNCT
fcis-3178	12	8	fer	fer	PROPN
fcis-3178	12	9	)	)	PUNCT
fcis-3178	12	10	is	be	AUX
fcis-3178	12	11	one	one	NUM
fcis-3178	12	12	of	of	ADP
fcis-3178	12	13	the	the	DET
fcis-3178	12	14	most	most	ADV
fcis-3178	12	15	effective	effective	ADJ
fcis-3178	12	16	,	,	PUNCT
fcis-3178	12	17	natural	natural	ADJ
fcis-3178	12	18	,	,	PUNCT
fcis-3178	12	19	and	and	CCONJ
fcis-3178	12	20	universal	universal	ADJ
fcis-3178	12	21	human	human	ADJ
fcis-3178	12	22	cues	cue	NOUN
fcis-3178	12	23	for	for	ADP
fcis-3178	12	24	conveying	convey	VERB
fcis-3178	12	25	emotional	emotional	ADJ
fcis-3178	12	26	reactions	reaction	NOUN
fcis-3178	12	27	and	and	CCONJ
fcis-3178	12	28	intentions	intention	NOUN
fcis-3178	12	29	.	.	PUNCT
fcis-3178	13	1	it	it	PRON
fcis-3178	13	2	is	be	AUX
fcis-3178	13	3	the	the	DET
fcis-3178	13	4	most	most	ADV
fcis-3178	13	5	natural	natural	ADJ
fcis-3178	13	6	way	way	NOUN
fcis-3178	13	7	to	to	PART
fcis-3178	13	8	express	express	VERB
fcis-3178	13	9	the	the	DET
fcis-3178	13	10	inner	inner	ADJ
fcis-3178	13	11	world	world	NOUN
fcis-3178	13	12	,	,	PUNCT
fcis-3178	13	13	and	and	CCONJ
fcis-3178	13	14	it	it	PRON
fcis-3178	13	15	plays	play	VERB
fcis-3178	13	16	a	a	DET
fcis-3178	13	17	crucial	crucial	ADJ
fcis-3178	13	18	role	role	NOUN
fcis-3178	13	19	in	in	ADP
fcis-3178	13	20	social	social	ADJ
fcis-3178	13	21	interactions	interaction	NOUN
fcis-3178	13	22	[	[	X
fcis-3178	13	23	1	1	NUM
fcis-3178	13	24	]	]	PUNCT
fcis-3178	13	25	.	.	PUNCT
fcis-3178	14	1	fer	fer	PROPN
fcis-3178	14	2	generally	generally	ADV
fcis-3178	14	3	includes	include	VERB
fcis-3178	14	4	static	static	ADJ
fcis-3178	14	5	image	image	NOUN
fcis-3178	14	6	fer	fer	NOUN
fcis-3178	14	7	and	and	CCONJ
fcis-3178	14	8	dynamic	dynamic	ADJ
fcis-3178	14	9	sequence	sequence	NOUN
fcis-3178	14	10	fer	fer	PROPN
fcis-3178	14	11	.	.	PUNCT
fcis-3178	15	1	the	the	DET
fcis-3178	15	2	former	former	ADJ
fcis-3178	15	3	one	one	NOUN
fcis-3178	15	4	generally	generally	ADV
fcis-3178	15	5	refers	refer	VERB
fcis-3178	15	6	to	to	ADP
fcis-3178	15	7	image	image	NOUN
fcis-3178	15	8	expression	expression	NOUN
fcis-3178	15	9	recognition	recognition	NOUN
fcis-3178	15	10	,	,	PUNCT
fcis-3178	15	11	and	and	CCONJ
fcis-3178	15	12	the	the	DET
fcis-3178	15	13	latter	latter	ADJ
fcis-3178	15	14	refers	refer	VERB
fcis-3178	15	15	to	to	ADP
fcis-3178	15	16	analysis	analysis	NOUN
fcis-3178	15	17	and	and	CCONJ
fcis-3178	15	18	modelling	modelling	NOUN
fcis-3178	15	19	based	base	VERB
fcis-3178	15	20	on	on	ADP
fcis-3178	15	21	video	video	NOUN
fcis-3178	15	22	sequence	sequence	NOUN
fcis-3178	15	23	.	.	PUNCT
fcis-3178	16	1	1.2	1.2	NUM
fcis-3178	16	2	.	.	X
fcis-3178	16	3	literature	literature	PROPN
fcis-3178	16	4	review	review	PROPN
fcis-3178	16	5	ekman	ekman	PROPN
fcis-3178	16	6	and	and	CCONJ
fcis-3178	16	7	friesen	friesen	PROPN
fcis-3178	16	8	named	name	VERB
fcis-3178	16	9	six	six	NUM
fcis-3178	16	10	fundamental	fundamental	ADJ
fcis-3178	16	11	emotions	emotion	NOUN
fcis-3178	16	12	in	in	ADP
fcis-3178	16	13	the	the	DET
fcis-3178	16	14	twentieth	twentieth	ADJ
fcis-3178	16	15	century	century	NOUN
fcis-3178	16	16	based	base	VERB
fcis-3178	16	17	on	on	ADP
fcis-3178	16	18	cross	cross	ADJ
fcis-3178	16	19	-	-	ADJ
fcis-3178	16	20	cultural	cultural	ADJ
fcis-3178	16	21	research	research	NOUN
fcis-3178	16	22	indicating	indicate	VERB
fcis-3178	16	23	that	that	SCONJ
fcis-3178	16	24	people	people	NOUN
fcis-3178	16	25	experience	experience	VERB
fcis-3178	16	26	certain	certain	ADJ
fcis-3178	16	27	basic	basic	ADJ
fcis-3178	16	28	emotions	emotion	NOUN
fcis-3178	16	29	in	in	ADP
fcis-3178	16	30	the	the	DET
fcis-3178	16	31	same	same	ADJ
fcis-3178	16	32	manner	manner	NOUN
fcis-3178	16	33	regardless	regardless	ADV
fcis-3178	16	34	of	of	ADP
fcis-3178	16	35	culture	culture	NOUN
fcis-3178	16	36	.	.	PUNCT
fcis-3178	17	1	[	[	X
fcis-3178	17	2	2	2	NUM
fcis-3178	17	3	]	]	PUNCT
fcis-3178	17	4	anger	anger	NOUN
fcis-3178	17	5	,	,	PUNCT
fcis-3178	17	6	disgust	disgust	NOUN
fcis-3178	17	7	,	,	PUNCT
fcis-3178	17	8	fear	fear	NOUN
fcis-3178	17	9	,	,	PUNCT
fcis-3178	17	10	happiness	happiness	NOUN
fcis-3178	17	11	,	,	PUNCT
fcis-3178	17	12	sadness	sadness	NOUN
fcis-3178	17	13	,	,	PUNCT
fcis-3178	17	14	and	and	CCONJ
fcis-3178	17	15	surprise	surprise	NOUN
fcis-3178	17	16	are	be	AUX
fcis-3178	17	17	the	the	DET
fcis-3178	17	18	classic	classic	ADJ
fcis-3178	17	19	facial	facial	ADJ
fcis-3178	17	20	expressions	expression	NOUN
fcis-3178	17	21	.	.	PUNCT
fcis-3178	18	1	subsequently	subsequently	ADV
fcis-3178	18	2	,	,	PUNCT
fcis-3178	18	3	contempt	contempt	NOUN
fcis-3178	18	4	was	be	AUX
fcis-3178	18	5	listed	list	VERB
fcis-3178	18	6	as	as	ADP
fcis-3178	18	7	one	one	NUM
fcis-3178	18	8	of	of	ADP
fcis-3178	18	9	the	the	DET
fcis-3178	18	10	fundamental	fundamental	ADJ
fcis-3178	18	11	emotions	emotion	NOUN
fcis-3178	18	12	.	.	PUNCT
fcis-3178	19	1	the	the	DET
fcis-3178	19	2	use	use	NOUN
fcis-3178	19	3	of	of	ADP
fcis-3178	19	4	cnn	cnn	PROPN
fcis-3178	19	5	set	set	VERB
fcis-3178	19	6	can	can	AUX
fcis-3178	19	7	outperform	outperform	VERB
fcis-3178	19	8	a	a	DET
fcis-3178	19	9	single	single	ADJ
fcis-3178	19	10	cnn	cnn	PROPN
fcis-3178	19	11	classifier	classifier	NOUN
fcis-3178	19	12	,	,	PUNCT
fcis-3178	19	13	which	which	PRON
fcis-3178	19	14	is	be	AUX
fcis-3178	19	15	often	often	ADV
fcis-3178	19	16	limited	limit	VERB
fcis-3178	19	17	by	by	ADP
fcis-3178	19	18	certain	certain	ADJ
fcis-3178	19	19	conditions	condition	NOUN
fcis-3178	19	20	in	in	ADP
fcis-3178	19	21	the	the	DET
fcis-3178	19	22	application	application	NOUN
fcis-3178	19	23	.	.	PUNCT
fcis-3178	20	1	while	while	SCONJ
fcis-3178	20	2	the	the	DET
fcis-3178	20	3	ensemble	ensemble	ADJ
fcis-3178	20	4	cnn	cnn	PROPN
fcis-3178	20	5	fuses	fuse	VERB
fcis-3178	20	6	the	the	DET
fcis-3178	20	7	discriminative	discriminative	NOUN
fcis-3178	20	8	information	information	NOUN
fcis-3178	20	9	of	of	ADP
fcis-3178	20	10	each	each	DET
fcis-3178	20	11	single	single	ADJ
fcis-3178	20	12	classifier	classifier	NOUN
fcis-3178	20	13	,	,	PUNCT
fcis-3178	20	14	it	it	PRON
fcis-3178	20	15	can	can	AUX
fcis-3178	20	16	realize	realize	VERB
fcis-3178	20	17	the	the	DET
fcis-3178	20	18	complementarities	complementarity	NOUN
fcis-3178	20	19	between	between	ADP
fcis-3178	20	20	the	the	DET
fcis-3178	20	21	advantages	advantage	NOUN
fcis-3178	20	22	and	and	CCONJ
fcis-3178	20	23	disadvantages	disadvantage	NOUN
fcis-3178	20	24	of	of	ADP
fcis-3178	20	25	each	each	DET
fcis-3178	20	26	classifier	classifier	NOUN
fcis-3178	20	27	.	.	PUNCT
fcis-3178	21	1	therefore	therefore	ADV
fcis-3178	21	2	,	,	PUNCT
fcis-3178	21	3	it	it	PRON
fcis-3178	21	4	is	be	AUX
fcis-3178	21	5	very	very	ADV
fcis-3178	21	6	important	important	ADJ
fcis-3178	21	7	to	to	PART
fcis-3178	21	8	find	find	VERB
fcis-3178	21	9	a	a	DET
fcis-3178	21	10	method	method	NOUN
fcis-3178	21	11	to	to	PART
fcis-3178	21	12	improve	improve	VERB
fcis-3178	21	13	the	the	DET
fcis-3178	21	14	classification	classification	NOUN
fcis-3178	21	15	performance	performance	NOUN
fcis-3178	21	16	.	.	PUNCT
fcis-3178	22	1	[	[	X
fcis-3178	22	2	3	3	X
fcis-3178	22	3	]	]	PUNCT
fcis-3178	22	4	in	in	ADP
fcis-3178	22	5	addition	addition	NOUN
fcis-3178	22	6	,	,	PUNCT
fcis-3178	22	7	the	the	DET
fcis-3178	22	8	facial	facial	ADJ
fcis-3178	22	9	micro	micro	NOUN
fcis-3178	22	10	-	-	NOUN
fcis-3178	22	11	expression	expression	NOUN
fcis-3178	22	12	can	can	AUX
fcis-3178	22	13	sometimes	sometimes	ADV
fcis-3178	22	14	express	express	VERB
fcis-3178	22	15	opposite	opposite	ADJ
fcis-3178	22	16	emotions	emotion	NOUN
fcis-3178	22	17	according	accord	VERB
fcis-3178	22	18	to	to	ADP
fcis-3178	22	19	psychological	psychological	ADJ
fcis-3178	22	20	changes	change	NOUN
fcis-3178	22	21	.	.	PUNCT
fcis-3178	23	1	it	it	PRON
fcis-3178	23	2	is	be	AUX
fcis-3178	23	3	necessary	necessary	ADJ
fcis-3178	23	4	to	to	PART
fcis-3178	23	5	accurately	accurately	ADV
fcis-3178	23	6	judge	judge	VERB
fcis-3178	23	7	the	the	DET
fcis-3178	23	8	emotions	emotion	NOUN
fcis-3178	23	9	to	to	PART
fcis-3178	23	10	be	be	AUX
fcis-3178	23	11	expressed	express	VERB
fcis-3178	23	12	by	by	ADP
fcis-3178	23	13	the	the	DET
fcis-3178	23	14	other	other	ADJ
fcis-3178	23	15	person	person	NOUN
fcis-3178	23	16	based	base	VERB
fcis-3178	23	17	on	on	ADP
fcis-3178	23	18	the	the	DET
fcis-3178	23	19	context	context	NOUN
fcis-3178	23	20	relationship	relationship	NOUN
fcis-3178	23	21	such	such	ADJ
fcis-3178	23	22	as	as	ADP
fcis-3178	23	23	body	body	NOUN
fcis-3178	23	24	movements	movement	NOUN
fcis-3178	23	25	,	,	PUNCT
fcis-3178	23	26	language	language	NOUN
fcis-3178	23	27	and	and	CCONJ
fcis-3178	23	28	events	event	NOUN
fcis-3178	23	29	.	.	PUNCT
fcis-3178	24	1	[	[	X
fcis-3178	24	2	4	4	X
fcis-3178	24	3	]	]	PUNCT
fcis-3178	24	4	this	this	DET
fcis-3178	24	5	paper	paper	NOUN
fcis-3178	24	6	presents	present	VERB
fcis-3178	24	7	a	a	DET
fcis-3178	24	8	fer	fer	PROPN
fcis-3178	24	9	method	method	NOUN
fcis-3178	24	10	based	base	VERB
fcis-3178	24	11	on	on	ADP
fcis-3178	24	12	cnn	cnn	PROPN
fcis-3178	24	13	,	,	PUNCT
fcis-3178	24	14	which	which	PRON
fcis-3178	24	15	can	can	AUX
fcis-3178	24	16	meet	meet	VERB
fcis-3178	24	17	the	the	DET
fcis-3178	24	18	requirements	requirement	NOUN
fcis-3178	24	19	of	of	ADP
fcis-3178	24	20	fer	fer	PROPN
fcis-3178	24	21	.	.	PUNCT
fcis-3178	25	1	this	this	DET
fcis-3178	25	2	research	research	NOUN
fcis-3178	25	3	can	can	AUX
fcis-3178	25	4	make	make	VERB
fcis-3178	25	5	fer	fer	PROPN
fcis-3178	25	6	applied	apply	VERB
fcis-3178	25	7	to	to	ADP
fcis-3178	25	8	many	many	ADJ
fcis-3178	25	9	fields	field	NOUN
fcis-3178	25	10	.	.	PUNCT
fcis-3178	26	1	2	2	X
fcis-3178	26	2	.	.	X
fcis-3178	26	3	materials	material	NOUN
fcis-3178	26	4	and	and	CCONJ
fcis-3178	26	5	methods	method	NOUN
fcis-3178	26	6	2.1	2.1	NUM
fcis-3178	26	7	.	.	PUNCT
fcis-3178	27	1	program	program	NOUN
fcis-3178	27	2	environment	environment	NOUN
fcis-3178	27	3	this	this	DET
fcis-3178	27	4	fer	fer	PROPN
fcis-3178	27	5	is	be	AUX
fcis-3178	27	6	based	base	VERB
fcis-3178	27	7	on	on	ADP
fcis-3178	27	8	python3	python3	PROPN
fcis-3178	27	9	and	and	CCONJ
fcis-3178	27	10	keras2	keras2	PROPN
fcis-3178	27	11	(	(	PUNCT
fcis-3178	27	12	tensorflow	tensorflow	NOUN
fcis-3178	27	13	backend	backend	NOUN
fcis-3178	27	14	)	)	PUNCT
fcis-3178	27	15	with	with	ADP
fcis-3178	27	16	the	the	DET
fcis-3178	27	17	following	follow	VERB
fcis-3178	27	18	installations	installation	NOUN
fcis-3178	27	19	(	(	PUNCT
fcis-3178	27	20	conda	conda	VERB
fcis-3178	27	21	virtual	virtual	ADJ
fcis-3178	27	22	environment	environment	NOUN
fcis-3178	27	23	is	be	AUX
fcis-3178	27	24	recommended	recommend	VERB
fcis-3178	27	25	)	)	PUNCT
fcis-3178	27	26	.	.	PUNCT
fcis-3178	28	1	environments	environment	NOUN
fcis-3178	28	2	:	:	PUNCT
fcis-3178	28	3	anaconda	anaconda	PROPN
fcis-3178	28	4	,	,	PUNCT
fcis-3178	28	5	python3.6	python3.6	NOUN
fcis-3178	28	6	,	,	PUNCT
fcis-3178	28	7	tensorflow3.x	tensorflow3.x	PROPN
fcis-3178	28	8	,	,	PUNCT
fcis-3178	28	9	pycharm	pycharm	NOUN
fcis-3178	28	10	;	;	PUNCT
fcis-3178	28	11	equipment	equipment	NOUN
fcis-3178	28	12	:	:	PUNCT
fcis-3178	28	13	personal	personal	ADJ
fcis-3178	28	14	computer	computer	NOUN
fcis-3178	28	15	pc	pc	NOUN
fcis-3178	28	16	,	,	PUNCT
fcis-3178	28	17	windows	window	VERB
fcis-3178	28	18	10	10	NUM
fcis-3178	28	19	;	;	PUNCT
fcis-3178	28	20	the	the	DET
fcis-3178	28	21	construction	construction	NOUN
fcis-3178	28	22	of	of	ADP
fcis-3178	28	23	the	the	DET
fcis-3178	28	24	model	model	NOUN
fcis-3178	28	25	mainly	mainly	ADV
fcis-3178	28	26	refers	refer	VERB
fcis-3178	28	27	to	to	ADP
fcis-3178	28	28	the	the	DET
fcis-3178	28	29	following	follow	VERB
fcis-3178	28	30	network	network	NOUN
fcis-3178	28	31	structure	structure	NOUN
fcis-3178	28	32	designed	design	VERB
fcis-3178	28	33	by	by	ADP
fcis-3178	28	34	going	go	VERB
fcis-3178	28	35	deeper	deep	ADJ
fcis-3178	28	36	.	.	PUNCT
fcis-3178	29	1	after	after	ADP
fcis-3178	29	2	the	the	DET
fcis-3178	29	3	input	input	NOUN
fcis-3178	29	4	layer	layer	NOUN
fcis-3178	29	5	,	,	PUNCT
fcis-3178	29	6	the	the	DET
fcis-3178	29	7	(	(	PUNCT
fcis-3178	29	8	1,1	1,1	NUM
fcis-3178	29	9	)	)	PUNCT
fcis-3178	29	10	convolutional	convolutional	ADJ
fcis-3178	29	11	layer	layer	NOUN
fcis-3178	29	12	is	be	AUX
fcis-3178	29	13	added	add	VERB
fcis-3178	29	14	to	to	PART
fcis-3178	29	15	increase	increase	VERB
fcis-3178	29	16	the	the	DET
fcis-3178	29	17	nonlinear	nonlinear	ADJ
fcis-3178	29	18	representation	representation	NOUN
fcis-3178	29	19	,	,	PUNCT
fcis-3178	29	20	and	and	CCONJ
fcis-3178	29	21	the	the	DET
fcis-3178	29	22	model	model	NOUN
fcis-3178	29	23	has	have	VERB
fcis-3178	29	24	a	a	DET
fcis-3178	29	25	shallow	shallow	ADJ
fcis-3178	29	26	level	level	NOUN
fcis-3178	29	27	with	with	ADP
fcis-3178	29	28	fewer	few	ADJ
fcis-3178	29	29	parameters	parameter	NOUN
fcis-3178	29	30	.	.	PUNCT
fcis-3178	30	1	2.2	2.2	NUM
fcis-3178	30	2	.	.	PUNCT
fcis-3178	30	3	data	datum	NOUN
fcis-3178	30	4	processing	processing	NOUN
fcis-3178	30	5	and	and	CCONJ
fcis-3178	30	6	analysis	analysis	NOUN
fcis-3178	30	7	in	in	ADP
fcis-3178	30	8	this	this	DET
fcis-3178	30	9	paper	paper	NOUN
fcis-3178	30	10	,	,	PUNCT
fcis-3178	30	11	the	the	DET
fcis-3178	30	12	datasets	dataset	NOUN
fcis-3178	30	13	used	use	VERB
fcis-3178	30	14	to	to	PART
fcis-3178	30	15	train	train	VERB
fcis-3178	30	16	the	the	DET
fcis-3178	30	17	model	model	NOUN
fcis-3178	30	18	are	be	AUX
fcis-3178	30	19	'	'	PUNCT
fcis-3178	30	20	ck+	ck+	NOUN
fcis-3178	30	21	'	'	PUNCT
fcis-3178	30	22	,	,	PUNCT
fcis-3178	30	23	'	'	PUNCT
fcis-3178	30	24	jaffe	jaffe	PROPN
fcis-3178	30	25	'	'	PUNCT
fcis-3178	30	26	and	and	CCONJ
fcis-3178	30	27	'	'	PUNCT
fcis-3178	30	28	fer2013	fer2013	ADJ
fcis-3178	30	29	'	'	PUNCT
fcis-3178	30	30	.	.	PUNCT
fcis-3178	31	1	we	we	PRON
fcis-3178	31	2	chose	choose	VERB
fcis-3178	31	3	these	these	DET
fcis-3178	31	4	three	three	NUM
fcis-3178	31	5	datasets	dataset	NOUN
fcis-3178	31	6	over	over	ADP
fcis-3178	31	7	a	a	DET
fcis-3178	31	8	single	single	ADJ
fcis-3178	31	9	one	one	NOUN
fcis-3178	31	10	mainly	mainly	ADV
fcis-3178	31	11	because	because	SCONJ
fcis-3178	31	12	a	a	DET
fcis-3178	31	13	single	single	ADJ
fcis-3178	31	14	dataset	dataset	NOUN
fcis-3178	31	15	is	be	AUX
fcis-3178	31	16	either	either	CCONJ
fcis-3178	31	17	too	too	ADV
fcis-3178	31	18	small	small	ADJ
fcis-3178	31	19	or	or	CCONJ
fcis-3178	31	20	inaccurate	inaccurate	ADJ
fcis-3178	31	21	,	,	PUNCT
fcis-3178	31	22	and	and	CCONJ
fcis-3178	31	23	the	the	DET
fcis-3178	31	24	trained	train	VERB
fcis-3178	31	25	model	model	NOUN
fcis-3178	31	26	is	be	AUX
fcis-3178	31	27	not	not	PART
fcis-3178	31	28	good	good	ADJ
fcis-3178	31	29	.	.	PUNCT
fcis-3178	32	1	therefore	therefore	ADV
fcis-3178	32	2	,	,	PUNCT
fcis-3178	32	3	we	we	PRON
fcis-3178	32	4	decided	decide	VERB
fcis-3178	32	5	to	to	PART
fcis-3178	32	6	take	take	VERB
fcis-3178	32	7	turns	turn	NOUN
fcis-3178	32	8	using	use	VERB
fcis-3178	32	9	three	three	NUM
fcis-3178	32	10	datasets	dataset	NOUN
fcis-3178	32	11	for	for	ADP
fcis-3178	32	12	training	training	NOUN
fcis-3178	32	13	and	and	CCONJ
fcis-3178	32	14	comparison	comparison	NOUN
fcis-3178	32	15	through	through	ADP
fcis-3178	32	16	several	several	ADJ
fcis-3178	32	17	trials	trial	NOUN
fcis-3178	32	18	,	,	PUNCT
fcis-3178	32	19	and	and	CCONJ
fcis-3178	32	20	tried	try	VERB
fcis-3178	32	21	to	to	PART
fcis-3178	32	22	analyse	analyse	VERB
fcis-3178	32	23	the	the	DET
fcis-3178	32	24	impact	impact	NOUN
fcis-3178	32	25	of	of	ADP
fcis-3178	32	26	different	different	ADJ
fcis-3178	32	27	datasets	dataset	NOUN
fcis-3178	32	28	on	on	ADP
fcis-3178	32	29	model	model	NOUN
fcis-3178	32	30	training	training	NOUN
fcis-3178	32	31	.	.	PUNCT
fcis-3178	33	1	in	in	ADP
fcis-3178	33	2	addition	addition	NOUN
fcis-3178	33	3	to	to	ADP
fcis-3178	33	4	the	the	DET
fcis-3178	33	5	dataset	dataset	NOUN
fcis-3178	33	6	selection	selection	NOUN
fcis-3178	33	7	,	,	PUNCT
fcis-3178	33	8	we	we	PRON
fcis-3178	33	9	also	also	ADV
fcis-3178	33	10	need	need	VERB
fcis-3178	33	11	to	to	PART
fcis-3178	33	12	preprocess	preprocess	VERB
fcis-3178	33	13	the	the	DET
fcis-3178	33	14	target	target	NOUN
fcis-3178	33	15	images	image	NOUN
fcis-3178	33	16	to	to	PART
fcis-3178	33	17	be	be	AUX
fcis-3178	33	18	recognized	recognize	VERB
fcis-3178	33	19	so	so	SCONJ
fcis-3178	33	20	that	that	SCONJ
fcis-3178	33	21	the	the	DET
fcis-3178	33	22	trained	train	VERB
fcis-3178	33	23	model	model	NOUN
fcis-3178	33	24	can	can	AUX
fcis-3178	33	25	be	be	AUX
fcis-3178	33	26	used	use	VERB
fcis-3178	33	27	.	.	PUNCT
fcis-3178	34	1	the	the	DET
fcis-3178	34	2	specific	specific	ADJ
fcis-3178	34	3	preprocessing	preprocessing	NOUN
fcis-3178	34	4	includes	include	VERB
fcis-3178	34	5	the	the	DET
fcis-3178	34	6	following	follow	VERB
fcis-3178	34	7	steps	step	NOUN
fcis-3178	34	8	:	:	PUNCT
fcis-3178	34	9	the	the	DET
fcis-3178	34	10	location	location	NOUN
fcis-3178	34	11	of	of	ADP
fcis-3178	34	12	the	the	DET
fcis-3178	34	13	face	face	NOUN
fcis-3178	34	14	,	,	PUNCT
fcis-3178	34	15	the	the	DET
fcis-3178	34	16	geometric	geometric	ADJ
fcis-3178	34	17	normalization	normalization	NOUN
fcis-3178	34	18	of	of	ADP
fcis-3178	34	19	the	the	DET
fcis-3178	34	20	face	face	NOUN
fcis-3178	34	21	region	region	NOUN
fcis-3178	34	22	and	and	CCONJ
fcis-3178	34	23	the	the	DET
fcis-3178	34	24	gray	gray	ADJ
fcis-3178	34	25	level	level	NOUN
fcis-3178	34	26	normalization	normalization	NOUN
fcis-3178	34	27	[	[	X
fcis-3178	34	28	5	5	NUM
fcis-3178	34	29	]	]	PUNCT
fcis-3178	34	30	.	.	PUNCT
fcis-3178	35	1	geometric	geometric	ADJ
fcis-3178	35	2	normalization	normalization	NOUN
fcis-3178	35	3	is	be	AUX
fcis-3178	35	4	to	to	PART
fcis-3178	35	5	determine	determine	VERB
fcis-3178	35	6	the	the	DET
fcis-3178	35	7	main	main	ADJ
fcis-3178	35	8	rectangular	rectangular	ADJ
fcis-3178	35	9	feature	feature	NOUN
fcis-3178	35	10	region	region	NOUN
fcis-3178	35	11	according	accord	VERB
fcis-3178	35	12	to	to	ADP
fcis-3178	35	13	the	the	DET
fcis-3178	35	14	sum	sum	NOUN
fcis-3178	35	15	of	of	ADP
fcis-3178	35	16	feature	feature	NOUN
fcis-3178	35	17	points	point	NOUN
fcis-3178	35	18	and	and	CCONJ
fcis-3178	35	19	geometric	geometric	ADJ
fcis-3178	35	20	model	model	NOUN
fcis-3178	35	21	of	of	ADP
fcis-3178	35	22	facial	facial	ADJ
fcis-3178	35	23	expression	expression	NOUN
fcis-3178	35	24	image	image	NOUN
fcis-3178	35	25	,	,	PUNCT
fcis-3178	35	26	and	and	CCONJ
fcis-3178	35	27	then	then	ADV
fcis-3178	35	28	cut	cut	VERB
fcis-3178	35	29	the	the	DET
fcis-3178	35	30	original	original	ADJ
fcis-3178	35	31	size	size	NOUN
fcis-3178	35	32	of	of	ADP
fcis-3178	35	33	256×256	256×256	NUM
fcis-3178	35	34	image	image	NOUN
fcis-3178	35	35	into	into	ADP
fcis-3178	35	36	150×100	150×100	NUM
fcis-3178	35	37	image	image	NOUN
fcis-3178	35	38	,	,	PUNCT
fcis-3178	35	39	so	so	SCONJ
fcis-3178	35	40	that	that	SCONJ
fcis-3178	35	41	the	the	DET
fcis-3178	35	42	size	size	NOUN
fcis-3178	35	43	of	of	ADP
fcis-3178	35	44	the	the	DET
fcis-3178	35	45	face	face	NOUN
fcis-3178	35	46	in	in	ADP
fcis-3178	35	47	the	the	DET
fcis-3178	35	48	image	image	NOUN
fcis-3178	35	49	is	be	AUX
fcis-3178	35	50	consistent	consistent	ADJ
fcis-3178	35	51	.	.	PUNCT
fcis-3178	36	1	117	117	NUM
fcis-3178	36	2	table	table	NOUN
fcis-3178	36	3	1	1	NUM
fcis-3178	36	4	.	.	PUNCT
fcis-3178	37	1	an	an	DET
fcis-3178	37	2	overview	overview	NOUN
fcis-3178	37	3	of	of	ADP
fcis-3178	37	4	the	the	DET
fcis-3178	37	5	facial	facial	ADJ
fcis-3178	37	6	expression	expression	NOUN
fcis-3178	37	7	datasets	dataset	NOUN
fcis-3178	37	8	database	database	NOUN
fcis-3178	37	9	samples	sample	VERB
fcis-3178	37	10	subject	subject	ADJ
fcis-3178	37	11	collection	collection	NOUN
fcis-3178	37	12	condition	condition	NOUN
fcis-3178	37	13	expression	expression	NOUN
fcis-3178	37	14	dirstribution	dirstribution	NOUN
fcis-3178	37	15	access	access	NOUN
fcis-3178	37	16	ck+	ck+	NOUN
fcis-3178	37	17	593	593	NUM
fcis-3178	37	18	image	image	NOUN
fcis-3178	37	19	sequences	sequence	NOUN
fcis-3178	37	20	123	123	NUM
fcis-3178	37	21	lab	lab	NOUN
fcis-3178	37	22	seven	seven	NUM
fcis-3178	37	23	basic	basic	ADJ
fcis-3178	37	24	expressions	expression	NOUN
fcis-3178	37	25	plus	plus	CCONJ
fcis-3178	37	26	contempt	contempt	NOUN
fcis-3178	37	27	http://www.consortium.ri.cmu	http://www.consortium.ri.cmu	NOUN
fcis-3178	37	28	.	.	PUNCT
fcis-3178	38	1	edu	edu	PROPN
fcis-3178	38	2	/	/	SYM
fcis-3178	38	3	ckagree/	ckagree/	PROPN
fcis-3178	38	4	jaffe	jaffe	PROPN
fcis-3178	38	5	213	213	NUM
fcis-3178	38	6	images	image	NOUN
fcis-3178	38	7	10	10	NUM
fcis-3178	38	8	lab	lab	NOUN
fcis-3178	38	9	seven	seven	NUM
fcis-3178	38	10	basic	basic	ADJ
fcis-3178	38	11	expressions	expression	NOUN
fcis-3178	38	12	http:/www.kasrl.org	http:/www.kasrl.org	X
fcis-3178	38	13	/	/	SYM
fcis-3178	38	14	jaffe.html	jaffe.html	PROPN
fcis-3178	38	15	fer-2013	fer-2013	NOUN
fcis-3178	38	16	35,887	35,887	NUM
fcis-3178	38	17	images	image	NOUN
fcis-3178	38	18	n	n	CCONJ
fcis-3178	38	19	/	/	SYM
fcis-3178	38	20	a	a	DET
fcis-3178	38	21	web	web	NOUN
fcis-3178	38	22	seven	seven	NUM
fcis-3178	38	23	basic	basic	ADJ
fcis-3178	38	24	expressions	expression	NOUN
fcis-3178	38	25	https://www.kaggle.com/c/chal	https://www.kaggle.com/c/chal	PROPN
fcis-3178	38	26	engesin	engesin	NOUN
fcis-3178	38	27	-	-	PUNCT
fcis-3178	38	28	representationlearning	representationlearning	NOUN
fcis-3178	38	29	-	-	NOUN
fcis-3178	38	30	facialexpression	facialexpression	NOUN
fcis-3178	38	31	2.3	2.3	NUM
fcis-3178	38	32	.	.	PUNCT
fcis-3178	39	1	theoretical	theoretical	ADJ
fcis-3178	39	2	analysis	analysis	NOUN
fcis-3178	39	3	of	of	ADP
fcis-3178	39	4	the	the	DET
fcis-3178	39	5	cnn	cnn	PROPN
fcis-3178	39	6	and	and	CCONJ
fcis-3178	39	7	vggnet	vggnet	PROPN
fcis-3178	39	8	2.3.1	2.3.1	PROPN
fcis-3178	39	9	.	.	PUNCT
fcis-3178	40	1	definition	definition	NOUN
fcis-3178	40	2	of	of	ADP
fcis-3178	40	3	cnn	cnn	PROPN
fcis-3178	40	4	convolutional	convolutional	ADJ
fcis-3178	40	5	neural	neural	ADJ
fcis-3178	40	6	network	network	NOUN
fcis-3178	40	7	(	(	PUNCT
fcis-3178	40	8	cnn	cnn	PROPN
fcis-3178	40	9	)	)	PUNCT
fcis-3178	40	10	structure	structure	NOUN
fcis-3178	40	11	is	be	AUX
fcis-3178	40	12	mainly	mainly	ADV
fcis-3178	40	13	composed	compose	VERB
fcis-3178	40	14	of	of	ADP
fcis-3178	40	15	convolutional	convolutional	ADJ
fcis-3178	40	16	layer	layer	NOUN
fcis-3178	40	17	,	,	PUNCT
fcis-3178	40	18	pooling	pool	VERB
fcis-3178	40	19	layer	layer	NOUN
fcis-3178	40	20	and	and	CCONJ
fcis-3178	40	21	fully	fully	ADV
fcis-3178	40	22	connected	connected	ADJ
fcis-3178	40	23	layer(fc)[6	layer(fc)[6	NOUN
fcis-3178	40	24	]	]	PUNCT
fcis-3178	40	25	.	.	PUNCT
fcis-3178	41	1	the	the	DET
fcis-3178	41	2	function	function	NOUN
fcis-3178	41	3	of	of	ADP
fcis-3178	41	4	the	the	DET
fcis-3178	41	5	convolution	convolution	NOUN
fcis-3178	41	6	layer	layer	NOUN
fcis-3178	41	7	is	be	AUX
fcis-3178	41	8	to	to	PART
fcis-3178	41	9	realize	realize	VERB
fcis-3178	41	10	the	the	DET
fcis-3178	41	11	feature	feature	NOUN
fcis-3178	41	12	extraction	extraction	NOUN
fcis-3178	41	13	composed	compose	VERB
fcis-3178	41	14	of	of	ADP
fcis-3178	41	15	some	some	DET
fcis-3178	41	16	convolution	convolution	NOUN
fcis-3178	41	17	kernels	kernel	NOUN
fcis-3178	41	18	,	,	PUNCT
fcis-3178	41	19	do	do	VERB
fcis-3178	41	20	convolution	convolution	NOUN
fcis-3178	41	21	operation	operation	NOUN
fcis-3178	41	22	on	on	ADP
fcis-3178	41	23	the	the	DET
fcis-3178	41	24	input	input	NOUN
fcis-3178	41	25	image	image	NOUN
fcis-3178	41	26	,	,	PUNCT
fcis-3178	41	27	add	add	VERB
fcis-3178	41	28	the	the	DET
fcis-3178	41	29	offset	offset	NOUN
fcis-3178	41	30	,	,	PUNCT
fcis-3178	41	31	and	and	CCONJ
fcis-3178	41	32	output	output	VERB
fcis-3178	41	33	the	the	DET
fcis-3178	41	34	result	result	NOUN
fcis-3178	41	35	to	to	ADP
fcis-3178	41	36	the	the	DET
fcis-3178	41	37	activation	activation	NOUN
fcis-3178	41	38	function	function	VERB
fcis-3178	41	39	to	to	PART
fcis-3178	41	40	obtain	obtain	VERB
fcis-3178	41	41	the	the	DET
fcis-3178	41	42	output	output	NOUN
fcis-3178	41	43	,	,	PUNCT
fcis-3178	41	44	which	which	PRON
fcis-3178	41	45	reduces	reduce	VERB
fcis-3178	41	46	the	the	DET
fcis-3178	41	47	number	number	NOUN
fcis-3178	41	48	of	of	ADP
fcis-3178	41	49	network	network	NOUN
fcis-3178	41	50	parameters	parameter	NOUN
fcis-3178	41	51	and	and	CCONJ
fcis-3178	41	52	reduces	reduce	VERB
fcis-3178	41	53	the	the	DET
fcis-3178	41	54	complexity	complexity	NOUN
fcis-3178	41	55	of	of	ADP
fcis-3178	41	56	parameter	parameter	NOUN
fcis-3178	41	57	selection	selection	NOUN
fcis-3178	41	58	.	.	PUNCT
fcis-3178	42	1	the	the	DET
fcis-3178	42	2	image	image	NOUN
fcis-3178	42	3	can	can	AUX
fcis-3178	42	4	be	be	AUX
fcis-3178	42	5	directly	directly	ADV
fcis-3178	42	6	used	use	VERB
fcis-3178	42	7	as	as	ADP
fcis-3178	42	8	the	the	DET
fcis-3178	42	9	network	network	NOUN
fcis-3178	42	10	input	input	NOUN
fcis-3178	42	11	,	,	PUNCT
fcis-3178	42	12	avoiding	avoid	VERB
fcis-3178	42	13	the	the	DET
fcis-3178	42	14	complex	complex	ADJ
fcis-3178	42	15	process	process	NOUN
fcis-3178	42	16	of	of	ADP
fcis-3178	42	17	feature	feature	NOUN
fcis-3178	42	18	extraction	extraction	NOUN
fcis-3178	42	19	and	and	CCONJ
fcis-3178	42	20	data	datum	NOUN
fcis-3178	42	21	reconstruction	reconstruction	NOUN
fcis-3178	42	22	in	in	ADP
fcis-3178	42	23	traditional	traditional	ADJ
fcis-3178	42	24	methods	method	NOUN
fcis-3178	42	25	.	.	PUNCT
fcis-3178	43	1	pooling	pool	VERB
fcis-3178	43	2	layers	layer	NOUN
fcis-3178	43	3	can	can	AUX
fcis-3178	43	4	be	be	AUX
fcis-3178	43	5	used	use	VERB
fcis-3178	43	6	to	to	PART
fcis-3178	43	7	preserve	preserve	VERB
fcis-3178	43	8	invariance	invariance	NOUN
fcis-3178	43	9	.	.	PUNCT
fcis-3178	44	1	the	the	DET
fcis-3178	44	2	role	role	NOUN
fcis-3178	44	3	of	of	ADP
fcis-3178	44	4	the	the	DET
fcis-3178	44	5	convolutional	convolutional	ADJ
fcis-3178	44	6	layer	layer	NOUN
fcis-3178	44	7	is	be	AUX
fcis-3178	44	8	to	to	PART
fcis-3178	44	9	detect	detect	VERB
fcis-3178	44	10	the	the	DET
fcis-3178	44	11	local	local	ADJ
fcis-3178	44	12	connections	connection	NOUN
fcis-3178	44	13	of	of	ADP
fcis-3178	44	14	the	the	DET
fcis-3178	44	15	features	feature	NOUN
fcis-3178	44	16	of	of	ADP
fcis-3178	44	17	the	the	DET
fcis-3178	44	18	previous	previous	ADJ
fcis-3178	44	19	layer	layer	NOUN
fcis-3178	44	20	to	to	PART
fcis-3178	44	21	achieve	achieve	VERB
fcis-3178	44	22	feature	feature	NOUN
fcis-3178	44	23	extraction	extraction	NOUN
fcis-3178	44	24	,	,	PUNCT
fcis-3178	44	25	while	while	SCONJ
fcis-3178	44	26	the	the	DET
fcis-3178	44	27	role	role	NOUN
fcis-3178	44	28	of	of	ADP
fcis-3178	44	29	the	the	DET
fcis-3178	44	30	pooling	pooling	NOUN
fcis-3178	44	31	layer	layer	NOUN
fcis-3178	44	32	is	be	AUX
fcis-3178	44	33	to	to	PART
fcis-3178	44	34	fuse	fuse	VERB
fcis-3178	44	35	the	the	DET
fcis-3178	44	36	similar	similar	ADJ
fcis-3178	44	37	features	feature	NOUN
fcis-3178	44	38	.	.	PUNCT
fcis-3178	45	1	pooling	pool	VERB
fcis-3178	45	2	layers	layer	NOUN
fcis-3178	45	3	are	be	AUX
fcis-3178	45	4	often	often	ADV
fcis-3178	45	5	used	use	VERB
fcis-3178	45	6	together	together	ADV
fcis-3178	45	7	with	with	ADP
fcis-3178	45	8	convolutional	convolutional	ADJ
fcis-3178	45	9	layers	layer	NOUN
fcis-3178	45	10	to	to	PART
fcis-3178	45	11	reduce	reduce	VERB
fcis-3178	45	12	the	the	DET
fcis-3178	45	13	size	size	NOUN
fcis-3178	45	14	by	by	ADP
fcis-3178	45	15	down	down	ADP
fcis-3178	45	16	sampling	sampling	NOUN
fcis-3178	45	17	,	,	PUNCT
fcis-3178	45	18	resulting	result	VERB
fcis-3178	45	19	in	in	ADP
fcis-3178	45	20	invariance	invariance	NOUN
fcis-3178	45	21	of	of	ADP
fcis-3178	45	22	the	the	DET
fcis-3178	45	23	features	feature	NOUN
fcis-3178	45	24	.	.	PUNCT
fcis-3178	46	1	[	[	X
fcis-3178	46	2	7	7	NUM
fcis-3178	46	3	]	]	PUNCT
fcis-3178	46	4	figure	figure	NOUN
fcis-3178	46	5	1	1	NUM
fcis-3178	46	6	.	.	PUNCT
fcis-3178	46	7	structure	structure	NOUN
fcis-3178	46	8	of	of	ADP
fcis-3178	46	9	cnn	cnn	PROPN
fcis-3178	46	10	used	use	VERB
fcis-3178	46	11	in	in	ADP
fcis-3178	46	12	this	this	DET
fcis-3178	46	13	paper	paper	NOUN
fcis-3178	46	14	2.3.2	2.3.2	NUM
fcis-3178	46	15	.	.	PUNCT
fcis-3178	47	1	vggnet	vggnet	NOUN
fcis-3178	47	2	the	the	DET
fcis-3178	47	3	vggnet	vggnet	NOUN
fcis-3178	47	4	used	use	VERB
fcis-3178	47	5	in	in	ADP
fcis-3178	47	6	this	this	DET
fcis-3178	47	7	paper	paper	NOUN
fcis-3178	47	8	was	be	AUX
fcis-3178	47	9	designed	design	VERB
fcis-3178	47	10	in	in	ADP
fcis-3178	47	11	2014	2014	NUM
fcis-3178	47	12	.	.	PUNCT
fcis-3178	48	1	according	accord	VERB
fcis-3178	48	2	to	to	ADP
fcis-3178	48	3	the	the	DET
fcis-3178	48	4	change	change	NOUN
fcis-3178	48	5	of	of	ADP
fcis-3178	48	6	the	the	DET
fcis-3178	48	7	feature	feature	NOUN
fcis-3178	48	8	map	map	NOUN
fcis-3178	48	9	size	size	NOUN
fcis-3178	48	10	,	,	PUNCT
fcis-3178	48	11	the	the	DET
fcis-3178	48	12	vgg16	vgg16	NOUN
fcis-3178	48	13	model	model	NOUN
fcis-3178	48	14	can	can	AUX
fcis-3178	48	15	be	be	AUX
fcis-3178	48	16	divided	divide	VERB
fcis-3178	48	17	into	into	ADP
fcis-3178	48	18	six	six	NUM
fcis-3178	48	19	parts	part	NOUN
fcis-3178	48	20	.	.	PUNCT
fcis-3178	49	1	the	the	DET
fcis-3178	49	2	side	side	NOUN
fcis-3178	49	3	length	length	NOUN
fcis-3178	49	4	of	of	ADP
fcis-3178	49	5	the	the	DET
fcis-3178	49	6	feature	feature	NOUN
fcis-3178	49	7	map	map	NOUN
fcis-3178	49	8	is	be	AUX
fcis-3178	49	9	reduced	reduce	VERB
fcis-3178	49	10	to	to	ADP
fcis-3178	49	11	1/2	1/2	NUM
fcis-3178	49	12	for	for	ADP
fcis-3178	49	13	each	each	DET
fcis-3178	49	14	pooling	pool	VERB
fcis-3178	49	15	operation	operation	NOUN
fcis-3178	49	16	,	,	PUNCT
fcis-3178	49	17	and	and	CCONJ
fcis-3178	49	18	the	the	DET
fcis-3178	49	19	other	other	ADJ
fcis-3178	49	20	operations	operation	NOUN
fcis-3178	49	21	do	do	AUX
fcis-3178	49	22	not	not	PART
fcis-3178	49	23	affect	affect	VERB
fcis-3178	49	24	the	the	DET
fcis-3178	49	25	feature	feature	NOUN
fcis-3178	49	26	map	map	NOUN
fcis-3178	49	27	size	size	NOUN
fcis-3178	49	28	.	.	PUNCT
fcis-3178	50	1	there	there	PRON
fcis-3178	50	2	is	be	VERB
fcis-3178	50	3	no	no	DET
fcis-3178	50	4	essential	essential	ADJ
fcis-3178	50	5	difference	difference	NOUN
fcis-3178	50	6	between	between	ADP
fcis-3178	50	7	vggnet16	vggnet16	NOUN
fcis-3178	50	8	and	and	CCONJ
fcis-3178	50	9	vggnet19	vggnet19	NOUN
fcis-3178	50	10	,	,	PUNCT
fcis-3178	50	11	only	only	ADV
fcis-3178	50	12	the	the	DET
fcis-3178	50	13	network	network	NOUN
fcis-3178	50	14	depth	depth	NOUN
fcis-3178	50	15	is	be	AUX
fcis-3178	50	16	different	different	ADJ
fcis-3178	50	17	,	,	PUNCT
fcis-3178	50	18	with	with	ADP
fcis-3178	50	19	the	the	DET
fcis-3178	50	20	former	former	ADJ
fcis-3178	50	21	having	have	VERB
fcis-3178	50	22	16	16	NUM
fcis-3178	50	23	layers	layer	NOUN
fcis-3178	50	24	(	(	PUNCT
fcis-3178	50	25	13	13	NUM
fcis-3178	50	26	convolutions	convolution	NOUN
fcis-3178	50	27	and	and	CCONJ
fcis-3178	50	28	3	3	NUM
fcis-3178	50	29	fully	fully	ADV
fcis-3178	50	30	connected	connected	ADJ
fcis-3178	50	31	layers	layer	NOUN
fcis-3178	50	32	)	)	PUNCT
fcis-3178	50	33	and	and	CCONJ
fcis-3178	50	34	the	the	DET
fcis-3178	50	35	latter	latter	ADJ
fcis-3178	50	36	having	have	VERB
fcis-3178	50	37	19	19	NUM
fcis-3178	50	38	layers	layer	NOUN
fcis-3178	50	39	(	(	PUNCT
fcis-3178	50	40	16	16	NUM
fcis-3178	50	41	convolutions	convolution	NOUN
fcis-3178	50	42	and	and	CCONJ
fcis-3178	50	43	3	3	NUM
fcis-3178	50	44	fully	fully	ADV
fcis-3178	50	45	connected	connected	ADJ
fcis-3178	50	46	layers	layer	NOUN
fcis-3178	50	47	)	)	PUNCT
fcis-3178	50	48	.	.	PUNCT
fcis-3178	51	1	[	[	X
fcis-3178	51	2	8	8	NUM
fcis-3178	51	3	]	]	PUNCT
fcis-3178	51	4	figure	figure	NOUN
fcis-3178	51	5	2	2	NUM
fcis-3178	51	6	.	.	PUNCT
fcis-3178	51	7	data	datum	NOUN
fcis-3178	51	8	configure	configure	NOUN
fcis-3178	51	9	in	in	ADP
fcis-3178	51	10	cnn	cnn	PROPN
fcis-3178	51	11	2.4	2.4	NUM
fcis-3178	51	12	.	.	PUNCT
fcis-3178	52	1	sampling	sampling	NOUN
fcis-3178	52	2	and	and	CCONJ
fcis-3178	52	3	implementation	implementation	NOUN
fcis-3178	52	4	we	we	PRON
fcis-3178	52	5	import	import	VERB
fcis-3178	52	6	three	three	NUM
fcis-3178	52	7	datasets	dataset	NOUN
fcis-3178	52	8	(	(	PUNCT
fcis-3178	52	9	jaffe	jaffe	PROPN
fcis-3178	52	10	,	,	PUNCT
fcis-3178	52	11	ck+	ck+	NOUN
fcis-3178	52	12	,	,	PUNCT
fcis-3178	52	13	and	and	CCONJ
fcis-3178	52	14	fer2013	fer2013	ADJ
fcis-3178	52	15	)	)	PUNCT
fcis-3178	52	16	and	and	CCONJ
fcis-3178	52	17	modify	modify	VERB
fcis-3178	52	18	the	the	DET
fcis-3178	52	19	model	model	NOUN
fcis-3178	52	20	until	until	SCONJ
fcis-3178	52	21	all	all	PRON
fcis-3178	52	22	are	be	AUX
fcis-3178	52	23	trained	train	VERB
fcis-3178	52	24	.	.	PUNCT
fcis-3178	53	1	normalized	normalize	VERB
fcis-3178	53	2	datasets	dataset	NOUN
fcis-3178	53	3	and	and	CCONJ
fcis-3178	53	4	three	three	NUM
fcis-3178	53	5	cnn	cnn	PROPN
fcis-3178	53	6	models	model	NOUN
fcis-3178	53	7	are	be	AUX
fcis-3178	53	8	trained	train	VERB
fcis-3178	53	9	.	.	PUNCT
fcis-3178	54	1	table	table	NOUN
fcis-3178	54	2	1	1	NUM
fcis-3178	54	3	compares	compare	VERB
fcis-3178	54	4	three	three	NUM
fcis-3178	54	5	cnn	cnn	PROPN
fcis-3178	54	6	subnetwork	subnetwork	NOUN
fcis-3178	54	7	setups	setup	NOUN
fcis-3178	54	8	.	.	PUNCT
fcis-3178	55	1	three	three	NUM
fcis-3178	55	2	networks	network	NOUN
fcis-3178	55	3	ensure	ensure	VERB
fcis-3178	55	4	network	network	NOUN
fcis-3178	55	5	diversity	diversity	NOUN
fcis-3178	55	6	.	.	PUNCT
fcis-3178	56	1	various	various	ADJ
fcis-3178	56	2	convolutional	convolutional	ADJ
fcis-3178	56	3	layers	layer	NOUN
fcis-3178	56	4	can	can	AUX
fcis-3178	56	5	learn	learn	VERB
fcis-3178	56	6	different	different	ADJ
fcis-3178	56	7	characteristics	characteristic	NOUN
fcis-3178	56	8	.	.	PUNCT
fcis-3178	57	1	cnn1	cnn1	PROPN
fcis-3178	57	2	to	to	ADP
fcis-3178	57	3	cnn3	cnn3	PROPN
fcis-3178	57	4	are	be	AUX
fcis-3178	57	5	three	three	NUM
fcis-3178	57	6	models	model	NOUN
fcis-3178	57	7	.	.	PUNCT
fcis-3178	58	1	cnn1	cnn1	PROPN
fcis-3178	58	2	controls	control	VERB
fcis-3178	58	3	the	the	DET
fcis-3178	58	4	range	range	NOUN
fcis-3178	58	5	of	of	ADP
fcis-3178	58	6	the	the	DET
fcis-3178	58	7	receptive	receptive	ADJ
fcis-3178	58	8	field	field	NOUN
fcis-3178	58	9	.	.	PUNCT
fcis-3178	59	1	if	if	SCONJ
fcis-3178	59	2	the	the	DET
fcis-3178	59	3	field	field	NOUN
fcis-3178	59	4	is	be	AUX
fcis-3178	59	5	too	too	ADV
fcis-3178	59	6	vast	vast	ADJ
fcis-3178	59	7	,	,	PUNCT
fcis-3178	59	8	noise	noise	NOUN
fcis-3178	59	9	will	will	AUX
fcis-3178	59	10	hamper	hamper	VERB
fcis-3178	59	11	performance	performance	NOUN
fcis-3178	59	12	.	.	PUNCT
fcis-3178	60	1	it	it	PRON
fcis-3178	60	2	has	have	VERB
fcis-3178	60	3	three	three	NUM
fcis-3178	60	4	convolutional	convolutional	ADJ
fcis-3178	60	5	layers	layer	NOUN
fcis-3178	60	6	and	and	CCONJ
fcis-3178	60	7	three	three	NUM
fcis-3178	60	8	max	max	NOUN
fcis-3178	60	9	-	-	PUNCT
fcis-3178	60	10	pooling	pool	VERB
fcis-3178	60	11	layers	layer	NOUN
fcis-3178	60	12	,	,	PUNCT
fcis-3178	60	13	32	32	NUM
fcis-3178	60	14	,	,	PUNCT
fcis-3178	60	15	64	64	NUM
fcis-3178	60	16	,	,	PUNCT
fcis-3178	60	17	128	128	NUM
fcis-3178	60	18	convolutional	convolutional	ADJ
fcis-3178	60	19	filters	filter	NOUN
fcis-3178	60	20	,	,	PUNCT
fcis-3178	60	21	and	and	CCONJ
fcis-3178	60	22	33	33	NUM
fcis-3178	60	23	filter	filter	NOUN
fcis-3178	60	24	windows	window	NOUN
fcis-3178	60	25	.	.	PUNCT
fcis-3178	61	1	cnn2	cnn2	PROPN
fcis-3178	61	2	has	have	VERB
fcis-3178	61	3	three	three	NUM
fcis-3178	61	4	convolutional	convolutional	ADJ
fcis-3178	61	5	layers	layer	NOUN
fcis-3178	61	6	,	,	PUNCT
fcis-3178	61	7	three	three	NUM
fcis-3178	61	8	max	max	NOUN
fcis-3178	61	9	-	-	PUNCT
fcis-3178	61	10	pooling	pool	VERB
fcis-3178	61	11	layers	layer	NOUN
fcis-3178	61	12	,	,	PUNCT
fcis-3178	61	13	and	and	CCONJ
fcis-3178	61	14	32,32,64	32,32,64	NUM
fcis-3178	61	15	convolutional	convolutional	ADJ
fcis-3178	61	16	filters	filter	NOUN
fcis-3178	61	17	.	.	PUNCT
fcis-3178	62	1	cnn1	cnn1	PROPN
fcis-3178	62	2	's	's	PART
fcis-3178	62	3	nonlinear	nonlinear	ADJ
fcis-3178	62	4	representation	representation	NOUN
fcis-3178	62	5	is	be	AUX
fcis-3178	62	6	improved	improve	VERB
fcis-3178	62	7	by	by	ADP
fcis-3178	62	8	adding	add	VERB
fcis-3178	62	9	an	an	DET
fcis-3178	62	10	11	11	NUM
fcis-3178	62	11	-	-	PUNCT
fcis-3178	62	12	convolution	convolution	NOUN
fcis-3178	62	13	layer	layer	NOUN
fcis-3178	62	14	after	after	ADP
fcis-3178	62	15	the	the	DET
fcis-3178	62	16	input	input	NOUN
fcis-3178	62	17	layer	layer	NOUN
fcis-3178	62	18	.	.	PUNCT
fcis-3178	63	1	all	all	DET
fcis-3178	63	2	three	three	NUM
fcis-3178	63	3	ad	ad	NOUN
fcis-3178	63	4	hoc	hoc	X
fcis-3178	63	5	networks	network	NOUN
fcis-3178	63	6	end	end	VERB
fcis-3178	63	7	up	up	ADP
fcis-3178	63	8	as	as	ADP
fcis-3178	63	9	two	two	NUM
fcis-3178	63	10	dense	dense	ADJ
fcis-3178	63	11	fully	fully	ADV
fcis-3178	63	12	linked	link	VERB
fcis-3178	63	13	layers	layer	NOUN
fcis-3178	63	14	in	in	ADP
fcis-3178	63	15	cnn3	cnn3	PROPN
fcis-3178	63	16	.	.	PUNCT
fcis-3178	64	1	all	all	PRON
fcis-3178	64	2	pooling	pool	VERB
fcis-3178	64	3	layers	layer	NOUN
fcis-3178	64	4	are	be	AUX
fcis-3178	64	5	22	22	NUM
fcis-3178	64	6	.	.	PUNCT
fcis-3178	65	1	the	the	DET
fcis-3178	65	2	maxpooling	maxpoole	VERB
fcis-3178	65	3	layer	layer	NOUN
fcis-3178	65	4	summarizes	summarize	VERB
fcis-3178	65	5	the	the	DET
fcis-3178	65	6	filter	filter	NOUN
fcis-3178	65	7	region	region	NOUN
fcis-3178	65	8	as	as	ADP
fcis-3178	65	9	a	a	DET
fcis-3178	65	10	nonlinear	nonlinear	NOUN
fcis-3178	65	11	down	down	ADP
fcis-3178	65	12	sampling	sample	VERB
fcis-3178	65	13	,	,	PUNCT
fcis-3178	65	14	providing	provide	VERB
fcis-3178	65	15	translation	translation	NOUN
fcis-3178	65	16	invariance	invariance	NOUN
fcis-3178	65	17	and	and	CCONJ
fcis-3178	65	18	lowering	lower	VERB
fcis-3178	65	19	computations	computation	NOUN
fcis-3178	65	20	.	.	PUNCT
fcis-3178	66	1	table	table	NOUN
fcis-3178	66	2	1	1	NUM
fcis-3178	66	3	.	.	PUNCT
fcis-3178	67	1	parameters	parameter	NOUN
fcis-3178	67	2	of	of	ADP
fcis-3178	67	3	the	the	DET
fcis-3178	67	4	models	model	NOUN
fcis-3178	67	5	layer	layer	NOUN
fcis-3178	67	6	kernel	kernel	PROPN
fcis-3178	67	7	k	k	ADJ
fcis-3178	67	8	-	-	PUNCT
fcis-3178	67	9	size	size	NOUN
fcis-3178	67	10	stride	stride	NOUN
fcis-3178	67	11	pad	pad	NOUN
fcis-3178	67	12	drop	drop	NOUN
fcis-3178	67	13	output	output	NOUN
fcis-3178	67	14	input	input	NOUN
fcis-3178	67	15	0	0	NUM
fcis-3178	67	16	0	0	NUM
fcis-3178	67	17	none	none	NOUN
fcis-3178	67	18	none	none	NOUN
fcis-3178	67	19	0	0	NUM
fcis-3178	67	20	48	48	NUM
fcis-3178	67	21	*	*	SYM
fcis-3178	67	22	48	48	NUM
fcis-3178	67	23	*	*	SYM
fcis-3178	67	24	1	1	NUM
fcis-3178	67	25	conv1	conv1	NOUN
fcis-3178	67	26	-	-	PUNCT
fcis-3178	67	27	1	1	NUM
fcis-3178	67	28	32	32	NUM
fcis-3178	67	29	1	1	NUM
fcis-3178	67	30	*	*	SYM
fcis-3178	67	31	1	1	NUM
fcis-3178	67	32	1	1	NUM
fcis-3178	67	33	0	0	NUM
fcis-3178	67	34	0	0	NUM
fcis-3178	67	35	48	48	NUM
fcis-3178	67	36	*	*	SYM
fcis-3178	67	37	48	48	NUM
fcis-3178	67	38	*	*	SYM
fcis-3178	67	39	32	32	NUM
fcis-3178	67	40	conv2	conv2	NOUN
fcis-3178	67	41	-	-	SYM
fcis-3178	67	42	1	1	NUM
fcis-3178	67	43	64	64	NUM
fcis-3178	67	44	3	3	NUM
fcis-3178	67	45	*	*	SYM
fcis-3178	67	46	3	3	NUM
fcis-3178	67	47	1	1	NUM
fcis-3178	67	48	1	1	NUM
fcis-3178	67	49	0	0	NUM
fcis-3178	67	50	48	48	NUM
fcis-3178	67	51	*	*	SYM
fcis-3178	67	52	48	48	NUM
fcis-3178	67	53	*	*	SYM
fcis-3178	67	54	64	64	NUM
fcis-3178	67	55	conv2	conv2	NOUN
fcis-3178	67	56	-	-	PUNCT
fcis-3178	67	57	2	2	NUM
fcis-3178	67	58	64	64	NUM
fcis-3178	67	59	5	5	NUM
fcis-3178	67	60	*	*	SYM
fcis-3178	67	61	5	5	NUM
fcis-3178	67	62	1	1	NUM
fcis-3178	67	63	2	2	NUM
fcis-3178	67	64	0	0	NUM
fcis-3178	67	65	48	48	NUM
fcis-3178	67	66	*	*	SYM
fcis-3178	67	67	48	48	NUM
fcis-3178	67	68	*	*	SYM
fcis-3178	67	69	64	64	NUM
fcis-3178	67	70	pool2	pool2	NOUN
fcis-3178	67	71	0	0	NUM
fcis-3178	67	72	2	2	NUM
fcis-3178	67	73	*	*	SYM
fcis-3178	67	74	2	2	NUM
fcis-3178	67	75	2	2	NUM
fcis-3178	67	76	0	0	NUM
fcis-3178	67	77	0	0	NUM
fcis-3178	67	78	24	24	NUM
fcis-3178	67	79	*	*	NUM
fcis-3178	67	80	24	24	NUM
fcis-3178	67	81	*	*	SYM
fcis-3178	67	82	64	64	NUM
fcis-3178	67	83	conv3	conv3	NOUN
fcis-3178	67	84	-	-	PUNCT
fcis-3178	67	85	1	1	NUM
fcis-3178	67	86	64	64	NUM
fcis-3178	67	87	3	3	NUM
fcis-3178	67	88	*	*	SYM
fcis-3178	67	89	3	3	NUM
fcis-3178	67	90	1	1	NUM
fcis-3178	67	91	1	1	NUM
fcis-3178	67	92	0	0	NUM
fcis-3178	67	93	24	24	NUM
fcis-3178	67	94	*	*	SYM
fcis-3178	67	95	24	24	NUM
fcis-3178	67	96	*	*	SYM
fcis-3178	67	97	64	64	NUM
fcis-3178	67	98	conv3	conv3	NOUN
fcis-3178	67	99	-	-	PUNCT
fcis-3178	67	100	2	2	NUM
fcis-3178	67	101	64	64	NUM
fcis-3178	67	102	5	5	NUM
fcis-3178	67	103	*	*	SYM
fcis-3178	67	104	5	5	NUM
fcis-3178	67	105	1	1	NUM
fcis-3178	67	106	2	2	NUM
fcis-3178	67	107	0	0	NUM
fcis-3178	67	108	24	24	NUM
fcis-3178	67	109	*	*	SYM
fcis-3178	67	110	24	24	NUM
fcis-3178	67	111	*	*	SYM
fcis-3178	67	112	64	64	NUM
fcis-3178	67	113	pool3	pool3	NOUN
fcis-3178	67	114	0	0	NUM
fcis-3178	67	115	2	2	NUM
fcis-3178	67	116	*	*	SYM
fcis-3178	67	117	2	2	NUM
fcis-3178	67	118	2	2	NUM
fcis-3178	67	119	0	0	NUM
fcis-3178	67	120	0	0	NUM
fcis-3178	67	121	12	12	NUM
fcis-3178	67	122	*	*	SYM
fcis-3178	67	123	12	12	NUM
fcis-3178	67	124	*	*	SYM
fcis-3178	67	125	64	64	NUM
fcis-3178	67	126	fc1	fc1	NOUN
fcis-3178	67	127	none	none	NOUN
fcis-3178	67	128	none	none	NOUN
fcis-3178	67	129	none	none	NOUN
fcis-3178	67	130	0	0	NUM
fcis-3178	67	131	50	50	NUM
fcis-3178	67	132	%	%	NOUN
fcis-3178	67	133	1	1	NUM
fcis-3178	67	134	*	*	SYM
fcis-3178	67	135	1	1	NUM
fcis-3178	67	136	*	*	SYM
fcis-3178	67	137	2048	2048	NUM
fcis-3178	67	138	fc2	fc2	NOUN
fcis-3178	67	139	none	none	NOUN
fcis-3178	67	140	none	none	NOUN
fcis-3178	67	141	none	none	NOUN
fcis-3178	67	142	0	0	NUM
fcis-3178	67	143	50	50	NUM
fcis-3178	67	144	%	%	NOUN
fcis-3178	67	145	1	1	NUM
fcis-3178	67	146	*	*	SYM
fcis-3178	67	147	1	1	NUM
fcis-3178	67	148	*	*	SYM
fcis-3178	67	149	1024	1024	NUM
fcis-3178	67	150	output	output	NOUN
fcis-3178	67	151	none	none	NOUN
fcis-3178	67	152	none	none	NOUN
fcis-3178	67	153	none	none	NOUN
fcis-3178	67	154	0	0	NUM
fcis-3178	67	155	0	0	NUM
fcis-3178	67	156	1	1	NUM
fcis-3178	67	157	*	*	NUM
fcis-3178	67	158	1	1	NUM
fcis-3178	67	159	*	*	SYM
fcis-3178	67	160	8	8	NUM
fcis-3178	67	161	3	3	NUM
fcis-3178	67	162	.	.	NOUN
fcis-3178	67	163	result	result	NOUN
fcis-3178	67	164	and	and	CCONJ
fcis-3178	67	165	discussion	discussion	VERB
fcis-3178	67	166	the	the	DET
fcis-3178	67	167	training	training	NOUN
fcis-3178	67	168	experimental	experimental	ADJ
fcis-3178	67	169	results	result	NOUN
fcis-3178	67	170	are	be	AUX
fcis-3178	67	171	shown	show	VERB
fcis-3178	67	172	in	in	ADP
fcis-3178	67	173	table	table	NOUN
fcis-3178	67	174	2	2	NUM
fcis-3178	67	175	:	:	PUNCT
fcis-3178	67	176	table	table	NOUN
fcis-3178	67	177	2	2	NUM
fcis-3178	67	178	.	.	PUNCT
fcis-3178	67	179	cnn	cnn	PROPN
fcis-3178	67	180	training	training	NOUN
fcis-3178	67	181	result	result	PROPN
fcis-3178	67	182	loop	loop	NOUN
fcis-3178	67	183	train	train	NOUN
fcis-3178	67	184	-	-	PUNCT
fcis-3178	67	185	acc	acc	PROPN
fcis-3178	67	186	test	test	NOUN
fcis-3178	67	187	-	-	PUNCT
fcis-3178	67	188	acc	acc	PROPN
fcis-3178	67	189	200	200	NUM
fcis-3178	67	190	32	32	NUM
fcis-3178	67	191	%	%	NOUN
fcis-3178	67	192	28.80	28.80	NUM
fcis-3178	67	193	%	%	NOUN
fcis-3178	67	194	400	400	NUM
fcis-3178	67	195	60	60	NUM
fcis-3178	67	196	%	%	NOUN
fcis-3178	67	197	34.90	34.90	NUM
fcis-3178	67	198	%	%	NOUN
fcis-3178	67	199	600	600	NUM
fcis-3178	67	200	81	81	NUM
fcis-3178	67	201	%	%	NOUN
fcis-3178	67	202	48.80	48.80	NUM
fcis-3178	67	203	%	%	NOUN
fcis-3178	67	204	1000	1000	NUM
fcis-3178	67	205	100	100	NUM
fcis-3178	67	206	%	%	NOUN
fcis-3178	67	207	63.10	63.10	NUM
fcis-3178	67	208	%	%	NOUN
fcis-3178	67	209	118	118	NUM
fcis-3178	67	210	3.1	3.1	NUM
fcis-3178	67	211	.	.	PUNCT
fcis-3178	68	1	result	result	VERB
fcis-3178	68	2	analysis	analysis	NOUN
fcis-3178	68	3	in	in	ADP
fcis-3178	68	4	this	this	DET
fcis-3178	68	5	paper	paper	NOUN
fcis-3178	68	6	,	,	PUNCT
fcis-3178	68	7	we	we	PRON
fcis-3178	68	8	train	train	VERB
fcis-3178	68	9	on	on	ADP
fcis-3178	68	10	fer2013	fer2013	PROPN
fcis-3178	68	11	,	,	PUNCT
fcis-3178	68	12	jaffe	jaffe	PROPN
fcis-3178	68	13	,	,	PUNCT
fcis-3178	68	14	ck+	ck+	NOUN
fcis-3178	68	15	.	.	PUNCT
fcis-3178	69	1	since	since	SCONJ
fcis-3178	69	2	the	the	DET
fcis-3178	69	3	jaffe	jaffe	PROPN
fcis-3178	69	4	dataset	dataset	PROPN
fcis-3178	69	5	provides	provide	VERB
fcis-3178	69	6	a	a	DET
fcis-3178	69	7	half	half	ADJ
fcis-3178	69	8	-	-	PUNCT
fcis-3178	69	9	body	body	NOUN
fcis-3178	69	10	map	map	NOUN
fcis-3178	69	11	,	,	PUNCT
fcis-3178	69	12	a	a	DET
fcis-3178	69	13	face	face	NOUN
fcis-3178	69	14	detection	detection	NOUN
fcis-3178	69	15	step	step	NOUN
fcis-3178	69	16	has	have	VERB
fcis-3178	69	17	to	to	PART
fcis-3178	69	18	be	be	AUX
fcis-3178	69	19	performed	perform	VERB
fcis-3178	69	20	.	.	PUNCT
fcis-3178	70	1	we	we	PRON
fcis-3178	70	2	finally	finally	ADV
fcis-3178	70	3	achieved	achieve	VERB
fcis-3178	70	4	about	about	ADV
fcis-3178	70	5	67	67	NUM
fcis-3178	70	6	%	%	NOUN
fcis-3178	70	7	accuracy	accuracy	NOUN
fcis-3178	70	8	for	for	ADP
fcis-3178	70	9	both	both	DET
fcis-3178	70	10	pub	pub	NOUN
fcis-3178	70	11	test	test	NOUN
fcis-3178	70	12	and	and	CCONJ
fcis-3178	70	13	pri	pri	NOUN
fcis-3178	70	14	test	test	NOUN
fcis-3178	70	15	on	on	ADP
fcis-3178	70	16	fer2013	fer2013	ADJ
fcis-3178	70	17	,	,	PUNCT
fcis-3178	70	18	and	and	CCONJ
fcis-3178	70	19	about	about	ADV
fcis-3178	70	20	99	99	NUM
fcis-3178	70	21	%	%	NOUN
fcis-3178	70	22	accuracy	accuracy	NOUN
fcis-3178	70	23	for	for	ADP
fcis-3178	70	24	both	both	CCONJ
fcis-3178	70	25	jaffe	jaffe	PROPN
fcis-3178	70	26	and	and	CCONJ
fcis-3178	70	27	ck+	ck+	VERB
fcis-3178	70	28	with	with	ADP
fcis-3178	70	29	five	five	NUM
fcis-3178	70	30	-	-	ADJ
fcis-3178	70	31	fold	fold	ADJ
fcis-3178	70	32	cross	cross	NOUN
fcis-3178	70	33	validation	validation	NOUN
fcis-3178	70	34	.	.	PUNCT
fcis-3178	71	1	training	training	NOUN
fcis-3178	71	2	on	on	ADP
fcis-3178	71	3	the	the	DET
fcis-3178	71	4	fer2013	fer2013	ADJ
fcis-3178	71	5	dataset	dataset	NOUN
fcis-3178	71	6	,	,	PUNCT
fcis-3178	71	7	when	when	SCONJ
fcis-3178	71	8	the	the	DET
fcis-3178	71	9	batch	batch	NOUN
fcis-3178	71	10	size	size	NOUN
fcis-3178	71	11	is	be	AUX
fcis-3178	71	12	200	200	NUM
fcis-3178	71	13	,	,	PUNCT
fcis-3178	71	14	the	the	DET
fcis-3178	71	15	accuracy	accuracy	NOUN
fcis-3178	71	16	is	be	AUX
fcis-3178	71	17	0.67	0.67	NUM
fcis-3178	71	18	.	.	PUNCT
fcis-3178	72	1	we	we	PRON
fcis-3178	72	2	can	can	AUX
fcis-3178	72	3	observe	observe	VERB
fcis-3178	72	4	a	a	DET
fcis-3178	72	5	sharp	sharp	ADJ
fcis-3178	72	6	drop	drop	NOUN
fcis-3178	72	7	in	in	ADP
fcis-3178	72	8	validation	validation	NOUN
fcis-3178	72	9	and	and	CCONJ
fcis-3178	72	10	training	training	NOUN
fcis-3178	72	11	accuracy	accuracy	NOUN
fcis-3178	72	12	as	as	SCONJ
fcis-3178	72	13	the	the	DET
fcis-3178	72	14	batch	batch	NOUN
fcis-3178	72	15	size	size	NOUN
fcis-3178	72	16	approaches	approach	VERB
fcis-3178	72	17	185	185	NUM
fcis-3178	72	18	,	,	PUNCT
fcis-3178	72	19	and	and	CCONJ
fcis-3178	72	20	then	then	ADV
fcis-3178	72	21	pick	pick	VERB
fcis-3178	72	22	up	up	ADP
fcis-3178	72	23	to	to	PART
fcis-3178	72	24	create	create	VERB
fcis-3178	72	25	a	a	DET
fcis-3178	72	26	gap	gap	NOUN
fcis-3178	72	27	.	.	PUNCT
fcis-3178	73	1	the	the	DET
fcis-3178	73	2	reason	reason	NOUN
fcis-3178	73	3	for	for	ADP
fcis-3178	73	4	this	this	DET
fcis-3178	73	5	phenomenon	phenomenon	NOUN
fcis-3178	73	6	is	be	AUX
fcis-3178	73	7	speculated	speculate	VERB
fcis-3178	73	8	to	to	PART
fcis-3178	73	9	be	be	AUX
fcis-3178	73	10	the	the	DET
fcis-3178	73	11	overfitting	overfitte	VERB
fcis-3178	73	12	phenomenon	phenomenon	NOUN
fcis-3178	73	13	caused	cause	VERB
fcis-3178	73	14	by	by	ADP
fcis-3178	73	15	limited	limited	ADJ
fcis-3178	73	16	datasets	dataset	NOUN
fcis-3178	73	17	.	.	PUNCT
fcis-3178	74	1	compared	compare	VERB
fcis-3178	74	2	with	with	ADP
fcis-3178	74	3	the	the	DET
fcis-3178	74	4	accuracy	accuracy	NOUN
fcis-3178	74	5	obtained	obtain	VERB
fcis-3178	74	6	by	by	ADP
fcis-3178	74	7	other	other	ADJ
fcis-3178	74	8	scholars	scholar	NOUN
fcis-3178	74	9	who	who	PRON
fcis-3178	74	10	also	also	ADV
fcis-3178	74	11	used	use	VERB
fcis-3178	74	12	the	the	DET
fcis-3178	74	13	fer2013	fer2013	ADJ
fcis-3178	74	14	dataset	dataset	NOUN
fcis-3178	74	15	,	,	PUNCT
fcis-3178	74	16	our	our	PRON
fcis-3178	74	17	accuracy	accuracy	NOUN
fcis-3178	74	18	in	in	ADP
fcis-3178	74	19	this	this	DET
fcis-3178	74	20	paper	paper	NOUN
fcis-3178	74	21	is	be	AUX
fcis-3178	74	22	not	not	PART
fcis-3178	74	23	the	the	DET
fcis-3178	74	24	highest	high	ADJ
fcis-3178	74	25	,	,	PUNCT
fcis-3178	74	26	but	but	CCONJ
fcis-3178	74	27	it	it	PRON
fcis-3178	74	28	is	be	AUX
fcis-3178	74	29	as	as	SCONJ
fcis-3178	74	30	expected	expect	VERB
fcis-3178	74	31	.	.	PUNCT
fcis-3178	75	1	although	although	SCONJ
fcis-3178	75	2	fer2013	fer2013	PROPN
fcis-3178	75	3	has	have	VERB
fcis-3178	75	4	the	the	DET
fcis-3178	75	5	largest	large	ADJ
fcis-3178	75	6	amount	amount	NOUN
fcis-3178	75	7	of	of	ADP
fcis-3178	75	8	data	datum	NOUN
fcis-3178	75	9	in	in	ADP
fcis-3178	75	10	the	the	DET
fcis-3178	75	11	three	three	NUM
fcis-3178	75	12	datasets	dataset	NOUN
fcis-3178	75	13	,	,	PUNCT
fcis-3178	75	14	the	the	DET
fcis-3178	75	15	crawler	crawler	NOUN
fcis-3178	75	16	collection	collection	NOUN
fcis-3178	75	17	of	of	ADP
fcis-3178	75	18	this	this	DET
fcis-3178	75	19	dataset	dataset	NOUN
fcis-3178	75	20	has	have	VERB
fcis-3178	75	21	problems	problem	NOUN
fcis-3178	75	22	such	such	ADJ
fcis-3178	75	23	as	as	ADP
fcis-3178	75	24	label	label	NOUN
fcis-3178	75	25	errors	error	NOUN
fcis-3178	75	26	,	,	PUNCT
fcis-3178	75	27	watermarks	watermark	NOUN
fcis-3178	75	28	,	,	PUNCT
fcis-3178	75	29	and	and	CCONJ
fcis-3178	75	30	animated	animated	ADJ
fcis-3178	75	31	pictures	picture	NOUN
fcis-3178	75	32	,	,	PUNCT
fcis-3178	75	33	which	which	PRON
fcis-3178	75	34	lead	lead	VERB
fcis-3178	75	35	to	to	ADP
fcis-3178	75	36	training	training	NOUN
fcis-3178	75	37	errors	error	NOUN
fcis-3178	75	38	and	and	CCONJ
fcis-3178	75	39	finally	finally	ADV
fcis-3178	75	40	lead	lead	VERB
fcis-3178	75	41	to	to	ADP
fcis-3178	75	42	obvious	obvious	ADJ
fcis-3178	75	43	training	training	NOUN
fcis-3178	75	44	inaccuracies	inaccuracy	NOUN
fcis-3178	75	45	.	.	PUNCT
fcis-3178	76	1	figure	figure	VERB
fcis-3178	76	2	3	3	NUM
fcis-3178	76	3	.	.	PUNCT
fcis-3178	76	4	fer2013	fer2013	ADJ
fcis-3178	76	5	training	training	NOUN
fcis-3178	76	6	accuracy	accuracy	NOUN
fcis-3178	76	7	and	and	CCONJ
fcis-3178	76	8	loss	loss	NOUN
fcis-3178	76	9	however	however	ADV
fcis-3178	76	10	,	,	PUNCT
fcis-3178	76	11	we	we	PRON
fcis-3178	76	12	used	use	VERB
fcis-3178	76	13	jaffe	jaffe	PROPN
fcis-3178	76	14	and	and	CCONJ
fcis-3178	76	15	ck+	ck+	VERB
fcis-3178	76	16	as	as	ADP
fcis-3178	76	17	datasets	dataset	NOUN
fcis-3178	76	18	for	for	ADP
fcis-3178	76	19	training	training	NOUN
fcis-3178	76	20	,	,	PUNCT
fcis-3178	76	21	and	and	CCONJ
fcis-3178	76	22	obtained	obtain	VERB
fcis-3178	76	23	much	much	ADV
fcis-3178	76	24	higher	high	ADJ
fcis-3178	76	25	accuracy	accuracy	NOUN
fcis-3178	76	26	than	than	ADP
fcis-3178	76	27	the	the	DET
fcis-3178	76	28	model	model	NOUN
fcis-3178	76	29	trained	train	VERB
fcis-3178	76	30	by	by	ADP
fcis-3178	76	31	fer2013	fer2013	ADJ
fcis-3178	76	32	.	.	PUNCT
fcis-3178	77	1	the	the	DET
fcis-3178	77	2	following	follow	VERB
fcis-3178	77	3	figure	figure	NOUN
fcis-3178	77	4	shows	show	VERB
fcis-3178	77	5	only	only	ADV
fcis-3178	77	6	the	the	DET
fcis-3178	77	7	training	training	NOUN
fcis-3178	77	8	results	result	NOUN
fcis-3178	77	9	of	of	ADP
fcis-3178	77	10	jaffe	jaffe	PROPN
fcis-3178	77	11	.	.	PUNCT
fcis-3178	78	1	the	the	DET
fcis-3178	78	2	main	main	ADJ
fcis-3178	78	3	reason	reason	NOUN
fcis-3178	78	4	is	be	AUX
fcis-3178	78	5	that	that	SCONJ
fcis-3178	78	6	the	the	DET
fcis-3178	78	7	amount	amount	NOUN
fcis-3178	78	8	of	of	ADP
fcis-3178	78	9	data	datum	NOUN
fcis-3178	78	10	in	in	ADP
fcis-3178	78	11	this	this	DET
fcis-3178	78	12	dataset	dataset	NOUN
fcis-3178	78	13	is	be	AUX
fcis-3178	78	14	very	very	ADV
fcis-3178	78	15	small	small	ADJ
fcis-3178	78	16	,	,	PUNCT
fcis-3178	78	17	and	and	CCONJ
fcis-3178	78	18	the	the	DET
fcis-3178	78	19	accuracy	accuracy	NOUN
fcis-3178	78	20	of	of	ADP
fcis-3178	78	21	the	the	DET
fcis-3178	78	22	data	datum	NOUN
fcis-3178	78	23	collected	collect	VERB
fcis-3178	78	24	in	in	ADP
fcis-3178	78	25	the	the	DET
fcis-3178	78	26	laboratory	laboratory	NOUN
fcis-3178	78	27	is	be	AUX
fcis-3178	78	28	high	high	ADJ
fcis-3178	78	29	enough	enough	ADV
fcis-3178	78	30	,	,	PUNCT
fcis-3178	78	31	and	and	CCONJ
fcis-3178	78	32	there	there	PRON
fcis-3178	78	33	are	be	VERB
fcis-3178	78	34	almost	almost	ADV
fcis-3178	78	35	no	no	PRON
fcis-3178	78	36	mislabeled	mislabele	VERB
fcis-3178	78	37	cases	case	NOUN
fcis-3178	78	38	,	,	PUNCT
fcis-3178	78	39	and	and	CCONJ
fcis-3178	78	40	only	only	ADV
fcis-3178	78	41	a	a	DET
fcis-3178	78	42	few	few	ADJ
fcis-3178	78	43	unavoidable	unavoidable	ADJ
fcis-3178	78	44	cases	case	NOUN
fcis-3178	78	45	of	of	ADP
fcis-3178	78	46	fer	fer	PROPN
fcis-3178	78	47	ambiguity	ambiguity	NOUN
fcis-3178	78	48	,	,	PUNCT
fcis-3178	78	49	so	so	CCONJ
fcis-3178	78	50	the	the	DET
fcis-3178	78	51	accuracy	accuracy	NOUN
fcis-3178	78	52	of	of	ADP
fcis-3178	78	53	both	both	DET
fcis-3178	78	54	training	training	NOUN
fcis-3178	78	55	and	and	CCONJ
fcis-3178	78	56	validation	validation	NOUN
fcis-3178	78	57	can	can	AUX
fcis-3178	78	58	reach	reach	VERB
fcis-3178	78	59	99	99	NUM
fcis-3178	78	60	%	%	NOUN
fcis-3178	78	61	.	.	PUNCT
fcis-3178	79	1	figure	figure	NOUN
fcis-3178	79	2	4	4	NUM
fcis-3178	79	3	.	.	PUNCT
fcis-3178	79	4	jaffe	jaffe	PROPN
fcis-3178	79	5	training	training	NOUN
fcis-3178	79	6	accuracy	accuracy	NOUN
fcis-3178	79	7	and	and	CCONJ
fcis-3178	79	8	loss	loss	NOUN
fcis-3178	79	9	we	we	PRON
fcis-3178	79	10	employed	employ	VERB
fcis-3178	79	11	a	a	DET
fcis-3178	79	12	visual	visual	ADJ
fcis-3178	79	13	training	training	NOUN
fcis-3178	79	14	process	process	NOUN
fcis-3178	79	15	to	to	PART
fcis-3178	79	16	evaluate	evaluate	VERB
fcis-3178	79	17	the	the	DET
fcis-3178	79	18	training	training	NOUN
fcis-3178	79	19	results	result	NOUN
fcis-3178	79	20	and	and	CCONJ
fcis-3178	79	21	accuracy	accuracy	NOUN
fcis-3178	79	22	of	of	ADP
fcis-3178	79	23	three	three	NUM
fcis-3178	79	24	datasets	dataset	NOUN
fcis-3178	79	25	.	.	PUNCT
fcis-3178	80	1	the	the	DET
fcis-3178	80	2	following	follow	VERB
fcis-3178	80	3	figure	figure	NOUN
fcis-3178	80	4	shows	show	VERB
fcis-3178	80	5	the	the	DET
fcis-3178	80	6	common	common	ADJ
fcis-3178	80	7	plot	plot	NOUN
fcis-3178	80	8	of	of	ADP
fcis-3178	80	9	training	training	NOUN
fcis-3178	80	10	on	on	ADP
fcis-3178	80	11	the	the	DET
fcis-3178	80	12	three	three	NUM
fcis-3178	80	13	datasets	dataset	NOUN
fcis-3178	80	14	with	with	ADP
fcis-3178	80	15	the	the	DET
fcis-3178	80	16	specified	specify	VERB
fcis-3178	80	17	rounds	round	NOUN
fcis-3178	80	18	of	of	ADP
fcis-3178	80	19	batch	batch	NOUN
fcis-3178	80	20	size	size	NOUN
fcis-3178	80	21	training	training	NOUN
fcis-3178	80	22	.	.	PUNCT
fcis-3178	81	1	figure	figure	NOUN
fcis-3178	81	2	5	5	NUM
fcis-3178	81	3	.	.	PUNCT
fcis-3178	81	4	common	common	ADJ
fcis-3178	81	5	plot	plot	NOUN
fcis-3178	81	6	of	of	ADP
fcis-3178	81	7	training	training	NOUN
fcis-3178	81	8	compared	compare	VERB
fcis-3178	81	9	to	to	ADP
fcis-3178	81	10	other	other	ADJ
fcis-3178	81	11	methods	method	NOUN
fcis-3178	81	12	,	,	PUNCT
fcis-3178	81	13	this	this	DET
fcis-3178	81	14	paper	paper	NOUN
fcis-3178	81	15	's	's	PART
fcis-3178	81	16	strategy	strategy	NOUN
fcis-3178	81	17	achieves	achieve	VERB
fcis-3178	81	18	good	good	ADJ
fcis-3178	81	19	results	result	NOUN
fcis-3178	81	20	on	on	ADP
fcis-3178	81	21	the	the	DET
fcis-3178	81	22	ck	ck	NOUN
fcis-3178	81	23	+	+	CCONJ
fcis-3178	81	24	and	and	CCONJ
fcis-3178	81	25	jaffe	jaffe	PROPN
fcis-3178	81	26	expression	expression	NOUN
fcis-3178	81	27	datasets	dataset	NOUN
fcis-3178	81	28	.	.	PUNCT
fcis-3178	82	1	in	in	ADP
fcis-3178	82	2	the	the	DET
fcis-3178	82	3	ck+	ck+	NOUN
fcis-3178	82	4	dataset	dataset	NOUN
fcis-3178	82	5	,	,	PUNCT
fcis-3178	82	6	the	the	DET
fcis-3178	82	7	suggested	suggest	VERB
fcis-3178	82	8	approach	approach	NOUN
fcis-3178	82	9	's	's	PART
fcis-3178	82	10	recognition	recognition	NOUN
fcis-3178	82	11	rate	rate	NOUN
fcis-3178	82	12	is	be	AUX
fcis-3178	82	13	higher	high	ADJ
fcis-3178	82	14	than	than	ADP
fcis-3178	82	15	other	other	ADJ
fcis-3178	82	16	classic	classic	ADJ
fcis-3178	82	17	machine	machine	NOUN
fcis-3178	82	18	learning	learn	VERB
fcis-3178	82	19	algorithms	algorithm	NOUN
fcis-3178	82	20	,	,	PUNCT
fcis-3178	82	21	including	include	VERB
fcis-3178	82	22	the	the	DET
fcis-3178	82	23	fundamental	fundamental	ADJ
fcis-3178	82	24	cnn	cnn	PROPN
fcis-3178	82	25	method	method	NOUN
fcis-3178	82	26	.	.	PUNCT
fcis-3178	83	1	in	in	ADP
fcis-3178	83	2	the	the	DET
fcis-3178	83	3	jaffe	jaffe	PROPN
fcis-3178	83	4	dataset	dataset	PROPN
fcis-3178	83	5	,	,	PUNCT
fcis-3178	83	6	the	the	DET
fcis-3178	83	7	proposed	propose	VERB
fcis-3178	83	8	technique	technique	NOUN
fcis-3178	83	9	has	have	VERB
fcis-3178	83	10	a	a	DET
fcis-3178	83	11	lower	low	ADJ
fcis-3178	83	12	recognition	recognition	NOUN
fcis-3178	83	13	rate	rate	NOUN
fcis-3178	83	14	than	than	ADP
fcis-3178	83	15	traditional	traditional	ADJ
fcis-3178	83	16	methods	method	NOUN
fcis-3178	83	17	,	,	PUNCT
fcis-3178	83	18	but	but	CCONJ
fcis-3178	83	19	because	because	SCONJ
fcis-3178	83	20	it	it	PRON
fcis-3178	83	21	uses	use	VERB
fcis-3178	83	22	cnn	cnn	PROPN
fcis-3178	83	23	,	,	PUNCT
fcis-3178	83	24	the	the	DET
fcis-3178	83	25	complexity	complexity	NOUN
fcis-3178	83	26	of	of	ADP
fcis-3178	83	27	manually	manually	ADV
fcis-3178	83	28	built	build	VERB
fcis-3178	83	29	features	feature	NOUN
fcis-3178	83	30	is	be	AUX
fcis-3178	83	31	avoided	avoid	VERB
fcis-3178	83	32	,	,	PUNCT
fcis-3178	83	33	and	and	CCONJ
fcis-3178	83	34	a	a	DET
fcis-3178	83	35	good	good	ADJ
fcis-3178	83	36	recognition	recognition	NOUN
fcis-3178	83	37	effect	effect	NOUN
fcis-3178	83	38	may	may	AUX
fcis-3178	83	39	be	be	AUX
fcis-3178	83	40	reached	reach	VERB
fcis-3178	83	41	by	by	ADP
fcis-3178	83	42	training	train	VERB
fcis-3178	83	43	a	a	DET
fcis-3178	83	44	few	few	ADJ
fcis-3178	83	45	layers	layer	NOUN
fcis-3178	83	46	of	of	ADP
fcis-3178	83	47	the	the	DET
fcis-3178	83	48	network	network	NOUN
fcis-3178	83	49	,	,	PUNCT
fcis-3178	83	50	saving	save	VERB
fcis-3178	83	51	a	a	DET
fcis-3178	83	52	lot	lot	NOUN
fcis-3178	83	53	of	of	ADP
fcis-3178	83	54	training	training	NOUN
fcis-3178	83	55	time	time	NOUN
fcis-3178	83	56	.	.	PUNCT
fcis-3178	84	1	jaffe	jaffe	PROPN
fcis-3178	84	2	has	have	VERB
fcis-3178	84	3	fewer	few	ADJ
fcis-3178	84	4	original	original	ADJ
fcis-3178	84	5	data	datum	NOUN
fcis-3178	84	6	,	,	PUNCT
fcis-3178	84	7	which	which	PRON
fcis-3178	84	8	affects	affect	VERB
fcis-3178	84	9	its	its	PRON
fcis-3178	84	10	recognition	recognition	NOUN
fcis-3178	84	11	rate	rate	NOUN
fcis-3178	84	12	.	.	PUNCT
fcis-3178	85	1	3.2	3.2	NUM
fcis-3178	85	2	.	.	PUNCT
fcis-3178	85	3	possible	possible	ADJ
fcis-3178	85	4	solution	solution	NOUN
fcis-3178	85	5	in	in	ADP
fcis-3178	85	6	recent	recent	ADJ
fcis-3178	85	7	years	year	NOUN
fcis-3178	85	8	,	,	PUNCT
fcis-3178	85	9	the	the	DET
fcis-3178	85	10	most	most	ADV
fcis-3178	85	11	popular	popular	ADJ
fcis-3178	85	12	strategy	strategy	NOUN
fcis-3178	85	13	to	to	PART
fcis-3178	85	14	alleviate	alleviate	VERB
fcis-3178	85	15	the	the	DET
fcis-3178	85	16	problem	problem	NOUN
fcis-3178	85	17	of	of	ADP
fcis-3178	85	18	expression	expression	NOUN
fcis-3178	85	19	database	database	NOUN
fcis-3178	85	20	size	size	NOUN
fcis-3178	85	21	is	be	AUX
fcis-3178	85	22	to	to	PART
fcis-3178	85	23	transfer	transfer	VERB
fcis-3178	85	24	the	the	DET
fcis-3178	85	25	object	object	NOUN
fcis-3178	85	26	recognition	recognition	NOUN
fcis-3178	85	27	model	model	NOUN
fcis-3178	85	28	or	or	CCONJ
fcis-3178	85	29	face	face	VERB
fcis-3178	85	30	recognition	recognition	NOUN
fcis-3178	85	31	model	model	NOUN
fcis-3178	85	32	to	to	ADP
fcis-3178	85	33	the	the	DET
fcis-3178	85	34	expression	expression	NOUN
fcis-3178	85	35	recognition	recognition	NOUN
fcis-3178	85	36	task	task	NOUN
fcis-3178	85	37	,	,	PUNCT
fcis-3178	85	38	namely	namely	ADV
fcis-3178	85	39	transfer	transfer	VERB
fcis-3178	85	40	learning	learning	NOUN
fcis-3178	85	41	method	method	NOUN
fcis-3178	85	42	.	.	PUNCT
fcis-3178	86	1	in	in	ADP
fcis-3178	86	2	addition	addition	NOUN
fcis-3178	86	3	to	to	PART
fcis-3178	86	4	transfer	transfer	VERB
fcis-3178	86	5	learning	learning	NOUN
fcis-3178	86	6	strategies	strategy	NOUN
fcis-3178	86	7	,	,	PUNCT
fcis-3178	86	8	the	the	DET
fcis-3178	86	9	use	use	NOUN
fcis-3178	86	10	of	of	ADP
fcis-3178	86	11	semi	semi	ADJ
fcis-3178	86	12	-	-	ADJ
fcis-3178	86	13	supervised	supervised	ADJ
fcis-3178	86	14	methods	method	NOUN
fcis-3178	86	15	is	be	AUX
fcis-3178	86	16	also	also	ADV
fcis-3178	86	17	a	a	DET
fcis-3178	86	18	possible	possible	ADJ
fcis-3178	86	19	development	development	NOUN
fcis-3178	86	20	trend	trend	NOUN
fcis-3178	86	21	in	in	ADP
fcis-3178	86	22	the	the	DET
fcis-3178	86	23	future	future	NOUN
fcis-3178	86	24	.	.	PUNCT
fcis-3178	87	1	the	the	DET
fcis-3178	87	2	main	main	ADJ
fcis-3178	87	3	reasons	reason	NOUN
fcis-3178	87	4	are	be	AUX
fcis-3178	87	5	as	as	SCONJ
fcis-3178	87	6	follows	follow	VERB
fcis-3178	87	7	,	,	PUNCT
fcis-3178	87	8	firstly	firstly	ADV
fcis-3178	87	9	a	a	DET
fcis-3178	87	10	large	large	ADJ
fcis-3178	87	11	-	-	PUNCT
fcis-3178	87	12	scale	scale	NOUN
fcis-3178	87	13	face	face	NOUN
fcis-3178	87	14	recognition	recognition	NOUN
fcis-3178	87	15	database	database	NOUN
fcis-3178	87	16	contains	contain	VERB
fcis-3178	87	17	a	a	DET
fcis-3178	87	18	large	large	ADJ
fcis-3178	87	19	number	number	NOUN
fcis-3178	87	20	of	of	ADP
fcis-3178	87	21	expressions	expression	NOUN
fcis-3178	87	22	faces	face	VERB
fcis-3178	87	23	.	.	PUNCT
fcis-3178	88	1	also	also	ADV
fcis-3178	88	2	,	,	PUNCT
fcis-3178	88	3	databases	database	NOUN
fcis-3178	88	4	like	like	ADP
fcis-3178	88	5	affectnet	affectnet	NOUN
fcis-3178	88	6	and	and	CCONJ
fcis-3178	88	7	emotionet	emotionet	NOUN
fcis-3178	88	8	still	still	ADV
fcis-3178	88	9	have	have	VERB
fcis-3178	88	10	a	a	DET
fcis-3178	88	11	large	large	ADJ
fcis-3178	88	12	proportion	proportion	NOUN
fcis-3178	88	13	of	of	ADP
fcis-3178	88	14	expression	expression	NOUN
fcis-3178	88	15	faces	face	NOUN
fcis-3178	88	16	that	that	PRON
fcis-3178	88	17	are	be	AUX
fcis-3178	88	18	not	not	PART
fcis-3178	88	19	annotated	annotate	VERB
fcis-3178	88	20	.	.	PUNCT
fcis-3178	89	1	4	4	X
fcis-3178	89	2	.	.	X
fcis-3178	89	3	conclusion	conclusion	VERB
fcis-3178	89	4	this	this	DET
fcis-3178	89	5	paper	paper	NOUN
fcis-3178	89	6	studies	study	NOUN
fcis-3178	89	7	the	the	DET
fcis-3178	89	8	use	use	NOUN
fcis-3178	89	9	of	of	ADP
fcis-3178	89	10	cnn	cnn	PROPN
fcis-3178	89	11	to	to	PART
fcis-3178	89	12	realize	realize	VERB
fcis-3178	89	13	fer	fer	PROPN
fcis-3178	89	14	under	under	ADP
fcis-3178	89	15	different	different	ADJ
fcis-3178	89	16	data	data	NOUN
fcis-3178	89	17	sets	set	NOUN
fcis-3178	89	18	,	,	PUNCT
fcis-3178	89	19	and	and	CCONJ
fcis-3178	89	20	obtains	obtain	VERB
fcis-3178	89	21	high	high	ADJ
fcis-3178	89	22	accuracy	accuracy	NOUN
fcis-3178	89	23	.	.	PUNCT
fcis-3178	90	1	under	under	ADP
fcis-3178	90	2	the	the	DET
fcis-3178	90	3	vggnet	vggnet	ADJ
fcis-3178	90	4	structure	structure	NOUN
fcis-3178	90	5	of	of	ADP
fcis-3178	90	6	cnn	cnn	PROPN
fcis-3178	90	7	,	,	PUNCT
fcis-3178	90	8	it	it	PRON
fcis-3178	90	9	is	be	AUX
fcis-3178	90	10	compared	compare	VERB
fcis-3178	90	11	with	with	ADP
fcis-3178	90	12	the	the	DET
fcis-3178	90	13	traditional	traditional	ADJ
fcis-3178	90	14	method	method	NOUN
fcis-3178	90	15	,	,	PUNCT
fcis-3178	90	16	and	and	CCONJ
fcis-3178	90	17	the	the	DET
fcis-3178	90	18	implementation	implementation	NOUN
fcis-3178	90	19	process	process	NOUN
fcis-3178	90	20	and	and	CCONJ
fcis-3178	90	21	results	result	NOUN
fcis-3178	90	22	are	be	AUX
fcis-3178	90	23	discussed	discuss	VERB
fcis-3178	90	24	.	.	PUNCT
fcis-3178	91	1	4.1	4.1	NUM
fcis-3178	91	2	.	.	PUNCT
fcis-3178	91	3	program	program	NOUN
fcis-3178	91	4	result	result	VERB
fcis-3178	91	5	our	our	PRON
fcis-3178	91	6	system	system	NOUN
fcis-3178	91	7	can	can	AUX
fcis-3178	91	8	also	also	ADV
fcis-3178	91	9	run	run	VERB
fcis-3178	91	10	cross	cross	ADJ
fcis-3178	91	11	-	-	ADJ
fcis-3178	91	12	platform	platform	ADJ
fcis-3178	91	13	internet	internet	NOUN
fcis-3178	91	14	applications	application	NOUN
fcis-3178	91	15	.	.	PUNCT
fcis-3178	92	1	parallel	parallel	ADJ
fcis-3178	92	2	development	development	NOUN
fcis-3178	92	3	of	of	ADP
fcis-3178	92	4	prototypes	prototype	NOUN
fcis-3178	92	5	and	and	CCONJ
fcis-3178	92	6	code	code	NOUN
fcis-3178	92	7	saves	save	VERB
fcis-3178	92	8	time	time	NOUN
fcis-3178	92	9	.	.	PUNCT
fcis-3178	93	1	hdf5	hdf5	PROPN
fcis-3178	93	2	files	file	NOUN
fcis-3178	93	3	can	can	AUX
fcis-3178	93	4	iterate	iterate	VERB
fcis-3178	93	5	and	and	CCONJ
fcis-3178	93	6	replace	replace	VERB
fcis-3178	93	7	kera’s	kera’s	ADV
fcis-3178	93	8	-	-	PUNCT
fcis-3178	93	9	trained	train	VERB
fcis-3178	93	10	models	model	NOUN
fcis-3178	93	11	,	,	PUNCT
fcis-3178	93	12	facilitating	facilitate	VERB
fcis-3178	93	13	engineering	engineering	NOUN
fcis-3178	93	14	implementation	implementation	NOUN
fcis-3178	93	15	.	.	PUNCT
fcis-3178	94	1	by	by	ADP
fcis-3178	94	2	using	use	VERB
fcis-3178	94	3	a	a	DET
fcis-3178	94	4	simple	simple	ADJ
fcis-3178	94	5	parallel	parallel	ADJ
fcis-3178	94	6	network	network	NOUN
fcis-3178	94	7	model	model	NOUN
fcis-3178	94	8	for	for	ADP
fcis-3178	94	9	fer	fer	PROPN
fcis-3178	94	10	,	,	PUNCT
fcis-3178	94	11	training	training	NOUN
fcis-3178	94	12	and	and	CCONJ
fcis-3178	94	13	testing	testing	NOUN
fcis-3178	94	14	may	may	AUX
fcis-3178	94	15	be	be	AUX
fcis-3178	94	16	done	do	VERB
fcis-3178	94	17	in	in	ADP
fcis-3178	94	18	45	45	NUM
fcis-3178	94	19	minutes	minute	NOUN
fcis-3178	94	20	.	.	PUNCT
fcis-3178	95	1	in	in	ADP
fcis-3178	95	2	future	future	ADJ
fcis-3178	95	3	research	research	NOUN
fcis-3178	95	4	,	,	PUNCT
fcis-3178	95	5	can	can	AUX
fcis-3178	95	6	try	try	VERB
fcis-3178	95	7	to	to	PART
fcis-3178	95	8	optimize	optimize	VERB
fcis-3178	95	9	the	the	DET
fcis-3178	95	10	key	key	ADJ
fcis-3178	95	11	area	area	NOUN
fcis-3178	95	12	extraction	extraction	NOUN
fcis-3178	95	13	accuracy	accuracy	NOUN
fcis-3178	95	14	and	and	CCONJ
fcis-3178	95	15	robustness	robustness	NOUN
fcis-3178	95	16	of	of	ADP
fcis-3178	95	17	the	the	DET
fcis-3178	95	18	model	model	NOUN
fcis-3178	95	19	,	,	PUNCT
fcis-3178	95	20	which	which	PRON
fcis-3178	95	21	can	can	AUX
fcis-3178	95	22	better	well	ADV
fcis-3178	95	23	meet	meet	VERB
fcis-3178	95	24	the	the	DET
fcis-3178	95	25	needs	need	NOUN
fcis-3178	95	26	of	of	ADP
fcis-3178	95	27	real	real	ADJ
fcis-3178	95	28	-	-	PUNCT
fcis-3178	95	29	life	life	NOUN
fcis-3178	95	30	scenes	scene	NOUN
fcis-3178	95	31	,	,	PUNCT
fcis-3178	95	32	centralized	centralized	ADJ
fcis-3178	95	33	data	datum	NOUN
fcis-3178	95	34	quantity	quantity	NOUN
fcis-3178	95	35	and	and	CCONJ
fcis-3178	95	36	accuracy	accuracy	NOUN
fcis-3178	95	37	are	be	AUX
fcis-3178	95	38	important	important	ADJ
fcis-3178	95	39	,	,	PUNCT
fcis-3178	95	40	but	but	CCONJ
fcis-3178	95	41	the	the	DET
fcis-3178	95	42	huge	huge	ADJ
fcis-3178	95	43	amount	amount	NOUN
fcis-3178	95	44	of	of	ADP
fcis-3178	95	45	data	datum	NOUN
fcis-3178	95	46	access	access	NOUN
fcis-3178	95	47	is	be	AUX
fcis-3178	95	48	not	not	PART
fcis-3178	95	49	the	the	DET
fcis-3178	95	50	best	good	ADJ
fcis-3178	95	51	solution	solution	NOUN
fcis-3178	95	52	,	,	PUNCT
fcis-3178	95	53	at	at	ADP
fcis-3178	95	54	the	the	DET
fcis-3178	95	55	same	same	ADJ
fcis-3178	95	56	time	time	NOUN
fcis-3178	95	57	need	need	VERB
fcis-3178	95	58	to	to	PART
fcis-3178	95	59	optimize	optimize	VERB
fcis-3178	95	60	the	the	DET
fcis-3178	95	61	model	model	NOUN
fcis-3178	95	62	framework	framework	NOUN
fcis-3178	95	63	,	,	PUNCT
fcis-3178	95	64	to	to	PART
fcis-3178	95	65	achieve	achieve	VERB
fcis-3178	95	66	the	the	DET
fcis-3178	95	67	most	most	ADV
fcis-3178	95	68	efficient	efficient	ADJ
fcis-3178	95	69	balance	balance	NOUN
fcis-3178	95	70	.	.	PUNCT
fcis-3178	96	1	4.2	4.2	NUM
fcis-3178	96	2	.	.	PUNCT
fcis-3178	96	3	fer	fer	PROPN
fcis-3178	96	4	datasets	dataset	NOUN
fcis-3178	96	5	as	as	ADP
fcis-3178	96	6	fer	fer	PROPN
fcis-3178	96	7	research	research	PROPN
fcis-3178	96	8	focuses	focus	VERB
fcis-3178	96	9	on	on	ADP
fcis-3178	96	10	difficult	difficult	ADJ
fcis-3178	96	11	in	in	ADP
fcis-3178	96	12	-	-	PUNCT
fcis-3178	96	13	the	the	DET
fcis-3178	96	14	-	-	PUNCT
fcis-3178	96	15	wild	wild	ADJ
fcis-3178	96	16	environmental	environmental	ADJ
fcis-3178	96	17	conditions	condition	NOUN
fcis-3178	96	18	,	,	PUNCT
fcis-3178	96	19	several	several	ADJ
fcis-3178	96	20	researchers	researcher	NOUN
fcis-3178	96	21	are	be	AUX
fcis-3178	96	22	employing	employ	VERB
fcis-3178	96	23	deep	deep	ADJ
fcis-3178	96	24	learning	learning	NOUN
fcis-3178	96	25	to	to	PART
fcis-3178	96	26	overcome	overcome	VERB
fcis-3178	96	27	illumination	illumination	NOUN
fcis-3178	96	28	variance	variance	NOUN
fcis-3178	96	29	,	,	PUNCT
fcis-3178	96	30	occlusions	occlusion	NOUN
fcis-3178	96	31	,	,	PUNCT
fcis-3178	96	32	non	non	ADJ
fcis-3178	96	33	-	-	ADJ
fcis-3178	96	34	frontal	frontal	ADJ
fcis-3178	96	35	head	head	NOUN
fcis-3178	96	36	orientations	orientation	NOUN
fcis-3178	96	37	,	,	PUNCT
fcis-3178	96	38	identification	identification	NOUN
fcis-3178	96	39	bias	bias	NOUN
fcis-3178	96	40	,	,	PUNCT
fcis-3178	96	41	and	and	CCONJ
fcis-3178	96	42	lowintensity	lowintensity	NOUN
fcis-3178	96	43	emotions	emotion	NOUN
fcis-3178	96	44	.	.	PUNCT
fcis-3178	97	1	fer	fer	PROPN
fcis-3178	97	2	is	be	AUX
fcis-3178	97	3	a	a	DET
fcis-3178	97	4	data	data	NOUN
fcis-3178	97	5	-	-	PUNCT
fcis-3178	97	6	driven	drive	VERB
fcis-3178	97	7	task	task	NOUN
fcis-3178	97	8	,	,	PUNCT
fcis-3178	97	9	hence	hence	ADV
fcis-3178	97	10	deep	deep	PROPN
fcis-3178	97	11	fer	fer	PROPN
fcis-3178	97	12	systems	system	NOUN
fcis-3178	97	13	face	face	VERB
fcis-3178	97	14	a	a	DET
fcis-3178	97	15	lack	lack	NOUN
fcis-3178	97	16	of	of	ADP
fcis-3178	97	17	both	both	PRON
fcis-3178	97	18	quantity	quantity	NOUN
fcis-3178	97	19	and	and	CCONJ
fcis-3178	97	20	quality	quality	NOUN
fcis-3178	97	21	training	training	NOUN
fcis-3178	97	22	data	datum	NOUN
fcis-3178	97	23	.	.	PUNCT
fcis-3178	98	1	different	different	ADJ
fcis-3178	98	2	ages	age	NOUN
fcis-3178	98	3	,	,	PUNCT
fcis-3178	98	4	cultures	culture	NOUN
fcis-3178	98	5	,	,	PUNCT
fcis-3178	98	6	and	and	CCONJ
fcis-3178	98	7	genders	gender	NOUN
fcis-3178	98	8	display	display	VERB
fcis-3178	98	9	and	and	CCONJ
fcis-3178	98	10	interpret	interpret	VERB
fcis-3178	98	11	facial	facial	ADJ
fcis-3178	98	12	expression	expression	NOUN
fcis-3178	98	13	differently	differently	ADV
fcis-3178	98	14	.	.	PUNCT
fcis-3178	99	1	an	an	DET
fcis-3178	99	2	ideal	ideal	ADJ
fcis-3178	99	3	facial	facial	ADJ
fcis-3178	99	4	expression	expression	NOUN
fcis-3178	99	5	dataset	dataset	NOUN
fcis-3178	99	6	should	should	AUX
fcis-3178	99	7	include	include	VERB
fcis-3178	99	8	abundant	abundant	ADJ
fcis-3178	99	9	sample	sample	NOUN
fcis-3178	99	10	images	image	NOUN
fcis-3178	99	11	with	with	ADP
fcis-3178	99	12	precise	precise	ADJ
fcis-3178	99	13	face	face	NOUN
fcis-3178	99	14	attribute	attribute	NOUN
fcis-3178	99	15	labels	label	NOUN
fcis-3178	99	16	,	,	PUNCT
fcis-3178	99	17	not	not	PART
fcis-3178	99	18	just	just	ADV
fcis-3178	99	19	expression	expression	NOUN
fcis-3178	99	20	but	but	CCONJ
fcis-3178	99	21	also	also	ADV
fcis-3178	99	22	age	age	NOUN
fcis-3178	99	23	,	,	PUNCT
fcis-3178	99	24	gender	gender	NOUN
fcis-3178	99	25	,	,	PUNCT
fcis-3178	99	26	and	and	CCONJ
fcis-3178	99	27	ethnicity	ethnicity	NOUN
fcis-3178	99	28	,	,	PUNCT
fcis-3178	99	29	to	to	PART
fcis-3178	99	30	facilitate	facilitate	VERB
fcis-3178	99	31	research	research	NOUN
fcis-3178	99	32	on	on	ADP
fcis-3178	99	33	cross	cross	ADJ
fcis-3178	99	34	-	-	ADJ
fcis-3178	99	35	age	age	ADJ
fcis-3178	99	36	range	range	NOUN
fcis-3178	99	37	,	,	PUNCT
fcis-3178	99	38	crossgender	crossgender	NOUN
fcis-3178	99	39	,	,	PUNCT
fcis-3178	99	40	and	and	CCONJ
fcis-3178	99	41	cross	cross	ADJ
fcis-3178	99	42	-	-	ADJ
fcis-3178	99	43	cultural	cultural	ADJ
fcis-3178	99	44	fer	fer	PROPN
fcis-3178	99	45	using	use	VERB
fcis-3178	99	46	deep	deep	ADJ
fcis-3178	99	47	learning	learning	NOUN
fcis-3178	99	48	techniques	technique	NOUN
fcis-3178	99	49	,	,	PUNCT
fcis-3178	99	50	such	such	ADJ
fcis-3178	99	51	as	as	ADP
fcis-3178	99	52	multitask	multitask	ADJ
fcis-3178	99	53	deep	deep	ADJ
fcis-3178	99	54	networks	network	NOUN
fcis-3178	99	55	.	.	PUNCT
fcis-3178	100	1	4.3	4.3	NUM
fcis-3178	100	2	.	.	PUNCT
fcis-3178	100	3	dataset	dataset	NOUN
fcis-3178	100	4	bias	bias	NOUN
fcis-3178	100	5	&	&	CCONJ
fcis-3178	100	6	imbalanced	imbalanced	ADJ
fcis-3178	100	7	distribution	distribution	NOUN
fcis-3178	100	8	different	different	ADJ
fcis-3178	100	9	collecting	collect	VERB
fcis-3178	100	10	circumstances	circumstance	NOUN
fcis-3178	100	11	and	and	CCONJ
fcis-3178	100	12	subjective	subjective	ADJ
fcis-3178	100	13	labelling	labelling	NOUN
fcis-3178	100	14	generate	generate	NOUN
fcis-3178	100	15	bias	bias	NOUN
fcis-3178	100	16	in	in	ADP
fcis-3178	100	17	facial	facial	ADJ
fcis-3178	100	18	expression	expression	NOUN
fcis-3178	100	19	databases	database	NOUN
fcis-3178	100	20	.	.	PUNCT
fcis-3178	101	1	recent	recent	ADJ
fcis-3178	101	2	research	research	NOUN
fcis-3178	101	3	successfully	successfully	ADV
fcis-3178	101	4	tests	test	VERB
fcis-3178	101	5	algorithms	algorithm	NOUN
fcis-3178	101	6	on	on	ADP
fcis-3178	101	7	a	a	DET
fcis-3178	101	8	specific	specific	ADJ
fcis-3178	101	9	dataset	dataset	NOUN
fcis-3178	101	10	.	.	PUNCT
fcis-3178	102	1	withindatabase	withindatabase	PROPN
fcis-3178	102	2	algorithms	algorithm	NOUN
fcis-3178	102	3	lack	lack	VERB
fcis-3178	102	4	generalizability	generalizability	NOUN
fcis-3178	102	5	on	on	ADP
fcis-3178	102	6	unseen	unseen	ADJ
fcis-3178	102	7	test	test	NOUN
fcis-3178	102	8	data	datum	NOUN
fcis-3178	102	9	,	,	PUNCT
fcis-3178	102	10	119	119	NUM
fcis-3178	102	11	and	and	CCONJ
fcis-3178	102	12	cross	cross	ADJ
fcis-3178	102	13	-	-	ADJ
fcis-3178	102	14	dataset	dataset	ADJ
fcis-3178	102	15	performance	performance	NOUN
fcis-3178	102	16	is	be	AUX
fcis-3178	102	17	worsened	worsen	VERB
fcis-3178	102	18	by	by	ADP
fcis-3178	102	19	inconsistencies	inconsistency	NOUN
fcis-3178	102	20	.	.	PUNCT
fcis-3178	103	1	due	due	ADP
fcis-3178	103	2	to	to	ADP
fcis-3178	103	3	unequal	unequal	ADJ
fcis-3178	103	4	expression	expression	NOUN
fcis-3178	103	5	annotations	annotation	NOUN
fcis-3178	103	6	,	,	PUNCT
fcis-3178	103	7	combining	combine	VERB
fcis-3178	103	8	several	several	ADJ
fcis-3178	103	9	datasets	dataset	NOUN
fcis-3178	103	10	can	can	AUX
fcis-3178	103	11	not	not	PART
fcis-3178	103	12	increase	increase	VERB
fcis-3178	103	13	fer	fer	PROPN
fcis-3178	103	14	performance	performance	NOUN
fcis-3178	103	15	.	.	PUNCT
fcis-3178	104	1	fer	fer	PROPN
fcis-3178	104	2	system	system	NOUN
fcis-3178	104	3	evaluation	evaluation	NOUN
fcis-3178	104	4	criteria	criterion	NOUN
fcis-3178	104	5	include	include	VERB
fcis-3178	104	6	cross	cross	ADJ
fcis-3178	104	7	-	-	ADJ
fcis-3178	104	8	database	database	ADJ
fcis-3178	104	9	performance	performance	NOUN
fcis-3178	104	10	.	.	PUNCT
fcis-3178	105	1	domain	domain	NOUN
fcis-3178	105	2	adaptability	adaptability	NOUN
fcis-3178	105	3	and	and	CCONJ
fcis-3178	105	4	knowledge	knowledge	NOUN
fcis-3178	105	5	distillation	distillation	NOUN
fcis-3178	105	6	minimize	minimize	VERB
fcis-3178	105	7	prejudice	prejudice	NOUN
fcis-3178	105	8	.	.	PUNCT
fcis-3178	106	1	sample	sample	NOUN
fcis-3178	106	2	acquisition	acquisition	NOUN
fcis-3178	106	3	causes	cause	VERB
fcis-3178	106	4	unbalanced	unbalanced	ADJ
fcis-3178	106	5	facial	facial	ADJ
fcis-3178	106	6	class	class	NOUN
fcis-3178	106	7	distribution	distribution	NOUN
fcis-3178	106	8	.	.	PUNCT
fcis-3178	107	1	identifying	identify	VERB
fcis-3178	107	2	disgust	disgust	NOUN
fcis-3178	107	3	,	,	PUNCT
fcis-3178	107	4	anxiety	anxiety	NOUN
fcis-3178	107	5	,	,	PUNCT
fcis-3178	107	6	and	and	CCONJ
fcis-3178	107	7	other	other	ADJ
fcis-3178	107	8	rare	rare	ADJ
fcis-3178	107	9	expressions	expression	NOUN
fcis-3178	107	10	is	be	AUX
fcis-3178	107	11	difficult	difficult	ADJ
fcis-3178	107	12	.	.	PUNCT
fcis-3178	108	1	resampling	resample	VERB
fcis-3178	108	2	and	and	CCONJ
fcis-3178	108	3	balancing	balance	VERB
fcis-3178	108	4	class	class	NOUN
fcis-3178	108	5	distribution	distribution	NOUN
fcis-3178	108	6	during	during	ADP
fcis-3178	108	7	pre	pre	ADJ
fcis-3178	108	8	-	-	ADJ
fcis-3178	108	9	processing	processing	NOUN
fcis-3178	108	10	is	be	AUX
fcis-3178	108	11	one	one	NUM
fcis-3178	108	12	way	way	NOUN
fcis-3178	108	13	.	.	PUNCT
fcis-3178	109	1	cost	cost	NOUN
fcis-3178	109	2	-	-	PUNCT
fcis-3178	109	3	sensitive	sensitive	ADJ
fcis-3178	109	4	loss	loss	NOUN
fcis-3178	109	5	layer	layer	NOUN
fcis-3178	109	6	for	for	ADP
fcis-3178	109	7	network	network	NOUN
fcis-3178	109	8	training	training	NOUN
fcis-3178	109	9	.	.	PUNCT
fcis-3178	110	1	references	reference	NOUN
fcis-3178	110	2	[	[	X
fcis-3178	110	3	1	1	NUM
fcis-3178	110	4	]	]	X
fcis-3178	110	5	tian	tian	ADJ
fcis-3178	110	6	,	,	PUNCT
fcis-3178	110	7	y.	y.	PROPN
fcis-3178	110	8	,	,	PUNCT
fcis-3178	110	9	kanade	kanade	PROPN
fcis-3178	110	10	,	,	PUNCT
fcis-3178	110	11	t.	t.	PROPN
fcis-3178	110	12	,	,	PUNCT
fcis-3178	110	13	and	and	CCONJ
fcis-3178	110	14	cohn	cohn	PROPN
fcis-3178	110	15	,	,	PUNCT
fcis-3178	110	16	j.	j.	PROPN
fcis-3178	110	17	f.	f.	PROPN
fcis-3178	110	18	(	(	PUNCT
fcis-3178	110	19	2001	2001	NUM
fcis-3178	110	20	)	)	PUNCT
fcis-3178	110	21	“	"	PUNCT
fcis-3178	110	22	recognizing	recognize	VERB
fcis-3178	110	23	action	action	NOUN
fcis-3178	110	24	units	unit	NOUN
fcis-3178	110	25	for	for	ADP
fcis-3178	110	26	facial	facial	ADJ
fcis-3178	110	27	expression	expression	NOUN
fcis-3178	110	28	analysis	analysis	NOUN
fcis-3178	110	29	,	,	PUNCT
fcis-3178	110	30	”	"	PUNCT
fcis-3178	110	31	ieee	ieee	NOUN
fcis-3178	110	32	transactions	transaction	NOUN
fcis-3178	110	33	on	on	ADP
fcis-3178	110	34	pattern	pattern	NOUN
fcis-3178	110	35	analysis	analysis	NOUN
fcis-3178	110	36	and	and	CCONJ
fcis-3178	110	37	machine	machine	NOUN
fcis-3178	110	38	intelligence	intelligence	NOUN
fcis-3178	110	39	,	,	PUNCT
fcis-3178	110	40	vol	vol	NOUN
fcis-3178	110	41	.	.	PROPN
fcis-3178	110	42	23	23	NUM
fcis-3178	110	43	,	,	PUNCT
fcis-3178	110	44	no	no	INTJ
fcis-3178	110	45	.	.	NOUN
fcis-3178	110	46	2	2	NUM
fcis-3178	110	47	,	,	PUNCT
fcis-3178	110	48	pp	pp	ADJ
fcis-3178	110	49	.	.	PUNCT
fcis-3178	111	1	97–115	97–115	NUM
fcis-3178	111	2	.	.	PUNCT
fcis-3178	112	1	[	[	X
fcis-3178	112	2	2	2	NUM
fcis-3178	112	3	]	]	PUNCT
fcis-3178	112	4	ekman	ekman	NOUN
fcis-3178	112	5	,	,	PUNCT
fcis-3178	112	6	p.	p.	PROPN
fcis-3178	112	7	,	,	PUNCT
fcis-3178	112	8	&	&	CCONJ
fcis-3178	112	9	friesen	friesen	PROPN
fcis-3178	112	10	,	,	PUNCT
fcis-3178	112	11	w.	w.	PROPN
fcis-3178	112	12	v.	v.	PROPN
fcis-3178	112	13	(	(	PUNCT
fcis-3178	112	14	1971	1971	NUM
fcis-3178	112	15	)	)	PUNCT
fcis-3178	112	16	.	.	PUNCT
fcis-3178	113	1	constants	constant	NOUN
fcis-3178	113	2	across	across	ADP
fcis-3178	113	3	cultures	culture	NOUN
fcis-3178	113	4	in	in	ADP
fcis-3178	113	5	the	the	DET
fcis-3178	113	6	face	face	NOUN
fcis-3178	113	7	and	and	CCONJ
fcis-3178	113	8	emotion	emotion	NOUN
fcis-3178	113	9	.	.	PUNCT
fcis-3178	114	1	journal	journal	NOUN
fcis-3178	114	2	of	of	ADP
fcis-3178	114	3	personality	personality	NOUN
fcis-3178	114	4	and	and	CCONJ
fcis-3178	114	5	social	social	ADJ
fcis-3178	114	6	psychology	psychology	NOUN
fcis-3178	114	7	,	,	PUNCT
fcis-3178	114	8	17(2	17(2	NUM
fcis-3178	114	9	)	)	PUNCT
fcis-3178	114	10	,	,	PUNCT
fcis-3178	114	11	124–129	124–129	NUM
fcis-3178	114	12	.	.	PUNCT
fcis-3178	115	1	https://doi.org/10.1037/h0030377	https://doi.org/10.1037/h0030377	NOUN
fcis-3178	115	2	.	.	PUNCT
fcis-3178	116	1	[	[	X
fcis-3178	116	2	3	3	NUM
fcis-3178	116	3	]	]	PUNCT
fcis-3178	116	4	m.	m.	NOUN
fcis-3178	116	5	shi	shi	PROPN
fcis-3178	116	6	,	,	PUNCT
fcis-3178	116	7	l.	l.	PROPN
fcis-3178	116	8	xu	xu	PROPN
fcis-3178	116	9	and	and	CCONJ
fcis-3178	116	10	x.	x.	PROPN
fcis-3178	116	11	chen	chen	PROPN
fcis-3178	116	12	,	,	PUNCT
fcis-3178	116	13	"	"	PUNCT
fcis-3178	116	14	a	a	DET
fcis-3178	116	15	novel	novel	ADJ
fcis-3178	116	16	facial	facial	ADJ
fcis-3178	116	17	expression	expression	NOUN
fcis-3178	116	18	intelligent	intelligent	ADJ
fcis-3178	116	19	recognition	recognition	NOUN
fcis-3178	116	20	method	method	NOUN
fcis-3178	116	21	using	use	VERB
fcis-3178	116	22	improved	improve	VERB
fcis-3178	116	23	convolutional	convolutional	ADJ
fcis-3178	116	24	neural	neural	ADJ
fcis-3178	116	25	network	network	NOUN
fcis-3178	116	26	,	,	PUNCT
fcis-3178	116	27	"	"	PUNCT
fcis-3178	116	28	in	in	ADP
fcis-3178	116	29	ieee	ieee	NOUN
fcis-3178	116	30	access	access	NOUN
fcis-3178	116	31	,	,	PUNCT
fcis-3178	116	32	vol	vol	NOUN
fcis-3178	116	33	.	.	PROPN
fcis-3178	116	34	8	8	NUM
fcis-3178	116	35	,	,	PUNCT
fcis-3178	116	36	pp	pp	ADJ
fcis-3178	116	37	.	.	PUNCT
fcis-3178	117	1	57606	57606	NUM
fcis-3178	117	2	-	-	SYM
fcis-3178	117	3	57614	57614	NUM
fcis-3178	117	4	,	,	PUNCT
fcis-3178	117	5	2020	2020	NUM
fcis-3178	117	6	,	,	PUNCT
fcis-3178	117	7	doi	doi	NOUN
fcis-3178	117	8	:	:	PUNCT
fcis-3178	117	9	10.1109	10.1109	NUM
fcis-3178	117	10	/	/	SYM
fcis-3178	117	11	access.2020.2982286	access.2020.2982286	ADJ
fcis-3178	117	12	.	.	PUNCT
fcis-3178	118	1	[	[	X
fcis-3178	118	2	4	4	X
fcis-3178	118	3	]	]	X
fcis-3178	118	4	lisa	lisa	PROPN
fcis-3178	118	5	f.	f.	PROPN
fcis-3178	118	6	b	b	PROPN
fcis-3178	118	7	,	,	PUNCT
fcis-3178	118	8	kristen	kristen	PROPN
fcis-3178	118	9	a.	a.	PROPN
fcis-3178	118	10	lt	lt	PROPN
fcis-3178	118	11	,	,	PUNCT
fcis-3178	118	12	maria	maria	PROPN
fcis-3178	118	13	g.	g.	PROPN
fcis-3178	118	14	language	language	NOUN
fcis-3178	118	15	as	as	ADP
fcis-3178	118	16	context	context	NOUN
fcis-3178	118	17	for	for	ADP
fcis-3178	118	18	the	the	DET
fcis-3178	118	19	perception	perception	NOUN
fcis-3178	118	20	of	of	ADP
fcis-3178	118	21	emotion	emotion	NOUN
fcis-3178	118	22	,	,	PUNCT
fcis-3178	118	23	trends	trend	NOUN
fcis-3178	118	24	in	in	ADP
fcis-3178	118	25	cognitive	cognitive	ADJ
fcis-3178	118	26	sciences	science	NOUN
fcis-3178	118	27	,	,	PUNCT
fcis-3178	118	28	volume	volume	NOUN
fcis-3178	118	29	11	11	NUM
fcis-3178	118	30	,	,	PUNCT
fcis-3178	118	31	issue	issue	NOUN
fcis-3178	118	32	8	8	NUM
fcis-3178	118	33	,	,	PUNCT
fcis-3178	118	34	2007	2007	NUM
fcis-3178	118	35	,	,	PUNCT
fcis-3178	118	36	pages	page	NOUN
fcis-3178	118	37	327	327	NUM
fcis-3178	118	38	-	-	SYM
fcis-3178	118	39	332	332	NUM
fcis-3178	118	40	,	,	PUNCT
fcis-3178	118	41	issn	issn	PROPN
fcis-3178	118	42	1364	1364	NUM
fcis-3178	118	43	-	-	SYM
fcis-3178	118	44	6613	6613	NUM
fcis-3178	118	45	,	,	PUNCT
fcis-3178	118	46	https://doi.org/10.1016/j.tics.2007.06.003	https://doi.org/10.1016/j.tics.2007.06.003	NOUN
fcis-3178	118	47	.	.	PUNCT
fcis-3178	119	1	(	(	PUNCT
fcis-3178	119	2	https://www.sciencedirect.com/science/article/pii/s13646613	https://www.sciencedirect.com/science/article/pii/s13646613	PROPN
fcis-3178	119	3	07001532	07001532	NUM
fcis-3178	119	4	)	)	PUNCT
fcis-3178	119	5	.	.	PUNCT
fcis-3178	120	1	[	[	X
fcis-3178	120	2	5	5	X
fcis-3178	120	3	]	]	PUNCT
fcis-3178	120	4	weimin	weimin	PROPN
fcis-3178	120	5	huang	huang	PROPN
fcis-3178	120	6	and	and	CCONJ
fcis-3178	120	7	r.	r.	PROPN
fcis-3178	120	8	mariani	mariani	PROPN
fcis-3178	120	9	,	,	PUNCT
fcis-3178	120	10	"	"	PUNCT
fcis-3178	120	11	face	face	NOUN
fcis-3178	120	12	detection	detection	NOUN
fcis-3178	120	13	and	and	CCONJ
fcis-3178	120	14	precise	precise	ADJ
fcis-3178	120	15	eyes	eye	NOUN
fcis-3178	120	16	location	location	NOUN
fcis-3178	120	17	,	,	PUNCT
fcis-3178	120	18	"	"	PUNCT
fcis-3178	120	19	proceedings	proceeding	NOUN
fcis-3178	120	20	15th	15th	ADJ
fcis-3178	120	21	international	international	ADJ
fcis-3178	120	22	conference	conference	NOUN
fcis-3178	120	23	on	on	ADP
fcis-3178	120	24	pattern	pattern	NOUN
fcis-3178	120	25	recognition	recognition	NOUN
fcis-3178	120	26	.	.	PUNCT
fcis-3178	121	1	icpr-2000	icpr-2000	NOUN
fcis-3178	121	2	,	,	PUNCT
fcis-3178	121	3	2000	2000	NUM
fcis-3178	121	4	,	,	PUNCT
fcis-3178	121	5	pp	pp	ADJ
fcis-3178	121	6	.	.	PUNCT
fcis-3178	122	1	722	722	NUM
fcis-3178	122	2	-	-	SYM
fcis-3178	122	3	727	727	NUM
fcis-3178	122	4	vol.4	vol.4	PROPN
fcis-3178	122	5	,	,	PUNCT
fcis-3178	122	6	doi	doi	NOUN
fcis-3178	122	7	:	:	PUNCT
fcis-3178	122	8	10.1109	10.1109	NUM
fcis-3178	122	9	/	/	SYM
fcis-3178	122	10	icpr.2000.903019	icpr.2000.903019	NOUN
fcis-3178	122	11	.	.	PUNCT
fcis-3178	123	1	[	[	X
fcis-3178	123	2	6	6	NUM
fcis-3178	123	3	]	]	PUNCT
fcis-3178	123	4	alzubaidi	alzubaidi	NOUN
fcis-3178	123	5	,	,	PUNCT
fcis-3178	123	6	l.	l.	PROPN
fcis-3178	123	7	,	,	PUNCT
fcis-3178	123	8	zhang	zhang	PROPN
fcis-3178	123	9	,	,	PUNCT
fcis-3178	123	10	j.	j.	PROPN
fcis-3178	123	11	,	,	PUNCT
fcis-3178	123	12	humaidi	humaidi	PROPN
fcis-3178	123	13	,	,	PUNCT
fcis-3178	123	14	a.j	a.j	PROPN
fcis-3178	123	15	.	.	PROPN
fcis-3178	123	16	et	et	PROPN
fcis-3178	124	1	al	al	PROPN
fcis-3178	124	2	.	.	PROPN
fcis-3178	124	3	review	review	NOUN
fcis-3178	124	4	of	of	ADP
fcis-3178	124	5	deep	deep	ADJ
fcis-3178	124	6	learning	learning	NOUN
fcis-3178	124	7	:	:	PUNCT
fcis-3178	125	1	concepts	concept	NOUN
fcis-3178	125	2	,	,	PUNCT
fcis-3178	125	3	cnn	cnn	PROPN
fcis-3178	125	4	architectures	architecture	NOUN
fcis-3178	125	5	,	,	PUNCT
fcis-3178	125	6	challenges	challenge	NOUN
fcis-3178	125	7	,	,	PUNCT
fcis-3178	125	8	applications	application	NOUN
fcis-3178	125	9	,	,	PUNCT
fcis-3178	125	10	future	future	ADJ
fcis-3178	125	11	directions	direction	NOUN
fcis-3178	125	12	.	.	PUNCT
fcis-3178	126	1	j	j	PROPN
fcis-3178	126	2	big	big	ADJ
fcis-3178	126	3	data	datum	NOUN
fcis-3178	126	4	8	8	NUM
fcis-3178	126	5	,	,	PUNCT
fcis-3178	126	6	53	53	NUM
fcis-3178	126	7	(	(	PUNCT
fcis-3178	126	8	2021	2021	NUM
fcis-3178	126	9	)	)	PUNCT
fcis-3178	126	10	.	.	PUNCT
fcis-3178	127	1	https://doi.org/10.1186/s40537-021-00444-8	https://doi.org/10.1186/s40537-021-00444-8	NOUN
fcis-3178	127	2	.	.	PUNCT
fcis-3178	128	1	[	[	X
fcis-3178	128	2	7	7	NUM
fcis-3178	128	3	]	]	X
fcis-3178	128	4	taylor	taylor	PROPN
fcis-3178	128	5	,	,	PUNCT
fcis-3178	128	6	g.w	g.w	PROPN
fcis-3178	128	7	.	.	PROPN
fcis-3178	128	8	,	,	PUNCT
fcis-3178	128	9	fergus	fergus	PROPN
fcis-3178	128	10	,	,	PUNCT
fcis-3178	128	11	r.	r.	PROPN
fcis-3178	128	12	,	,	PUNCT
fcis-3178	128	13	lecun	lecun	ADJ
fcis-3178	128	14	,	,	PUNCT
fcis-3178	128	15	y.	y.	NOUN
fcis-3178	128	16	,	,	PUNCT
fcis-3178	128	17	bregler	bregler	NOUN
fcis-3178	128	18	,	,	PUNCT
fcis-3178	128	19	c.	c.	PROPN
fcis-3178	128	20	(	(	PUNCT
fcis-3178	128	21	2010	2010	NUM
fcis-3178	128	22	)	)	PUNCT
fcis-3178	128	23	.	.	PUNCT
fcis-3178	129	1	convolutional	convolutional	ADJ
fcis-3178	129	2	learning	learning	NOUN
fcis-3178	129	3	of	of	ADP
fcis-3178	129	4	spatio	spatio	PROPN
fcis-3178	129	5	-	-	PUNCT
fcis-3178	129	6	temporal	temporal	ADJ
fcis-3178	129	7	features	feature	NOUN
fcis-3178	129	8	.	.	PUNCT
fcis-3178	130	1	in	in	ADP
fcis-3178	130	2	:	:	PUNCT
fcis-3178	130	3	daniilidis	daniilidis	PROPN
fcis-3178	130	4	,	,	PUNCT
fcis-3178	130	5	k.	k.	PROPN
fcis-3178	130	6	,	,	PUNCT
fcis-3178	130	7	maragos	marago	NOUN
fcis-3178	130	8	,	,	PUNCT
fcis-3178	130	9	p.	p.	NOUN
fcis-3178	130	10	,	,	PUNCT
fcis-3178	130	11	paragios	paragio	NOUN
fcis-3178	130	12	,	,	PUNCT
fcis-3178	130	13	n.	n.	NOUN
fcis-3178	130	14	(	(	PUNCT
fcis-3178	130	15	eds	ed	NOUN
fcis-3178	130	16	)	)	PUNCT
fcis-3178	130	17	computer	computer	NOUN
fcis-3178	130	18	vision	vision	NOUN
fcis-3178	130	19	–	–	PUNCT
fcis-3178	130	20	eccv	eccv	ADV
fcis-3178	130	21	2010	2010	NUM
fcis-3178	130	22	.	.	PUNCT
fcis-3178	131	1	eccv	eccv	ADJ
fcis-3178	131	2	2010	2010	NUM
fcis-3178	131	3	.	.	PUNCT
fcis-3178	132	1	lecture	lecture	NOUN
fcis-3178	132	2	notes	note	NOUN
fcis-3178	132	3	in	in	ADP
fcis-3178	132	4	computer	computer	NOUN
fcis-3178	132	5	science	science	NOUN
fcis-3178	132	6	,	,	PUNCT
fcis-3178	132	7	vol	vol	NOUN
fcis-3178	132	8	6316	6316	NUM
fcis-3178	132	9	.	.	PUNCT
fcis-3178	133	1	springer	springer	NOUN
fcis-3178	133	2	,	,	PUNCT
fcis-3178	133	3	berlin	berlin	PROPN
fcis-3178	133	4	,	,	PUNCT
fcis-3178	133	5	heidelberg	heidelberg	NOUN
fcis-3178	133	6	.	.	PUNCT
fcis-3178	134	1	https://doi.org/10.1007/978-3-642-15567-3_11	https://doi.org/10.1007/978-3-642-15567-3_11	PROPN
fcis-3178	134	2	.	.	PUNCT
fcis-3178	135	1	[	[	X
fcis-3178	135	2	8	8	NUM
fcis-3178	135	3	]	]	X
fcis-3178	135	4	h.	h.	PROPN
fcis-3178	135	5	jun	jun	PROPN
fcis-3178	135	6	,	,	PUNCT
fcis-3178	135	7	l.	l.	PROPN
fcis-3178	135	8	shuai	shuai	PROPN
fcis-3178	135	9	,	,	PUNCT
fcis-3178	135	10	s.	s.	PROPN
fcis-3178	135	11	jinming	jinming	PROPN
fcis-3178	135	12	,	,	PUNCT
fcis-3178	135	13	l.	l.	PROPN
fcis-3178	135	14	yue	yue	PROPN
fcis-3178	135	15	,	,	PUNCT
fcis-3178	135	16	w.	w.	PROPN
fcis-3178	135	17	jingwei	jingwei	PROPN
fcis-3178	135	18	and	and	CCONJ
fcis-3178	135	19	j.	j.	PROPN
fcis-3178	135	20	peng	peng	PROPN
fcis-3178	135	21	,	,	PUNCT
fcis-3178	135	22	"	"	PUNCT
fcis-3178	135	23	facial	facial	ADJ
fcis-3178	135	24	expression	expression	NOUN
fcis-3178	135	25	recognition	recognition	NOUN
fcis-3178	135	26	based	base	VERB
fcis-3178	135	27	on	on	ADP
fcis-3178	135	28	vggnet	vggnet	ADJ
fcis-3178	135	29	convolutional	convolutional	ADJ
fcis-3178	135	30	neural	neural	ADJ
fcis-3178	135	31	network	network	NOUN
fcis-3178	135	32	,	,	PUNCT
fcis-3178	135	33	"	"	PUNCT
fcis-3178	135	34	2018	2018	NUM
fcis-3178	135	35	chinese	chinese	ADJ
fcis-3178	135	36	automation	automation	NOUN
fcis-3178	135	37	congress	congress	PROPN
fcis-3178	135	38	(	(	PUNCT
fcis-3178	135	39	cac	cac	PROPN
fcis-3178	135	40	)	)	PUNCT
fcis-3178	135	41	,	,	PUNCT
fcis-3178	135	42	2018	2018	NUM
fcis-3178	135	43	,	,	PUNCT
fcis-3178	135	44	pp	pp	ADJ
fcis-3178	135	45	.	.	PUNCT
fcis-3178	135	46	4146	4146	NUM
fcis-3178	135	47	-	-	SYM
fcis-3178	135	48	4151	4151	NUM
fcis-3178	135	49	,	,	PUNCT
fcis-3178	135	50	doi	doi	NOUN
fcis-3178	135	51	:	:	PUNCT
fcis-3178	135	52	10.1109	10.1109	NUM
fcis-3178	135	53	/	/	SYM
fcis-3178	135	54	cac.2018.8623238	cac.2018.8623238	ADJ
fcis-3178	135	55	.	.	PUNCT
fcis-3178	136	1	https://doi.org/10.1037/h0030377	https://doi.org/10.1037/h0030377	NUM
fcis-3178	136	2	https://doi.org/10.1016/j.tics.2007.06.003	https://doi.org/10.1016/j.tics.2007.06.003	NOUN
fcis-3178	136	3	https://www.sciencedirect.com/science/article/pii/s1364661307001532	https://www.sciencedirect.com/science/article/pii/s1364661307001532	NOUN
fcis-3178	136	4	https://www.sciencedirect.com/science/article/pii/s1364661307001532	https://www.sciencedirect.com/science/article/pii/s1364661307001532	NOUN
fcis-3178	136	5	https://doi.org/10.1186/s40537-021-00444-8	https://doi.org/10.1186/s40537-021-00444-8	VERB