id	sid	tid	token	lemma	pos
fcis-14054	1	1	frontiers	frontier	NOUN
fcis-14054	1	2	in	in	ADP
fcis-14054	1	3	computing	computing	NOUN
fcis-14054	1	4	and	and	CCONJ
fcis-14054	1	5	intelligent	intelligent	ADJ
fcis-14054	1	6	systems	system	NOUN
fcis-14054	1	7	issn	issn	VERB
fcis-14054	1	8	:	:	PUNCT
fcis-14054	1	9	2832	2832	NUM
fcis-14054	1	10	-	-	SYM
fcis-14054	1	11	6024	6024	NUM
fcis-14054	1	12	|	|	NOUN
fcis-14054	1	13	vol	vol	NOUN
fcis-14054	1	14	.	.	PROPN
fcis-14054	2	1	5	5	NUM
fcis-14054	2	2	,	,	PUNCT
fcis-14054	2	3	no	no	INTJ
fcis-14054	2	4	.	.	NOUN
fcis-14054	2	5	3	3	NUM
fcis-14054	2	6	,	,	PUNCT
fcis-14054	2	7	2023	2023	NUM
fcis-14054	2	8	154	154	NUM
fcis-14054	2	9	image	image	NOUN
fcis-14054	2	10	retrieval	retrieval	NOUN
fcis-14054	2	11	based	base	VERB
fcis-14054	2	12	on	on	ADP
fcis-14054	2	13	deep	deep	ADJ
fcis-14054	2	14	learning	learn	VERB
fcis-14054	2	15	yuru	yuru	PROPN
fcis-14054	2	16	gao	gao	PROPN
fcis-14054	2	17	*	*	PUNCT
fcis-14054	2	18	school	school	NOUN
fcis-14054	2	19	of	of	ADP
fcis-14054	2	20	mathematics	mathematic	NOUN
fcis-14054	2	21	,	,	PUNCT
fcis-14054	2	22	xi’an	xi’an	PROPN
fcis-14054	2	23	university	university	PROPN
fcis-14054	2	24	of	of	ADP
fcis-14054	2	25	finance	finance	NOUN
fcis-14054	2	26	and	and	CCONJ
fcis-14054	2	27	economics	economic	NOUN
fcis-14054	2	28	,	,	PUNCT
fcis-14054	2	29	xi’an	xi’an	PROPN
fcis-14054	2	30	,	,	PUNCT
fcis-14054	2	31	710000	710000	NUM
fcis-14054	2	32	,	,	PUNCT
fcis-14054	2	33	china	china	PROPN
fcis-14054	2	34	*	*	PUNCT
fcis-14054	2	35	corresponding	correspond	VERB
fcis-14054	2	36	author	author	NOUN
fcis-14054	2	37	email	email	NOUN
fcis-14054	2	38	:	:	PUNCT
fcis-14054	3	1	gaoyyyu@outlook.com	gaoyyyu@outlook.com	X
fcis-14054	3	2	abstract	abstract	NOUN
fcis-14054	3	3	:	:	PUNCT
fcis-14054	3	4	with	with	ADP
fcis-14054	3	5	the	the	DET
fcis-14054	3	6	rapid	rapid	ADJ
fcis-14054	3	7	growth	growth	NOUN
fcis-14054	3	8	of	of	ADP
fcis-14054	3	9	image	image	NOUN
fcis-14054	3	10	data	datum	NOUN
fcis-14054	3	11	,	,	PUNCT
fcis-14054	3	12	how	how	SCONJ
fcis-14054	3	13	to	to	PART
fcis-14054	3	14	efficiently	efficiently	ADV
fcis-14054	3	15	and	and	CCONJ
fcis-14054	3	16	accurately	accurately	ADV
fcis-14054	3	17	extract	extract	VERB
fcis-14054	3	18	useful	useful	ADJ
fcis-14054	3	19	features	feature	NOUN
fcis-14054	3	20	from	from	ADP
fcis-14054	3	21	massive	massive	ADJ
fcis-14054	3	22	image	image	NOUN
fcis-14054	3	23	data	datum	NOUN
fcis-14054	3	24	and	and	CCONJ
fcis-14054	3	25	perform	perform	VERB
fcis-14054	3	26	fast	fast	ADJ
fcis-14054	3	27	image	image	NOUN
fcis-14054	3	28	retrieval	retrieval	NOUN
fcis-14054	3	29	has	have	AUX
fcis-14054	3	30	become	become	VERB
fcis-14054	3	31	an	an	DET
fcis-14054	3	32	important	important	ADJ
fcis-14054	3	33	research	research	NOUN
fcis-14054	3	34	direction	direction	NOUN
fcis-14054	3	35	.	.	PUNCT
fcis-14054	4	1	this	this	DET
fcis-14054	4	2	study	study	NOUN
fcis-14054	4	3	focuses	focus	VERB
fcis-14054	4	4	on	on	ADP
fcis-14054	4	5	the	the	DET
fcis-14054	4	6	design	design	NOUN
fcis-14054	4	7	and	and	CCONJ
fcis-14054	4	8	training	training	NOUN
fcis-14054	4	9	of	of	ADP
fcis-14054	4	10	deep	deep	ADJ
fcis-14054	4	11	learning	learning	NOUN
fcis-14054	4	12	-	-	PUNCT
fcis-14054	4	13	based	base	VERB
fcis-14054	4	14	image	image	NOUN
fcis-14054	4	15	feature	feature	NOUN
fcis-14054	4	16	extraction	extraction	NOUN
fcis-14054	4	17	networks	network	NOUN
fcis-14054	4	18	to	to	PART
fcis-14054	4	19	improve	improve	VERB
fcis-14054	4	20	the	the	DET
fcis-14054	4	21	robustness	robustness	NOUN
fcis-14054	4	22	and	and	CCONJ
fcis-14054	4	23	generalization	generalization	NOUN
fcis-14054	4	24	of	of	ADP
fcis-14054	4	25	image	image	NOUN
fcis-14054	4	26	features	feature	NOUN
fcis-14054	4	27	by	by	ADP
fcis-14054	4	28	optimizing	optimize	VERB
fcis-14054	4	29	the	the	DET
fcis-14054	4	30	network	network	NOUN
fcis-14054	4	31	structure	structure	NOUN
fcis-14054	4	32	and	and	CCONJ
fcis-14054	4	33	loss	loss	NOUN
fcis-14054	4	34	function	function	NOUN
fcis-14054	4	35	.	.	PUNCT
fcis-14054	5	1	in	in	ADP
fcis-14054	5	2	order	order	NOUN
fcis-14054	5	3	to	to	PART
fcis-14054	5	4	evaluate	evaluate	VERB
fcis-14054	5	5	the	the	DET
fcis-14054	5	6	performance	performance	NOUN
fcis-14054	5	7	of	of	ADP
fcis-14054	5	8	the	the	DET
fcis-14054	5	9	system	system	NOUN
fcis-14054	5	10	,	,	PUNCT
fcis-14054	5	11	this	this	DET
fcis-14054	5	12	study	study	NOUN
fcis-14054	5	13	also	also	ADV
fcis-14054	5	14	designs	design	VERB
fcis-14054	5	15	appropriate	appropriate	ADJ
fcis-14054	5	16	evaluation	evaluation	NOUN
fcis-14054	5	17	indicators	indicator	NOUN
fcis-14054	5	18	and	and	CCONJ
fcis-14054	5	19	conducts	conduct	VERB
fcis-14054	5	20	corresponding	correspond	VERB
fcis-14054	5	21	experiments	experiment	NOUN
fcis-14054	5	22	.	.	PUNCT
fcis-14054	6	1	through	through	ADP
fcis-14054	6	2	experimental	experimental	ADJ
fcis-14054	6	3	verification	verification	NOUN
fcis-14054	6	4	,	,	PUNCT
fcis-14054	6	5	the	the	DET
fcis-14054	6	6	results	result	NOUN
fcis-14054	6	7	show	show	VERB
fcis-14054	6	8	that	that	SCONJ
fcis-14054	6	9	these	these	DET
fcis-14054	6	10	methods	method	NOUN
fcis-14054	6	11	can	can	AUX
fcis-14054	6	12	effectively	effectively	ADV
fcis-14054	6	13	improve	improve	VERB
fcis-14054	6	14	the	the	DET
fcis-14054	6	15	performance	performance	NOUN
fcis-14054	6	16	of	of	ADP
fcis-14054	6	17	image	image	NOUN
fcis-14054	6	18	feature	feature	NOUN
fcis-14054	6	19	extraction	extraction	NOUN
fcis-14054	6	20	and	and	CCONJ
fcis-14054	6	21	image	image	NOUN
fcis-14054	6	22	retrieval	retrieval	NOUN
fcis-14054	6	23	,	,	PUNCT
fcis-14054	6	24	and	and	CCONJ
fcis-14054	6	25	have	have	VERB
fcis-14054	6	26	broad	broad	ADJ
fcis-14054	6	27	potential	potential	NOUN
fcis-14054	6	28	in	in	ADP
fcis-14054	6	29	practical	practical	ADJ
fcis-14054	6	30	applications	application	NOUN
fcis-14054	6	31	.	.	PUNCT
fcis-14054	7	1	keywords	keyword	NOUN
fcis-14054	7	2	:	:	PUNCT
fcis-14054	7	3	deep	deep	ADJ
fcis-14054	7	4	learning	learning	NOUN
fcis-14054	7	5	;	;	PUNCT
fcis-14054	7	6	image	image	NOUN
fcis-14054	7	7	feature	feature	NOUN
fcis-14054	7	8	extraction	extraction	NOUN
fcis-14054	7	9	;	;	PUNCT
fcis-14054	7	10	image	image	NOUN
fcis-14054	7	11	retrieval	retrieval	NOUN
fcis-14054	7	12	;	;	PUNCT
fcis-14054	7	13	application	application	NOUN
fcis-14054	7	14	scenarios	scenario	NOUN
fcis-14054	7	15	.	.	PUNCT
fcis-14054	8	1	1	1	X
fcis-14054	8	2	.	.	X
fcis-14054	8	3	introduction	introduction	NOUN
fcis-14054	8	4	1.1	1.1	NUM
fcis-14054	8	5	.	.	PUNCT
fcis-14054	9	1	research	research	NOUN
fcis-14054	9	2	background	background	NOUN
fcis-14054	9	3	and	and	CCONJ
fcis-14054	9	4	significance	significance	NOUN
fcis-14054	9	5	1.1.1	1.1.1	NUM
fcis-14054	9	6	.	.	PUNCT
fcis-14054	9	7	significance	significance	NOUN
fcis-14054	9	8	of	of	ADP
fcis-14054	9	9	image	image	NOUN
fcis-14054	9	10	retrieval	retrieval	NOUN
fcis-14054	9	11	in	in	ADP
fcis-14054	9	12	recent	recent	ADJ
fcis-14054	9	13	years	year	NOUN
fcis-14054	9	14	with	with	ADP
fcis-14054	9	15	the	the	DET
fcis-14054	9	16	rapid	rapid	ADJ
fcis-14054	9	17	development	development	NOUN
fcis-14054	9	18	of	of	ADP
fcis-14054	9	19	information	information	NOUN
fcis-14054	9	20	technology	technology	NOUN
fcis-14054	9	21	and	and	CCONJ
fcis-14054	9	22	the	the	DET
fcis-14054	9	23	popularization	popularization	NOUN
fcis-14054	9	24	of	of	ADP
fcis-14054	9	25	the	the	DET
fcis-14054	9	26	internet	internet	NOUN
fcis-14054	10	1	,	,	PUNCT
fcis-14054	10	2	image	image	NOUN
fcis-14054	10	3	data	datum	NOUN
fcis-14054	10	4	shows	show	VERB
fcis-14054	10	5	an	an	DET
fcis-14054	10	6	explosive	explosive	ADJ
fcis-14054	10	7	growth	growth	NOUN
fcis-14054	10	8	trend	trend	NOUN
fcis-14054	10	9	.	.	PUNCT
fcis-14054	11	1	however	however	ADV
fcis-14054	11	2	,	,	PUNCT
fcis-14054	11	3	due	due	ADP
fcis-14054	11	4	to	to	ADP
fcis-14054	11	5	the	the	DET
fcis-14054	11	6	high	high	ADJ
fcis-14054	11	7	feature	feature	NOUN
fcis-14054	11	8	dimension	dimension	NOUN
fcis-14054	11	9	of	of	ADP
fcis-14054	11	10	images	image	NOUN
fcis-14054	11	11	and	and	CCONJ
fcis-14054	11	12	the	the	DET
fcis-14054	11	13	large	large	ADJ
fcis-14054	11	14	uncertainty	uncertainty	NOUN
fcis-14054	11	15	of	of	ADP
fcis-14054	11	16	image	image	NOUN
fcis-14054	11	17	content	content	NOUN
fcis-14054	11	18	,	,	PUNCT
fcis-14054	11	19	effective	effective	ADJ
fcis-14054	11	20	and	and	CCONJ
fcis-14054	11	21	efficient	efficient	ADJ
fcis-14054	11	22	image	image	NOUN
fcis-14054	11	23	retrieval	retrieval	NOUN
fcis-14054	11	24	has	have	AUX
fcis-14054	11	25	become	become	VERB
fcis-14054	11	26	a	a	DET
fcis-14054	11	27	challenge	challenge	NOUN
fcis-14054	11	28	.	.	PUNCT
fcis-14054	12	1	therefore	therefore	ADV
fcis-14054	12	2	,	,	PUNCT
fcis-14054	12	3	it	it	PRON
fcis-14054	12	4	is	be	AUX
fcis-14054	12	5	of	of	ADP
fcis-14054	12	6	great	great	ADJ
fcis-14054	12	7	significance	significance	NOUN
fcis-14054	12	8	to	to	PART
fcis-14054	12	9	study	study	VERB
fcis-14054	12	10	image	image	NOUN
fcis-14054	12	11	retrieval	retrieval	NOUN
fcis-14054	12	12	methods	method	NOUN
fcis-14054	12	13	to	to	PART
fcis-14054	12	14	meet	meet	VERB
fcis-14054	12	15	people	people	NOUN
fcis-14054	12	16	's	's	PART
fcis-14054	12	17	needs	need	NOUN
fcis-14054	12	18	for	for	ADP
fcis-14054	12	19	image	image	NOUN
fcis-14054	12	20	information	information	NOUN
fcis-14054	12	21	.	.	PUNCT
fcis-14054	13	1	1.1.2	1.1.2	X
fcis-14054	13	2	.	.	PUNCT
fcis-14054	14	1	the	the	DET
fcis-14054	14	2	rapid	rapid	ADJ
fcis-14054	14	3	development	development	NOUN
fcis-14054	14	4	of	of	ADP
fcis-14054	14	5	deep	deep	ADJ
fcis-14054	14	6	learning	learning	NOUN
fcis-14054	14	7	in	in	ADP
fcis-14054	14	8	the	the	DET
fcis-14054	14	9	field	field	NOUN
fcis-14054	14	10	of	of	ADP
fcis-14054	14	11	image	image	NOUN
fcis-14054	14	12	processing	processing	NOUN
fcis-14054	14	13	deep	deep	ADJ
fcis-14054	14	14	learning	learning	NOUN
fcis-14054	14	15	is	be	AUX
fcis-14054	14	16	a	a	DET
fcis-14054	14	17	machine	machine	NOUN
fcis-14054	14	18	learning	learning	NOUN
fcis-14054	14	19	method	method	NOUN
fcis-14054	14	20	based	base	VERB
fcis-14054	14	21	on	on	ADP
fcis-14054	14	22	artificial	artificial	ADJ
fcis-14054	14	23	neural	neural	ADJ
fcis-14054	14	24	networks	network	NOUN
fcis-14054	14	25	.	.	PUNCT
fcis-14054	15	1	by	by	ADP
fcis-14054	15	2	using	use	VERB
fcis-14054	15	3	depth	depth	NOUN
fcis-14054	15	4	learning	learning	NOUN
fcis-14054	15	5	models	model	NOUN
fcis-14054	15	6	,	,	PUNCT
fcis-14054	15	7	feature	feature	VERB
fcis-14054	15	8	representations	representation	NOUN
fcis-14054	15	9	with	with	ADP
fcis-14054	15	10	more	more	ADJ
fcis-14054	15	11	semantic	semantic	ADJ
fcis-14054	15	12	information	information	NOUN
fcis-14054	15	13	can	can	AUX
fcis-14054	15	14	be	be	AUX
fcis-14054	15	15	extracted	extract	VERB
fcis-14054	15	16	,	,	PUNCT
fcis-14054	15	17	thereby	thereby	ADV
fcis-14054	15	18	improving	improve	VERB
fcis-14054	15	19	the	the	DET
fcis-14054	15	20	accuracy	accuracy	NOUN
fcis-14054	15	21	and	and	CCONJ
fcis-14054	15	22	efficiency	efficiency	NOUN
fcis-14054	15	23	of	of	ADP
fcis-14054	15	24	image	image	NOUN
fcis-14054	15	25	retrieval	retrieval	NOUN
fcis-14054	15	26	.	.	PUNCT
fcis-14054	16	1	this	this	DET
fcis-14054	16	2	paper	paper	NOUN
fcis-14054	16	3	will	will	AUX
fcis-14054	16	4	study	study	VERB
fcis-14054	16	5	deep	deep	ADJ
fcis-14054	16	6	learning	learning	NOUN
fcis-14054	16	7	-	-	PUNCT
fcis-14054	16	8	based	base	VERB
fcis-14054	16	9	image	image	NOUN
fcis-14054	16	10	retrieval	retrieval	NOUN
fcis-14054	16	11	methods	method	NOUN
fcis-14054	16	12	and	and	CCONJ
fcis-14054	16	13	optimize	optimize	VERB
fcis-14054	16	14	them	they	PRON
fcis-14054	16	15	for	for	ADP
fcis-14054	16	16	the	the	DET
fcis-14054	16	17	problems	problem	NOUN
fcis-14054	16	18	existing	exist	VERB
fcis-14054	16	19	in	in	ADP
fcis-14054	16	20	existing	exist	VERB
fcis-14054	16	21	methods	method	NOUN
fcis-14054	16	22	.	.	PUNCT
fcis-14054	17	1	through	through	ADP
fcis-14054	17	2	in	in	ADP
fcis-14054	17	3	-	-	PUNCT
fcis-14054	17	4	depth	depth	NOUN
fcis-14054	17	5	analysis	analysis	NOUN
fcis-14054	17	6	and	and	CCONJ
fcis-14054	17	7	experimental	experimental	ADJ
fcis-14054	17	8	verification	verification	NOUN
fcis-14054	17	9	,	,	PUNCT
fcis-14054	17	10	an	an	DET
fcis-14054	17	11	efficient	efficient	ADJ
fcis-14054	17	12	and	and	CCONJ
fcis-14054	17	13	accurate	accurate	ADJ
fcis-14054	17	14	image	image	NOUN
fcis-14054	17	15	retrieval	retrieval	NOUN
fcis-14054	17	16	scheme	scheme	NOUN
fcis-14054	17	17	is	be	AUX
fcis-14054	17	18	proposed	propose	VERB
fcis-14054	17	19	to	to	PART
fcis-14054	17	20	meet	meet	VERB
fcis-14054	17	21	the	the	DET
fcis-14054	17	22	needs	need	NOUN
fcis-14054	17	23	of	of	ADP
fcis-14054	17	24	practical	practical	ADJ
fcis-14054	17	25	applications	application	NOUN
fcis-14054	17	26	.	.	PUNCT
fcis-14054	18	1	first	first	ADV
fcis-14054	18	2	,	,	PUNCT
fcis-14054	18	3	this	this	DET
fcis-14054	18	4	paper	paper	NOUN
fcis-14054	18	5	will	will	AUX
fcis-14054	18	6	introduce	introduce	VERB
fcis-14054	18	7	the	the	DET
fcis-14054	18	8	basic	basic	ADJ
fcis-14054	18	9	principles	principle	NOUN
fcis-14054	18	10	and	and	CCONJ
fcis-14054	18	11	technologies	technology	NOUN
fcis-14054	18	12	of	of	ADP
fcis-14054	18	13	deep	deep	ADJ
fcis-14054	18	14	learning	learning	NOUN
fcis-14054	18	15	in	in	ADP
fcis-14054	18	16	image	image	NOUN
fcis-14054	18	17	processing	processing	NOUN
fcis-14054	18	18	,	,	PUNCT
fcis-14054	18	19	and	and	CCONJ
fcis-14054	18	20	discuss	discuss	VERB
fcis-14054	18	21	its	its	PRON
fcis-14054	18	22	advantages	advantage	NOUN
fcis-14054	18	23	and	and	CCONJ
fcis-14054	18	24	challenges	challenge	NOUN
fcis-14054	18	25	;	;	PUNCT
fcis-14054	18	26	then	then	ADV
fcis-14054	18	27	,	,	PUNCT
fcis-14054	18	28	an	an	DET
fcis-14054	18	29	improved	improved	ADJ
fcis-14054	18	30	image	image	NOUN
fcis-14054	18	31	retrieval	retrieval	NOUN
fcis-14054	18	32	method	method	NOUN
fcis-14054	18	33	will	will	AUX
fcis-14054	18	34	be	be	AUX
fcis-14054	18	35	proposed	propose	VERB
fcis-14054	18	36	,	,	PUNCT
fcis-14054	18	37	and	and	CCONJ
fcis-14054	18	38	experimental	experimental	ADJ
fcis-14054	18	39	verification	verification	NOUN
fcis-14054	18	40	and	and	CCONJ
fcis-14054	18	41	performance	performance	NOUN
fcis-14054	18	42	evaluation	evaluation	NOUN
fcis-14054	18	43	will	will	AUX
fcis-14054	18	44	be	be	AUX
fcis-14054	18	45	carried	carry	VERB
fcis-14054	18	46	out	out	ADP
fcis-14054	18	47	;	;	PUNCT
fcis-14054	18	48	finally	finally	ADV
fcis-14054	18	49	,	,	PUNCT
fcis-14054	18	50	this	this	DET
fcis-14054	18	51	paper	paper	NOUN
fcis-14054	18	52	will	will	AUX
fcis-14054	18	53	summarize	summarize	VERB
fcis-14054	18	54	the	the	DET
fcis-14054	18	55	research	research	NOUN
fcis-14054	18	56	results	result	NOUN
fcis-14054	18	57	and	and	CCONJ
fcis-14054	18	58	put	put	VERB
fcis-14054	18	59	forward	forward	ADP
fcis-14054	18	60	prospects	prospect	NOUN
fcis-14054	18	61	for	for	ADP
fcis-14054	18	62	future	future	ADJ
fcis-14054	18	63	research	research	NOUN
fcis-14054	18	64	directions	direction	NOUN
fcis-14054	18	65	.	.	PUNCT
fcis-14054	19	1	1.2	1.2	NUM
fcis-14054	19	2	.	.	PUNCT
fcis-14054	19	3	research	research	NOUN
fcis-14054	19	4	status	status	NOUN
fcis-14054	19	5	at	at	ADP
fcis-14054	19	6	home	home	ADV
fcis-14054	19	7	and	and	CCONJ
fcis-14054	19	8	abroad	abroad	ADV
fcis-14054	19	9	in	in	ADP
fcis-14054	19	10	order	order	NOUN
fcis-14054	19	11	to	to	PART
fcis-14054	19	12	accurately	accurately	ADV
fcis-14054	19	13	retrieve	retrieve	VERB
fcis-14054	19	14	the	the	DET
fcis-14054	19	15	corresponding	correspond	VERB
fcis-14054	19	16	image	image	NOUN
fcis-14054	19	17	,	,	PUNCT
fcis-14054	19	18	it	it	PRON
fcis-14054	19	19	is	be	AUX
fcis-14054	19	20	necessary	necessary	ADJ
fcis-14054	19	21	to	to	PART
fcis-14054	19	22	accurately	accurately	ADV
fcis-14054	19	23	extract	extract	VERB
fcis-14054	19	24	its	its	PRON
fcis-14054	19	25	image	image	NOUN
fcis-14054	19	26	features	feature	NOUN
fcis-14054	19	27	,	,	PUNCT
fcis-14054	19	28	such	such	ADJ
fcis-14054	19	29	as	as	ADP
fcis-14054	19	30	traditional	traditional	ADJ
fcis-14054	19	31	manual	manual	ADJ
fcis-14054	19	32	features	feature	NOUN
fcis-14054	19	33	and	and	CCONJ
fcis-14054	19	34	deep	deep	ADJ
fcis-14054	19	35	feature	feature	NOUN
fcis-14054	19	36	learning	learn	VERB
fcis-14054	19	37	methods	method	NOUN
fcis-14054	19	38	that	that	PRON
fcis-14054	19	39	are	be	AUX
fcis-14054	19	40	widely	widely	ADV
fcis-14054	19	41	used	use	VERB
fcis-14054	19	42	now	now	ADV
fcis-14054	19	43	.	.	PUNCT
fcis-14054	20	1	many	many	ADJ
fcis-14054	20	2	scientific	scientific	ADJ
fcis-14054	20	3	research	research	NOUN
fcis-14054	20	4	institutions	institution	NOUN
fcis-14054	20	5	and	and	CCONJ
fcis-14054	20	6	universities	university	NOUN
fcis-14054	20	7	abroad	abroad	ADV
fcis-14054	20	8	have	have	AUX
fcis-14054	20	9	conducted	conduct	VERB
fcis-14054	20	10	in	in	ADP
fcis-14054	20	11	-	-	PUNCT
fcis-14054	20	12	depth	depth	NOUN
fcis-14054	20	13	research	research	NOUN
fcis-14054	20	14	on	on	ADP
fcis-14054	20	15	image	image	NOUN
fcis-14054	20	16	retrieval	retrieval	NOUN
fcis-14054	20	17	technology	technology	NOUN
fcis-14054	20	18	,	,	PUNCT
fcis-14054	20	19	including	include	VERB
fcis-14054	20	20	ibm	ibm	PROPN
fcis-14054	20	21	,	,	PUNCT
fcis-14054	20	22	mit	mit	NOUN
fcis-14054	20	23	,	,	PUNCT
fcis-14054	20	24	etc	etc	X
fcis-14054	20	25	.	.	X
fcis-14054	20	26	well	well	ADV
fcis-14054	20	27	-	-	PUNCT
fcis-14054	20	28	known	know	VERB
fcis-14054	20	29	universities	university	NOUN
fcis-14054	20	30	such	such	ADJ
fcis-14054	20	31	as	as	ADP
fcis-14054	20	32	columbia	columbia	PROPN
fcis-14054	20	33	,	,	PUNCT
fcis-14054	20	34	stanford	stanford	PROPN
fcis-14054	20	35	,	,	PUNCT
fcis-14054	20	36	uiuc	uiuc	PROPN
fcis-14054	20	37	,	,	PUNCT
fcis-14054	20	38	etc	etc	X
fcis-14054	20	39	.	.	X
fcis-14054	20	40	have	have	AUX
fcis-14054	20	41	also	also	ADV
fcis-14054	20	42	done	do	VERB
fcis-14054	20	43	a	a	DET
fcis-14054	20	44	lot	lot	NOUN
fcis-14054	20	45	of	of	ADP
fcis-14054	20	46	related	related	ADJ
fcis-14054	20	47	research	research	NOUN
fcis-14054	20	48	in	in	ADP
fcis-14054	20	49	this	this	DET
fcis-14054	20	50	field	field	NOUN
fcis-14054	21	1	[	[	X
fcis-14054	21	2	1	1	NUM
fcis-14054	21	3	]	]	PUNCT
fcis-14054	21	4	.	.	PUNCT
fcis-14054	22	1	in	in	ADP
fcis-14054	22	2	china	china	PROPN
fcis-14054	22	3	,	,	PUNCT
fcis-14054	22	4	cbir	cbir	PROPN
fcis-14054	22	5	has	have	AUX
fcis-14054	22	6	also	also	ADV
fcis-14054	22	7	been	be	AUX
fcis-14054	22	8	successively	successively	ADV
fcis-14054	22	9	studied	study	VERB
fcis-14054	22	10	by	by	ADP
fcis-14054	22	11	major	major	ADJ
fcis-14054	22	12	universities	university	NOUN
fcis-14054	22	13	and	and	CCONJ
fcis-14054	22	14	scientific	scientific	ADJ
fcis-14054	22	15	research	research	NOUN
fcis-14054	22	16	institutions	institution	NOUN
fcis-14054	22	17	in	in	ADP
fcis-14054	22	18	the	the	DET
fcis-14054	22	19	fields	field	NOUN
fcis-14054	22	20	of	of	ADP
fcis-14054	22	21	audio	audio	NOUN
fcis-14054	22	22	and	and	CCONJ
fcis-14054	22	23	video	video	NOUN
fcis-14054	22	24	.	.	PUNCT
fcis-14054	23	1	among	among	ADP
fcis-14054	23	2	them	they	PRON
fcis-14054	23	3	,	,	PUNCT
fcis-14054	23	4	tsinghua	tsinghua	PROPN
fcis-14054	23	5	university	university	PROPN
fcis-14054	23	6	has	have	AUX
fcis-14054	23	7	carried	carry	VERB
fcis-14054	23	8	out	out	ADP
fcis-14054	23	9	research	research	NOUN
fcis-14054	23	10	on	on	ADP
fcis-14054	23	11	"	"	PUNCT
fcis-14054	23	12	content	content	NOUN
fcis-14054	23	13	-	-	PUNCT
fcis-14054	23	14	based	base	VERB
fcis-14054	23	15	image	image	NOUN
fcis-14054	23	16	retrieval	retrieval	NOUN
fcis-14054	23	17	on	on	ADP
fcis-14054	23	18	the	the	DET
fcis-14054	23	19	internet	internet	NOUN
fcis-14054	23	20	"	"	PUNCT
fcis-14054	23	21	;	;	PUNCT
fcis-14054	23	22	zhejiang	zhejiang	PROPN
fcis-14054	23	23	university	university	PROPN
fcis-14054	23	24	has	have	AUX
fcis-14054	23	25	also	also	ADV
fcis-14054	23	26	developed	develop	VERB
fcis-14054	23	27	image	image	NOUN
fcis-14054	23	28	retrieval	retrieval	NOUN
fcis-14054	23	29	systems	system	NOUN
fcis-14054	23	30	such	such	ADJ
fcis-14054	23	31	as	as	ADP
fcis-14054	23	32	photo	photo	NOUN
fcis-14054	23	33	navigator	navigator	NOUN
fcis-14054	23	34	and	and	CCONJ
fcis-14054	23	35	web	web	NOUN
fcis-14054	23	36	scope	scope	NOUN
fcis-14054	23	37	cbr	cbr	PROPN
fcis-14054	23	38	here	here	ADV
fcis-14054	23	39	.	.	PUNCT
fcis-14054	24	1	1.3	1.3	NUM
fcis-14054	24	2	.	.	PUNCT
fcis-14054	25	1	the	the	DET
fcis-14054	25	2	main	main	ADJ
fcis-14054	25	3	content	content	NOUN
fcis-14054	25	4	of	of	ADP
fcis-14054	25	5	the	the	DET
fcis-14054	25	6	paper	paper	NOUN
fcis-14054	25	7	the	the	DET
fcis-14054	25	8	purpose	purpose	NOUN
fcis-14054	25	9	of	of	ADP
fcis-14054	25	10	this	this	DET
fcis-14054	25	11	study	study	NOUN
fcis-14054	25	12	is	be	AUX
fcis-14054	25	13	to	to	PART
fcis-14054	25	14	explore	explore	VERB
fcis-14054	25	15	the	the	DET
fcis-14054	25	16	development	development	NOUN
fcis-14054	25	17	and	and	CCONJ
fcis-14054	25	18	application	application	NOUN
fcis-14054	25	19	of	of	ADP
fcis-14054	25	20	image	image	NOUN
fcis-14054	25	21	retrieval	retrieval	NOUN
fcis-14054	25	22	methods	method	NOUN
fcis-14054	25	23	based	base	VERB
fcis-14054	25	24	on	on	ADP
fcis-14054	25	25	deep	deep	ADJ
fcis-14054	25	26	learning	learning	NOUN
fcis-14054	25	27	.	.	PUNCT
fcis-14054	26	1	the	the	DET
fcis-14054	26	2	specific	specific	ADJ
fcis-14054	26	3	main	main	ADJ
fcis-14054	26	4	contents	content	NOUN
fcis-14054	26	5	include	include	VERB
fcis-14054	26	6	:	:	PUNCT
fcis-14054	26	7	(	(	PUNCT
fcis-14054	26	8	1	1	X
fcis-14054	26	9	)	)	PUNCT
fcis-14054	26	10	the	the	DET
fcis-14054	26	11	application	application	NOUN
fcis-14054	26	12	of	of	ADP
fcis-14054	26	13	deep	deep	ADJ
fcis-14054	26	14	learning	learning	NOUN
fcis-14054	26	15	in	in	ADP
fcis-14054	26	16	image	image	NOUN
fcis-14054	26	17	retrieval	retrieval	NOUN
fcis-14054	26	18	.	.	PUNCT
fcis-14054	27	1	(	(	PUNCT
fcis-14054	27	2	2	2	X
fcis-14054	27	3	)	)	PUNCT
fcis-14054	27	4	image	image	NOUN
fcis-14054	27	5	retrieval	retrieval	NOUN
fcis-14054	27	6	design	design	NOUN
fcis-14054	27	7	and	and	CCONJ
fcis-14054	27	8	optimization	optimization	NOUN
fcis-14054	27	9	based	base	VERB
fcis-14054	27	10	on	on	ADP
fcis-14054	27	11	deep	deep	ADJ
fcis-14054	27	12	learning	learning	NOUN
fcis-14054	27	13	.	.	PUNCT
fcis-14054	28	1	(	(	PUNCT
fcis-14054	28	2	3	3	X
fcis-14054	28	3	)	)	PUNCT
fcis-14054	28	4	conclusions	conclusion	NOUN
fcis-14054	28	5	and	and	CCONJ
fcis-14054	28	6	prospects	prospect	NOUN
fcis-14054	28	7	.	.	PUNCT
fcis-14054	29	1	2	2	X
fcis-14054	29	2	.	.	X
fcis-14054	29	3	application	application	NOUN
fcis-14054	29	4	of	of	ADP
fcis-14054	29	5	deep	deep	ADJ
fcis-14054	29	6	learning	learning	NOUN
fcis-14054	29	7	in	in	ADP
fcis-14054	29	8	image	image	NOUN
fcis-14054	29	9	retrieval	retrieval	NOUN
fcis-14054	29	10	2.1	2.1	NUM
fcis-14054	29	11	.	.	PUNCT
fcis-14054	29	12	basic	basic	ADJ
fcis-14054	29	13	principles	principle	NOUN
fcis-14054	29	14	and	and	CCONJ
fcis-14054	29	15	common	common	ADJ
fcis-14054	29	16	models	model	NOUN
fcis-14054	29	17	of	of	ADP
fcis-14054	29	18	deep	deep	ADJ
fcis-14054	29	19	learning	learn	VERB
fcis-14054	29	20	the	the	DET
fcis-14054	29	21	basic	basic	ADJ
fcis-14054	29	22	principle	principle	NOUN
fcis-14054	29	23	of	of	ADP
fcis-14054	29	24	deep	deep	ADJ
fcis-14054	29	25	learning	learning	NOUN
fcis-14054	29	26	is	be	AUX
fcis-14054	29	27	to	to	PART
fcis-14054	29	28	simulate	simulate	VERB
fcis-14054	29	29	the	the	DET
fcis-14054	29	30	connection	connection	NOUN
fcis-14054	29	31	relationship	relationship	NOUN
fcis-14054	29	32	between	between	ADP
fcis-14054	29	33	nerve	nerve	NOUN
fcis-14054	29	34	cells	cell	NOUN
fcis-14054	29	35	in	in	ADP
fcis-14054	29	36	the	the	DET
fcis-14054	29	37	human	human	ADJ
fcis-14054	29	38	brain	brain	NOUN
fcis-14054	29	39	by	by	ADP
fcis-14054	29	40	constructing	construct	VERB
fcis-14054	29	41	multi	multi	ADJ
fcis-14054	29	42	-	-	ADJ
fcis-14054	29	43	layer	layer	ADJ
fcis-14054	29	44	neural	neural	ADJ
fcis-14054	29	45	networks	network	NOUN
fcis-14054	29	46	,	,	PUNCT
fcis-14054	29	47	and	and	CCONJ
fcis-14054	29	48	train	train	VERB
fcis-14054	29	49	and	and	CCONJ
fcis-14054	29	50	learn	learn	VERB
fcis-14054	29	51	through	through	ADP
fcis-14054	29	52	a	a	DET
fcis-14054	29	53	large	large	ADJ
fcis-14054	29	54	amount	amount	NOUN
fcis-14054	29	55	of	of	ADP
fcis-14054	29	56	data	datum	NOUN
fcis-14054	29	57	,	,	PUNCT
fcis-14054	29	58	so	so	SCONJ
fcis-14054	29	59	as	as	SCONJ
fcis-14054	29	60	to	to	PART
fcis-14054	29	61	achieve	achieve	VERB
fcis-14054	29	62	efficient	efficient	ADJ
fcis-14054	29	63	processing	processing	NOUN
fcis-14054	29	64	and	and	CCONJ
fcis-14054	29	65	accurate	accurate	ADJ
fcis-14054	29	66	prediction	prediction	NOUN
fcis-14054	29	67	of	of	ADP
fcis-14054	29	68	complex	complex	ADJ
fcis-14054	29	69	tasks	task	NOUN
fcis-14054	29	70	.	.	PUNCT
fcis-14054	30	1	2.1.1	2.1.1	X
fcis-14054	30	2	.	.	PUNCT
fcis-14054	30	3	fundamentals	fundamental	NOUN
fcis-14054	30	4	of	of	ADP
fcis-14054	30	5	deep	deep	ADJ
fcis-14054	30	6	learning	learn	VERB
fcis-14054	30	7	the	the	DET
fcis-14054	30	8	core	core	ADJ
fcis-14054	30	9	idea	idea	NOUN
fcis-14054	30	10	of	of	ADP
fcis-14054	30	11	deep	deep	ADJ
fcis-14054	30	12	learning	learning	NOUN
fcis-14054	30	13	is	be	AUX
fcis-14054	30	14	to	to	PART
fcis-14054	30	15	realize	realize	VERB
fcis-14054	30	16	feature	feature	NOUN
fcis-14054	30	17	extraction	extraction	NOUN
fcis-14054	30	18	and	and	CCONJ
fcis-14054	30	19	representation	representation	NOUN
fcis-14054	30	20	learning	learning	NOUN
fcis-14054	30	21	of	of	ADP
fcis-14054	30	22	data	datum	NOUN
fcis-14054	30	23	through	through	ADP
fcis-14054	30	24	multilayer	multilayer	ADJ
fcis-14054	30	25	neural	neural	ADJ
fcis-14054	30	26	networks	network	NOUN
fcis-14054	30	27	.	.	PUNCT
fcis-14054	31	1	after	after	ADP
fcis-14054	31	2	multi	multi	ADJ
fcis-14054	31	3	-	-	ADJ
fcis-14054	31	4	layer	layer	ADJ
fcis-14054	31	5	nonlinear	nonlinear	ADJ
fcis-14054	31	6	transformation	transformation	NOUN
fcis-14054	31	7	and	and	CCONJ
fcis-14054	31	8	mapping	mapping	NOUN
fcis-14054	31	9	of	of	ADP
fcis-14054	31	10	input	input	NOUN
fcis-14054	31	11	data	datum	NOUN
fcis-14054	31	12	,	,	PUNCT
fcis-14054	31	13	advanced	advanced	ADJ
fcis-14054	31	14	feature	feature	NOUN
fcis-14054	31	15	representations	representation	NOUN
fcis-14054	31	16	are	be	AUX
fcis-14054	31	17	finally	finally	ADV
fcis-14054	31	18	obtained	obtain	VERB
fcis-14054	31	19	,	,	PUNCT
fcis-14054	31	20	which	which	PRON
fcis-14054	31	21	can	can	AUX
fcis-14054	31	22	be	be	AUX
fcis-14054	31	23	used	use	VERB
fcis-14054	31	24	for	for	ADP
fcis-14054	31	25	tasks	task	NOUN
fcis-14054	31	26	such	such	ADJ
fcis-14054	31	27	as	as	ADP
fcis-14054	31	28	classification	classification	NOUN
fcis-14054	31	29	and	and	CCONJ
fcis-14054	31	30	recognition	recognition	NOUN
fcis-14054	31	31	.	.	PUNCT
fcis-14054	32	1	2.1.2	2.1.2	X
fcis-14054	32	2	.	.	PUNCT
fcis-14054	32	3	common	common	ADJ
fcis-14054	32	4	deep	deep	ADJ
fcis-14054	32	5	learning	learning	NOUN
fcis-14054	32	6	models	model	NOUN
fcis-14054	32	7	common	common	ADJ
fcis-14054	32	8	deep	deep	ADJ
fcis-14054	32	9	learning	learning	NOUN
fcis-14054	32	10	models	model	NOUN
fcis-14054	32	11	include	include	VERB
fcis-14054	32	12	convolutional	convolutional	ADJ
fcis-14054	32	13	neural	neural	ADJ
fcis-14054	32	14	networks	network	NOUN
fcis-14054	32	15	(	(	PUNCT
fcis-14054	32	16	cnns	cnns	PROPN
fcis-14054	32	17	)	)	PUNCT
fcis-14054	32	18	,	,	PUNCT
fcis-14054	32	19	recurrent	recurrent	ADJ
fcis-14054	32	20	neural	neural	ADJ
fcis-14054	32	21	networks	network	NOUN
fcis-14054	32	22	(	(	PUNCT
fcis-14054	32	23	rnns	rnns	PROPN
fcis-14054	32	24	)	)	PUNCT
fcis-14054	32	25	,	,	PUNCT
fcis-14054	32	26	and	and	CCONJ
fcis-14054	32	27	deep	deep	ADJ
fcis-14054	32	28	neural	neural	ADJ
fcis-14054	32	29	networks	network	NOUN
fcis-14054	32	30	(	(	PUNCT
fcis-14054	32	31	dnns	dnn	NOUN
fcis-14054	32	32	)	)	PUNCT
fcis-14054	33	1	[	[	X
fcis-14054	33	2	2	2	NUM
fcis-14054	33	3	]	]	PUNCT
fcis-14054	33	4	.	.	PUNCT
fcis-14054	34	1	2.2	2.2	NUM
fcis-14054	34	2	.	.	PUNCT
fcis-14054	34	3	application	application	NOUN
fcis-14054	34	4	of	of	ADP
fcis-14054	34	5	deep	deep	ADJ
fcis-14054	34	6	learning	learning	NOUN
fcis-14054	34	7	in	in	ADP
fcis-14054	34	8	image	image	NOUN
fcis-14054	34	9	feature	feature	NOUN
fcis-14054	34	10	extraction	extraction	NOUN
fcis-14054	34	11	image	image	NOUN
fcis-14054	34	12	feature	feature	NOUN
fcis-14054	34	13	extraction	extraction	NOUN
fcis-14054	34	14	is	be	AUX
fcis-14054	34	15	a	a	DET
fcis-14054	34	16	key	key	ADJ
fcis-14054	34	17	step	step	NOUN
fcis-14054	34	18	in	in	ADP
fcis-14054	34	19	image	image	NOUN
fcis-14054	34	20	retrieval	retrieval	NOUN
fcis-14054	34	21	,	,	PUNCT
fcis-14054	34	22	155	155	NUM
fcis-14054	34	23	and	and	CCONJ
fcis-14054	34	24	its	its	PRON
fcis-14054	34	25	goal	goal	NOUN
fcis-14054	34	26	is	be	AUX
fcis-14054	34	27	to	to	PART
fcis-14054	34	28	convert	convert	VERB
fcis-14054	34	29	the	the	DET
fcis-14054	34	30	information	information	NOUN
fcis-14054	34	31	in	in	ADP
fcis-14054	34	32	the	the	DET
fcis-14054	34	33	image	image	NOUN
fcis-14054	34	34	into	into	ADP
fcis-14054	34	35	a	a	DET
fcis-14054	34	36	numerical	numerical	ADJ
fcis-14054	34	37	representation	representation	NOUN
fcis-14054	34	38	that	that	SCONJ
fcis-14054	34	39	the	the	DET
fcis-14054	34	40	machine	machine	NOUN
fcis-14054	34	41	can	can	AUX
fcis-14054	34	42	process	process	VERB
fcis-14054	34	43	and	and	CCONJ
fcis-14054	34	44	understand	understand	VERB
fcis-14054	34	45	.	.	PUNCT
fcis-14054	35	1	2.2.1	2.2.1	NUM
fcis-14054	35	2	.	.	X
fcis-14054	36	1	review	review	NOUN
fcis-14054	36	2	of	of	ADP
fcis-14054	36	3	image	image	NOUN
fcis-14054	36	4	feature	feature	NOUN
fcis-14054	36	5	extraction	extraction	NOUN
fcis-14054	36	6	methods	method	NOUN
fcis-14054	36	7	based	base	VERB
fcis-14054	36	8	on	on	ADP
fcis-14054	36	9	deep	deep	ADJ
fcis-14054	36	10	learning	learn	VERB
fcis-14054	36	11	the	the	DET
fcis-14054	36	12	most	most	ADV
fcis-14054	36	13	classic	classic	ADJ
fcis-14054	36	14	image	image	NOUN
fcis-14054	36	15	feature	feature	NOUN
fcis-14054	36	16	extraction	extraction	NOUN
fcis-14054	36	17	method	method	NOUN
fcis-14054	36	18	based	base	VERB
fcis-14054	36	19	on	on	ADP
fcis-14054	36	20	deep	deep	ADJ
fcis-14054	36	21	learning	learning	NOUN
fcis-14054	36	22	is	be	AUX
fcis-14054	36	23	to	to	PART
fcis-14054	36	24	use	use	VERB
fcis-14054	36	25	convolutional	convolutional	ADJ
fcis-14054	36	26	neural	neural	ADJ
fcis-14054	36	27	networks	network	NOUN
fcis-14054	36	28	(	(	PUNCT
fcis-14054	36	29	cnn	cnn	PROPN
fcis-14054	36	30	)	)	PUNCT
fcis-14054	36	31	for	for	ADP
fcis-14054	36	32	feature	feature	NOUN
fcis-14054	36	33	extraction	extraction	NOUN
fcis-14054	36	34	.	.	PUNCT
fcis-14054	37	1	by	by	ADP
fcis-14054	37	2	training	train	VERB
fcis-14054	37	3	a	a	DET
fcis-14054	37	4	deep	deep	ADJ
fcis-14054	37	5	learning	learning	NOUN
fcis-14054	37	6	model	model	NOUN
fcis-14054	37	7	,	,	PUNCT
fcis-14054	37	8	lowlevel	lowlevel	VERB
fcis-14054	37	9	features	feature	NOUN
fcis-14054	37	10	in	in	ADP
fcis-14054	37	11	the	the	DET
fcis-14054	37	12	image	image	NOUN
fcis-14054	37	13	can	can	AUX
fcis-14054	37	14	be	be	AUX
fcis-14054	37	15	gradually	gradually	ADV
fcis-14054	37	16	fused	fuse	VERB
fcis-14054	37	17	and	and	CCONJ
fcis-14054	37	18	combined	combine	VERB
fcis-14054	37	19	to	to	PART
fcis-14054	37	20	obtain	obtain	VERB
fcis-14054	37	21	more	more	ADV
fcis-14054	37	22	advanced	advanced	ADJ
fcis-14054	37	23	and	and	CCONJ
fcis-14054	37	24	abstract	abstract	ADJ
fcis-14054	37	25	feature	feature	NOUN
fcis-14054	37	26	representations	representation	NOUN
fcis-14054	37	27	.	.	PUNCT
fcis-14054	38	1	2.2.2	2.2.2	X
fcis-14054	38	2	.	.	PUNCT
fcis-14054	38	3	advantages	advantage	NOUN
fcis-14054	38	4	and	and	CCONJ
fcis-14054	38	5	limitations	limitation	NOUN
fcis-14054	38	6	of	of	ADP
fcis-14054	38	7	deep	deep	ADJ
fcis-14054	38	8	learning	learning	NOUN
fcis-14054	38	9	in	in	ADP
fcis-14054	38	10	image	image	NOUN
fcis-14054	38	11	feature	feature	NOUN
fcis-14054	38	12	extraction	extraction	NOUN
fcis-14054	38	13	deep	deep	ADJ
fcis-14054	38	14	learning	learning	NOUN
fcis-14054	38	15	has	have	VERB
fcis-14054	38	16	many	many	ADJ
fcis-14054	38	17	advantages	advantage	NOUN
fcis-14054	38	18	in	in	ADP
fcis-14054	38	19	image	image	NOUN
fcis-14054	38	20	feature	feature	NOUN
fcis-14054	38	21	extraction	extraction	NOUN
fcis-14054	38	22	,	,	PUNCT
fcis-14054	38	23	such	such	ADJ
fcis-14054	38	24	as	as	ADP
fcis-14054	38	25	being	be	AUX
fcis-14054	38	26	able	able	ADJ
fcis-14054	38	27	to	to	PART
fcis-14054	38	28	automatically	automatically	ADV
fcis-14054	38	29	learn	learn	VERB
fcis-14054	38	30	feature	feature	NOUN
fcis-14054	38	31	representations	representation	NOUN
fcis-14054	38	32	from	from	ADP
fcis-14054	38	33	data	datum	NOUN
fcis-14054	38	34	,	,	PUNCT
fcis-14054	38	35	with	with	ADP
fcis-14054	38	36	better	well	ADJ
fcis-14054	38	37	generalization	generalization	NOUN
fcis-14054	38	38	ability	ability	NOUN
fcis-14054	38	39	and	and	CCONJ
fcis-14054	38	40	robustness	robustness	NOUN
fcis-14054	38	41	;	;	PUNCT
fcis-14054	38	42	however	however	ADV
fcis-14054	38	43	,	,	PUNCT
fcis-14054	38	44	there	there	PRON
fcis-14054	38	45	are	be	VERB
fcis-14054	38	46	also	also	ADV
fcis-14054	38	47	some	some	DET
fcis-14054	38	48	limitations	limitation	NOUN
fcis-14054	38	49	,	,	PUNCT
fcis-14054	38	50	such	such	ADJ
fcis-14054	38	51	as	as	ADP
fcis-14054	38	52	high	high	ADJ
fcis-14054	38	53	dependence	dependence	NOUN
fcis-14054	38	54	on	on	ADP
fcis-14054	38	55	a	a	DET
fcis-14054	38	56	large	large	ADJ
fcis-14054	38	57	number	number	NOUN
fcis-14054	38	58	of	of	ADP
fcis-14054	38	59	labeled	label	VERB
fcis-14054	38	60	data	datum	NOUN
fcis-14054	38	61	,	,	PUNCT
fcis-14054	38	62	strong	strong	ADJ
fcis-14054	38	63	sensitivity	sensitivity	NOUN
fcis-14054	38	64	to	to	ADP
fcis-14054	38	65	hyperparameters	hyperparameter	NOUN
fcis-14054	38	66	,	,	PUNCT
fcis-14054	38	67	etc	etc	X
fcis-14054	38	68	.	.	X
fcis-14054	39	1	2.3	2.3	NUM
fcis-14054	39	2	.	.	PUNCT
fcis-14054	39	3	advantages	advantage	NOUN
fcis-14054	39	4	and	and	CCONJ
fcis-14054	39	5	challenges	challenge	NOUN
fcis-14054	39	6	of	of	ADP
fcis-14054	39	7	deep	deep	ADJ
fcis-14054	39	8	learning	learning	NOUN
fcis-14054	39	9	in	in	ADP
fcis-14054	39	10	image	image	NOUN
fcis-14054	39	11	retrieval	retrieval	NOUN
fcis-14054	39	12	deep	deep	ADJ
fcis-14054	39	13	learning	learning	NOUN
fcis-14054	39	14	has	have	VERB
fcis-14054	39	15	unique	unique	ADJ
fcis-14054	39	16	advantages	advantage	NOUN
fcis-14054	39	17	and	and	CCONJ
fcis-14054	39	18	challenges	challenge	NOUN
fcis-14054	39	19	in	in	ADP
fcis-14054	39	20	image	image	NOUN
fcis-14054	39	21	retrieval	retrieval	NOUN
fcis-14054	39	22	,	,	PUNCT
fcis-14054	39	23	and	and	CCONJ
fcis-14054	39	24	understanding	understanding	NOUN
fcis-14054	39	25	and	and	CCONJ
fcis-14054	39	26	solving	solve	VERB
fcis-14054	39	27	these	these	DET
fcis-14054	39	28	problems	problem	NOUN
fcis-14054	39	29	is	be	AUX
fcis-14054	39	30	of	of	ADP
fcis-14054	39	31	great	great	ADJ
fcis-14054	39	32	significance	significance	NOUN
fcis-14054	39	33	to	to	PART
fcis-14054	39	34	further	far	ADV
fcis-14054	39	35	improve	improve	VERB
fcis-14054	39	36	the	the	DET
fcis-14054	39	37	performance	performance	NOUN
fcis-14054	39	38	of	of	ADP
fcis-14054	39	39	image	image	NOUN
fcis-14054	39	40	retrieval	retrieval	NOUN
fcis-14054	39	41	.	.	PUNCT
fcis-14054	40	1	2.3.1	2.3.1	X
fcis-14054	40	2	.	.	PUNCT
fcis-14054	41	1	the	the	DET
fcis-14054	41	2	unique	unique	ADJ
fcis-14054	41	3	contribution	contribution	NOUN
fcis-14054	41	4	of	of	ADP
fcis-14054	41	5	deep	deep	ADJ
fcis-14054	41	6	learning	learning	NOUN
fcis-14054	41	7	in	in	ADP
fcis-14054	41	8	image	image	NOUN
fcis-14054	41	9	retrieval	retrieval	NOUN
fcis-14054	41	10	first	first	ADV
fcis-14054	41	11	,	,	PUNCT
fcis-14054	41	12	deep	deep	ADJ
fcis-14054	41	13	learning	learning	NOUN
fcis-14054	41	14	can	can	AUX
fcis-14054	41	15	extract	extract	VERB
fcis-14054	41	16	more	more	ADV
fcis-14054	41	17	advanced	advanced	ADJ
fcis-14054	41	18	and	and	CCONJ
fcis-14054	41	19	semantic	semantic	ADJ
fcis-14054	41	20	feature	feature	NOUN
fcis-14054	41	21	representations	representation	NOUN
fcis-14054	41	22	,	,	PUNCT
fcis-14054	41	23	thereby	thereby	ADV
fcis-14054	41	24	improving	improve	VERB
fcis-14054	41	25	the	the	DET
fcis-14054	41	26	accuracy	accuracy	NOUN
fcis-14054	41	27	and	and	CCONJ
fcis-14054	41	28	effect	effect	NOUN
fcis-14054	41	29	of	of	ADP
fcis-14054	41	30	image	image	NOUN
fcis-14054	41	31	retrieval	retrieval	NOUN
fcis-14054	41	32	.	.	PUNCT
fcis-14054	42	1	secondly	secondly	ADV
fcis-14054	42	2	,	,	PUNCT
fcis-14054	42	3	deep	deep	ADJ
fcis-14054	42	4	learning	learning	NOUN
fcis-14054	42	5	can	can	AUX
fcis-14054	42	6	also	also	ADV
fcis-14054	42	7	use	use	VERB
fcis-14054	42	8	large	large	ADJ
fcis-14054	42	9	-	-	PUNCT
fcis-14054	42	10	scale	scale	NOUN
fcis-14054	42	11	unlabeled	unlabele	VERB
fcis-14054	42	12	data	datum	NOUN
fcis-14054	42	13	for	for	ADP
fcis-14054	42	14	unsupervised	unsupervised	ADJ
fcis-14054	42	15	learning	learning	NOUN
fcis-14054	42	16	to	to	PART
fcis-14054	42	17	solve	solve	VERB
fcis-14054	42	18	the	the	DET
fcis-14054	42	19	problem	problem	NOUN
fcis-14054	42	20	of	of	ADP
fcis-14054	42	21	data	datum	NOUN
fcis-14054	42	22	annotation	annotation	NOUN
fcis-14054	42	23	and	and	CCONJ
fcis-14054	42	24	improve	improve	VERB
fcis-14054	42	25	the	the	DET
fcis-14054	42	26	efficiency	efficiency	NOUN
fcis-14054	42	27	of	of	ADP
fcis-14054	42	28	image	image	NOUN
fcis-14054	42	29	retrieval	retrieval	NOUN
fcis-14054	42	30	.	.	PUNCT
fcis-14054	43	1	2.3.2	2.3.2	X
fcis-14054	43	2	.	.	X
fcis-14054	43	3	deep	deep	ADJ
fcis-14054	43	4	learning	learning	NOUN
fcis-14054	43	5	challenges	challenge	NOUN
fcis-14054	43	6	and	and	CCONJ
fcis-14054	43	7	limitations	limitation	NOUN
fcis-14054	43	8	in	in	ADP
fcis-14054	43	9	image	image	NOUN
fcis-14054	43	10	retrieval	retrieval	NOUN
fcis-14054	43	11	the	the	DET
fcis-14054	43	12	training	training	NOUN
fcis-14054	43	13	of	of	ADP
fcis-14054	43	14	deep	deep	ADJ
fcis-14054	43	15	learning	learning	NOUN
fcis-14054	43	16	models	model	NOUN
fcis-14054	43	17	requires	require	VERB
fcis-14054	43	18	a	a	DET
fcis-14054	43	19	lot	lot	NOUN
fcis-14054	43	20	of	of	ADP
fcis-14054	43	21	computing	compute	VERB
fcis-14054	43	22	resources	resource	NOUN
fcis-14054	43	23	and	and	CCONJ
fcis-14054	43	24	time	time	NOUN
fcis-14054	43	25	,	,	PUNCT
fcis-14054	43	26	and	and	CCONJ
fcis-14054	43	27	may	may	AUX
fcis-14054	43	28	not	not	PART
fcis-14054	43	29	be	be	AUX
fcis-14054	43	30	suitable	suitable	ADJ
fcis-14054	43	31	for	for	ADP
fcis-14054	43	32	some	some	DET
fcis-14054	43	33	scenarios	scenario	NOUN
fcis-14054	43	34	with	with	ADP
fcis-14054	43	35	limited	limited	ADJ
fcis-14054	43	36	resources	resource	NOUN
fcis-14054	43	37	.	.	PUNCT
fcis-14054	44	1	secondly	secondly	ADV
fcis-14054	44	2	,	,	PUNCT
fcis-14054	44	3	the	the	DET
fcis-14054	44	4	interpretability	interpretability	NOUN
fcis-14054	44	5	of	of	ADP
fcis-14054	44	6	deep	deep	ADJ
fcis-14054	44	7	learning	learning	NOUN
fcis-14054	44	8	models	model	NOUN
fcis-14054	44	9	is	be	AUX
fcis-14054	44	10	relatively	relatively	ADV
fcis-14054	44	11	weak	weak	ADJ
fcis-14054	44	12	,	,	PUNCT
fcis-14054	44	13	and	and	CCONJ
fcis-14054	44	14	it	it	PRON
fcis-14054	44	15	is	be	AUX
fcis-14054	44	16	difficult	difficult	ADJ
fcis-14054	44	17	to	to	PART
fcis-14054	44	18	explain	explain	VERB
fcis-14054	44	19	the	the	DET
fcis-14054	44	20	reasons	reason	NOUN
fcis-14054	44	21	and	and	CCONJ
fcis-14054	44	22	basis	basis	NOUN
fcis-14054	44	23	for	for	ADP
fcis-14054	44	24	their	their	PRON
fcis-14054	44	25	prediction	prediction	NOUN
fcis-14054	44	26	results	result	NOUN
fcis-14054	44	27	.	.	PUNCT
fcis-14054	45	1	further	further	ADJ
fcis-14054	45	2	exploration	exploration	NOUN
fcis-14054	45	3	and	and	CCONJ
fcis-14054	45	4	research	research	NOUN
fcis-14054	45	5	are	be	AUX
fcis-14054	45	6	needed	need	VERB
fcis-14054	45	7	to	to	PART
fcis-14054	45	8	solve	solve	VERB
fcis-14054	45	9	these	these	DET
fcis-14054	45	10	problems	problem	NOUN
fcis-14054	45	11	and	and	CCONJ
fcis-14054	45	12	better	well	ADV
fcis-14054	45	13	apply	apply	VERB
fcis-14054	45	14	deep	deep	ADJ
fcis-14054	45	15	learning	learning	NOUN
fcis-14054	45	16	technology	technology	NOUN
fcis-14054	45	17	to	to	ADP
fcis-14054	45	18	the	the	DET
fcis-14054	45	19	field	field	NOUN
fcis-14054	45	20	of	of	ADP
fcis-14054	45	21	image	image	NOUN
fcis-14054	45	22	retrieval	retrieval	NOUN
fcis-14054	45	23	.	.	PUNCT
fcis-14054	46	1	3	3	X
fcis-14054	46	2	.	.	X
fcis-14054	46	3	design	design	NOUN
fcis-14054	46	4	and	and	CCONJ
fcis-14054	46	5	optimization	optimization	NOUN
fcis-14054	46	6	of	of	ADP
fcis-14054	46	7	image	image	NOUN
fcis-14054	46	8	retrieval	retrieval	NOUN
fcis-14054	46	9	method	method	NOUN
fcis-14054	46	10	based	base	VERB
fcis-14054	46	11	on	on	ADP
fcis-14054	46	12	deep	deep	ADJ
fcis-14054	46	13	learning	learn	VERB
fcis-14054	46	14	3.1	3.1	NUM
fcis-14054	46	15	.	.	PUNCT
fcis-14054	47	1	design	design	NOUN
fcis-14054	47	2	and	and	CCONJ
fcis-14054	47	3	training	training	NOUN
fcis-14054	47	4	of	of	ADP
fcis-14054	47	5	image	image	NOUN
fcis-14054	47	6	feature	feature	NOUN
fcis-14054	47	7	extraction	extraction	NOUN
fcis-14054	47	8	network	network	NOUN
fcis-14054	47	9	image	image	NOUN
fcis-14054	47	10	feature	feature	NOUN
fcis-14054	47	11	extraction	extraction	NOUN
fcis-14054	47	12	network	network	NOUN
fcis-14054	47	13	is	be	AUX
fcis-14054	47	14	the	the	DET
fcis-14054	47	15	core	core	NOUN
fcis-14054	47	16	component	component	NOUN
fcis-14054	47	17	of	of	ADP
fcis-14054	47	18	image	image	NOUN
fcis-14054	47	19	retrieval	retrieval	NOUN
fcis-14054	47	20	system	system	NOUN
fcis-14054	47	21	,	,	PUNCT
fcis-14054	47	22	and	and	CCONJ
fcis-14054	47	23	its	its	PRON
fcis-14054	47	24	design	design	NOUN
fcis-14054	47	25	and	and	CCONJ
fcis-14054	47	26	training	training	NOUN
fcis-14054	47	27	are	be	AUX
fcis-14054	47	28	crucial	crucial	ADJ
fcis-14054	47	29	to	to	ADP
fcis-14054	47	30	the	the	DET
fcis-14054	47	31	performance	performance	NOUN
fcis-14054	47	32	and	and	CCONJ
fcis-14054	47	33	effect	effect	NOUN
fcis-14054	47	34	of	of	ADP
fcis-14054	47	35	image	image	NOUN
fcis-14054	47	36	retrieval	retrieval	NOUN
fcis-14054	47	37	.	.	PUNCT
fcis-14054	48	1	training	training	NOUN
fcis-14054	48	2	strategy	strategy	NOUN
fcis-14054	48	3	and	and	CCONJ
fcis-14054	48	4	optimization	optimization	NOUN
fcis-14054	48	5	method	method	NOUN
fcis-14054	48	6	of	of	ADP
fcis-14054	48	7	image	image	NOUN
fcis-14054	48	8	feature	feature	NOUN
fcis-14054	48	9	extraction	extraction	NOUN
fcis-14054	48	10	network	network	NOUN
fcis-14054	48	11	(	(	PUNCT
fcis-14054	48	12	1	1	NUM
fcis-14054	48	13	)	)	PUNCT
fcis-14054	48	14	data	datum	NOUN
fcis-14054	48	15	preprocessing	preprocessing	NOUN
fcis-14054	48	16	,	,	PUNCT
fcis-14054	48	17	such	such	ADJ
fcis-14054	48	18	as	as	ADP
fcis-14054	48	19	image	image	NOUN
fcis-14054	48	20	size	size	NOUN
fcis-14054	48	21	adjustment	adjustment	NOUN
fcis-14054	48	22	,	,	PUNCT
fcis-14054	48	23	data	data	NOUN
fcis-14054	48	24	enhancement	enhancement	NOUN
fcis-14054	48	25	and	and	CCONJ
fcis-14054	48	26	normalization	normalization	NOUN
fcis-14054	48	27	,	,	PUNCT
fcis-14054	48	28	etc	etc	X
fcis-14054	48	29	.	.	X
fcis-14054	48	30	,	,	PUNCT
fcis-14054	48	31	to	to	PART
fcis-14054	48	32	improve	improve	VERB
fcis-14054	48	33	the	the	DET
fcis-14054	48	34	robustness	robustness	NOUN
fcis-14054	48	35	and	and	CCONJ
fcis-14054	48	36	generalization	generalization	NOUN
fcis-14054	48	37	ability	ability	NOUN
fcis-14054	48	38	of	of	ADP
fcis-14054	48	39	the	the	DET
fcis-14054	48	40	network	network	NOUN
fcis-14054	48	41	.	.	PUNCT
fcis-14054	49	1	(	(	PUNCT
fcis-14054	49	2	2	2	X
fcis-14054	49	3	)	)	PUNCT
fcis-14054	49	4	network	network	NOUN
fcis-14054	49	5	training	training	NOUN
fcis-14054	49	6	using	use	VERB
fcis-14054	49	7	large	large	ADJ
fcis-14054	49	8	-	-	PUNCT
fcis-14054	49	9	scale	scale	NOUN
fcis-14054	49	10	labeled	label	VERB
fcis-14054	49	11	datasets	dataset	NOUN
fcis-14054	49	12	to	to	PART
fcis-14054	49	13	increase	increase	VERB
fcis-14054	49	14	the	the	DET
fcis-14054	49	15	learning	learning	NOUN
fcis-14054	49	16	ability	ability	NOUN
fcis-14054	49	17	and	and	CCONJ
fcis-14054	49	18	generalization	generalization	NOUN
fcis-14054	49	19	performance	performance	NOUN
fcis-14054	49	20	of	of	ADP
fcis-14054	49	21	the	the	DET
fcis-14054	49	22	model	model	NOUN
fcis-14054	49	23	.	.	PUNCT
fcis-14054	50	1	(	(	PUNCT
fcis-14054	50	2	3	3	X
fcis-14054	50	3	)	)	PUNCT
fcis-14054	50	4	use	use	VERB
fcis-14054	50	5	appropriate	appropriate	ADJ
fcis-14054	50	6	loss	loss	NOUN
fcis-14054	50	7	functions	function	NOUN
fcis-14054	50	8	and	and	CCONJ
fcis-14054	50	9	regularization	regularization	NOUN
fcis-14054	50	10	methods	method	NOUN
fcis-14054	50	11	,	,	PUNCT
fcis-14054	50	12	such	such	ADJ
fcis-14054	50	13	as	as	ADP
fcis-14054	50	14	cross	cross	NOUN
fcis-14054	50	15	entropy	entropy	PROPN
fcis-14054	50	16	loss	loss	PROPN
fcis-14054	50	17	and	and	CCONJ
fcis-14054	50	18	l1	l1	PROPN
fcis-14054	50	19	/	/	SYM
fcis-14054	50	20	l2	l2	NOUN
fcis-14054	50	21	regularization	regularization	NOUN
fcis-14054	50	22	,	,	PUNCT
fcis-14054	50	23	to	to	PART
fcis-14054	50	24	constrain	constrain	VERB
fcis-14054	50	25	the	the	DET
fcis-14054	50	26	range	range	NOUN
fcis-14054	50	27	of	of	ADP
fcis-14054	50	28	network	network	NOUN
fcis-14054	50	29	outputs	output	NOUN
fcis-14054	50	30	and	and	CCONJ
fcis-14054	50	31	parameters	parameter	NOUN
fcis-14054	50	32	.	.	PUNCT
fcis-14054	51	1	(	(	PUNCT
fcis-14054	51	2	4	4	X
fcis-14054	51	3	)	)	PUNCT
fcis-14054	51	4	adopt	adopt	VERB
fcis-14054	51	5	appropriate	appropriate	ADJ
fcis-14054	51	6	optimizers	optimizer	NOUN
fcis-14054	51	7	and	and	CCONJ
fcis-14054	51	8	learning	learn	VERB
fcis-14054	51	9	rate	rate	NOUN
fcis-14054	51	10	scheduling	scheduling	NOUN
fcis-14054	51	11	strategies	strategy	NOUN
fcis-14054	51	12	,	,	PUNCT
fcis-14054	51	13	such	such	ADJ
fcis-14054	51	14	as	as	ADP
fcis-14054	51	15	random	random	ADJ
fcis-14054	51	16	layer	layer	NOUN
fcis-14054	51	17	descent	descent	NOUN
fcis-14054	51	18	(	(	PUNCT
fcis-14054	51	19	sgd	sgd	NOUN
fcis-14054	51	20	)	)	PUNCT
fcis-14054	51	21	and	and	CCONJ
fcis-14054	51	22	adaptive	adaptive	ADJ
fcis-14054	51	23	learning	learning	NOUN
fcis-14054	51	24	rate	rate	NOUN
fcis-14054	51	25	algorithms	algorithm	NOUN
fcis-14054	51	26	(	(	PUNCT
fcis-14054	51	27	such	such	ADJ
fcis-14054	51	28	as	as	ADP
fcis-14054	51	29	adam	adam	PROPN
fcis-14054	51	30	)	)	PUNCT
fcis-14054	51	31	,	,	PUNCT
fcis-14054	51	32	to	to	PART
fcis-14054	51	33	speed	speed	VERB
fcis-14054	51	34	up	up	ADP
fcis-14054	51	35	the	the	DET
fcis-14054	51	36	network	network	NOUN
fcis-14054	51	37	convergence	convergence	NOUN
fcis-14054	51	38	and	and	CCONJ
fcis-14054	51	39	optimization	optimization	NOUN
fcis-14054	51	40	process	process	NOUN
fcis-14054	51	41	.	.	PUNCT
fcis-14054	52	1	3.2	3.2	NUM
fcis-14054	52	2	.	.	PUNCT
fcis-14054	52	3	improvement	improvement	NOUN
fcis-14054	52	4	and	and	CCONJ
fcis-14054	52	5	optimization	optimization	NOUN
fcis-14054	52	6	of	of	ADP
fcis-14054	52	7	search	search	NOUN
fcis-14054	52	8	result	result	NOUN
fcis-14054	52	9	sorting	sort	VERB
fcis-14054	52	10	algorithm	algorithm	NOUN
fcis-14054	52	11	the	the	DET
fcis-14054	52	12	ranking	rank	VERB
fcis-14054	52	13	algorithms	algorithm	NOUN
fcis-14054	52	14	of	of	ADP
fcis-14054	52	15	image	image	NOUN
fcis-14054	52	16	retrieval	retrieval	NOUN
fcis-14054	52	17	results	result	NOUN
fcis-14054	52	18	include	include	VERB
fcis-14054	52	19	traditional	traditional	ADJ
fcis-14054	52	20	methods	method	NOUN
fcis-14054	52	21	based	base	VERB
fcis-14054	52	22	on	on	ADP
fcis-14054	52	23	keyword	keyword	NOUN
fcis-14054	52	24	matching	matching	NOUN
fcis-14054	52	25	,	,	PUNCT
fcis-14054	52	26	feature	feature	NOUN
fcis-14054	52	27	vector	vector	NOUN
fcis-14054	52	28	comparison	comparison	NOUN
fcis-14054	52	29	and	and	CCONJ
fcis-14054	52	30	image	image	NOUN
fcis-14054	52	31	similarity	similarity	NOUN
fcis-14054	52	32	measurement	measurement	NOUN
fcis-14054	52	33	,	,	PUNCT
fcis-14054	52	34	as	as	ADV
fcis-14054	52	35	well	well	ADV
fcis-14054	52	36	as	as	ADP
fcis-14054	52	37	improved	improved	ADJ
fcis-14054	52	38	algorithms	algorithm	NOUN
fcis-14054	52	39	based	base	VERB
fcis-14054	52	40	on	on	ADP
fcis-14054	52	41	deep	deep	ADJ
fcis-14054	52	42	learning	learning	NOUN
fcis-14054	52	43	.	.	PUNCT
fcis-14054	53	1	the	the	DET
fcis-14054	53	2	commonly	commonly	ADV
fcis-14054	53	3	used	use	VERB
fcis-14054	53	4	sorting	sort	VERB
fcis-14054	53	5	algorithms	algorithm	NOUN
fcis-14054	53	6	in	in	ADP
fcis-14054	53	7	traditional	traditional	ADJ
fcis-14054	53	8	methods	method	NOUN
fcis-14054	53	9	include	include	VERB
fcis-14054	53	10	vector	vector	NOUN
fcis-14054	53	11	space	space	NOUN
fcis-14054	53	12	model	model	NOUN
fcis-14054	53	13	,	,	PUNCT
fcis-14054	53	14	inverted	inverted	ADJ
fcis-14054	53	15	index	index	NOUN
fcis-14054	53	16	and	and	CCONJ
fcis-14054	53	17	bm25	bm25	PROPN
fcis-14054	53	18	.	.	PUNCT
fcis-14054	54	1	the	the	DET
fcis-14054	54	2	research	research	NOUN
fcis-14054	54	3	on	on	ADP
fcis-14054	54	4	the	the	DET
fcis-14054	54	5	improvement	improvement	NOUN
fcis-14054	54	6	of	of	ADP
fcis-14054	54	7	the	the	DET
fcis-14054	54	8	search	search	NOUN
fcis-14054	54	9	result	result	NOUN
fcis-14054	54	10	sorting	sort	VERB
fcis-14054	54	11	algorithm	algorithm	NOUN
fcis-14054	54	12	mainly	mainly	ADV
fcis-14054	54	13	includes	include	VERB
fcis-14054	54	14	the	the	DET
fcis-14054	54	15	following	follow	VERB
fcis-14054	54	16	aspects	aspect	NOUN
fcis-14054	54	17	:	:	PUNCT
fcis-14054	54	18	(	(	PUNCT
fcis-14054	54	19	1	1	X
fcis-14054	54	20	)	)	PUNCT
fcis-14054	54	21	introduce	introduce	VERB
fcis-14054	54	22	image	image	NOUN
fcis-14054	54	23	semantic	semantic	ADJ
fcis-14054	54	24	representation	representation	NOUN
fcis-14054	54	25	and	and	CCONJ
fcis-14054	54	26	context	context	NOUN
fcis-14054	54	27	information	information	NOUN
fcis-14054	54	28	,	,	PUNCT
fcis-14054	54	29	and	and	CCONJ
fcis-14054	54	30	improve	improve	VERB
fcis-14054	54	31	the	the	DET
fcis-14054	54	32	accuracy	accuracy	NOUN
fcis-14054	54	33	and	and	CCONJ
fcis-14054	54	34	effect	effect	NOUN
fcis-14054	54	35	of	of	ADP
fcis-14054	54	36	sorting	sort	VERB
fcis-14054	54	37	by	by	ADP
fcis-14054	54	38	improving	improve	VERB
fcis-14054	54	39	network	network	NOUN
fcis-14054	54	40	structure	structure	NOUN
fcis-14054	54	41	and	and	CCONJ
fcis-14054	54	42	loss	loss	NOUN
fcis-14054	54	43	function	function	NOUN
fcis-14054	54	44	.	.	PUNCT
fcis-14054	55	1	(	(	PUNCT
fcis-14054	55	2	2	2	X
fcis-14054	55	3	)	)	PUNCT
fcis-14054	55	4	increase	increase	VERB
fcis-14054	55	5	the	the	DET
fcis-14054	55	6	correlation	correlation	NOUN
fcis-14054	55	7	modeling	modeling	NOUN
fcis-14054	55	8	between	between	ADP
fcis-14054	55	9	images	image	NOUN
fcis-14054	55	10	and	and	CCONJ
fcis-14054	55	11	queries	query	NOUN
fcis-14054	55	12	,	,	PUNCT
fcis-14054	55	13	and	and	CCONJ
fcis-14054	55	14	improve	improve	VERB
fcis-14054	55	15	the	the	DET
fcis-14054	55	16	robustness	robustness	NOUN
fcis-14054	55	17	and	and	CCONJ
fcis-14054	55	18	generalization	generalization	NOUN
fcis-14054	55	19	ability	ability	NOUN
fcis-14054	55	20	of	of	ADP
fcis-14054	55	21	the	the	DET
fcis-14054	55	22	sorting	sort	VERB
fcis-14054	55	23	algorithm	algorithm	NOUN
fcis-14054	55	24	by	by	ADP
fcis-14054	55	25	using	use	VERB
fcis-14054	55	26	methods	method	NOUN
fcis-14054	55	27	such	such	ADJ
fcis-14054	55	28	as	as	ADP
fcis-14054	55	29	attention	attention	NOUN
fcis-14054	55	30	mechanism	mechanism	NOUN
fcis-14054	55	31	and	and	CCONJ
fcis-14054	55	32	multimodal	multimodal	NOUN
fcis-14054	55	33	information	information	NOUN
fcis-14054	55	34	fusion	fusion	NOUN
fcis-14054	55	35	.	.	PUNCT
fcis-14054	56	1	(	(	PUNCT
fcis-14054	56	2	3	3	X
fcis-14054	56	3	)	)	PUNCT
fcis-14054	56	4	using	use	VERB
fcis-14054	56	5	the	the	DET
fcis-14054	56	6	reinforcement	reinforcement	NOUN
fcis-14054	56	7	learning	learning	NOUN
fcis-14054	56	8	method	method	NOUN
fcis-14054	56	9	,	,	PUNCT
fcis-14054	56	10	a	a	DET
fcis-14054	56	11	reinforcement	reinforcement	NOUN
fcis-14054	56	12	learning	learning	NOUN
fcis-14054	56	13	model	model	NOUN
fcis-14054	56	14	is	be	AUX
fcis-14054	56	15	established	establish	VERB
fcis-14054	56	16	to	to	PART
fcis-14054	56	17	optimize	optimize	VERB
fcis-14054	56	18	the	the	DET
fcis-14054	56	19	sorting	sort	VERB
fcis-14054	56	20	strategy	strategy	NOUN
fcis-14054	56	21	to	to	PART
fcis-14054	56	22	improve	improve	VERB
fcis-14054	56	23	the	the	DET
fcis-14054	56	24	performance	performance	NOUN
fcis-14054	56	25	of	of	ADP
fcis-14054	56	26	image	image	NOUN
fcis-14054	56	27	retrieval	retrieval	NOUN
fcis-14054	56	28	.	.	PUNCT
fcis-14054	57	1	3.3	3.3	NUM
fcis-14054	57	2	.	.	PUNCT
fcis-14054	58	1	experimental	experimental	ADJ
fcis-14054	58	2	design	design	NOUN
fcis-14054	58	3	of	of	ADP
fcis-14054	58	4	image	image	NOUN
fcis-14054	58	5	retrieval	retrieval	NOUN
fcis-14054	58	6	system	system	NOUN
fcis-14054	58	7	the	the	DET
fcis-14054	58	8	performance	performance	NOUN
fcis-14054	58	9	evaluation	evaluation	NOUN
fcis-14054	58	10	of	of	ADP
fcis-14054	58	11	image	image	NOUN
fcis-14054	58	12	retrieval	retrieval	NOUN
fcis-14054	58	13	system	system	NOUN
fcis-14054	58	14	is	be	AUX
fcis-14054	58	15	an	an	DET
fcis-14054	58	16	important	important	ADJ
fcis-14054	58	17	means	mean	NOUN
fcis-14054	58	18	to	to	PART
fcis-14054	58	19	measure	measure	VERB
fcis-14054	58	20	the	the	DET
fcis-14054	58	21	performance	performance	NOUN
fcis-14054	58	22	and	and	CCONJ
fcis-14054	58	23	effect	effect	NOUN
fcis-14054	58	24	of	of	ADP
fcis-14054	58	25	the	the	DET
fcis-14054	58	26	system	system	NOUN
fcis-14054	58	27	.	.	PUNCT
fcis-14054	59	1	it	it	PRON
fcis-14054	59	2	needs	need	VERB
fcis-14054	59	3	to	to	PART
fcis-14054	59	4	select	select	VERB
fcis-14054	59	5	appropriate	appropriate	ADJ
fcis-14054	59	6	evaluation	evaluation	NOUN
fcis-14054	59	7	indicators	indicator	NOUN
fcis-14054	59	8	and	and	CCONJ
fcis-14054	59	9	design	design	VERB
fcis-14054	59	10	reasonable	reasonable	ADJ
fcis-14054	59	11	experiments	experiment	NOUN
fcis-14054	59	12	.	.	PUNCT
fcis-14054	60	1	3.3.1	3.3.1	X
fcis-14054	60	2	.	.	PUNCT
fcis-14054	60	3	performance	performance	NOUN
fcis-14054	60	4	evaluation	evaluation	NOUN
fcis-14054	60	5	indicators	indicator	NOUN
fcis-14054	60	6	commonly	commonly	ADV
fcis-14054	60	7	used	use	VERB
fcis-14054	60	8	image	image	NOUN
fcis-14054	60	9	retrieval	retrieval	NOUN
fcis-14054	60	10	performance	performance	NOUN
fcis-14054	60	11	evaluation	evaluation	NOUN
fcis-14054	60	12	indicators	indicator	NOUN
fcis-14054	60	13	include	include	VERB
fcis-14054	60	14	accuracy	accuracy	NOUN
fcis-14054	60	15	,	,	PUNCT
fcis-14054	60	16	recall	recall	NOUN
fcis-14054	60	17	,	,	PUNCT
fcis-14054	60	18	precision	precision	NOUN
fcis-14054	60	19	,	,	PUNCT
fcis-14054	60	20	and	and	CCONJ
fcis-14054	60	21	average	average	ADJ
fcis-14054	60	22	accuracy	accuracy	NOUN
fcis-14054	60	23	.	.	PUNCT
fcis-14054	61	1	accuracy	accuracy	NOUN
fcis-14054	61	2	measures	measure	VERB
fcis-14054	61	3	the	the	DET
fcis-14054	61	4	proportion	proportion	NOUN
fcis-14054	61	5	of	of	ADP
fcis-14054	61	6	relevant	relevant	ADJ
fcis-14054	61	7	images	image	NOUN
fcis-14054	61	8	in	in	ADP
fcis-14054	61	9	the	the	DET
fcis-14054	61	10	first	first	ADJ
fcis-14054	61	11	n	n	PRON
fcis-14054	61	12	results	result	NOUN
fcis-14054	61	13	returned	return	VERB
fcis-14054	61	14	by	by	ADP
fcis-14054	61	15	the	the	DET
fcis-14054	61	16	system	system	NOUN
fcis-14054	61	17	;	;	PUNCT
fcis-14054	61	18	recall	recall	VERB
fcis-14054	61	19	measures	measure	NOUN
fcis-14054	61	20	the	the	DET
fcis-14054	61	21	proportion	proportion	NOUN
fcis-14054	61	22	of	of	ADP
fcis-14054	61	23	relevant	relevant	ADJ
fcis-14054	61	24	images	image	NOUN
fcis-14054	61	25	that	that	SCONJ
fcis-14054	61	26	the	the	DET
fcis-14054	61	27	system	system	NOUN
fcis-14054	61	28	can	can	AUX
fcis-14054	61	29	retrieve	retrieve	VERB
fcis-14054	61	30	to	to	ADP
fcis-14054	61	31	all	all	DET
fcis-14054	61	32	relevant	relevant	ADJ
fcis-14054	61	33	images	image	NOUN
fcis-14054	61	34	[	[	X
fcis-14054	61	35	3	3	NUM
fcis-14054	61	36	]	]	PUNCT
fcis-14054	61	37	;	;	PUNCT
fcis-14054	61	38	precision	precision	NOUN
fcis-14054	61	39	measures	measure	VERB
fcis-14054	61	40	the	the	DET
fcis-14054	61	41	proportion	proportion	NOUN
fcis-14054	61	42	of	of	ADP
fcis-14054	61	43	relevant	relevant	ADJ
fcis-14054	61	44	images	image	NOUN
fcis-14054	61	45	in	in	ADP
fcis-14054	61	46	the	the	DET
fcis-14054	61	47	first	first	ADJ
fcis-14054	61	48	n	n	PRON
fcis-14054	61	49	results	result	NOUN
fcis-14054	61	50	returned	return	VERB
fcis-14054	61	51	by	by	ADP
fcis-14054	61	52	the	the	DET
fcis-14054	61	53	system	system	NOUN
fcis-14054	61	54	;	;	PUNCT
fcis-14054	61	55	average	average	ADJ
fcis-14054	61	56	accuracy	accuracy	NOUN
fcis-14054	61	57	measures	measure	VERB
fcis-14054	61	58	the	the	DET
fcis-14054	61	59	average	average	ADJ
fcis-14054	61	60	accuracy	accuracy	NOUN
fcis-14054	61	61	of	of	ADP
fcis-14054	61	62	relevant	relevant	ADJ
fcis-14054	61	63	images	image	NOUN
fcis-14054	61	64	returned	return	VERB
fcis-14054	61	65	by	by	ADP
fcis-14054	61	66	the	the	DET
fcis-14054	61	67	system	system	NOUN
fcis-14054	61	68	.	.	PUNCT
fcis-14054	62	1	3.3.2	3.3.2	X
fcis-14054	62	2	.	.	PUNCT
fcis-14054	62	3	experimental	experimental	ADJ
fcis-14054	62	4	design	design	NOUN
fcis-14054	62	5	and	and	CCONJ
fcis-14054	62	6	evaluation	evaluation	NOUN
fcis-14054	62	7	methods	method	NOUN
fcis-14054	62	8	the	the	DET
fcis-14054	62	9	experimental	experimental	ADJ
fcis-14054	62	10	design	design	NOUN
fcis-14054	62	11	includes	include	VERB
fcis-14054	62	12	the	the	DET
fcis-14054	62	13	selection	selection	NOUN
fcis-14054	62	14	and	and	CCONJ
fcis-14054	62	15	division	division	NOUN
fcis-14054	62	16	of	of	ADP
fcis-14054	62	17	datasets	dataset	NOUN
fcis-14054	62	18	,	,	PUNCT
fcis-14054	62	19	the	the	DET
fcis-14054	62	20	setting	setting	NOUN
fcis-14054	62	21	of	of	ADP
fcis-14054	62	22	evaluation	evaluation	NOUN
fcis-14054	62	23	indicators	indicator	NOUN
fcis-14054	62	24	and	and	CCONJ
fcis-14054	62	25	the	the	DET
fcis-14054	62	26	design	design	NOUN
fcis-14054	62	27	of	of	ADP
fcis-14054	62	28	specific	specific	ADJ
fcis-14054	62	29	experiments	experiment	NOUN
fcis-14054	62	30	.	.	PUNCT
fcis-14054	63	1	the	the	DET
fcis-14054	63	2	selection	selection	NOUN
fcis-14054	63	3	of	of	ADP
fcis-14054	63	4	datasets	dataset	NOUN
fcis-14054	63	5	should	should	AUX
fcis-14054	63	6	consider	consider	VERB
fcis-14054	63	7	representativeness	representativeness	ADJ
fcis-14054	63	8	and	and	CCONJ
fcis-14054	63	9	diversity	diversity	NOUN
fcis-14054	63	10	to	to	PART
fcis-14054	63	11	ensure	ensure	VERB
fcis-14054	63	12	the	the	DET
fcis-14054	63	13	reliability	reliability	NOUN
fcis-14054	63	14	and	and	CCONJ
fcis-14054	63	15	generalization	generalization	NOUN
fcis-14054	63	16	of	of	ADP
fcis-14054	63	17	experimental	experimental	ADJ
fcis-14054	63	18	results	result	NOUN
fcis-14054	63	19	.	.	PUNCT
fcis-14054	64	1	the	the	DET
fcis-14054	64	2	setting	setting	NOUN
fcis-14054	64	3	of	of	ADP
fcis-14054	64	4	evaluation	evaluation	NOUN
fcis-14054	64	5	indicators	indicator	NOUN
fcis-14054	64	6	should	should	AUX
fcis-14054	64	7	be	be	AUX
fcis-14054	64	8	selected	select	VERB
fcis-14054	64	9	according	accord	VERB
fcis-14054	64	10	to	to	ADP
fcis-14054	64	11	specific	specific	ADJ
fcis-14054	64	12	tasks	task	NOUN
fcis-14054	64	13	and	and	CCONJ
fcis-14054	64	14	needs	need	NOUN
fcis-14054	64	15	,	,	PUNCT
fcis-14054	64	16	and	and	CCONJ
fcis-14054	64	17	measured	measure	VERB
fcis-14054	64	18	in	in	ADP
fcis-14054	64	19	combination	combination	NOUN
fcis-14054	64	20	with	with	ADP
fcis-14054	64	21	practical	practical	ADJ
fcis-14054	64	22	application	application	NOUN
fcis-14054	64	23	scenarios	scenario	NOUN
fcis-14054	64	24	.	.	PUNCT
fcis-14054	65	1	the	the	DET
fcis-14054	65	2	design	design	NOUN
fcis-14054	65	3	of	of	ADP
fcis-14054	65	4	experiments	experiment	NOUN
fcis-14054	65	5	should	should	AUX
fcis-14054	65	6	consider	consider	VERB
fcis-14054	65	7	control	control	NOUN
fcis-14054	65	8	variables	variable	NOUN
fcis-14054	65	9	and	and	CCONJ
fcis-14054	65	10	randomness	randomness	NOUN
fcis-14054	65	11	to	to	PART
fcis-14054	65	12	ensure	ensure	VERB
fcis-14054	65	13	the	the	DET
fcis-14054	65	14	reliability	reliability	NOUN
fcis-14054	65	15	and	and	CCONJ
fcis-14054	65	16	statistical	statistical	ADJ
fcis-14054	65	17	significance	significance	NOUN
fcis-14054	65	18	of	of	ADP
fcis-14054	65	19	experimental	experimental	ADJ
fcis-14054	65	20	results	result	NOUN
fcis-14054	65	21	.	.	PUNCT
fcis-14054	66	1	the	the	DET
fcis-14054	66	2	design	design	NOUN
fcis-14054	66	3	and	and	CCONJ
fcis-14054	66	4	optimization	optimization	NOUN
fcis-14054	66	5	of	of	ADP
fcis-14054	66	6	image	image	NOUN
fcis-14054	66	7	retrieval	retrieval	NOUN
fcis-14054	66	8	methods	method	NOUN
fcis-14054	66	9	based	base	VERB
fcis-14054	66	10	on	on	ADP
fcis-14054	66	11	deep	deep	ADJ
fcis-14054	66	12	learning	learning	NOUN
fcis-14054	66	13	involves	involve	VERB
fcis-14054	66	14	the	the	DET
fcis-14054	66	15	design	design	NOUN
fcis-14054	66	16	and	and	CCONJ
fcis-14054	66	17	training	training	NOUN
fcis-14054	66	18	of	of	ADP
fcis-14054	66	19	image	image	NOUN
fcis-14054	66	20	feature	feature	NOUN
fcis-14054	66	21	extraction	extraction	NOUN
fcis-14054	66	22	networks	network	NOUN
fcis-14054	66	23	,	,	PUNCT
fcis-14054	66	24	the	the	DET
fcis-14054	66	25	improvement	improvement	NOUN
fcis-14054	66	26	and	and	CCONJ
fcis-14054	66	27	optimization	optimization	NOUN
fcis-14054	66	28	of	of	ADP
fcis-14054	66	29	retrieval	retrieval	NOUN
fcis-14054	66	30	result	result	NOUN
fcis-14054	66	31	ranking	ranking	ADJ
fcis-14054	66	32	algorithms	algorithm	NOUN
fcis-14054	66	33	,	,	PUNCT
fcis-14054	66	34	and	and	CCONJ
fcis-14054	66	35	the	the	DET
fcis-14054	66	36	performance	performance	NOUN
fcis-14054	66	37	evaluation	evaluation	NOUN
fcis-14054	66	38	indicators	indicator	NOUN
fcis-14054	66	39	and	and	CCONJ
fcis-14054	66	40	experimental	experimental	ADJ
fcis-14054	66	41	design	design	NOUN
fcis-14054	66	42	of	of	ADP
fcis-14054	66	43	image	image	NOUN
fcis-14054	66	44	retrieval	retrieval	NOUN
fcis-14054	66	45	systems	system	NOUN
fcis-14054	66	46	.	.	PUNCT
fcis-14054	67	1	the	the	DET
fcis-14054	67	2	design	design	NOUN
fcis-14054	67	3	of	of	ADP
fcis-14054	67	4	experiments	experiment	NOUN
fcis-14054	67	5	should	should	AUX
fcis-14054	67	6	156	156	NUM
fcis-14054	67	7	consider	consider	VERB
fcis-14054	67	8	control	control	NOUN
fcis-14054	67	9	variables	variable	NOUN
fcis-14054	67	10	and	and	CCONJ
fcis-14054	67	11	randomness	randomness	NOUN
fcis-14054	67	12	to	to	PART
fcis-14054	67	13	ensure	ensure	VERB
fcis-14054	67	14	the	the	DET
fcis-14054	67	15	reliability	reliability	NOUN
fcis-14054	67	16	and	and	CCONJ
fcis-14054	67	17	statistical	statistical	ADJ
fcis-14054	67	18	significance	significance	NOUN
fcis-14054	67	19	of	of	ADP
fcis-14054	67	20	experimental	experimental	ADJ
fcis-14054	67	21	results	result	NOUN
fcis-14054	67	22	.	.	PUNCT
fcis-14054	68	1	by	by	ADP
fcis-14054	68	2	rationally	rationally	ADV
fcis-14054	68	3	designing	design	VERB
fcis-14054	68	4	and	and	CCONJ
fcis-14054	68	5	optimizing	optimize	VERB
fcis-14054	68	6	these	these	DET
fcis-14054	68	7	links	link	NOUN
fcis-14054	68	8	,	,	PUNCT
fcis-14054	68	9	the	the	DET
fcis-14054	68	10	accuracy	accuracy	NOUN
fcis-14054	68	11	and	and	CCONJ
fcis-14054	68	12	effect	effect	NOUN
fcis-14054	68	13	of	of	ADP
fcis-14054	68	14	the	the	DET
fcis-14054	68	15	image	image	NOUN
fcis-14054	68	16	retrieval	retrieval	NOUN
fcis-14054	68	17	system	system	NOUN
fcis-14054	68	18	can	can	AUX
fcis-14054	68	19	be	be	AUX
fcis-14054	68	20	improved	improve	VERB
fcis-14054	68	21	,	,	PUNCT
fcis-14054	68	22	and	and	CCONJ
fcis-14054	68	23	the	the	DET
fcis-14054	68	24	user	user	NOUN
fcis-14054	68	25	experience	experience	NOUN
fcis-14054	68	26	and	and	CCONJ
fcis-14054	68	27	practicability	practicability	NOUN
fcis-14054	68	28	of	of	ADP
fcis-14054	68	29	the	the	DET
fcis-14054	68	30	system	system	NOUN
fcis-14054	68	31	can	can	AUX
fcis-14054	68	32	be	be	AUX
fcis-14054	68	33	enhanced	enhance	VERB
fcis-14054	68	34	.	.	PUNCT
fcis-14054	69	1	4	4	X
fcis-14054	69	2	.	.	X
fcis-14054	69	3	conclusion	conclusion	NOUN
fcis-14054	69	4	and	and	CCONJ
fcis-14054	69	5	prospects	prospect	NOUN
fcis-14054	69	6	4.1	4.1	NUM
fcis-14054	69	7	.	.	PUNCT
fcis-14054	70	1	summary	summary	NOUN
fcis-14054	70	2	of	of	ADP
fcis-14054	70	3	research	research	NOUN
fcis-14054	70	4	work	work	NOUN
fcis-14054	70	5	this	this	DET
fcis-14054	70	6	paper	paper	NOUN
fcis-14054	70	7	comprehensively	comprehensively	ADV
fcis-14054	70	8	studies	study	NOUN
fcis-14054	70	9	and	and	CCONJ
fcis-14054	70	10	optimizes	optimize	VERB
fcis-14054	70	11	the	the	DET
fcis-14054	70	12	image	image	NOUN
fcis-14054	70	13	retrieval	retrieval	NOUN
fcis-14054	70	14	method	method	NOUN
fcis-14054	70	15	based	base	VERB
fcis-14054	70	16	on	on	ADP
fcis-14054	70	17	deep	deep	ADJ
fcis-14054	70	18	learning	learning	NOUN
fcis-14054	70	19	,	,	PUNCT
fcis-14054	70	20	and	and	CCONJ
fcis-14054	70	21	has	have	AUX
fcis-14054	70	22	achieved	achieve	VERB
fcis-14054	70	23	certain	certain	ADJ
fcis-14054	70	24	achievements	achievement	NOUN
fcis-14054	70	25	and	and	CCONJ
fcis-14054	70	26	contributions	contribution	NOUN
fcis-14054	70	27	.	.	PUNCT
fcis-14054	71	1	(	(	PUNCT
fcis-14054	71	2	1	1	X
fcis-14054	71	3	)	)	PUNCT
fcis-14054	71	4	this	this	DET
fcis-14054	71	5	paper	paper	NOUN
fcis-14054	71	6	summarizes	summarize	VERB
fcis-14054	71	7	the	the	DET
fcis-14054	71	8	development	development	NOUN
fcis-14054	71	9	and	and	CCONJ
fcis-14054	71	10	comparison	comparison	NOUN
fcis-14054	71	11	of	of	ADP
fcis-14054	71	12	image	image	NOUN
fcis-14054	71	13	feature	feature	NOUN
fcis-14054	71	14	description	description	NOUN
fcis-14054	71	15	methods	method	NOUN
fcis-14054	71	16	,	,	PUNCT
fcis-14054	71	17	and	and	CCONJ
fcis-14054	71	18	sorts	sort	NOUN
fcis-14054	71	19	out	out	ADP
fcis-14054	71	20	the	the	DET
fcis-14054	71	21	characteristics	characteristic	NOUN
fcis-14054	71	22	and	and	CCONJ
fcis-14054	71	23	advantages	advantage	NOUN
fcis-14054	71	24	of	of	ADP
fcis-14054	71	25	traditional	traditional	ADJ
fcis-14054	71	26	methods	method	NOUN
fcis-14054	71	27	and	and	CCONJ
fcis-14054	71	28	methods	method	NOUN
fcis-14054	71	29	based	base	VERB
fcis-14054	71	30	on	on	ADP
fcis-14054	71	31	deep	deep	ADJ
fcis-14054	71	32	learning	learning	NOUN
fcis-14054	71	33	.	.	PUNCT
fcis-14054	72	1	(	(	PUNCT
fcis-14054	72	2	2	2	X
fcis-14054	72	3	)	)	PUNCT
fcis-14054	72	4	in	in	ADP
fcis-14054	72	5	the	the	DET
fcis-14054	72	6	research	research	NOUN
fcis-14054	72	7	of	of	ADP
fcis-14054	72	8	image	image	NOUN
fcis-14054	72	9	feature	feature	NOUN
fcis-14054	72	10	extraction	extraction	NOUN
fcis-14054	72	11	algorithm	algorithm	NOUN
fcis-14054	72	12	based	base	VERB
fcis-14054	72	13	on	on	ADP
fcis-14054	72	14	deep	deep	ADJ
fcis-14054	72	15	learning	learning	NOUN
fcis-14054	72	16	,	,	PUNCT
fcis-14054	72	17	this	this	DET
fcis-14054	72	18	paper	paper	NOUN
fcis-14054	72	19	introduces	introduce	VERB
fcis-14054	72	20	the	the	DET
fcis-14054	72	21	commonly	commonly	ADV
fcis-14054	72	22	used	use	VERB
fcis-14054	72	23	algorithms	algorithm	NOUN
fcis-14054	72	24	and	and	CCONJ
fcis-14054	72	25	their	their	PRON
fcis-14054	72	26	characteristics	characteristic	NOUN
fcis-14054	72	27	,	,	PUNCT
fcis-14054	72	28	as	as	ADV
fcis-14054	72	29	well	well	ADV
fcis-14054	72	30	as	as	ADP
fcis-14054	72	31	the	the	DET
fcis-14054	72	32	advantages	advantage	NOUN
fcis-14054	72	33	and	and	CCONJ
fcis-14054	72	34	limitations	limitation	NOUN
fcis-14054	72	35	of	of	ADP
fcis-14054	72	36	different	different	ADJ
fcis-14054	72	37	algorithms	algorithm	NOUN
fcis-14054	72	38	.	.	PUNCT
fcis-14054	73	1	4.2	4.2	NUM
fcis-14054	73	2	.	.	PUNCT
fcis-14054	74	1	existing	exist	VERB
fcis-14054	74	2	deficiencies	deficiency	NOUN
fcis-14054	74	3	and	and	CCONJ
fcis-14054	74	4	improvement	improvement	NOUN
fcis-14054	74	5	directions	direction	NOUN
fcis-14054	74	6	although	although	SCONJ
fcis-14054	74	7	this	this	DET
fcis-14054	74	8	paper	paper	NOUN
fcis-14054	74	9	has	have	AUX
fcis-14054	74	10	achieved	achieve	VERB
fcis-14054	74	11	certain	certain	ADJ
fcis-14054	74	12	results	result	NOUN
fcis-14054	74	13	in	in	ADP
fcis-14054	74	14	the	the	DET
fcis-14054	74	15	research	research	NOUN
fcis-14054	74	16	and	and	CCONJ
fcis-14054	74	17	optimization	optimization	NOUN
fcis-14054	74	18	of	of	ADP
fcis-14054	74	19	image	image	NOUN
fcis-14054	74	20	retrieval	retrieval	NOUN
fcis-14054	74	21	methods	method	NOUN
fcis-14054	74	22	based	base	VERB
fcis-14054	74	23	on	on	ADP
fcis-14054	74	24	deep	deep	ADJ
fcis-14054	74	25	learning	learning	NOUN
fcis-14054	74	26	,	,	PUNCT
fcis-14054	74	27	there	there	PRON
fcis-14054	74	28	are	be	VERB
fcis-14054	74	29	still	still	ADV
fcis-14054	74	30	some	some	DET
fcis-14054	74	31	shortcomings	shortcoming	NOUN
fcis-14054	74	32	.	.	PUNCT
fcis-14054	75	1	(	(	PUNCT
fcis-14054	75	2	1	1	X
fcis-14054	75	3	)	)	PUNCT
fcis-14054	75	4	image	image	NOUN
fcis-14054	75	5	retrieval	retrieval	NOUN
fcis-14054	75	6	methods	method	NOUN
fcis-14054	75	7	based	base	VERB
fcis-14054	75	8	on	on	ADP
fcis-14054	75	9	deep	deep	ADJ
fcis-14054	75	10	learning	learning	NOUN
fcis-14054	75	11	may	may	AUX
fcis-14054	75	12	have	have	VERB
fcis-14054	75	13	problems	problem	NOUN
fcis-14054	75	14	of	of	ADP
fcis-14054	75	15	low	low	ADJ
fcis-14054	75	16	accuracy	accuracy	NOUN
fcis-14054	75	17	or	or	CCONJ
fcis-14054	75	18	unstable	unstable	ADJ
fcis-14054	75	19	effect	effect	NOUN
fcis-14054	75	20	in	in	ADP
fcis-14054	75	21	some	some	DET
fcis-14054	75	22	cases	case	NOUN
fcis-14054	75	23	,	,	PUNCT
fcis-14054	75	24	and	and	CCONJ
fcis-14054	75	25	need	need	VERB
fcis-14054	75	26	further	further	ADJ
fcis-14054	75	27	improvement	improvement	NOUN
fcis-14054	75	28	.	.	PUNCT
fcis-14054	76	1	(	(	PUNCT
fcis-14054	76	2	2	2	X
fcis-14054	76	3	)	)	PUNCT
fcis-14054	76	4	current	current	ADJ
fcis-14054	76	5	methods	method	NOUN
fcis-14054	76	6	may	may	AUX
fcis-14054	76	7	be	be	AUX
fcis-14054	76	8	inefficient	inefficient	ADJ
fcis-14054	76	9	for	for	ADP
fcis-14054	76	10	large	large	ADJ
fcis-14054	76	11	-	-	PUNCT
fcis-14054	76	12	scale	scale	NOUN
fcis-14054	76	13	datasets	dataset	NOUN
fcis-14054	76	14	,	,	PUNCT
fcis-14054	76	15	and	and	CCONJ
fcis-14054	76	16	it	it	PRON
fcis-14054	76	17	is	be	AUX
fcis-14054	76	18	necessary	necessary	ADJ
fcis-14054	76	19	to	to	PART
fcis-14054	76	20	find	find	VERB
fcis-14054	76	21	faster	fast	ADV
fcis-14054	76	22	and	and	CCONJ
fcis-14054	76	23	more	more	ADV
fcis-14054	76	24	efficient	efficient	ADJ
fcis-14054	76	25	algorithms	algorithm	NOUN
fcis-14054	76	26	and	and	CCONJ
fcis-14054	76	27	techniques	technique	NOUN
fcis-14054	76	28	to	to	PART
fcis-14054	76	29	improve	improve	VERB
fcis-14054	76	30	the	the	DET
fcis-14054	76	31	speed	speed	NOUN
fcis-14054	76	32	and	and	CCONJ
fcis-14054	76	33	practicality	practicality	NOUN
fcis-14054	76	34	of	of	ADP
fcis-14054	76	35	image	image	NOUN
fcis-14054	76	36	retrieval	retrieval	NOUN
fcis-14054	76	37	.	.	PUNCT
fcis-14054	77	1	4.3	4.3	NUM
fcis-14054	77	2	.	.	PUNCT
fcis-14054	77	3	directions	direction	NOUN
fcis-14054	77	4	and	and	CCONJ
fcis-14054	77	5	potential	potential	ADJ
fcis-14054	77	6	application	application	NOUN
fcis-14054	77	7	scenarios	scenario	NOUN
fcis-14054	77	8	for	for	ADP
fcis-14054	77	9	follow	follow	VERB
fcis-14054	77	10	-	-	PUNCT
fcis-14054	77	11	up	up	ADP
fcis-14054	77	12	research	research	NOUN
fcis-14054	77	13	given	give	VERB
fcis-14054	77	14	the	the	DET
fcis-14054	77	15	limitations	limitation	NOUN
fcis-14054	77	16	and	and	CCONJ
fcis-14054	77	17	challenges	challenge	NOUN
fcis-14054	77	18	of	of	ADP
fcis-14054	77	19	current	current	ADJ
fcis-14054	77	20	research	research	NOUN
fcis-14054	77	21	,	,	PUNCT
fcis-14054	77	22	there	there	PRON
fcis-14054	77	23	are	be	VERB
fcis-14054	77	24	several	several	ADJ
fcis-14054	77	25	possible	possible	ADJ
fcis-14054	77	26	follow	follow	VERB
fcis-14054	77	27	-	-	PUNCT
fcis-14054	77	28	up	up	ADP
fcis-14054	77	29	research	research	NOUN
fcis-14054	77	30	directions	direction	NOUN
fcis-14054	77	31	and	and	CCONJ
fcis-14054	77	32	potential	potential	ADJ
fcis-14054	77	33	application	application	NOUN
fcis-14054	77	34	scenarios	scenario	NOUN
fcis-14054	77	35	worth	worth	ADJ
fcis-14054	77	36	exploring	explore	VERB
fcis-14054	77	37	:	:	PUNCT
fcis-14054	77	38	(	(	PUNCT
fcis-14054	77	39	1	1	X
fcis-14054	77	40	)	)	PUNCT
fcis-14054	77	41	image	image	NOUN
fcis-14054	77	42	retrieval	retrieval	NOUN
fcis-14054	77	43	methods	method	NOUN
fcis-14054	77	44	based	base	VERB
fcis-14054	77	45	on	on	ADP
fcis-14054	77	46	deep	deep	ADJ
fcis-14054	77	47	learning	learning	NOUN
fcis-14054	77	48	can	can	AUX
fcis-14054	77	49	be	be	AUX
fcis-14054	77	50	further	far	ADV
fcis-14054	77	51	studied	study	VERB
fcis-14054	77	52	and	and	CCONJ
fcis-14054	77	53	optimized	optimize	VERB
fcis-14054	77	54	,	,	PUNCT
fcis-14054	77	55	especially	especially	ADV
fcis-14054	77	56	for	for	ADP
fcis-14054	77	57	large	large	ADJ
fcis-14054	77	58	-	-	PUNCT
fcis-14054	77	59	scale	scale	NOUN
fcis-14054	77	60	datasets	dataset	NOUN
fcis-14054	77	61	and	and	CCONJ
fcis-14054	77	62	complex	complex	ADJ
fcis-14054	77	63	scenes	scene	NOUN
fcis-14054	77	64	.	.	PUNCT
fcis-14054	78	1	(	(	PUNCT
fcis-14054	78	2	2	2	X
fcis-14054	78	3	)	)	PUNCT
fcis-14054	78	4	multimodal	multimodal	NOUN
fcis-14054	78	5	retrieval	retrieval	NOUN
fcis-14054	78	6	can	can	AUX
fcis-14054	78	7	be	be	AUX
fcis-14054	78	8	explored	explore	VERB
fcis-14054	78	9	,	,	PUNCT
fcis-14054	78	10	and	and	CCONJ
fcis-14054	78	11	images	image	NOUN
fcis-14054	78	12	can	can	AUX
fcis-14054	78	13	be	be	AUX
fcis-14054	78	14	associated	associate	VERB
fcis-14054	78	15	with	with	ADP
fcis-14054	78	16	other	other	ADJ
fcis-14054	78	17	data	datum	NOUN
fcis-14054	78	18	types	type	NOUN
fcis-14054	78	19	(	(	PUNCT
fcis-14054	78	20	such	such	ADJ
fcis-14054	78	21	as	as	ADP
fcis-14054	78	22	text	text	NOUN
fcis-14054	78	23	,	,	PUNCT
fcis-14054	78	24	audio	audio	NOUN
fcis-14054	78	25	)	)	PUNCT
fcis-14054	78	26	to	to	PART
fcis-14054	78	27	achieve	achieve	VERB
fcis-14054	78	28	richer	rich	ADJ
fcis-14054	78	29	and	and	CCONJ
fcis-14054	78	30	more	more	ADV
fcis-14054	78	31	comprehensive	comprehensive	ADJ
fcis-14054	78	32	retrieval	retrieval	NOUN
fcis-14054	78	33	results	result	NOUN
fcis-14054	78	34	.	.	PUNCT
fcis-14054	79	1	(	(	PUNCT
fcis-14054	79	2	3	3	X
fcis-14054	79	3	)	)	PUNCT
fcis-14054	79	4	for	for	ADP
fcis-14054	79	5	specific	specific	ADJ
fcis-14054	79	6	application	application	NOUN
fcis-14054	79	7	fields	field	NOUN
fcis-14054	79	8	,	,	PUNCT
fcis-14054	79	9	such	such	ADJ
fcis-14054	79	10	as	as	ADP
fcis-14054	79	11	medical	medical	ADJ
fcis-14054	79	12	images	image	NOUN
fcis-14054	79	13	,	,	PUNCT
fcis-14054	79	14	remote	remote	ADJ
fcis-14054	79	15	sensing	sensing	NOUN
fcis-14054	79	16	images	image	NOUN
fcis-14054	79	17	,	,	PUNCT
fcis-14054	79	18	etc	etc	X
fcis-14054	79	19	.	.	X
fcis-14054	79	20	,	,	PUNCT
fcis-14054	79	21	customized	customize	VERB
fcis-14054	79	22	image	image	NOUN
fcis-14054	79	23	retrieval	retrieval	NOUN
fcis-14054	79	24	methods	method	NOUN
fcis-14054	79	25	for	for	ADP
fcis-14054	79	26	specific	specific	ADJ
fcis-14054	79	27	tasks	task	NOUN
fcis-14054	79	28	and	and	CCONJ
fcis-14054	79	29	needs	need	NOUN
fcis-14054	79	30	can	can	AUX
fcis-14054	79	31	be	be	AUX
fcis-14054	79	32	researched	research	VERB
fcis-14054	79	33	and	and	CCONJ
fcis-14054	79	34	developed	develop	VERB
fcis-14054	79	35	.	.	PUNCT
fcis-14054	80	1	in	in	ADP
fcis-14054	80	2	a	a	DET
fcis-14054	80	3	word	word	NOUN
fcis-14054	80	4	,	,	PUNCT
fcis-14054	80	5	image	image	NOUN
fcis-14054	80	6	retrieval	retrieval	NOUN
fcis-14054	80	7	methods	method	NOUN
fcis-14054	80	8	based	base	VERB
fcis-14054	80	9	on	on	ADP
fcis-14054	80	10	deep	deep	ADJ
fcis-14054	80	11	learning	learning	NOUN
fcis-14054	80	12	still	still	ADV
fcis-14054	80	13	have	have	VERB
fcis-14054	80	14	great	great	ADJ
fcis-14054	80	15	research	research	NOUN
fcis-14054	80	16	space	space	NOUN
fcis-14054	80	17	and	and	CCONJ
fcis-14054	80	18	application	application	NOUN
fcis-14054	80	19	prospects	prospect	NOUN
fcis-14054	80	20	in	in	ADP
fcis-14054	80	21	the	the	DET
fcis-14054	80	22	future	future	NOUN
fcis-14054	80	23	.	.	PUNCT
fcis-14054	81	1	through	through	ADP
fcis-14054	81	2	further	further	ADJ
fcis-14054	81	3	improvement	improvement	NOUN
fcis-14054	81	4	and	and	CCONJ
fcis-14054	81	5	optimization	optimization	NOUN
fcis-14054	81	6	,	,	PUNCT
fcis-14054	81	7	the	the	DET
fcis-14054	81	8	accuracy	accuracy	NOUN
fcis-14054	81	9	,	,	PUNCT
fcis-14054	81	10	efficiency	efficiency	NOUN
fcis-14054	81	11	and	and	CCONJ
fcis-14054	81	12	adaptability	adaptability	NOUN
fcis-14054	81	13	of	of	ADP
fcis-14054	81	14	image	image	NOUN
fcis-14054	81	15	retrieval	retrieval	NOUN
fcis-14054	81	16	can	can	AUX
fcis-14054	81	17	be	be	AUX
fcis-14054	81	18	improved	improve	VERB
fcis-14054	81	19	,	,	PUNCT
fcis-14054	81	20	and	and	CCONJ
fcis-14054	81	21	better	well	ADJ
fcis-14054	81	22	solutions	solution	NOUN
fcis-14054	81	23	can	can	AUX
fcis-14054	81	24	be	be	AUX
fcis-14054	81	25	provided	provide	VERB
fcis-14054	81	26	for	for	ADP
fcis-14054	81	27	practical	practical	ADJ
fcis-14054	81	28	applications	application	NOUN
fcis-14054	81	29	in	in	ADP
fcis-14054	81	30	various	various	ADJ
fcis-14054	81	31	fields	field	NOUN
fcis-14054	81	32	.	.	PUNCT
fcis-14054	82	1	references	reference	NOUN
fcis-14054	82	2	[	[	X
fcis-14054	82	3	1	1	NUM
fcis-14054	82	4	]	]	X
fcis-14054	82	5	gordo	gordo	NOUN
fcis-14054	82	6	,	,	PUNCT
fcis-14054	82	7	a.	a.	NOUN
fcis-14054	82	8	,	,	PUNCT
fcis-14054	82	9	almazán	almazán	PROPN
fcis-14054	82	10	,	,	PUNCT
fcis-14054	82	11	j.	j.	PROPN
fcis-14054	82	12	,	,	PUNCT
fcis-14054	82	13	revaud	revaud	PROPN
fcis-14054	82	14	,	,	PUNCT
fcis-14054	82	15	j.	j.	PROPN
fcis-14054	82	16	,	,	PUNCT
fcis-14054	82	17	et	et	PROPN
fcis-14054	82	18	al	al	PROPN
fcis-14054	82	19	.	.	PROPN
fcis-14054	83	1	(	(	PUNCT
fcis-14054	83	2	2017	2017	NUM
fcis-14054	83	3	)	)	PUNCT
fcis-14054	83	4	end	end	NOUN
fcis-14054	83	5	-	-	PUNCT
fcis-14054	83	6	to	to	ADP
fcis-14054	83	7	-	-	PUNCT
fcis-14054	83	8	end	end	NOUN
fcis-14054	83	9	learning	learning	NOUN
fcis-14054	83	10	of	of	ADP
fcis-14054	83	11	deep	deep	ADJ
fcis-14054	83	12	visual	visual	ADJ
fcis-14054	83	13	representations	representation	NOUN
fcis-14054	83	14	for	for	ADP
fcis-14054	83	15	image	image	NOUN
fcis-14054	83	16	retrieval	retrieval	NOUN
fcis-14054	83	17	.	.	PUNCT
fcis-14054	84	1	international	international	ADJ
fcis-14054	84	2	journal	journal	PROPN
fcis-14054	84	3	of	of	ADP
fcis-14054	84	4	computer	computer	NOUN
fcis-14054	84	5	vision	vision	NOUN
fcis-14054	84	6	,	,	PUNCT
fcis-14054	84	7	124	124	NUM
fcis-14054	84	8	(	(	PUNCT
fcis-14054	84	9	2	2	NUM
fcis-14054	84	10	):	):	PUNCT
fcis-14054	84	11	237	237	NUM
fcis-14054	84	12	-254	-254	NOUN
fcis-14054	84	13	.	.	PUNCT
fcis-14054	85	1	[	[	X
fcis-14054	85	2	2	2	NUM
fcis-14054	85	3	]	]	PUNCT
fcis-14054	85	4	xu	xu	PROPN
fcis-14054	85	5	,	,	PUNCT
fcis-14054	85	6	x.	x.	NOUN
fcis-14054	85	7	(	(	PUNCT
fcis-14054	85	8	2019	2019	NUM
fcis-14054	85	9	)	)	PUNCT
fcis-14054	85	10	research	research	NOUN
fcis-14054	85	11	on	on	ADP
fcis-14054	85	12	question	question	NOUN
fcis-14054	85	13	answering	answer	VERB
fcis-14054	85	14	system	system	NOUN
fcis-14054	85	15	based	base	VERB
fcis-14054	85	16	on	on	ADP
fcis-14054	85	17	deep	deep	ADJ
fcis-14054	85	18	learning	learning	NOUN
fcis-14054	85	19	.	.	PUNCT
fcis-14054	86	1	journal	journal	PROPN
fcis-14054	86	2	of	of	ADP
fcis-14054	86	3	hubei	hubei	PROPN
fcis-14054	86	4	normal	normal	ADJ
fcis-14054	86	5	university	university	NOUN
fcis-14054	86	6	(	(	PUNCT
fcis-14054	86	7	natural	natural	ADJ
fcis-14054	86	8	science	science	NOUN
fcis-14054	86	9	edition	edition	NOUN
fcis-14054	86	10	)	)	PUNCT
fcis-14054	86	11	,	,	PUNCT
fcis-14054	86	12	39	39	NUM
fcis-14054	86	13	(	(	PUNCT
fcis-14054	86	14	01	01	NUM
fcis-14054	86	15	):	):	PUNCT
fcis-14054	86	16	10	10	NUM
fcis-14054	86	17	-	-	SYM
fcis-14054	86	18	18	18	NUM
fcis-14054	86	19	.	.	PUNCT
fcis-14054	87	1	[	[	X
fcis-14054	87	2	3	3	NUM
fcis-14054	87	3	]	]	X
fcis-14054	87	4	liu	liu	PROPN
fcis-14054	87	5	,	,	PUNCT
fcis-14054	87	6	y.	y.	PROPN
fcis-14054	87	7	,	,	PUNCT
fcis-14054	87	8	guo	guo	PROPN
fcis-14054	87	9	,	,	PUNCT
fcis-14054	87	10	c.	c.	PROPN
fcis-14054	87	11	,	,	PUNCT
fcis-14054	87	12	feng	feng	PROPN
fcis-14054	87	13	,	,	PUNCT
fcis-14054	87	14	s.	s.	PROPN
fcis-14054	87	15	,	,	PUNCT
fcis-14054	87	16	et	et	PROPN
fcis-14054	87	17	al	al	PROPN
fcis-14054	87	18	.	.	PROPN
fcis-14054	87	19	(	(	PUNCT
fcis-14054	87	20	2021	2021	NUM
fcis-14054	87	21	)	)	PUNCT
fcis-14054	87	22	research	research	NOUN
fcis-14054	87	23	on	on	ADP
fcis-14054	87	24	crossmodal	crossmodal	ADJ
fcis-14054	87	25	graphic	graphic	ADJ
fcis-14054	87	26	content	content	NOUN
fcis-14054	87	27	screening	screening	NOUN
fcis-14054	87	28	and	and	CCONJ
fcis-14054	87	29	storage	storage	NOUN
fcis-14054	87	30	mechanism	mechanism	NOUN
fcis-14054	87	31	based	base	VERB
fcis-14054	87	32	on	on	ADP
fcis-14054	87	33	semantic	semantic	ADJ
fcis-14054	87	34	similarity	similarity	NOUN
fcis-14054	87	35	.	.	PUNCT
fcis-14054	88	1	computer	computer	NOUN
fcis-14054	88	2	research	research	NOUN
fcis-14054	88	3	and	and	CCONJ
fcis-14054	88	4	development	development	NOUN
fcis-14054	88	5	,	,	PUNCT
fcis-14054	88	6	58	58	NUM
fcis-14054	88	7	(	(	PUNCT
fcis-14054	88	8	02	02	NUM
fcis-14054	88	9	):	):	PUNCT
fcis-14054	88	10	338	338	NUM
fcis-14054	88	11	-	-	SYM
fcis-14054	88	12	355	355	NUM
fcis-14054	88	13	.	.	PUNCT