id	sid	tid	token	lemma	pos
fcis-30019	1	1	frontiers	frontier	NOUN
fcis-30019	1	2	in	in	ADP
fcis-30019	1	3	computing	computing	NOUN
fcis-30019	1	4	and	and	CCONJ
fcis-30019	1	5	intelligent	intelligent	ADJ
fcis-30019	1	6	systems	system	NOUN
fcis-30019	1	7	issn	issn	VERB
fcis-30019	1	8	:	:	PUNCT
fcis-30019	1	9	2832	2832	NUM
fcis-30019	1	10	-	-	SYM
fcis-30019	1	11	6024	6024	NUM
fcis-30019	1	12	|	|	NOUN
fcis-30019	1	13	vol	vol	NOUN
fcis-30019	1	14	.	.	PROPN
fcis-30019	2	1	11	11	NUM
fcis-30019	2	2	,	,	PUNCT
fcis-30019	2	3	no	no	INTJ
fcis-30019	2	4	.	.	NOUN
fcis-30019	2	5	3	3	NUM
fcis-30019	2	6	,	,	PUNCT
fcis-30019	2	7	2025	2025	NUM
fcis-30019	2	8	54	54	NUM
fcis-30019	2	9	deepfake	deepfake	NOUN
fcis-30019	2	10	detection	detection	NOUN
fcis-30019	2	11	technology	technology	NOUN
fcis-30019	2	12	integrating	integrate	VERB
fcis-30019	2	13	spatial	spatial	ADJ
fcis-30019	2	14	domain	domain	NOUN
fcis-30019	2	15	and	and	CCONJ
fcis-30019	2	16	frequency	frequency	NOUN
fcis-30019	2	17	domain	domain	NOUN
fcis-30019	2	18	haoqi	haoqi	PROPN
fcis-30019	2	19	geng	geng	PROPN
fcis-30019	3	1	*	*	PUNCT
fcis-30019	3	2	,	,	PUNCT
fcis-30019	3	3	tianliang	tianliang	PROPN
fcis-30019	3	4	lu	lu	PROPN
fcis-30019	3	5	,	,	PUNCT
fcis-30019	3	6	wanxin	wanxin	PROPN
fcis-30019	3	7	huang	huang	PROPN
fcis-30019	3	8	and	and	CCONJ
fcis-30019	3	9	bowen	bowen	PROPN
fcis-30019	3	10	ding	ding	PROPN
fcis-30019	3	11	people	people	NOUN
fcis-30019	3	12	’s	’s	PART
fcis-30019	3	13	public	public	ADJ
fcis-30019	3	14	security	security	NOUN
fcis-30019	3	15	university	university	PROPN
fcis-30019	3	16	of	of	ADP
fcis-30019	3	17	china	china	PROPN
fcis-30019	3	18	,	,	PUNCT
fcis-30019	3	19	beijing	beijing	PROPN
fcis-30019	3	20	,	,	PUNCT
fcis-30019	3	21	china	china	PROPN
fcis-30019	3	22	*	*	PUNCT
fcis-30019	3	23	corresponding	correspond	VERB
fcis-30019	3	24	author	author	NOUN
fcis-30019	3	25	:	:	PUNCT
fcis-30019	3	26	haoqi	haoqi	PROPN
fcis-30019	3	27	geng	geng	PROPN
fcis-30019	3	28	abstract	abstract	PROPN
fcis-30019	3	29	:	:	PUNCT
fcis-30019	3	30	in	in	ADP
fcis-30019	3	31	recent	recent	ADJ
fcis-30019	3	32	years	year	NOUN
fcis-30019	3	33	,	,	PUNCT
fcis-30019	3	34	with	with	ADP
fcis-30019	3	35	the	the	DET
fcis-30019	3	36	rapid	rapid	ADJ
fcis-30019	3	37	development	development	NOUN
fcis-30019	3	38	of	of	ADP
fcis-30019	3	39	deep	deep	ADJ
fcis-30019	3	40	forgery	forgery	NOUN
fcis-30019	3	41	technology	technology	NOUN
fcis-30019	3	42	,	,	PUNCT
fcis-30019	3	43	its	its	PRON
fcis-30019	3	44	verisimilitude	verisimilitude	NOUN
fcis-30019	3	45	is	be	AUX
fcis-30019	3	46	increasing	increase	VERB
fcis-30019	3	47	day	day	NOUN
fcis-30019	3	48	by	by	ADP
fcis-30019	3	49	day	day	NOUN
fcis-30019	3	50	,	,	PUNCT
fcis-30019	3	51	and	and	CCONJ
fcis-30019	3	52	its	its	PRON
fcis-30019	3	53	social	social	ADJ
fcis-30019	3	54	impact	impact	NOUN
fcis-30019	3	55	is	be	AUX
fcis-30019	3	56	becoming	become	VERB
fcis-30019	3	57	more	more	ADV
fcis-30019	3	58	and	and	CCONJ
fcis-30019	3	59	more	more	ADV
fcis-30019	3	60	serious	serious	ADJ
fcis-30019	3	61	.	.	PUNCT
fcis-30019	4	1	however	however	ADV
fcis-30019	4	2	,	,	PUNCT
fcis-30019	4	3	although	although	SCONJ
fcis-30019	4	4	a	a	DET
fcis-30019	4	5	variety	variety	NOUN
fcis-30019	4	6	of	of	ADP
fcis-30019	4	7	face	face	NOUN
fcis-30019	4	8	deep	deep	ADJ
fcis-30019	4	9	forgery	forgery	NOUN
fcis-30019	4	10	video	video	NOUN
fcis-30019	4	11	detection	detection	NOUN
fcis-30019	4	12	algorithms	algorithm	NOUN
fcis-30019	4	13	have	have	AUX
fcis-30019	4	14	been	be	AUX
fcis-30019	4	15	proposed	propose	VERB
fcis-30019	4	16	and	and	CCONJ
fcis-30019	4	17	have	have	AUX
fcis-30019	4	18	shown	show	VERB
fcis-30019	4	19	certain	certain	ADJ
fcis-30019	4	20	detection	detection	NOUN
fcis-30019	4	21	capabilities	capability	NOUN
fcis-30019	4	22	on	on	ADP
fcis-30019	4	23	open	open	ADJ
fcis-30019	4	24	source	source	NOUN
fcis-30019	4	25	data	data	NOUN
fcis-30019	4	26	sets	set	NOUN
fcis-30019	4	27	,	,	PUNCT
fcis-30019	4	28	in	in	ADP
fcis-30019	4	29	the	the	DET
fcis-30019	4	30	face	face	NOUN
fcis-30019	4	31	of	of	ADP
fcis-30019	4	32	increasingly	increasingly	ADV
fcis-30019	4	33	sophisticated	sophisticated	ADJ
fcis-30019	4	34	deep	deep	ADJ
fcis-30019	4	35	forgery	forgery	NOUN
fcis-30019	4	36	technology	technology	NOUN
fcis-30019	4	37	,	,	PUNCT
fcis-30019	4	38	the	the	DET
fcis-30019	4	39	differences	difference	NOUN
fcis-30019	4	40	between	between	ADP
fcis-30019	4	41	genuine	genuine	ADJ
fcis-30019	4	42	and	and	CCONJ
fcis-30019	4	43	fake	fake	ADJ
fcis-30019	4	44	videos	video	NOUN
fcis-30019	4	45	are	be	AUX
fcis-30019	4	46	gradually	gradually	ADV
fcis-30019	4	47	difficult	difficult	ADJ
fcis-30019	4	48	to	to	PART
fcis-30019	4	49	be	be	AUX
fcis-30019	4	50	captured	capture	VERB
fcis-30019	4	51	by	by	ADP
fcis-30019	4	52	the	the	DET
fcis-30019	4	53	naked	naked	ADJ
fcis-30019	4	54	eye	eye	NOUN
fcis-30019	4	55	,	,	PUNCT
fcis-30019	4	56	and	and	CCONJ
fcis-30019	4	57	existing	exist	VERB
fcis-30019	4	58	detection	detection	NOUN
fcis-30019	4	59	methods	method	NOUN
fcis-30019	4	60	generally	generally	ADV
fcis-30019	4	61	have	have	VERB
fcis-30019	4	62	problems	problem	NOUN
fcis-30019	4	63	such	such	ADJ
fcis-30019	4	64	as	as	ADP
fcis-30019	4	65	low	low	ADJ
fcis-30019	4	66	cross	cross	ADJ
fcis-30019	4	67	-	-	ADJ
fcis-30019	4	68	compressibility	compressibility	ADJ
fcis-30019	4	69	detection	detection	NOUN
fcis-30019	4	70	and	and	CCONJ
fcis-30019	4	71	poor	poor	ADJ
fcis-30019	4	72	robustness	robustness	NOUN
fcis-30019	4	73	.	.	PUNCT
fcis-30019	5	1	therefore	therefore	ADV
fcis-30019	5	2	,	,	PUNCT
fcis-30019	5	3	in	in	ADP
fcis-30019	5	4	order	order	NOUN
fcis-30019	5	5	to	to	PART
fcis-30019	5	6	improve	improve	VERB
fcis-30019	5	7	the	the	DET
fcis-30019	5	8	detection	detection	NOUN
fcis-30019	5	9	accuracy	accuracy	NOUN
fcis-30019	5	10	and	and	CCONJ
fcis-30019	5	11	model	model	NOUN
fcis-30019	5	12	robustness	robustness	NOUN
fcis-30019	5	13	,	,	PUNCT
fcis-30019	5	14	a	a	DET
fcis-30019	5	15	deep	deep	ADJ
fcis-30019	5	16	forged	forge	VERB
fcis-30019	5	17	video	video	NOUN
fcis-30019	5	18	detection	detection	NOUN
fcis-30019	5	19	method	method	NOUN
fcis-30019	5	20	named	name	VERB
fcis-30019	5	21	sfdt	sfdt	NOUN
fcis-30019	5	22	is	be	AUX
fcis-30019	5	23	proposed	propose	VERB
fcis-30019	5	24	.	.	PUNCT
fcis-30019	6	1	in	in	ADP
fcis-30019	6	2	this	this	DET
fcis-30019	6	3	scheme	scheme	NOUN
fcis-30019	6	4	,	,	PUNCT
fcis-30019	6	5	the	the	DET
fcis-30019	6	6	framework	framework	NOUN
fcis-30019	6	7	structure	structure	NOUN
fcis-30019	6	8	of	of	ADP
fcis-30019	6	9	fusion	fusion	NOUN
fcis-30019	6	10	of	of	ADP
fcis-30019	6	11	air	air	NOUN
fcis-30019	6	12	frequency	frequency	NOUN
fcis-30019	6	13	domain	domain	NOUN
fcis-30019	6	14	is	be	AUX
fcis-30019	6	15	adopted	adopt	VERB
fcis-30019	6	16	.	.	PUNCT
fcis-30019	7	1	firstly	firstly	ADV
fcis-30019	7	2	,	,	PUNCT
fcis-30019	7	3	feature	feature	NOUN
fcis-30019	7	4	extraction	extraction	NOUN
fcis-30019	7	5	is	be	AUX
fcis-30019	7	6	enhanced	enhance	VERB
fcis-30019	7	7	by	by	ADP
fcis-30019	7	8	improving	improve	VERB
fcis-30019	7	9	mvit	mvit	NOUN
fcis-30019	7	10	in	in	ADP
fcis-30019	7	11	the	the	DET
fcis-30019	7	12	airspace	airspace	NOUN
fcis-30019	7	13	and	and	CCONJ
fcis-30019	7	14	using	use	VERB
fcis-30019	7	15	asff	asff	PROPN
fcis-30019	7	16	adaptive	adaptive	ADJ
fcis-30019	7	17	module	module	NOUN
fcis-30019	7	18	;	;	PUNCT
fcis-30019	7	19	secondly	secondly	ADV
fcis-30019	7	20	,	,	PUNCT
fcis-30019	7	21	frequency	frequency	NOUN
fcis-30019	7	22	domain	domain	NOUN
fcis-30019	7	23	features	feature	NOUN
fcis-30019	7	24	are	be	AUX
fcis-30019	7	25	extracted	extract	VERB
fcis-30019	7	26	by	by	ADP
fcis-30019	7	27	dynamic	dynamic	ADJ
fcis-30019	7	28	filter	filter	NOUN
fcis-30019	7	29	in	in	ADP
fcis-30019	7	30	the	the	DET
fcis-30019	7	31	frequency	frequency	NOUN
fcis-30019	7	32	domain	domain	NOUN
fcis-30019	7	33	;	;	PUNCT
fcis-30019	7	34	then	then	ADV
fcis-30019	7	35	fecam	fecam	PROPN
fcis-30019	7	36	is	be	AUX
fcis-30019	7	37	used	use	VERB
fcis-30019	7	38	to	to	PART
fcis-30019	7	39	reduce	reduce	VERB
fcis-30019	7	40	the	the	DET
fcis-30019	7	41	loss	loss	NOUN
fcis-30019	7	42	caused	cause	VERB
fcis-30019	7	43	by	by	ADP
fcis-30019	7	44	frequency	frequency	NOUN
fcis-30019	7	45	domain	domain	NOUN
fcis-30019	7	46	information	information	NOUN
fcis-30019	7	47	transmission	transmission	NOUN
fcis-30019	7	48	;	;	PUNCT
fcis-30019	7	49	finally	finally	ADV
fcis-30019	7	50	,	,	PUNCT
fcis-30019	7	51	multi	multi	ADJ
fcis-30019	7	52	-	-	ADJ
fcis-30019	7	53	mode	mode	ADJ
fcis-30019	7	54	fusion	fusion	NOUN
fcis-30019	7	55	module	module	NOUN
fcis-30019	7	56	is	be	AUX
fcis-30019	7	57	used	use	VERB
fcis-30019	7	58	for	for	ADP
fcis-30019	7	59	feature	feature	NOUN
fcis-30019	7	60	fusion	fusion	NOUN
fcis-30019	7	61	.	.	PUNCT
fcis-30019	8	1	this	this	DET
fcis-30019	8	2	air	air	NOUN
fcis-30019	8	3	-	-	PUNCT
fcis-30019	8	4	frequency	frequency	NOUN
fcis-30019	8	5	fusion	fusion	NOUN
fcis-30019	8	6	detection	detection	NOUN
fcis-30019	8	7	scheme	scheme	NOUN
fcis-30019	8	8	can	can	AUX
fcis-30019	8	9	not	not	PART
fcis-30019	8	10	only	only	ADV
fcis-30019	8	11	improve	improve	VERB
fcis-30019	8	12	the	the	DET
fcis-30019	8	13	cross	cross	ADJ
fcis-30019	8	14	-	-	ADJ
fcis-30019	8	15	compression	compression	ADJ
fcis-30019	8	16	detection	detection	NOUN
fcis-30019	8	17	performance	performance	NOUN
fcis-30019	8	18	of	of	ADP
fcis-30019	8	19	the	the	DET
fcis-30019	8	20	model	model	NOUN
fcis-30019	8	21	,	,	PUNCT
fcis-30019	8	22	but	but	CCONJ
fcis-30019	8	23	also	also	ADV
fcis-30019	8	24	effectively	effectively	ADV
fcis-30019	8	25	deal	deal	VERB
fcis-30019	8	26	with	with	ADP
fcis-30019	8	27	various	various	ADJ
fcis-30019	8	28	interference	interference	NOUN
fcis-30019	8	29	in	in	ADP
fcis-30019	8	30	the	the	DET
fcis-30019	8	31	transmission	transmission	NOUN
fcis-30019	8	32	of	of	ADP
fcis-30019	8	33	video	video	NOUN
fcis-30019	8	34	,	,	PUNCT
fcis-30019	8	35	and	and	CCONJ
fcis-30019	8	36	improve	improve	VERB
fcis-30019	8	37	the	the	DET
fcis-30019	8	38	robustness	robustness	NOUN
fcis-30019	8	39	of	of	ADP
fcis-30019	8	40	the	the	DET
fcis-30019	8	41	model	model	NOUN
fcis-30019	8	42	.	.	PUNCT
fcis-30019	9	1	keywords	keyword	NOUN
fcis-30019	9	2	:	:	PUNCT
fcis-30019	9	3	deepfake	deepfake	VERB
fcis-30019	9	4	detection	detection	NOUN
fcis-30019	9	5	;	;	PUNCT
fcis-30019	9	6	spatial	spatial	ADJ
fcis-30019	9	7	domain	domain	NOUN
fcis-30019	9	8	;	;	PUNCT
fcis-30019	9	9	frequency	frequency	NOUN
fcis-30019	9	10	domain	domain	NOUN
fcis-30019	9	11	;	;	PUNCT
fcis-30019	9	12	fecam	fecam	NOUN
fcis-30019	9	13	.	.	PUNCT
fcis-30019	10	1	1	1	X
fcis-30019	10	2	.	.	X
fcis-30019	10	3	introduction	introduction	NOUN
fcis-30019	10	4	deepfake	deepfake	NOUN
fcis-30019	10	5	technology	technology	NOUN
fcis-30019	10	6	[	[	X
fcis-30019	10	7	1	1	NUM
fcis-30019	10	8	]	]	PUNCT
fcis-30019	10	9	is	be	AUX
fcis-30019	10	10	one	one	NUM
fcis-30019	10	11	of	of	ADP
fcis-30019	10	12	the	the	DET
fcis-30019	10	13	key	key	ADJ
fcis-30019	10	14	technologies	technology	NOUN
fcis-30019	10	15	for	for	ADP
fcis-30019	10	16	making	make	VERB
fcis-30019	10	17	fake	fake	ADJ
fcis-30019	10	18	videos	video	NOUN
fcis-30019	10	19	,	,	PUNCT
fcis-30019	10	20	also	also	ADV
fcis-30019	10	21	known	know	VERB
fcis-30019	10	22	as	as	ADP
fcis-30019	10	23	ai	ai	ADP
fcis-30019	10	24	face	face	NOUN
fcis-30019	10	25	changing	change	VERB
fcis-30019	10	26	technology	technology	NOUN
fcis-30019	10	27	.	.	PUNCT
fcis-30019	11	1	once	once	ADV
fcis-30019	11	2	this	this	DET
fcis-30019	11	3	technology	technology	NOUN
fcis-30019	11	4	is	be	AUX
fcis-30019	11	5	maliciously	maliciously	ADV
fcis-30019	11	6	used	use	VERB
fcis-30019	11	7	by	by	ADP
fcis-30019	11	8	attackers	attacker	NOUN
fcis-30019	11	9	,	,	PUNCT
fcis-30019	11	10	it	it	PRON
fcis-30019	11	11	will	will	AUX
fcis-30019	11	12	cause	cause	VERB
fcis-30019	11	13	unpredictable	unpredictable	ADJ
fcis-30019	11	14	consequences	consequence	NOUN
fcis-30019	11	15	to	to	ADP
fcis-30019	11	16	the	the	DET
fcis-30019	11	17	politics	politic	NOUN
fcis-30019	11	18	and	and	CCONJ
fcis-30019	11	19	economy	economy	NOUN
fcis-30019	11	20	of	of	ADP
fcis-30019	11	21	the	the	DET
fcis-30019	11	22	society	society	NOUN
fcis-30019	11	23	.	.	PUNCT
fcis-30019	12	1	a	a	DET
fcis-30019	12	2	typical	typical	ADJ
fcis-30019	12	3	case	case	NOUN
fcis-30019	12	4	is	be	AUX
fcis-30019	12	5	that	that	SCONJ
fcis-30019	12	6	the	the	DET
fcis-30019	12	7	deep	deep	ADJ
fcis-30019	12	8	fake	fake	ADJ
fcis-30019	12	9	videos	video	NOUN
fcis-30019	12	10	of	of	ADP
fcis-30019	12	11	the	the	DET
fcis-30019	12	12	presidents	president	NOUN
fcis-30019	12	13	of	of	ADP
fcis-30019	12	14	russia	russia	PROPN
fcis-30019	12	15	and	and	CCONJ
fcis-30019	12	16	ukraine	ukraine	NOUN
fcis-30019	12	17	are	be	AUX
fcis-30019	12	18	maliciously	maliciously	ADV
fcis-30019	12	19	uploaded	upload	VERB
fcis-30019	12	20	to	to	ADP
fcis-30019	12	21	social	social	ADJ
fcis-30019	12	22	media	medium	NOUN
fcis-30019	12	23	,	,	PUNCT
fcis-30019	12	24	thus	thus	ADV
fcis-30019	12	25	promoting	promote	VERB
fcis-30019	12	26	false	false	ADJ
fcis-30019	12	27	narratives	narrative	NOUN
fcis-30019	12	28	against	against	ADP
fcis-30019	12	29	the	the	DET
fcis-30019	12	30	people	people	NOUN
fcis-30019	12	31	and	and	CCONJ
fcis-30019	12	32	national	national	ADJ
fcis-30019	12	33	interests	interest	NOUN
fcis-30019	12	34	of	of	ADP
fcis-30019	12	35	the	the	DET
fcis-30019	12	36	target	target	NOUN
fcis-30019	12	37	countries	country	NOUN
fcis-30019	12	38	and	and	CCONJ
fcis-30019	12	39	impacting	impact	VERB
fcis-30019	12	40	the	the	DET
fcis-30019	12	41	psychology	psychology	NOUN
fcis-30019	12	42	of	of	ADP
fcis-30019	12	43	the	the	DET
fcis-30019	12	44	people	people	NOUN
fcis-30019	12	45	of	of	ADP
fcis-30019	12	46	the	the	DET
fcis-30019	12	47	target	target	NOUN
fcis-30019	12	48	countries	country	NOUN
fcis-30019	12	49	.	.	PUNCT
fcis-30019	13	1	what	what	PRON
fcis-30019	13	2	is	be	AUX
fcis-30019	13	3	more	more	ADJ
fcis-30019	13	4	,	,	PUNCT
fcis-30019	13	5	the	the	DET
fcis-30019	13	6	faces	face	NOUN
fcis-30019	13	7	of	of	ADP
fcis-30019	13	8	some	some	DET
fcis-30019	13	9	wellknown	wellknown	ADJ
fcis-30019	13	10	singers	singer	NOUN
fcis-30019	13	11	,	,	PUNCT
fcis-30019	13	12	movie	movie	NOUN
fcis-30019	13	13	stars	star	NOUN
fcis-30019	13	14	and	and	CCONJ
fcis-30019	13	15	other	other	ADJ
fcis-30019	13	16	public	public	ADJ
fcis-30019	13	17	figures	figure	NOUN
fcis-30019	13	18	are	be	AUX
fcis-30019	13	19	"	"	PUNCT
fcis-30019	13	20	transferred	transfer	VERB
fcis-30019	13	21	"	"	PUNCT
fcis-30019	13	22	to	to	ADP
fcis-30019	13	23	porn	porn	NOUN
fcis-30019	13	24	stars	star	NOUN
fcis-30019	13	25	,	,	PUNCT
fcis-30019	13	26	falsifying	falsify	VERB
fcis-30019	13	27	porn	porn	NOUN
fcis-30019	13	28	for	for	ADP
fcis-30019	13	29	illegal	illegal	ADJ
fcis-30019	13	30	profit	profit	NOUN
fcis-30019	13	31	,	,	PUNCT
fcis-30019	13	32	which	which	PRON
fcis-30019	13	33	constitutes	constitute	VERB
fcis-30019	13	34	a	a	DET
fcis-30019	13	35	serious	serious	ADJ
fcis-30019	13	36	violation	violation	NOUN
fcis-30019	13	37	of	of	ADP
fcis-30019	13	38	personal	personal	ADJ
fcis-30019	13	39	reputation	reputation	NOUN
fcis-30019	13	40	and	and	CCONJ
fcis-30019	13	41	portrait	portrait	NOUN
fcis-30019	13	42	rights	right	NOUN
fcis-30019	13	43	.	.	PUNCT
fcis-30019	14	1	at	at	ADP
fcis-30019	14	2	present	present	ADJ
fcis-30019	14	3	,	,	PUNCT
fcis-30019	14	4	the	the	DET
fcis-30019	14	5	mainstream	mainstream	NOUN
fcis-30019	14	6	method	method	NOUN
fcis-30019	14	7	of	of	ADP
fcis-30019	14	8	forgery	forgery	NOUN
fcis-30019	14	9	is	be	AUX
fcis-30019	14	10	to	to	PART
fcis-30019	14	11	use	use	VERB
fcis-30019	14	12	machine	machine	NOUN
fcis-30019	14	13	learning	learn	VERB
fcis-30019	14	14	algorithms	algorithm	NOUN
fcis-30019	14	15	such	such	ADJ
fcis-30019	14	16	as	as	ADP
fcis-30019	14	17	generative	generative	ADJ
fcis-30019	14	18	adversarial	adversarial	ADJ
fcis-30019	14	19	network	network	NOUN
fcis-30019	14	20	(	(	PUNCT
fcis-30019	14	21	gan	gan	PROPN
fcis-30019	14	22	)	)	PUNCT
fcis-30019	15	1	[	[	X
fcis-30019	15	2	2	2	NUM
fcis-30019	15	3	]	]	PUNCT
fcis-30019	15	4	or	or	CCONJ
fcis-30019	15	5	convolutional	convolutional	ADJ
fcis-30019	15	6	neural	neural	ADJ
fcis-30019	15	7	network	network	NOUN
fcis-30019	15	8	(	(	PUNCT
fcis-30019	15	9	cnn	cnn	PROPN
fcis-30019	15	10	)	)	PUNCT
fcis-30019	15	11	through	through	ADP
fcis-30019	15	12	a	a	DET
fcis-30019	15	13	large	large	ADJ
fcis-30019	15	14	number	number	NOUN
fcis-30019	15	15	of	of	ADP
fcis-30019	15	16	coding	code	VERB
fcis-30019	15	17	and	and	CCONJ
fcis-30019	15	18	decoding	decode	VERB
fcis-30019	15	19	training	training	NOUN
fcis-30019	15	20	.	.	PUNCT
fcis-30019	16	1	the	the	DET
fcis-30019	16	2	robustness	robustness	NOUN
fcis-30019	16	3	of	of	ADP
fcis-30019	16	4	the	the	DET
fcis-30019	16	5	model	model	NOUN
fcis-30019	16	6	is	be	AUX
fcis-30019	16	7	enhanced	enhance	VERB
fcis-30019	16	8	,	,	PUNCT
fcis-30019	16	9	and	and	CCONJ
fcis-30019	16	10	finally	finally	ADV
fcis-30019	16	11	the	the	DET
fcis-30019	16	12	facial	facial	ADJ
fcis-30019	16	13	features	feature	NOUN
fcis-30019	16	14	of	of	ADP
fcis-30019	16	15	one	one	NUM
fcis-30019	16	16	target	target	NOUN
fcis-30019	16	17	individual	individual	NOUN
fcis-30019	16	18	are	be	AUX
fcis-30019	16	19	successfully	successfully	ADV
fcis-30019	16	20	"	"	PUNCT
fcis-30019	16	21	transplanted	transplant	VERB
fcis-30019	16	22	"	"	PUNCT
fcis-30019	16	23	to	to	ADP
fcis-30019	16	24	the	the	DET
fcis-30019	16	25	face	face	NOUN
fcis-30019	16	26	of	of	ADP
fcis-30019	16	27	another	another	DET
fcis-30019	16	28	individual	individual	NOUN
fcis-30019	16	29	.	.	PUNCT
fcis-30019	17	1	since	since	SCONJ
fcis-30019	17	2	videos	video	NOUN
fcis-30019	17	3	are	be	AUX
fcis-30019	17	4	made	make	VERB
fcis-30019	17	5	up	up	ADP
fcis-30019	17	6	of	of	ADP
fcis-30019	17	7	successive	successive	ADJ
fcis-30019	17	8	frames	frame	NOUN
fcis-30019	17	9	,	,	PUNCT
fcis-30019	17	10	simply	simply	ADV
fcis-30019	17	11	replacing	replace	VERB
fcis-30019	17	12	the	the	DET
fcis-30019	17	13	faces	face	NOUN
fcis-30019	17	14	in	in	ADP
fcis-30019	17	15	each	each	DET
fcis-30019	17	16	picture	picture	NOUN
fcis-30019	17	17	can	can	AUX
fcis-30019	17	18	result	result	VERB
fcis-30019	17	19	in	in	ADP
fcis-30019	17	20	a	a	DET
fcis-30019	17	21	new	new	ADJ
fcis-30019	17	22	,	,	PUNCT
fcis-30019	17	23	very	very	ADV
fcis-30019	17	24	realistic	realistic	ADJ
fcis-30019	17	25	fake	fake	ADJ
fcis-30019	17	26	video	video	NOUN
fcis-30019	17	27	.	.	PUNCT
fcis-30019	18	1	to	to	PART
fcis-30019	18	2	be	be	AUX
fcis-30019	18	3	specific	specific	ADJ
fcis-30019	18	4	,	,	PUNCT
fcis-30019	18	5	first	first	ADV
fcis-30019	18	6	of	of	ADP
fcis-30019	18	7	all	all	PRON
fcis-30019	18	8	,	,	PUNCT
fcis-30019	18	9	the	the	DET
fcis-30019	18	10	video	video	NOUN
fcis-30019	18	11	of	of	ADP
fcis-30019	18	12	the	the	DET
fcis-30019	18	13	imitated	imitate	VERB
fcis-30019	18	14	object	object	NOUN
fcis-30019	18	15	is	be	AUX
fcis-30019	18	16	transformed	transform	VERB
fcis-30019	18	17	frame	frame	NOUN
fcis-30019	18	18	by	by	ADP
fcis-30019	18	19	frame	frame	NOUN
fcis-30019	18	20	to	to	PART
fcis-30019	18	21	obtain	obtain	VERB
fcis-30019	18	22	a	a	DET
fcis-30019	18	23	large	large	ADJ
fcis-30019	18	24	number	number	NOUN
fcis-30019	18	25	of	of	ADP
fcis-30019	18	26	photos	photo	NOUN
fcis-30019	18	27	,	,	PUNCT
fcis-30019	18	28	then	then	ADV
fcis-30019	18	29	the	the	DET
fcis-30019	18	30	face	face	NOUN
fcis-30019	18	31	of	of	ADP
fcis-30019	18	32	the	the	DET
fcis-30019	18	33	object	object	NOUN
fcis-30019	18	34	is	be	AUX
fcis-30019	18	35	replaced	replace	VERB
fcis-30019	18	36	by	by	ADP
fcis-30019	18	37	the	the	DET
fcis-30019	18	38	face	face	NOUN
fcis-30019	18	39	of	of	ADP
fcis-30019	18	40	the	the	DET
fcis-30019	18	41	imitated	imitate	VERB
fcis-30019	18	42	object	object	NOUN
fcis-30019	18	43	,	,	PUNCT
fcis-30019	18	44	and	and	CCONJ
fcis-30019	18	45	finally	finally	ADV
fcis-30019	18	46	the	the	DET
fcis-30019	18	47	fake	fake	ADJ
fcis-30019	18	48	video	video	NOUN
fcis-30019	18	49	resynthesis	resynthesis	NOUN
fcis-30019	18	50	of	of	ADP
fcis-30019	18	51	the	the	DET
fcis-30019	18	52	replaced	replace	VERB
fcis-30019	18	53	photo	photo	NOUN
fcis-30019	18	54	is	be	AUX
fcis-30019	18	55	carried	carry	VERB
fcis-30019	18	56	out	out	ADP
fcis-30019	18	57	to	to	PART
fcis-30019	18	58	deceive	deceive	VERB
fcis-30019	18	59	the	the	DET
fcis-30019	18	60	human	human	ADJ
fcis-30019	18	61	eye	eye	NOUN
fcis-30019	18	62	and	and	CCONJ
fcis-30019	18	63	produce	produce	VERB
fcis-30019	18	64	wrong	wrong	ADJ
fcis-30019	18	65	cognition	cognition	NOUN
fcis-30019	18	66	.	.	PUNCT
fcis-30019	19	1	although	although	SCONJ
fcis-30019	19	2	the	the	DET
fcis-30019	19	3	current	current	ADJ
fcis-30019	19	4	mainstream	mainstream	NOUN
fcis-30019	19	5	detection	detection	NOUN
fcis-30019	19	6	methods	method	NOUN
fcis-30019	19	7	are	be	AUX
fcis-30019	19	8	largely	largely	ADV
fcis-30019	19	9	based	base	VERB
fcis-30019	19	10	on	on	ADP
fcis-30019	19	11	deep	deep	ADJ
fcis-30019	19	12	learning	learning	NOUN
fcis-30019	19	13	technology	technology	NOUN
fcis-30019	19	14	,	,	PUNCT
fcis-30019	19	15	the	the	DET
fcis-30019	19	16	inherent	inherent	ADJ
fcis-30019	19	17	limitations	limitation	NOUN
fcis-30019	19	18	of	of	ADP
fcis-30019	19	19	deep	deep	ADJ
fcis-30019	19	20	learning	learning	NOUN
fcis-30019	19	21	have	have	AUX
fcis-30019	19	22	become	become	VERB
fcis-30019	19	23	a	a	DET
fcis-30019	19	24	key	key	ADJ
fcis-30019	19	25	factor	factor	NOUN
fcis-30019	19	26	restricting	restrict	VERB
fcis-30019	19	27	the	the	DET
fcis-30019	19	28	progress	progress	NOUN
fcis-30019	19	29	of	of	ADP
fcis-30019	19	30	deep	deep	ADJ
fcis-30019	19	31	fake	fake	ADJ
fcis-30019	19	32	video	video	NOUN
fcis-30019	19	33	detection	detection	NOUN
fcis-30019	19	34	technology	technology	NOUN
fcis-30019	19	35	.	.	PUNCT
fcis-30019	20	1	the	the	DET
fcis-30019	20	2	method	method	NOUN
fcis-30019	20	3	can	can	AUX
fcis-30019	20	4	be	be	AUX
fcis-30019	20	5	classified	classify	VERB
fcis-30019	20	6	as	as	ADP
fcis-30019	20	7	"	"	PUNCT
fcis-30019	20	8	feature	feature	NOUN
fcis-30019	20	9	extraction	extraction	NOUN
fcis-30019	20	10	network	network	NOUN
fcis-30019	20	11	+	+	CCONJ
fcis-30019	20	12	classification	classification	NOUN
fcis-30019	20	13	network	network	NOUN
fcis-30019	20	14	"	"	PUNCT
fcis-30019	20	15	,	,	PUNCT
fcis-30019	20	16	where	where	SCONJ
fcis-30019	20	17	in	in	ADP
fcis-30019	20	18	feature	feature	NOUN
fcis-30019	20	19	extraction	extraction	NOUN
fcis-30019	20	20	network	network	NOUN
fcis-30019	20	21	is	be	AUX
fcis-30019	20	22	generally	generally	ADV
fcis-30019	20	23	a	a	DET
fcis-30019	20	24	deep	deep	ADJ
fcis-30019	20	25	structure	structure	NOUN
fcis-30019	20	26	and	and	CCONJ
fcis-30019	20	27	has	have	VERB
fcis-30019	20	28	the	the	DET
fcis-30019	20	29	inherent	inherent	ADJ
fcis-30019	20	30	defect	defect	NOUN
fcis-30019	20	31	of	of	ADP
fcis-30019	20	32	low	low	ADJ
fcis-30019	20	33	learning	learning	NOUN
fcis-30019	20	34	priority	priority	NOUN
fcis-30019	20	35	for	for	ADP
fcis-30019	20	36	high	high	ADJ
fcis-30019	20	37	-	-	PUNCT
fcis-30019	20	38	frequency	frequency	NOUN
fcis-30019	20	39	components	component	NOUN
fcis-30019	20	40	,	,	PUNCT
fcis-30019	20	41	so	so	SCONJ
fcis-30019	20	42	the	the	DET
fcis-30019	20	43	utilization	utilization	NOUN
fcis-30019	20	44	rate	rate	NOUN
fcis-30019	20	45	of	of	ADP
fcis-30019	20	46	high	high	ADJ
fcis-30019	20	47	-	-	PUNCT
fcis-30019	20	48	frequency	frequency	NOUN
fcis-30019	20	49	information	information	NOUN
fcis-30019	20	50	is	be	AUX
fcis-30019	20	51	low	low	ADJ
fcis-30019	20	52	in	in	ADP
fcis-30019	20	53	the	the	DET
fcis-30019	20	54	final	final	ADJ
fcis-30019	20	55	classification	classification	NOUN
fcis-30019	20	56	decision	decision	NOUN
fcis-30019	20	57	.	.	PUNCT
fcis-30019	21	1	however	however	ADV
fcis-30019	21	2	,	,	PUNCT
fcis-30019	21	3	due	due	ADP
fcis-30019	21	4	to	to	ADP
fcis-30019	21	5	the	the	DET
fcis-30019	21	6	special	special	ADJ
fcis-30019	21	7	synthesis	synthesis	NOUN
fcis-30019	21	8	method	method	NOUN
fcis-30019	21	9	of	of	ADP
fcis-30019	21	10	deepfake	deepfake	NOUN
fcis-30019	21	11	technology	technology	NOUN
fcis-30019	21	12	,	,	PUNCT
fcis-30019	21	13	discontinuous	discontinuous	ADJ
fcis-30019	21	14	features	feature	NOUN
fcis-30019	21	15	will	will	AUX
fcis-30019	21	16	be	be	AUX
fcis-30019	21	17	generated	generate	VERB
fcis-30019	21	18	between	between	ADP
fcis-30019	21	19	the	the	DET
fcis-30019	21	20	synthetic	synthetic	ADJ
fcis-30019	21	21	face	face	NOUN
fcis-30019	21	22	and	and	CCONJ
fcis-30019	21	23	the	the	DET
fcis-30019	21	24	surrounding	surround	VERB
fcis-30019	21	25	pixels	pixel	NOUN
fcis-30019	21	26	,	,	PUNCT
fcis-30019	21	27	and	and	CCONJ
fcis-30019	21	28	such	such	ADJ
fcis-30019	21	29	edge	edge	NOUN
fcis-30019	21	30	discontinuous	discontinuous	ADJ
fcis-30019	21	31	features	feature	NOUN
fcis-30019	21	32	often	often	ADV
fcis-30019	21	33	exist	exist	VERB
fcis-30019	21	34	in	in	ADP
fcis-30019	21	35	the	the	DET
fcis-30019	21	36	high	high	ADJ
fcis-30019	21	37	-	-	PUNCT
fcis-30019	21	38	frequency	frequency	NOUN
fcis-30019	21	39	information	information	NOUN
fcis-30019	21	40	in	in	ADP
fcis-30019	21	41	the	the	DET
fcis-30019	21	42	frequency	frequency	NOUN
fcis-30019	21	43	domain	domain	NOUN
fcis-30019	21	44	.	.	PUNCT
fcis-30019	22	1	see	see	VERB
fcis-30019	22	2	fig	fig	NOUN
fcis-30019	22	3	.	.	PUNCT
fcis-30019	23	1	1	1	NUM
fcis-30019	23	2	with	with	ADP
fcis-30019	23	3	the	the	DET
fcis-30019	23	4	development	development	NOUN
fcis-30019	23	5	and	and	CCONJ
fcis-30019	23	6	iteration	iteration	NOUN
fcis-30019	23	7	of	of	ADP
fcis-30019	23	8	deepfake	deepfake	NOUN
fcis-30019	23	9	technology	technology	NOUN
fcis-30019	23	10	,	,	PUNCT
fcis-30019	23	11	it	it	PRON
fcis-30019	23	12	is	be	AUX
fcis-30019	23	13	difficult	difficult	ADJ
fcis-30019	23	14	to	to	PART
fcis-30019	23	15	identify	identify	VERB
fcis-30019	23	16	such	such	ADJ
fcis-30019	23	17	edge	edge	NOUN
fcis-30019	23	18	discontinuity	discontinuity	NOUN
fcis-30019	23	19	features	feature	NOUN
fcis-30019	23	20	based	base	VERB
fcis-30019	23	21	on	on	ADP
fcis-30019	23	22	the	the	DET
fcis-30019	23	23	spatial	spatial	ADJ
fcis-30019	23	24	domain	domain	NOUN
fcis-30019	23	25	.	.	PUNCT
fcis-30019	24	1	therefore	therefore	ADV
fcis-30019	24	2	,	,	PUNCT
fcis-30019	24	3	it	it	PRON
fcis-30019	24	4	is	be	AUX
fcis-30019	24	5	an	an	DET
fcis-30019	24	6	effective	effective	ADJ
fcis-30019	24	7	and	and	CCONJ
fcis-30019	24	8	feasible	feasible	ADJ
fcis-30019	24	9	method	method	NOUN
fcis-30019	24	10	to	to	PART
fcis-30019	24	11	detect	detect	VERB
fcis-30019	24	12	deep	deep	ADJ
fcis-30019	24	13	forged	forge	VERB
fcis-30019	24	14	face	face	NOUN
fcis-30019	24	15	video	video	NOUN
fcis-30019	24	16	by	by	ADP
fcis-30019	24	17	using	use	VERB
fcis-30019	24	18	spatial	spatial	ADJ
fcis-30019	24	19	information	information	NOUN
fcis-30019	24	20	and	and	CCONJ
fcis-30019	24	21	frequency	frequency	NOUN
fcis-30019	24	22	domain	domain	NOUN
fcis-30019	24	23	information	information	NOUN
fcis-30019	24	24	fusion	fusion	NOUN
fcis-30019	24	25	fig	fig	NOUN
fcis-30019	24	26	1	1	NUM
fcis-30019	24	27	.	.	PUNCT
fcis-30019	24	28	example	example	NOUN
fcis-30019	24	29	of	of	ADP
fcis-30019	24	30	separating	separate	VERB
fcis-30019	24	31	low	low	ADJ
fcis-30019	24	32	frequency	frequency	NOUN
fcis-30019	24	33	information	information	NOUN
fcis-30019	24	34	from	from	ADP
fcis-30019	24	35	high	high	ADJ
fcis-30019	24	36	frequency	frequency	NOUN
fcis-30019	24	37	information	information	NOUN
fcis-30019	24	38	2	2	NUM
fcis-30019	24	39	.	.	PUNCT
fcis-30019	25	1	related	relate	VERB
fcis-30019	25	2	work	work	NOUN
fcis-30019	25	3	the	the	DET
fcis-30019	25	4	detection	detection	NOUN
fcis-30019	25	5	[	[	X
fcis-30019	25	6	3	3	X
fcis-30019	25	7	]	]	PUNCT
fcis-30019	25	8	and	and	CCONJ
fcis-30019	25	9	prevention	prevention	NOUN
fcis-30019	25	10	of	of	ADP
fcis-30019	25	11	deepfake	deepfake	NOUN
fcis-30019	25	12	videos	video	NOUN
fcis-30019	25	13	has	have	AUX
fcis-30019	25	14	become	become	VERB
fcis-30019	25	15	a	a	DET
fcis-30019	25	16	key	key	ADJ
fcis-30019	25	17	research	research	NOUN
fcis-30019	25	18	in	in	ADP
fcis-30019	25	19	the	the	DET
fcis-30019	25	20	field	field	NOUN
fcis-30019	25	21	of	of	ADP
fcis-30019	25	22	network	network	NOUN
fcis-30019	25	23	security	security	NOUN
fcis-30019	25	24	,	,	PUNCT
fcis-30019	25	25	and	and	CCONJ
fcis-30019	25	26	its	its	PRON
fcis-30019	25	27	core	core	NOUN
fcis-30019	25	28	task	task	NOUN
fcis-30019	25	29	is	be	AUX
fcis-30019	25	30	to	to	PART
fcis-30019	25	31	quickly	quickly	ADV
fcis-30019	25	32	and	and	CCONJ
fcis-30019	25	33	accurately	accurately	ADV
fcis-30019	25	34	identify	identify	VERB
fcis-30019	25	35	the	the	DET
fcis-30019	25	36	traces	trace	NOUN
fcis-30019	25	37	of	of	ADP
fcis-30019	25	38	55	55	NUM
fcis-30019	25	39	forgery	forgery	NOUN
fcis-30019	25	40	in	in	ADP
fcis-30019	25	41	videos	video	NOUN
fcis-30019	25	42	.	.	PUNCT
fcis-30019	26	1	for	for	ADP
fcis-30019	26	2	mainstream	mainstream	NOUN
fcis-30019	26	3	forged	forge	VERB
fcis-30019	26	4	audio	audio	NOUN
fcis-30019	26	5	and	and	CCONJ
fcis-30019	26	6	video	video	NOUN
fcis-30019	26	7	,	,	PUNCT
fcis-30019	26	8	traditional	traditional	ADJ
fcis-30019	26	9	detection	detection	NOUN
fcis-30019	26	10	methods	method	NOUN
fcis-30019	26	11	usually	usually	ADV
fcis-30019	26	12	use	use	VERB
fcis-30019	26	13	face	face	NOUN
fcis-30019	26	14	recognition	recognition	NOUN
fcis-30019	26	15	algorithms	algorithm	NOUN
fcis-30019	26	16	,	,	PUNCT
fcis-30019	26	17	convert	convert	VERB
fcis-30019	26	18	facial	facial	ADJ
fcis-30019	26	19	data	datum	NOUN
fcis-30019	26	20	into	into	ADP
fcis-30019	26	21	feature	feature	NOUN
fcis-30019	26	22	vectors	vector	NOUN
fcis-30019	26	23	,	,	PUNCT
fcis-30019	26	24	and	and	CCONJ
fcis-30019	26	25	distinguish	distinguish	VERB
fcis-30019	26	26	the	the	DET
fcis-30019	26	27	authenticity	authenticity	NOUN
fcis-30019	26	28	of	of	ADP
fcis-30019	26	29	the	the	DET
fcis-30019	26	30	video	video	NOUN
fcis-30019	26	31	by	by	ADP
fcis-30019	26	32	machine	machine	NOUN
fcis-30019	26	33	learning	learn	VERB
fcis-30019	26	34	techniques	technique	NOUN
fcis-30019	26	35	such	such	ADJ
fcis-30019	26	36	as	as	ADP
fcis-30019	26	37	support	support	NOUN
fcis-30019	26	38	vector	vector	NOUN
fcis-30019	26	39	machine	machine	NOUN
fcis-30019	26	40	(	(	PUNCT
fcis-30019	26	41	svm)[4	svm)[4	NOUN
fcis-30019	26	42	]	]	PUNCT
fcis-30019	26	43	.	.	PUNCT
fcis-30019	27	1	several	several	ADJ
fcis-30019	27	2	techniques	technique	NOUN
fcis-30019	27	3	that	that	PRON
fcis-30019	27	4	have	have	AUX
fcis-30019	27	5	been	be	AUX
fcis-30019	27	6	introduced	introduce	VERB
fcis-30019	27	7	rely	rely	ADV
fcis-30019	27	8	on	on	ADP
fcis-30019	27	9	cnns	cnn	NOUN
fcis-30019	27	10	for	for	ADP
fcis-30019	27	11	a	a	DET
fcis-30019	27	12	frame	frame	NOUN
fcis-30019	27	13	-	-	PUNCT
fcis-30019	27	14	by	by	ADP
fcis-30019	27	15	-	-	PUNCT
fcis-30019	27	16	frame	frame	NOUN
fcis-30019	27	17	examination	examination	NOUN
fcis-30019	27	18	.	.	PUNCT
fcis-30019	28	1	for	for	ADP
fcis-30019	28	2	example	example	NOUN
fcis-30019	28	3	,	,	PUNCT
fcis-30019	28	4	a	a	DET
fcis-30019	28	5	method	method	NOUN
fcis-30019	28	6	known	know	VERB
fcis-30019	28	7	as	as	ADP
fcis-30019	28	8	mesonet[5	mesonet[5	NOUN
fcis-30019	28	9	]	]	PUNCT
fcis-30019	28	10	has	have	AUX
fcis-30019	28	11	been	be	AUX
fcis-30019	28	12	developed	develop	VERB
fcis-30019	28	13	,	,	PUNCT
fcis-30019	28	14	which	which	PRON
fcis-30019	28	15	is	be	AUX
fcis-30019	28	16	a	a	DET
fcis-30019	28	17	less	less	ADV
fcis-30019	28	18	complex	complex	ADJ
fcis-30019	28	19	cnn	cnn	NOUN
fcis-30019	28	20	designed	design	VERB
fcis-30019	28	21	to	to	PART
fcis-30019	28	22	identify	identify	VERB
fcis-30019	28	23	synthetic	synthetic	ADJ
fcis-30019	28	24	faces	face	NOUN
fcis-30019	28	25	.	.	PUNCT
fcis-30019	29	1	however	however	ADV
fcis-30019	29	2	,	,	PUNCT
fcis-30019	29	3	a	a	DET
fcis-30019	29	4	research	research	NOUN
fcis-30019	29	5	[	[	X
fcis-30019	29	6	6	6	NUM
fcis-30019	29	7	]	]	PUNCT
fcis-30019	29	8	demonstrated	demonstrate	VERB
fcis-30019	29	9	that	that	SCONJ
fcis-30019	29	10	this	this	DET
fcis-30019	29	11	network	network	NOUN
fcis-30019	29	12	is	be	AUX
fcis-30019	29	13	surpassed	surpass	VERB
fcis-30019	29	14	in	in	ADP
fcis-30019	29	15	performance	performance	NOUN
fcis-30019	29	16	by	by	ADP
fcis-30019	29	17	a	a	DET
fcis-30019	29	18	retrained	retrain	VERB
fcis-30019	29	19	xceptionnet	xceptionnet	NOUN
fcis-30019	29	20	designed	design	VERB
fcis-30019	29	21	for	for	ADP
fcis-30019	29	22	the	the	DET
fcis-30019	29	23	same	same	ADJ
fcis-30019	29	24	purpose	purpose	NOUN
fcis-30019	29	25	.	.	PUNCT
fcis-30019	30	1	recently	recently	ADV
fcis-30019	30	2	,	,	PUNCT
fcis-30019	30	3	the	the	DET
fcis-30019	30	4	f3	f3	ADJ
fcis-30019	30	5	-	-	PUNCT
fcis-30019	30	6	net	net	NOUN
fcis-30019	30	7	[	[	X
fcis-30019	30	8	7]employs	7]employ	VERB
fcis-30019	30	9	a	a	DET
fcis-30019	30	10	dual	dual	ADJ
fcis-30019	30	11	-	-	PUNCT
fcis-30019	30	12	branch	branch	NOUN
fcis-30019	30	13	framework	framework	NOUN
fcis-30019	30	14	,	,	PUNCT
fcis-30019	30	15	with	with	ADP
fcis-30019	30	16	one	one	NUM
fcis-30019	30	17	branch	branch	NOUN
fcis-30019	30	18	dedicated	dedicate	VERB
fcis-30019	30	19	to	to	ADP
fcis-30019	30	20	identifying	identify	VERB
fcis-30019	30	21	forgery	forgery	NOUN
fcis-30019	30	22	patterns	pattern	NOUN
fcis-30019	30	23	through	through	ADP
fcis-30019	30	24	frequency	frequency	NOUN
fcis-30019	30	25	cues	cue	NOUN
fcis-30019	30	26	,	,	PUNCT
fcis-30019	30	27	while	while	SCONJ
fcis-30019	30	28	the	the	DET
fcis-30019	30	29	other	other	ADJ
fcis-30019	30	30	branch	branch	NOUN
fcis-30019	30	31	focuses	focus	VERB
fcis-30019	30	32	on	on	ADP
fcis-30019	30	33	discerning	discern	VERB
fcis-30019	30	34	the	the	DET
fcis-30019	30	35	discrepancies	discrepancy	NOUN
fcis-30019	30	36	in	in	ADP
fcis-30019	30	37	frequency	frequency	NOUN
fcis-30019	30	38	statistical	statistical	ADJ
fcis-30019	30	39	properties	property	NOUN
fcis-30019	30	40	between	between	ADP
fcis-30019	30	41	authentic	authentic	ADJ
fcis-30019	30	42	and	and	CCONJ
fcis-30019	30	43	counterfeit	counterfeit	ADJ
fcis-30019	30	44	images	image	NOUN
fcis-30019	30	45	.	.	PUNCT
fcis-30019	31	1	other	other	ADJ
fcis-30019	31	2	methods	method	NOUN
fcis-30019	31	3	leverage	leverage	VERB
fcis-30019	31	4	the	the	DET
fcis-30019	31	5	temporal	temporal	ADJ
fcis-30019	31	6	progression	progression	NOUN
fcis-30019	31	7	of	of	ADP
fcis-30019	31	8	video	video	NOUN
fcis-30019	31	9	frames	frame	NOUN
fcis-30019	31	10	by	by	ADP
fcis-30019	31	11	utilizing	utilize	VERB
fcis-30019	31	12	long	long	ADJ
fcis-30019	31	13	short	short	ADJ
fcis-30019	31	14	-	-	PUNCT
fcis-30019	31	15	term	term	NOUN
fcis-30019	31	16	memory	memory	NOUN
fcis-30019	31	17	(	(	PUNCT
fcis-30019	31	18	lstm	lstm	NOUN
fcis-30019	31	19	)	)	PUNCT
fcis-30019	32	1	[	[	X
fcis-30019	32	2	8]networks	8]network	NOUN
fcis-30019	32	3	for	for	ADP
fcis-30019	32	4	analysis	analysis	NOUN
fcis-30019	32	5	,	,	PUNCT
fcis-30019	32	6	where	where	SCONJ
fcis-30019	32	7	frame	frame	NOUN
fcis-30019	32	8	-	-	PUNCT
fcis-30019	32	9	specific	specific	ADJ
fcis-30019	32	10	features	feature	NOUN
fcis-30019	32	11	are	be	AUX
fcis-30019	32	12	initially	initially	ADV
fcis-30019	32	13	extracted	extract	VERB
fcis-30019	32	14	and	and	CCONJ
fcis-30019	32	15	subsequently	subsequently	ADV
fcis-30019	32	16	integrated	integrate	VERB
fcis-30019	32	17	using	use	VERB
fcis-30019	32	18	a	a	DET
fcis-30019	32	19	recurrent	recurrent	ADJ
fcis-30019	32	20	neural	neural	ADJ
fcis-30019	32	21	network	network	NOUN
fcis-30019	32	22	architecture	architecture	NOUN
fcis-30019	32	23	.	.	PUNCT
fcis-30019	33	1	this	this	DET
fcis-30019	33	2	paper	paper	NOUN
fcis-30019	33	3	implements	implement	VERB
fcis-30019	33	4	a	a	DET
fcis-30019	33	5	model	model	NOUN
fcis-30019	33	6	framework	framework	NOUN
fcis-30019	33	7	that	that	PRON
fcis-30019	33	8	fuses	fuse	VERB
fcis-30019	33	9	the	the	DET
fcis-30019	33	10	spatial	spatial	ADJ
fcis-30019	33	11	-	-	PUNCT
fcis-30019	33	12	frequency	frequency	NOUN
fcis-30019	33	13	domain	domain	NOUN
fcis-30019	33	14	to	to	PART
fcis-30019	33	15	detect	detect	VERB
fcis-30019	33	16	local	local	ADJ
fcis-30019	33	17	inconsistencies	inconsistency	NOUN
fcis-30019	33	18	at	at	ADP
fcis-30019	33	19	various	various	ADJ
fcis-30019	33	20	scales	scale	NOUN
fcis-30019	33	21	for	for	ADP
fcis-30019	33	22	forgery	forgery	NOUN
fcis-30019	33	23	identification	identification	NOUN
fcis-30019	33	24	.	.	PUNCT
fcis-30019	34	1	furthermore	furthermore	ADV
fcis-30019	34	2	,	,	PUNCT
fcis-30019	34	3	we	we	PRON
fcis-30019	34	4	incorporate	incorporate	VERB
fcis-30019	34	5	frequency	frequency	NOUN
fcis-30019	34	6	modality	modality	NOUN
fcis-30019	34	7	to	to	PART
fcis-30019	34	8	enhance	enhance	VERB
fcis-30019	34	9	the	the	DET
fcis-30019	34	10	resilience	resilience	NOUN
fcis-30019	34	11	of	of	ADP
fcis-30019	34	12	our	our	PRON
fcis-30019	34	13	approach	approach	NOUN
fcis-30019	34	14	against	against	ADP
fcis-30019	34	15	diverse	diverse	ADJ
fcis-30019	34	16	image	image	NOUN
fcis-30019	34	17	compression	compression	NOUN
fcis-30019	34	18	techniques	technique	NOUN
fcis-30019	34	19	.	.	PUNCT
fcis-30019	35	1	3	3	X
fcis-30019	35	2	.	.	X
fcis-30019	35	3	approh	approh	NOUN
fcis-30019	35	4	in	in	ADP
fcis-30019	35	5	this	this	DET
fcis-30019	35	6	paper	paper	NOUN
fcis-30019	35	7	,	,	PUNCT
fcis-30019	35	8	we	we	PRON
fcis-30019	35	9	propose	propose	VERB
fcis-30019	35	10	a	a	DET
fcis-30019	35	11	novel	novel	ADJ
fcis-30019	35	12	deepfake	deepfake	NOUN
fcis-30019	35	13	video	video	NOUN
fcis-30019	35	14	detection	detection	NOUN
fcis-30019	35	15	scheme	scheme	NOUN
fcis-30019	35	16	fusing	fuse	VERB
fcis-30019	35	17	spatial	spatial	ADJ
fcis-30019	35	18	-	-	PUNCT
fcis-30019	35	19	frequency	frequency	NOUN
fcis-30019	35	20	domain	domain	NOUN
fcis-30019	35	21	features	feature	NOUN
fcis-30019	35	22	,	,	PUNCT
fcis-30019	35	23	named	name	VERB
fcis-30019	35	24	sfdt	sfdt	NOUN
fcis-30019	35	25	.	.	PUNCT
fcis-30019	36	1	the	the	DET
fcis-30019	36	2	model	model	NOUN
fcis-30019	36	3	consists	consist	VERB
fcis-30019	36	4	of	of	ADP
fcis-30019	36	5	two	two	NUM
fcis-30019	36	6	key	key	ADJ
fcis-30019	36	7	layers	layer	NOUN
fcis-30019	36	8	:	:	PUNCT
fcis-30019	36	9	spatial	spatial	ADJ
fcis-30019	36	10	domain	domain	NOUN
fcis-30019	36	11	feature	feature	NOUN
fcis-30019	36	12	extraction	extraction	NOUN
fcis-30019	36	13	layer	layer	NOUN
fcis-30019	36	14	and	and	CCONJ
fcis-30019	36	15	frequency	frequency	NOUN
fcis-30019	36	16	domain	domain	NOUN
fcis-30019	36	17	feature	feature	NOUN
fcis-30019	36	18	extraction	extraction	NOUN
fcis-30019	36	19	layer	layer	NOUN
fcis-30019	36	20	.	.	PUNCT
fcis-30019	37	1	firstly	firstly	ADV
fcis-30019	37	2	,	,	PUNCT
fcis-30019	37	3	several	several	ADJ
fcis-30019	37	4	convolution	convolution	NOUN
fcis-30019	37	5	layers	layer	NOUN
fcis-30019	37	6	are	be	AUX
fcis-30019	37	7	used	use	VERB
fcis-30019	37	8	to	to	PART
fcis-30019	37	9	extract	extract	VERB
fcis-30019	37	10	the	the	DET
fcis-30019	37	11	features	feature	NOUN
fcis-30019	37	12	of	of	ADP
fcis-30019	37	13	the	the	DET
fcis-30019	37	14	input	input	NOUN
fcis-30019	37	15	image	image	NOUN
fcis-30019	37	16	as	as	ADP
fcis-30019	37	17	m1	m1	PROPN
fcis-30019	37	18	.	.	PUNCT
fcis-30019	38	1	the	the	DET
fcis-30019	38	2	purpose	purpose	NOUN
fcis-30019	38	3	of	of	ADP
fcis-30019	38	4	convolution	convolution	NOUN
fcis-30019	38	5	is	be	AUX
fcis-30019	38	6	mainly	mainly	ADV
fcis-30019	38	7	to	to	PART
fcis-30019	38	8	speed	speed	VERB
fcis-30019	38	9	up	up	ADP
fcis-30019	38	10	the	the	DET
fcis-30019	38	11	convergence	convergence	NOUN
fcis-30019	38	12	speed	speed	NOUN
fcis-30019	38	13	,	,	PUNCT
fcis-30019	38	14	and	and	CCONJ
fcis-30019	38	15	then	then	ADV
fcis-30019	38	16	they	they	PRON
fcis-30019	38	17	are	be	AUX
fcis-30019	38	18	sent	send	VERB
fcis-30019	38	19	to	to	ADP
fcis-30019	38	20	the	the	DET
fcis-30019	38	21	space	space	NOUN
fcis-30019	38	22	-	-	PUNCT
fcis-30019	38	23	frequency	frequency	NOUN
fcis-30019	38	24	domain	domain	NOUN
fcis-30019	38	25	for	for	ADP
fcis-30019	38	26	feature	feature	NOUN
fcis-30019	38	27	processing	processing	NOUN
fcis-30019	38	28	.	.	PUNCT
fcis-30019	39	1	the	the	DET
fcis-30019	39	2	spatial	spatial	ADJ
fcis-30019	39	3	feature	feature	NOUN
fcis-30019	39	4	extraction	extraction	NOUN
fcis-30019	39	5	layer	layer	NOUN
fcis-30019	39	6	is	be	AUX
fcis-30019	39	7	mainly	mainly	ADV
fcis-30019	39	8	composed	compose	VERB
fcis-30019	39	9	of	of	ADP
fcis-30019	39	10	the	the	DET
fcis-30019	39	11	backbone	backbone	NOUN
fcis-30019	39	12	network	network	NOUN
fcis-30019	39	13	mvit	mvit	NOUN
fcis-30019	39	14	.	.	PUNCT
fcis-30019	40	1	by	by	ADP
fcis-30019	40	2	embedding	embed	VERB
fcis-30019	40	3	the	the	DET
fcis-30019	40	4	asff	asff	NOUN
fcis-30019	40	5	spatial	spatial	ADJ
fcis-30019	40	6	adaptive	adaptive	ADJ
fcis-30019	40	7	module	module	NOUN
fcis-30019	40	8	in	in	ADP
fcis-30019	40	9	each	each	DET
fcis-30019	40	10	patch	patch	ADJ
fcis-30019	40	11	output	output	NOUN
fcis-30019	40	12	position	position	NOUN
fcis-30019	40	13	in	in	ADP
fcis-30019	40	14	the	the	DET
fcis-30019	40	15	mvit	mvit	PROPN
fcis-30019	40	16	structure	structure	NOUN
fcis-30019	40	17	,	,	PUNCT
fcis-30019	40	18	the	the	DET
fcis-30019	40	19	network	network	NOUN
fcis-30019	40	20	can	can	AUX
fcis-30019	40	21	better	well	ADV
fcis-30019	40	22	capture	capture	VERB
fcis-30019	40	23	the	the	DET
fcis-30019	40	24	multi	multi	ADJ
fcis-30019	40	25	-	-	ADJ
fcis-30019	40	26	scale	scale	ADJ
fcis-30019	40	27	information	information	NOUN
fcis-30019	40	28	of	of	ADP
fcis-30019	40	29	the	the	DET
fcis-30019	40	30	target	target	NOUN
fcis-30019	40	31	by	by	ADP
fcis-30019	40	32	dynamically	dynamically	ADV
fcis-30019	40	33	selecting	select	VERB
fcis-30019	40	34	and	and	CCONJ
fcis-30019	40	35	fusing	fuse	VERB
fcis-30019	40	36	the	the	DET
fcis-30019	40	37	multi	multi	ADJ
fcis-30019	40	38	-	-	ADJ
fcis-30019	40	39	layer	layer	ADJ
fcis-30019	40	40	feature	feature	NOUN
fcis-30019	40	41	maps	map	NOUN
fcis-30019	40	42	.	.	PUNCT
fcis-30019	41	1	in	in	ADP
fcis-30019	41	2	the	the	DET
fcis-30019	41	3	frequency	frequency	NOUN
fcis-30019	41	4	domain	domain	NOUN
fcis-30019	41	5	feature	feature	NOUN
fcis-30019	41	6	extraction	extraction	NOUN
fcis-30019	41	7	layer	layer	NOUN
fcis-30019	41	8	,	,	PUNCT
fcis-30019	41	9	the	the	DET
fcis-30019	41	10	spatial	spatial	ADJ
fcis-30019	41	11	domain	domain	NOUN
fcis-30019	41	12	is	be	AUX
fcis-30019	41	13	transformed	transform	VERB
fcis-30019	41	14	by	by	ADP
fcis-30019	41	15	two	two	NUM
fcis-30019	41	16	-	-	PUNCT
fcis-30019	41	17	dimensional	dimensional	ADJ
fcis-30019	41	18	fft	fft	NOUN
fcis-30019	41	19	,	,	PUNCT
fcis-30019	41	20	which	which	PRON
fcis-30019	41	21	is	be	AUX
fcis-30019	41	22	combined	combine	VERB
fcis-30019	41	23	with	with	ADP
fcis-30019	41	24	a	a	DET
fcis-30019	41	25	filter	filter	NOUN
fcis-30019	41	26	to	to	PART
fcis-30019	41	27	simulate	simulate	VERB
fcis-30019	41	28	the	the	DET
fcis-30019	41	29	correlation	correlation	NOUN
fcis-30019	41	30	of	of	ADP
fcis-30019	41	31	different	different	ADJ
fcis-30019	41	32	frequency	frequency	NOUN
fcis-30019	41	33	band	band	NOUN
fcis-30019	41	34	components	component	NOUN
fcis-30019	41	35	.	.	PUNCT
fcis-30019	42	1	by	by	ADP
fcis-30019	42	2	introducing	introduce	VERB
fcis-30019	42	3	the	the	DET
fcis-30019	42	4	fecam	fecam	NOUN
fcis-30019	42	5	frequency	frequency	NOUN
fcis-30019	42	6	enhancement	enhancement	NOUN
fcis-30019	42	7	module	module	NOUN
fcis-30019	42	8	,	,	PUNCT
fcis-30019	42	9	the	the	DET
fcis-30019	42	10	problem	problem	NOUN
fcis-30019	42	11	of	of	ADP
fcis-30019	42	12	high	high	ADJ
fcis-30019	42	13	frequency	frequency	NOUN
fcis-30019	42	14	noise	noise	NOUN
fcis-30019	42	15	caused	cause	VERB
fcis-30019	42	16	by	by	ADP
fcis-30019	42	17	gibbs	gibbs	PROPN
fcis-30019	42	18	phenomenon	phenomenon	NOUN
fcis-30019	42	19	in	in	ADP
fcis-30019	42	20	fast	fast	ADJ
fcis-30019	42	21	fourier	fourier	NOUN
fcis-30019	42	22	transform	transform	NOUN
fcis-30019	42	23	fft	fft	PROPN
fcis-30019	42	24	was	be	AUX
fcis-30019	42	25	improved	improve	VERB
fcis-30019	42	26	,	,	PUNCT
fcis-30019	42	27	so	so	SCONJ
fcis-30019	42	28	as	as	SCONJ
fcis-30019	42	29	to	to	PART
fcis-30019	42	30	optimize	optimize	VERB
fcis-30019	42	31	the	the	DET
fcis-30019	42	32	feature	feature	NOUN
fcis-30019	42	33	output	output	NOUN
fcis-30019	42	34	.	.	PUNCT
fcis-30019	43	1	finally	finally	ADV
fcis-30019	43	2	,	,	PUNCT
fcis-30019	43	3	the	the	DET
fcis-30019	43	4	cmf	cmf	PROPN
fcis-30019	43	5	feature	feature	NOUN
fcis-30019	43	6	fusion	fusion	NOUN
fcis-30019	43	7	module	module	NOUN
fcis-30019	43	8	was	be	AUX
fcis-30019	43	9	used	use	VERB
fcis-30019	43	10	to	to	PART
fcis-30019	43	11	fuse	fuse	VERB
fcis-30019	43	12	m1	m1	PROPN
fcis-30019	43	13	and	and	CCONJ
fcis-30019	43	14	spatial	spatial	ADJ
fcis-30019	43	15	-	-	PUNCT
fcis-30019	43	16	frequency	frequency	NOUN
fcis-30019	43	17	domain	domain	NOUN
fcis-30019	43	18	features	feature	NOUN
fcis-30019	43	19	for	for	ADP
fcis-30019	43	20	discrimination	discrimination	NOUN
fcis-30019	43	21	.	.	PUNCT
fcis-30019	44	1	fig	fig	NOUN
fcis-30019	44	2	2	2	NUM
fcis-30019	44	3	.	.	PUNCT
fcis-30019	44	4	sfdt	sfdt	PROPN
fcis-30019	44	5	overall	overall	ADJ
fcis-30019	44	6	model	model	NOUN
fcis-30019	44	7	framework	framework	NOUN
fcis-30019	44	8	3.1	3.1	NUM
fcis-30019	44	9	.	.	NOUN
fcis-30019	44	10	mvit_r	mvit_r	NOUN
fcis-30019	44	11	while	while	SCONJ
fcis-30019	44	12	mvit	mvit	PROPN
fcis-30019	44	13	have	have	AUX
fcis-30019	44	14	shown	show	VERB
fcis-30019	44	15	potential	potential	NOUN
fcis-30019	44	16	in	in	ADP
fcis-30019	44	17	modeling	model	VERB
fcis-30019	44	18	interactions	interaction	NOUN
fcis-30019	44	19	between	between	ADP
fcis-30019	44	20	tokens	token	NOUN
fcis-30019	44	21	,	,	PUNCT
fcis-30019	44	22	they	they	PRON
fcis-30019	44	23	focus	focus	VERB
fcis-30019	44	24	on	on	ADP
fcis-30019	44	25	content	content	NOUN
fcis-30019	44	26	rather	rather	ADV
fcis-30019	44	27	than	than	ADP
fcis-30019	44	28	structure	structure	NOUN
fcis-30019	44	29	.	.	PUNCT
fcis-30019	45	1	spatial	spatial	ADJ
fcis-30019	45	2	-	-	PUNCT
fcis-30019	45	3	temporal	temporal	ADJ
fcis-30019	45	4	structure	structure	NOUN
fcis-30019	45	5	modeling	modeling	NOUN
fcis-30019	45	6	relies	rely	VERB
fcis-30019	45	7	only	only	ADV
fcis-30019	45	8	on	on	ADP
fcis-30019	45	9	absolute	absolute	ADJ
fcis-30019	45	10	location	location	NOUN
fcis-30019	45	11	embeddings	embedding	NOUN
fcis-30019	45	12	to	to	PART
fcis-30019	45	13	provide	provide	VERB
fcis-30019	45	14	location	location	NOUN
fcis-30019	45	15	information	information	NOUN
fcis-30019	45	16	.	.	PUNCT
fcis-30019	46	1	this	this	PRON
fcis-30019	46	2	ignores	ignore	VERB
fcis-30019	46	3	the	the	DET
fcis-30019	46	4	fundamental	fundamental	ADJ
fcis-30019	46	5	principle	principle	NOUN
fcis-30019	46	6	of	of	ADP
fcis-30019	46	7	translation	translation	NOUN
fcis-30019	46	8	invariance	invariance	NOUN
fcis-30019	46	9	of	of	ADP
fcis-30019	46	10	vision	vision	NOUN
fcis-30019	46	11	.	.	PUNCT
fcis-30019	47	1	that	that	PRON
fcis-30019	47	2	is	be	AUX
fcis-30019	47	3	,	,	PUNCT
fcis-30019	47	4	the	the	DET
fcis-30019	47	5	way	way	NOUN
fcis-30019	47	6	mvit	mvit	NOUN
fcis-30019	47	7	models	model	VERB
fcis-30019	47	8	the	the	DET
fcis-30019	47	9	interaction	interaction	NOUN
fcis-30019	47	10	between	between	ADP
fcis-30019	47	11	two	two	NUM
fcis-30019	47	12	patches	patch	NOUN
fcis-30019	47	13	changes	change	NOUN
fcis-30019	47	14	with	with	ADP
fcis-30019	47	15	their	their	PRON
fcis-30019	47	16	absolute	absolute	ADJ
fcis-30019	47	17	position	position	NOUN
fcis-30019	47	18	in	in	ADP
fcis-30019	47	19	the	the	DET
fcis-30019	47	20	image	image	NOUN
fcis-30019	47	21	,	,	PUNCT
fcis-30019	47	22	even	even	ADV
fcis-30019	47	23	if	if	SCONJ
fcis-30019	47	24	the	the	DET
fcis-30019	47	25	relative	relative	ADJ
fcis-30019	47	26	position	position	NOUN
fcis-30019	47	27	does	do	AUX
fcis-30019	47	28	not	not	PART
fcis-30019	47	29	change	change	VERB
fcis-30019	47	30	.	.	PUNCT
fcis-30019	48	1	to	to	PART
fcis-30019	48	2	solve	solve	VERB
fcis-30019	48	3	this	this	DET
fcis-30019	48	4	problem	problem	NOUN
fcis-30019	48	5	,	,	PUNCT
fcis-30019	48	6	the	the	DET
fcis-30019	48	7	self	self	NOUN
fcis-30019	48	8	-	-	PUNCT
fcis-30019	48	9	attention	attention	NOUN
fcis-30019	48	10	module	module	NOUN
fcis-30019	48	11	is	be	AUX
fcis-30019	48	12	embedded	embed	VERB
fcis-30019	48	13	by	by	ADP
fcis-30019	48	14	inputing	input	VERB
fcis-30019	48	15	the	the	DET
fcis-30019	48	16	relative	relative	ADJ
fcis-30019	48	17	position	position	NOUN
fcis-30019	48	18	between	between	ADP
fcis-30019	48	19	elements	element	NOUN
fcis-30019	48	20	i	i	PRON
fcis-30019	48	21	and	and	CCONJ
fcis-30019	48	22	j	j	PROPN
fcis-30019	48	23	:	:	PUNCT
fcis-30019	48	24	atten	atten	VERB
fcis-30019	48	25	,	,	PUNCT
fcis-30019	48	26	,	,	PUNCT
fcis-30019	48	27	softmax	softmax	X
fcis-30019	48	28	√	√	ADV
fcis-30019	48	29	1	1	NUM
fcis-30019	48	30	q	q	SYM
fcis-30019	48	31	2	2	NUM
fcis-30019	48	32	where	where	SCONJ
fcis-30019	48	33	p(i	p(i	PROPN
fcis-30019	48	34	)	)	PUNCT
fcis-30019	48	35	and	and	CCONJ
fcis-30019	48	36	p(j	p(j	NOUN
fcis-30019	48	37	)	)	PUNCT
fcis-30019	48	38	represent	represent	VERB
fcis-30019	48	39	the	the	DET
fcis-30019	48	40	spatial	spatial	ADJ
fcis-30019	48	41	-	-	PUNCT
fcis-30019	48	42	temporal	temporal	ADJ
fcis-30019	48	43	positions	position	NOUN
fcis-30019	48	44	of	of	ADP
fcis-30019	48	45	elements	element	NOUN
fcis-30019	48	46	i	i	PRON
fcis-30019	48	47	and	and	CCONJ
fcis-30019	48	48	j.mvit	j.mvit	PROPN
fcis-30019	48	49	has	have	VERB
fcis-30019	48	50	a	a	DET
fcis-30019	48	51	larger	large	ADJ
fcis-30019	48	52	step	step	NOUN
fcis-30019	48	53	size	size	NOUN
fcis-30019	48	54	on	on	ADP
fcis-30019	48	55	the	the	DET
fcis-30019	48	56	k	k	PROPN
fcis-30019	48	57	and	and	CCONJ
fcis-30019	48	58	v	v	ADJ
fcis-30019	48	59	tensors	tensor	NOUN
fcis-30019	48	60	than	than	ADP
fcis-30019	48	61	on	on	ADP
fcis-30019	48	62	the	the	DET
fcis-30019	48	63	q	q	NOUN
fcis-30019	48	64	tensor	tensor	NOUN
fcis-30019	48	65	,	,	PUNCT
fcis-30019	48	66	which	which	PRON
fcis-30019	48	67	is	be	AUX
fcis-30019	48	68	downsampled	downsample	VERB
fcis-30019	48	69	only	only	ADV
fcis-30019	48	70	if	if	SCONJ
fcis-30019	48	71	the	the	DET
fcis-30019	48	72	resolution	resolution	NOUN
fcis-30019	48	73	of	of	ADP
fcis-30019	48	74	the	the	DET
fcis-30019	48	75	output	output	NOUN
fcis-30019	48	76	sequence	sequence	NOUN
fcis-30019	48	77	changes	change	VERB
fcis-30019	48	78	across	across	ADP
fcis-30019	48	79	stages	stage	NOUN
fcis-30019	48	80	.	.	PUNCT
fcis-30019	49	1	this	this	PRON
fcis-30019	49	2	motivates	motivate	VERB
fcis-30019	49	3	the	the	DET
fcis-30019	49	4	addition	addition	NOUN
fcis-30019	49	5	of	of	ADP
fcis-30019	49	6	residual	residual	ADJ
fcis-30019	49	7	pooling	pooling	NOUN
fcis-30019	49	8	connections	connection	NOUN
fcis-30019	49	9	to	to	ADP
fcis-30019	49	10	the	the	DET
fcis-30019	49	11	q	q	NOUN
fcis-30019	49	12	tensor	tensor	NOUN
fcis-30019	49	13	to	to	PART
fcis-30019	49	14	increase	increase	VERB
fcis-30019	49	15	the	the	DET
fcis-30019	49	16	information	information	NOUN
fcis-30019	49	17	flow	flow	NOUN
fcis-30019	49	18	,	,	PUNCT
fcis-30019	49	19	driving	drive	VERB
fcis-30019	49	20	the	the	DET
fcis-30019	49	21	training	training	NOUN
fcis-30019	49	22	and	and	CCONJ
fcis-30019	49	23	convergence	convergence	NOUN
fcis-30019	49	24	of	of	ADP
fcis-30019	49	25	attention	attention	NOUN
fcis-30019	49	26	.	.	PUNCT
fcis-30019	50	1	3.2	3.2	NUM
fcis-30019	50	2	.	.	PUNCT
fcis-30019	51	1	asff	asff	PROPN
fcis-30019	51	2	asff	asff	PROPN
fcis-30019	51	3	is	be	AUX
fcis-30019	51	4	an	an	DET
fcis-30019	51	5	improvement	improvement	NOUN
fcis-30019	51	6	based	base	VERB
fcis-30019	51	7	on	on	ADP
fcis-30019	51	8	the	the	DET
fcis-30019	51	9	faster	fast	ADJ
fcis-30019	51	10	r	r	NOUN
fcis-30019	51	11	-	-	PUNCT
fcis-30019	51	12	cnn	cnn	PROPN
fcis-30019	51	13	framework	framework	NOUN
fcis-30019	51	14	,	,	PUNCT
fcis-30019	51	15	which	which	PRON
fcis-30019	51	16	aims	aim	VERB
fcis-30019	51	17	to	to	PART
fcis-30019	51	18	improve	improve	VERB
fcis-30019	51	19	the	the	DET
fcis-30019	51	20	effect	effect	NOUN
fcis-30019	51	21	of	of	ADP
fcis-30019	51	22	feature	feature	NOUN
fcis-30019	51	23	fusion	fusion	NOUN
fcis-30019	51	24	,	,	PUNCT
fcis-30019	51	25	thereby	thereby	ADV
fcis-30019	51	26	improving	improve	VERB
fcis-30019	51	27	the	the	DET
fcis-30019	51	28	accuracy	accuracy	NOUN
fcis-30019	51	29	and	and	CCONJ
fcis-30019	51	30	performance	performance	NOUN
fcis-30019	51	31	of	of	ADP
fcis-30019	51	32	object	object	NOUN
fcis-30019	51	33	detection[9	detection[9	PROPN
fcis-30019	51	34	]	]	X
fcis-30019	51	35	.	.	PUNCT
fcis-30019	52	1	the	the	DET
fcis-30019	52	2	asff	asff	PROPN
fcis-30019	52	3	network	network	NOUN
fcis-30019	52	4	introduces	introduce	VERB
fcis-30019	52	5	an	an	DET
fcis-30019	52	6	adaptive	adaptive	ADJ
fcis-30019	52	7	structural	structural	ADJ
fcis-30019	52	8	feature	feature	NOUN
fcis-30019	52	9	fusion	fusion	NOUN
fcis-30019	52	10	module	module	NOUN
fcis-30019	52	11	to	to	PART
fcis-30019	52	12	fuse	fuse	VERB
fcis-30019	52	13	multiple	multiple	ADJ
fcis-30019	52	14	feature	feature	NOUN
fcis-30019	52	15	maps	map	NOUN
fcis-30019	52	16	at	at	ADP
fcis-30019	52	17	different	different	ADJ
fcis-30019	52	18	levels.traditional	levels.traditional	ADJ
fcis-30019	52	19	object	object	NOUN
fcis-30019	52	20	detection	detection	NOUN
fcis-30019	52	21	networks	network	NOUN
fcis-30019	52	22	usually	usually	ADV
fcis-30019	52	23	use	use	VERB
fcis-30019	52	24	only	only	ADV
fcis-30019	52	25	a	a	DET
fcis-30019	52	26	single	single	ADJ
fcis-30019	52	27	level	level	NOUN
fcis-30019	52	28	of	of	ADP
fcis-30019	52	29	feature	feature	NOUN
fcis-30019	52	30	maps	map	NOUN
fcis-30019	52	31	,	,	PUNCT
fcis-30019	52	32	while	while	SCONJ
fcis-30019	52	33	asff	asff	PROPN
fcis-30019	52	34	network	network	NOUN
fcis-30019	52	35	dynamically	dynamically	ADV
fcis-30019	52	36	selects	select	NOUN
fcis-30019	52	37	and	and	CCONJ
fcis-30019	52	38	fuses	fuse	VERB
fcis-30019	52	39	multi	multi	ADJ
fcis-30019	52	40	-	-	ADJ
fcis-30019	52	41	level	level	ADJ
fcis-30019	52	42	feature	feature	NOUN
fcis-30019	52	43	maps	map	NOUN
fcis-30019	52	44	,	,	PUNCT
fcis-30019	52	45	so	so	SCONJ
fcis-30019	52	46	that	that	SCONJ
fcis-30019	52	47	the	the	DET
fcis-30019	52	48	network	network	NOUN
fcis-30019	52	49	can	can	AUX
fcis-30019	52	50	better	well	ADV
fcis-30019	52	51	capture	capture	VERB
fcis-30019	52	52	multi	multi	ADJ
fcis-30019	52	53	-	-	ADJ
fcis-30019	52	54	scale	scale	ADJ
fcis-30019	52	55	information	information	NOUN
fcis-30019	52	56	of	of	ADP
fcis-30019	52	57	the	the	DET
fcis-30019	52	58	target	target	NOUN
fcis-30019	52	59	.	.	PUNCT
fcis-30019	53	1	the	the	DET
fcis-30019	53	2	asff	asff	PROPN
fcis-30019	53	3	network	network	NOUN
fcis-30019	53	4	first	first	ADJ
fcis-30019	53	5	extracts	extract	NOUN
fcis-30019	53	6	feature	feature	NOUN
fcis-30019	53	7	maps	map	NOUN
fcis-30019	53	8	at	at	ADP
fcis-30019	53	9	different	different	ADJ
fcis-30019	53	10	levels	level	NOUN
fcis-30019	53	11	through	through	ADP
fcis-30019	53	12	the	the	DET
fcis-30019	53	13	backbone	backbone	NOUN
fcis-30019	53	14	network	network	NOUN
fcis-30019	53	15	.	.	PUNCT
fcis-30019	54	1	then	then	ADV
fcis-30019	54	2	,	,	PUNCT
fcis-30019	54	3	it	it	PRON
fcis-30019	54	4	uses	use	VERB
fcis-30019	54	5	an	an	DET
fcis-30019	54	6	adaptive	adaptive	ADJ
fcis-30019	54	7	attention	attention	NOUN
fcis-30019	54	8	mechanism	mechanism	NOUN
fcis-30019	54	9	to	to	PART
fcis-30019	54	10	perform	perform	VERB
fcis-30019	54	11	weighted	weighted	ADJ
fcis-30019	54	12	fusion	fusion	NOUN
fcis-30019	54	13	of	of	ADP
fcis-30019	54	14	these	these	DET
fcis-30019	54	15	feature	feature	NOUN
fcis-30019	54	16	maps	map	NOUN
fcis-30019	54	17	,	,	PUNCT
fcis-30019	54	18	so	so	SCONJ
fcis-30019	54	19	as	as	SCONJ
fcis-30019	54	20	to	to	PART
fcis-30019	54	21	better	well	ADV
fcis-30019	54	22	retain	retain	VERB
fcis-30019	54	23	important	important	ADJ
fcis-30019	54	24	feature	feature	NOUN
fcis-30019	54	25	information	information	NOUN
fcis-30019	54	26	.	.	PUNCT
fcis-30019	55	1	finally	finally	ADV
fcis-30019	55	2	,	,	PUNCT
fcis-30019	55	3	the	the	DET
fcis-30019	55	4	fused	fuse	VERB
fcis-30019	55	5	features	feature	NOUN
fcis-30019	55	6	are	be	AUX
fcis-30019	55	7	passed	pass	VERB
fcis-30019	55	8	to	to	ADP
fcis-30019	55	9	the	the	DET
fcis-30019	55	10	object	object	NOUN
fcis-30019	55	11	detection	detection	NOUN
fcis-30019	55	12	head	head	NOUN
fcis-30019	55	13	for	for	ADP
fcis-30019	55	14	object	object	NOUN
fcis-30019	55	15	classification	classification	NOUN
fcis-30019	55	16	and	and	CCONJ
fcis-30019	55	17	localization	localization	NOUN
fcis-30019	55	18	.	.	PUNCT
fcis-30019	56	1	the	the	DET
fcis-30019	56	2	key	key	ADJ
fcis-30019	56	3	idea	idea	NOUN
fcis-30019	56	4	is	be	AUX
fcis-30019	56	5	to	to	PART
fcis-30019	56	6	adaptively	adaptively	ADV
fcis-30019	56	7	learn	learn	VERB
fcis-30019	56	8	the	the	DET
fcis-30019	56	9	fusion	fusion	NOUN
fcis-30019	56	10	spatial	spatial	ADJ
fcis-30019	56	11	weight	weight	NOUN
fcis-30019	56	12	of	of	ADP
fcis-30019	56	13	each	each	DET
fcis-30019	56	14	scale	scale	NOUN
fcis-30019	56	15	feature	feature	NOUN
fcis-30019	56	16	map	map	NOUN
fcis-30019	56	17	in	in	ADP
fcis-30019	56	18	two	two	NUM
fcis-30019	56	19	steps	step	NOUN
fcis-30019	56	20	:	:	PUNCT
fcis-30019	56	21	identity	identity	NOUN
fcis-30019	56	22	scaling	scaling	NOUN
fcis-30019	56	23	and	and	CCONJ
fcis-30019	56	24	adaptive	adaptive	ADJ
fcis-30019	56	25	fusion	fusion	NOUN
fcis-30019	56	26	.	.	PUNCT
fcis-30019	57	1	56	56	NUM
fcis-30019	57	2	fig	fig	NOUN
fcis-30019	57	3	3	3	NUM
fcis-30019	57	4	.	.	PUNCT
fcis-30019	57	5	overall	overall	ADJ
fcis-30019	57	6	structure	structure	NOUN
fcis-30019	57	7	of	of	ADP
fcis-30019	57	8	asff	asff	NOUN
fcis-30019	57	9	taking	take	VERB
fcis-30019	57	10	asff-3	asff-3	PROPN
fcis-30019	57	11	as	as	ADP
fcis-30019	57	12	an	an	DET
fcis-30019	57	13	example	example	NOUN
fcis-30019	57	14	,	,	PUNCT
fcis-30019	57	15	the	the	DET
fcis-30019	57	16	green	green	ADJ
fcis-30019	57	17	part	part	NOUN
fcis-30019	57	18	of	of	ADP
fcis-30019	57	19	the	the	DET
fcis-30019	57	20	figure	figure	NOUN
fcis-30019	57	21	describes	describe	VERB
fcis-30019	57	22	the	the	DET
fcis-30019	57	23	process	process	NOUN
fcis-30019	57	24	of	of	ADP
fcis-30019	57	25	feature	feature	NOUN
fcis-30019	57	26	fusion	fusion	NOUN
fcis-30019	57	27	as	as	ADP
fcis-30019	57	28	:	:	PUNCT
fcis-30019	57	29	→	→	SYM
fcis-30019	57	30	→	→	SYM
fcis-30019	57	31	→	→	SYM
fcis-30019	57	32	3	3	NUM
fcis-30019	57	33	where	where	SCONJ
fcis-30019	57	34	i	i	PRON
fcis-30019	57	35	and	and	CCONJ
fcis-30019	57	36	j	j	PROPN
fcis-30019	57	37	represent	represent	VERB
fcis-30019	57	38	the	the	DET
fcis-30019	57	39	spatial	spatial	ADJ
fcis-30019	57	40	position	position	NOUN
fcis-30019	57	41	,	,	PUNCT
fcis-30019	57	42	x	x	X
fcis-30019	57	43	is	be	AUX
fcis-30019	57	44	the	the	DET
fcis-30019	57	45	feature	feature	NOUN
fcis-30019	57	46	from	from	ADP
fcis-30019	57	47	different	different	ADJ
fcis-30019	57	48	levels	level	NOUN
fcis-30019	57	49	,	,	PUNCT
fcis-30019	57	50	and	and	CCONJ
fcis-30019	57	51	is	be	AUX
fcis-30019	57	52	the	the	DET
fcis-30019	57	53	feature	feature	NOUN
fcis-30019	57	54	from	from	ADP
fcis-30019	57	55	different	different	ADJ
fcis-30019	57	56	levels	level	NOUN
fcis-30019	57	57	multiplied	multiply	VERB
fcis-30019	57	58	by	by	ADP
fcis-30019	57	59	the	the	DET
fcis-30019	57	60	weight	weight	NOUN
fcis-30019	57	61	parameter	parameter	NOUN
fcis-30019	57	62	and	and	CCONJ
fcis-30019	57	63	added	add	VERB
fcis-30019	57	64	,	,	PUNCT
fcis-30019	57	65	the	the	DET
fcis-30019	57	66	new	new	ADJ
fcis-30019	57	67	fusion	fusion	NOUN
fcis-30019	57	68	feature	feature	NOUN
fcis-30019	57	69	asff-3	asff-3	PROPN
fcis-30019	57	70	can	can	AUX
fcis-30019	57	71	be	be	AUX
fcis-30019	57	72	obtained	obtain	VERB
fcis-30019	57	73	.	.	PUNCT
fcis-30019	58	1	the	the	DET
fcis-30019	58	2	output	output	NOUN
fcis-30019	58	3	adaptive	adaptive	ADJ
fcis-30019	58	4	spatial	spatial	ADJ
fcis-30019	58	5	feature	feature	NOUN
fcis-30019	58	6	is	be	AUX
fcis-30019	58	7	defined	define	VERB
fcis-30019	58	8	as	as	ADP
fcis-30019	58	9	f_s	f_s	X
fcis-30019	58	10	.	.	PROPN
fcis-30019	58	11	3.3	3.3	NUM
fcis-30019	58	12	.	.	PUNCT
fcis-30019	59	1	dynamic	dynamic	ADJ
fcis-30019	59	2	filter	filter	NOUN
fcis-30019	59	3	to	to	PART
fcis-30019	59	4	leverage	leverage	NOUN
fcis-30019	59	5	frequency	frequency	NOUN
fcis-30019	59	6	information	information	NOUN
fcis-30019	59	7	,	,	PUNCT
fcis-30019	59	8	auto	auto	NOUN
fcis-30019	59	9	-	-	PUNCT
fcis-30019	59	10	former	former	ADJ
fcis-30019	59	11	[	[	X
fcis-30019	59	12	10	10	NUM
fcis-30019	59	13	]	]	PUNCT
fcis-30019	59	14	uses	use	VERB
fcis-30019	59	15	fft	fft	PROPN
fcis-30019	59	16	to	to	PART
fcis-30019	59	17	efficiently	efficiently	ADV
fcis-30019	59	18	compute	compute	VERB
fcis-30019	59	19	autocorrelation	autocorrelation	NOUN
fcis-30019	59	20	functions	function	NOUN
fcis-30019	59	21	,	,	PUNCT
fcis-30019	59	22	fno	fno	PROPN
fcis-30019	60	1	[	[	X
fcis-30019	60	2	11	11	NUM
fcis-30019	60	3	]	]	PUNCT
fcis-30019	60	4	is	be	AUX
fcis-30019	60	5	used	use	VERB
fcis-30019	60	6	as	as	ADP
fcis-30019	60	7	an	an	DET
fcis-30019	60	8	internal	internal	ADJ
fcis-30019	60	9	block	block	NOUN
fcis-30019	60	10	of	of	ADP
fcis-30019	60	11	the	the	DET
fcis-30019	60	12	network	network	NOUN
fcis-30019	60	13	to	to	PART
fcis-30019	60	14	perform	perform	VERB
fcis-30019	60	15	representation	representation	NOUN
fcis-30019	60	16	learning	learning	NOUN
fcis-30019	60	17	in	in	ADP
fcis-30019	60	18	the	the	DET
fcis-30019	60	19	low	low	ADJ
fcis-30019	60	20	frequency	frequency	NOUN
fcis-30019	60	21	domain	domain	NOUN
fcis-30019	60	22	,	,	PUNCT
fcis-30019	60	23	and	and	CCONJ
fcis-30019	60	24	dctnet	dctnet	NOUN
fcis-30019	61	1	[	[	X
fcis-30019	61	2	12	12	NUM
fcis-30019	61	3	]	]	PUNCT
fcis-30019	61	4	uses	use	VERB
fcis-30019	61	5	discrete	discrete	ADJ
fcis-30019	61	6	cosine	cosine	NOUN
fcis-30019	61	7	transform	transform	NOUN
fcis-30019	61	8	to	to	PART
fcis-30019	61	9	compress	compress	VERB
fcis-30019	61	10	information	information	NOUN
fcis-30019	61	11	to	to	PART
fcis-30019	61	12	retain	retain	VERB
fcis-30019	61	13	more	more	ADJ
fcis-30019	61	14	original	original	ADJ
fcis-30019	61	15	image	image	NOUN
fcis-30019	61	16	information	information	NOUN
fcis-30019	61	17	in	in	ADP
fcis-30019	61	18	cv	cv	PROPN
fcis-30019	61	19	tasks	task	NOUN
fcis-30019	61	20	.	.	PUNCT
fcis-30019	62	1	most	most	ADJ
fcis-30019	62	2	of	of	ADP
fcis-30019	62	3	these	these	DET
fcis-30019	62	4	works	work	NOUN
fcis-30019	62	5	are	be	AUX
fcis-30019	62	6	based	base	VERB
fcis-30019	62	7	on	on	ADP
fcis-30019	62	8	fourier	fourier	NOUN
fcis-30019	62	9	transform	transform	NOUN
fcis-30019	62	10	,	,	PUNCT
fcis-30019	62	11	which	which	PRON
fcis-30019	62	12	is	be	AUX
fcis-30019	62	13	helpful	helpful	ADJ
fcis-30019	62	14	to	to	PART
fcis-30019	62	15	extract	extract	VERB
fcis-30019	62	16	frequency	frequency	NOUN
fcis-30019	62	17	features	feature	NOUN
fcis-30019	62	18	.	.	PUNCT
fcis-30019	63	1	however	however	ADV
fcis-30019	63	2	,	,	PUNCT
fcis-30019	63	3	most	most	ADJ
fcis-30019	63	4	fourier	fourier	NOUN
fcis-30019	63	5	transform	transform	NOUN
fcis-30019	63	6	-	-	PUNCT
fcis-30019	63	7	based	base	VERB
fcis-30019	63	8	methods	method	NOUN
fcis-30019	63	9	use	use	VERB
fcis-30019	63	10	the	the	DET
fcis-30019	63	11	fourier	fourier	NOUN
fcis-30019	63	12	transform	transform	NOUN
fcis-30019	63	13	to	to	PART
fcis-30019	63	14	obtain	obtain	VERB
fcis-30019	63	15	the	the	DET
fcis-30019	63	16	frequency	frequency	NOUN
fcis-30019	63	17	information	information	NOUN
fcis-30019	63	18	and	and	CCONJ
fcis-30019	63	19	the	the	DET
fcis-30019	63	20	inverse	inverse	ADJ
fcis-30019	63	21	fourier	fourier	NOUN
fcis-30019	63	22	transform	transform	NOUN
fcis-30019	63	23	to	to	PART
fcis-30019	63	24	reconstruct	reconstruct	VERB
fcis-30019	63	25	the	the	DET
fcis-30019	63	26	time	time	NOUN
fcis-30019	63	27	information	information	NOUN
fcis-30019	63	28	to	to	PART
fcis-30019	63	29	avoid	avoid	VERB
fcis-30019	63	30	complex	complex	ADJ
fcis-30019	63	31	training	training	NOUN
fcis-30019	63	32	,	,	PUNCT
fcis-30019	63	33	which	which	PRON
fcis-30019	63	34	introduces	introduce	VERB
fcis-30019	63	35	a	a	DET
fcis-30019	63	36	new	new	ADJ
fcis-30019	63	37	amount	amount	NOUN
fcis-30019	63	38	of	of	ADP
fcis-30019	63	39	computation	computation	NOUN
fcis-30019	63	40	that	that	PRON
fcis-30019	63	41	can	can	AUX
fcis-30019	63	42	be	be	AUX
fcis-30019	63	43	avoided	avoid	VERB
fcis-30019	63	44	if	if	SCONJ
fcis-30019	63	45	the	the	DET
fcis-30019	63	46	time	time	NOUN
fcis-30019	63	47	-	-	PUNCT
fcis-30019	63	48	frequency	frequency	NOUN
fcis-30019	63	49	transform	transform	NOUN
fcis-30019	63	50	is	be	AUX
fcis-30019	63	51	performed	perform	VERB
fcis-30019	63	52	using	use	VERB
fcis-30019	63	53	dct	dct	PROPN
fcis-30019	63	54	.	.	PUNCT
fcis-30019	64	1	more	more	ADV
fcis-30019	64	2	importantly	importantly	ADV
fcis-30019	64	3	,	,	PUNCT
fcis-30019	64	4	the	the	DET
fcis-30019	64	5	implicit	implicit	ADJ
fcis-30019	64	6	periodicity	periodicity	NOUN
fcis-30019	64	7	of	of	ADP
fcis-30019	64	8	the	the	DET
fcis-30019	64	9	dft	dft	PROPN
fcis-30019	64	10	causes	cause	VERB
fcis-30019	64	11	boundary	boundary	ADJ
fcis-30019	64	12	discontinuities	discontinuity	NOUN
fcis-30019	64	13	.	.	PUNCT
fcis-30019	65	1	the	the	DET
fcis-30019	65	2	high	high	ADJ
fcis-30019	65	3	-	-	PUNCT
fcis-30019	65	4	frequency	frequency	NOUN
fcis-30019	65	5	component	component	NOUN
fcis-30019	65	6	that	that	PRON
fcis-30019	65	7	leads	lead	VERB
fcis-30019	65	8	to	to	ADP
fcis-30019	65	9	significant	significant	ADJ
fcis-30019	65	10	is	be	AUX
fcis-30019	65	11	known	know	VERB
fcis-30019	65	12	as	as	ADP
fcis-30019	65	13	the	the	DET
fcis-30019	65	14	gibbson	gibbson	NOUN
fcis-30019	65	15	phenomenon	phenomenon	NOUN
fcis-30019	65	16	.	.	PUNCT
fcis-30019	66	1	after	after	ADP
fcis-30019	66	2	quantization	quantization	NOUN
fcis-30019	66	3	,	,	PUNCT
fcis-30019	66	4	gibbs	gibbs	PROPN
fcis-30019	66	5	phenomenon	phenomenon	NOUN
fcis-30019	66	6	causes	cause	VERB
fcis-30019	66	7	boundary	boundary	ADJ
fcis-30019	66	8	points	point	NOUN
fcis-30019	66	9	to	to	PART
fcis-30019	66	10	take	take	VERB
fcis-30019	66	11	wrong	wrong	ADJ
fcis-30019	66	12	values	value	NOUN
fcis-30019	66	13	.	.	PUNCT
fcis-30019	67	1	image	image	NOUN
fcis-30019	67	2	classification	classification	NOUN
fcis-30019	67	3	and	and	CCONJ
fcis-30019	67	4	downstream	downstream	ADJ
fcis-30019	67	5	tasks	task	NOUN
fcis-30019	67	6	dynamic	dynamic	ADJ
fcis-30019	67	7	filters	filter	NOUN
fcis-30019	67	8	are	be	AUX
fcis-30019	67	9	generated	generate	VERB
fcis-30019	67	10	based	base	VERB
fcis-30019	67	11	on	on	ADP
fcis-30019	67	12	global	global	ADJ
fcis-30019	67	13	filters	filter	NOUN
fcis-30019	67	14	.	.	PUNCT
fcis-30019	68	1	for	for	ADP
fcis-30019	68	2	the	the	DET
fcis-30019	68	3	2d	2d	NUM
fcis-30019	68	4	discrete	discrete	ADJ
fcis-30019	68	5	fourier	fourier	NOUN
fcis-30019	68	6	transform	transform	NOUN
fcis-30019	68	7	,	,	PUNCT
fcis-30019	68	8	given	give	VERB
fcis-30019	68	9	a	a	DET
fcis-30019	68	10	2d	2d	NUM
fcis-30019	68	11	signal	signal	NOUN
fcis-30019	68	12	x(h	x(h	PROPN
fcis-30019	68	13	,	,	PUNCT
fcis-30019	68	14	w	w	NOUN
fcis-30019	68	15	)	)	PUNCT
fcis-30019	68	16	,	,	PUNCT
fcis-30019	68	17	the	the	DET
fcis-30019	68	18	2d	2d	NOUN
fcis-30019	68	19	-	-	PUNCT
fcis-30019	68	20	dft	dft	NOUN
fcis-30019	68	21	is	be	AUX
fcis-30019	68	22	defined	define	VERB
fcis-30019	68	23	as	as	ADP
fcis-30019	68	24	follows	follow	VERB
fcis-30019	68	25	.	.	PUNCT
fcis-30019	69	1	,	,	PUNCT
fcis-30019	69	2	    	    	SPACE
fcis-30019	69	3	,	,	PUNCT
fcis-30019	69	4	√	√	PROPN
fcis-30019	69	5	4	4	NUM
fcis-30019	69	6	where	where	SCONJ
fcis-30019	69	7	,	,	PUNCT
fcis-30019	69	8	is	be	AUX
fcis-30019	69	9	the	the	DET
fcis-30019	69	10	result	result	NOUN
fcis-30019	69	11	of	of	ADP
fcis-30019	69	12	the	the	DET
fcis-30019	69	13	transformation	transformation	NOUN
fcis-30019	69	14	and	and	CCONJ
fcis-30019	69	15	is	be	AUX
fcis-30019	69	16	usually	usually	ADV
fcis-30019	69	17	expressed	express	VERB
fcis-30019	69	18	as	as	ADP
fcis-30019	69	19	a	a	DET
fcis-30019	69	20	value	value	NOUN
fcis-30019	69	21	in	in	ADP
fcis-30019	69	22	the	the	DET
fcis-30019	69	23	frequency	frequency	NOUN
fcis-30019	69	24	domain	domain	NOUN
fcis-30019	69	25	of	of	ADP
fcis-30019	69	26	the	the	DET
fcis-30019	69	27	complex	complex	ADJ
fcis-30019	69	28	number	number	NOUN
fcis-30019	69	29	,	,	PUNCT
fcis-30019	69	30	h	h	NOUN
fcis-30019	69	31	and	and	CCONJ
fcis-30019	69	32	w	w	PROPN
fcis-30019	69	33	are	be	AUX
fcis-30019	69	34	the	the	DET
fcis-30019	69	35	row	row	NOUN
fcis-30019	69	36	and	and	CCONJ
fcis-30019	69	37	column	column	NOUN
fcis-30019	69	38	indices	indice	VERB
fcis-30019	69	39	in	in	ADP
fcis-30019	69	40	the	the	DET
fcis-30019	69	41	spatial	spatial	ADJ
fcis-30019	69	42	domain	domain	NOUN
fcis-30019	69	43	,	,	PUNCT
fcis-30019	69	44	h	h	NOUN
fcis-30019	69	45	and	and	CCONJ
fcis-30019	69	46	w	w	AUX
fcis-30019	69	47	represent	represent	VERB
fcis-30019	69	48	the	the	DET
fcis-30019	69	49	number	number	NOUN
fcis-30019	69	50	of	of	ADP
fcis-30019	69	51	rows	row	NOUN
fcis-30019	69	52	and	and	CCONJ
fcis-30019	69	53	columns	column	NOUN
fcis-30019	69	54	of	of	ADP
fcis-30019	69	55	the	the	DET
fcis-30019	69	56	image	image	NOUN
fcis-30019	69	57	in	in	ADP
fcis-30019	69	58	the	the	DET
fcis-30019	69	59	spatial	spatial	ADJ
fcis-30019	69	60	domain	domain	NOUN
fcis-30019	69	61	,	,	PUNCT
fcis-30019	69	62	j	j	PROPN
fcis-30019	69	63	is	be	AUX
fcis-30019	69	64	the	the	DET
fcis-30019	69	65	imaginary	imaginary	ADJ
fcis-30019	69	66	unit	unit	NOUN
fcis-30019	69	67	.	.	PUNCT
fcis-30019	70	1	this	this	DET
fcis-30019	70	2	exponential	exponential	ADJ
fcis-30019	70	3	function	function	NOUN
fcis-30019	70	4	is	be	AUX
fcis-30019	70	5	used	use	VERB
fcis-30019	70	6	to	to	PART
fcis-30019	70	7	associate	associate	VERB
fcis-30019	70	8	points	point	NOUN
fcis-30019	70	9	in	in	ADP
fcis-30019	70	10	the	the	DET
fcis-30019	70	11	spatial	spatial	ADJ
fcis-30019	70	12	domain	domain	NOUN
fcis-30019	70	13	with	with	ADP
fcis-30019	70	14	points	point	NOUN
fcis-30019	70	15	in	in	ADP
fcis-30019	70	16	the	the	DET
fcis-30019	70	17	frequency	frequency	NOUN
fcis-30019	70	18	domain	domain	NOUN
fcis-30019	70	19	.	.	PUNCT
fcis-30019	71	1	√	√	PROPN
fcis-30019	71	2	is	be	AUX
fcis-30019	71	3	the	the	DET
fcis-30019	71	4	normalization	normalization	NOUN
fcis-30019	71	5	factor	factor	NOUN
fcis-30019	71	6	used	use	VERB
fcis-30019	71	7	to	to	PART
fcis-30019	71	8	ensure	ensure	VERB
fcis-30019	71	9	that	that	SCONJ
fcis-30019	71	10	the	the	DET
fcis-30019	71	11	energy	energy	NOUN
fcis-30019	71	12	remains	remain	VERB
fcis-30019	71	13	the	the	DET
fcis-30019	71	14	same	same	ADJ
fcis-30019	71	15	before	before	ADV
fcis-30019	71	16	and	and	CCONJ
fcis-30019	71	17	after	after	ADP
fcis-30019	71	18	the	the	DET
fcis-30019	71	19	transformation	transformation	NOUN
fcis-30019	71	20	.	.	PUNCT
fcis-30019	72	1	the	the	DET
fcis-30019	72	2	inverse	inverse	NOUN
fcis-30019	72	3	2d	2d	NOUN
fcis-30019	72	4	-	-	PUNCT
fcis-30019	72	5	dft	dft	NOUN
fcis-30019	72	6	exists	exist	NOUN
fcis-30019	72	7	and	and	CCONJ
fcis-30019	72	8	is	be	AUX
fcis-30019	72	9	known	know	VERB
fcis-30019	72	10	as	as	ADP
fcis-30019	72	11	the	the	DET
fcis-30019	72	12	twodimensional	twodimensional	ADJ
fcis-30019	72	13	inverse	inverse	NOUN
fcis-30019	72	14	discrete	discrete	NOUN
fcis-30019	72	15	fourier	fourier	NOUN
fcis-30019	72	16	transform	transform	NOUN
fcis-30019	72	17	(	(	PUNCT
fcis-30019	72	18	2didft	2didft	NUM
fcis-30019	72	19	)	)	PUNCT
fcis-30019	72	20	.	.	PUNCT
fcis-30019	73	1	moreover	moreover	ADV
fcis-30019	73	2	,	,	PUNCT
fcis-30019	73	3	the	the	DET
fcis-30019	73	4	frequency	frequency	NOUN
fcis-30019	73	5	domain	domain	NOUN
fcis-30019	73	6	has	have	AUX
fcis-30019	73	7	an	an	DET
fcis-30019	73	8	important	important	ADJ
fcis-30019	73	9	property	property	NOUN
fcis-30019	73	10	that	that	DET
fcis-30019	73	11	multiplication	multiplication	NOUN
fcis-30019	73	12	in	in	ADP
fcis-30019	73	13	the	the	DET
fcis-30019	73	14	frequency	frequency	NOUN
fcis-30019	73	15	domain	domain	NOUN
fcis-30019	73	16	is	be	AUX
fcis-30019	73	17	equivalent	equivalent	ADJ
fcis-30019	73	18	to	to	ADP
fcis-30019	73	19	cyclic	cyclic	ADJ
fcis-30019	73	20	convolution	convolution	NOUN
fcis-30019	73	21	in	in	ADP
fcis-30019	73	22	the	the	DET
fcis-30019	73	23	original	original	ADJ
fcis-30019	73	24	domain	domain	NOUN
fcis-30019	73	25	,	,	PUNCT
fcis-30019	73	26	known	know	VERB
fcis-30019	73	27	as	as	SCONJ
fcis-30019	73	28	the	the	DET
fcis-30019	73	29	convolution	convolution	NOUN
fcis-30019	73	30	theorem	theorem	VERB
fcis-30019	73	31	.	.	PUNCT
fcis-30019	73	32	define	define	VERB
fcis-30019	73	33	a	a	DET
fcis-30019	73	34	global	global	ADJ
fcis-30019	73	35	filter	filter	NOUN
fcis-30019	73	36	g	g	NOUN
fcis-30019	73	37	for	for	ADP
fcis-30019	73	38	feature	feature	NOUN
fcis-30019	73	39	x	x	NOUN
fcis-30019	73	40	,	,	PUNCT
fcis-30019	73	41	and	and	CCONJ
fcis-30019	73	42	the	the	DET
fcis-30019	73	43	global	global	ADJ
fcis-30019	73	44	filter	filter	NOUN
fcis-30019	73	45	is	be	AUX
fcis-30019	73	46	defined	define	VERB
fcis-30019	73	47	as	as	ADP
fcis-30019	73	48	:	:	PUNCT
fcis-30019	73	49	where	where	SCONJ
fcis-30019	73	50	g(x	g(x	NOUN
fcis-30019	73	51	)	)	PUNCT
fcis-30019	73	52	denotes	denote	VERB
fcis-30019	73	53	the	the	DET
fcis-30019	73	54	output	output	NOUN
fcis-30019	73	55	filter	filter	NOUN
fcis-30019	73	56	,	,	PUNCT
fcis-30019	73	57	⊙	⊙	PROPN
fcis-30019	73	58	signifies	signify	VERB
fcis-30019	73	59	matrix	matrix	NOUN
fcis-30019	73	60	point	point	NOUN
fcis-30019	73	61	-	-	PUNCT
fcis-30019	73	62	wise	wise	ADJ
fcis-30019	73	63	multiplication	multiplication	NOUN
fcis-30019	73	64	,	,	PUNCT
fcis-30019	73	65	is	be	AUX
fcis-30019	73	66	a	a	DET
fcis-30019	73	67	learnable	learnable	ADJ
fcis-30019	73	68	filter	filter	NOUN
fcis-30019	73	69	,	,	PUNCT
fcis-30019	73	70	and	and	CCONJ
fcis-30019	73	71	represents	represent	VERB
fcis-30019	73	72	the	the	DET
fcis-30019	73	73	inverse	inverse	ADJ
fcis-30019	73	74	transformation	transformation	NOUN
fcis-30019	73	75	of	of	ADP
fcis-30019	73	76	2d	2d	NOUN
fcis-30019	73	77	-	-	PUNCT
fcis-30019	73	78	fft	fft	NOUN
fcis-30019	73	79	.	.	PUNCT
fcis-30019	74	1	the	the	DET
fcis-30019	74	2	neural	neural	ADJ
fcis-30019	74	3	network	network	NOUN
fcis-30019	74	4	within	within	ADP
fcis-30019	74	5	the	the	DET
fcis-30019	74	6	dynamic	dynamic	ADJ
fcis-30019	74	7	filter	filter	NOUN
fcis-30019	74	8	dynamically	dynamically	ADV
fcis-30019	74	9	determines	determine	VERB
fcis-30019	74	10	a	a	DET
fcis-30019	74	11	sufficient	sufficient	ADJ
fcis-30019	74	12	number	number	NOUN
fcis-30019	74	13	of	of	ADP
fcis-30019	74	14	global	global	ADJ
fcis-30019	74	15	filters	filter	NOUN
fcis-30019	74	16	to	to	PART
fcis-30019	74	17	be	be	AUX
fcis-30019	74	18	included	include	VERB
fcis-30019	74	19	,	,	PUNCT
fcis-30019	74	20	and	and	CCONJ
fcis-30019	74	21	the	the	DET
fcis-30019	74	22	overall	overall	ADJ
fcis-30019	74	23	structure	structure	NOUN
fcis-30019	74	24	can	can	AUX
fcis-30019	74	25	be	be	AUX
fcis-30019	74	26	seen	see	VERB
fcis-30019	74	27	as	as	ADP
fcis-30019	74	28	a	a	DET
fcis-30019	74	29	dual	dual	ADJ
fcis-30019	74	30	-	-	PUNCT
fcis-30019	74	31	stream	stream	NOUN
fcis-30019	74	32	architecture	architecture	NOUN
fcis-30019	74	33	.	.	PUNCT
fcis-30019	75	1	first	first	ADV
fcis-30019	75	2	,	,	PUNCT
fcis-30019	75	3	the	the	DET
fcis-30019	75	4	input	input	NOUN
fcis-30019	75	5	is	be	AUX
fcis-30019	75	6	processed	process	VERB
fcis-30019	75	7	through	through	ADP
fcis-30019	75	8	multiple	multiple	ADJ
fcis-30019	75	9	layers	layer	NOUN
fcis-30019	75	10	of	of	ADP
fcis-30019	75	11	perception	perception	NOUN
fcis-30019	75	12	and	and	CCONJ
fcis-30019	75	13	a	a	DET
fcis-30019	75	14	softmax	softmax	NOUN
fcis-30019	75	15	function	function	NOUN
fcis-30019	75	16	to	to	PART
fcis-30019	75	17	extract	extract	VERB
fcis-30019	75	18	features	feature	NOUN
fcis-30019	75	19	.	.	PUNCT
fcis-30019	76	1	then	then	ADV
fcis-30019	76	2	,	,	PUNCT
fcis-30019	76	3	a	a	DET
fcis-30019	76	4	filter	filter	NOUN
fcis-30019	76	5	of	of	ADP
fcis-30019	76	6	size	size	NOUN
fcis-30019	76	7	n	n	NOUN
fcis-30019	76	8	is	be	AUX
fcis-30019	76	9	used	use	VERB
fcis-30019	76	10	to	to	PART
fcis-30019	76	11	dynamically	dynamically	ADV
fcis-30019	76	12	adjust	adjust	VERB
fcis-30019	76	13	the	the	DET
fcis-30019	76	14	weights	weight	NOUN
fcis-30019	76	15	,	,	PUNCT
fcis-30019	76	16	which	which	PRON
fcis-30019	76	17	is	be	AUX
fcis-30019	76	18	set	set	VERB
fcis-30019	76	19	to	to	ADP
fcis-30019	76	20	4	4	NUM
fcis-30019	76	21	in	in	ADP
fcis-30019	76	22	the	the	DET
fcis-30019	76	23	experiments	experiment	NOUN
fcis-30019	76	24	of	of	ADP
fcis-30019	76	25	this	this	DET
fcis-30019	76	26	chapter	chapter	NOUN
fcis-30019	76	27	.	.	PUNCT
fcis-30019	77	1	next	next	ADV
fcis-30019	77	2	,	,	PUNCT
fcis-30019	77	3	highfrequency	highfrequency	NOUN
fcis-30019	77	4	features	feature	NOUN
fcis-30019	77	5	are	be	AUX
fcis-30019	77	6	extracted	extract	VERB
fcis-30019	77	7	through	through	ADP
fcis-30019	77	8	2d	2d	NUM
fcis-30019	77	9	inverse	inverse	NOUN
fcis-30019	77	10	discrete	discrete	ADJ
fcis-30019	77	11	fourier	fourier	NOUN
fcis-30019	77	12	transform	transform	NOUN
fcis-30019	77	13	.	.	PUNCT
fcis-30019	78	1	finally	finally	ADV
fcis-30019	78	2	,	,	PUNCT
fcis-30019	78	3	the	the	DET
fcis-30019	78	4	output	output	NOUN
fcis-30019	78	5	is	be	AUX
fcis-30019	78	6	fused	fuse	VERB
fcis-30019	78	7	by	by	ADP
fcis-30019	78	8	weighted	weight	VERB
fcis-30019	78	9	summation	summation	NOUN
fcis-30019	78	10	and	and	CCONJ
fcis-30019	78	11	2d	2d	NOUN
fcis-30019	78	12	inverse	inverse	NOUN
fcis-30019	78	13	discrete	discrete	ADJ
fcis-30019	78	14	fourier	fourier	NOUN
fcis-30019	78	15	transform	transform	NOUN
fcis-30019	78	16	.	.	PUNCT
fcis-30019	79	1	the	the	DET
fcis-30019	79	2	definition	definition	NOUN
fcis-30019	79	3	of	of	ADP
fcis-30019	79	4	the	the	DET
fcis-30019	79	5	dynamic	dynamic	ADJ
fcis-30019	79	6	filter	filter	NOUN
fcis-30019	79	7	is	be	AUX
fcis-30019	79	8	as	as	SCONJ
fcis-30019	79	9	follows	follow	VERB
fcis-30019	79	10	:	:	PUNCT
fcis-30019	79	11	⊙	⊙	PROPN
fcis-30019	79	12	∘	∘	PROPN
fcis-30019	79	13	5	5	NUM
fcis-30019	79	14	where	where	SCONJ
fcis-30019	79	15	denotes	denote	VERB
fcis-30019	79	16	the	the	DET
fcis-30019	79	17	function	function	NOUN
fcis-30019	79	18	that	that	PRON
fcis-30019	79	19	determines	determine	VERB
fcis-30019	79	20	the	the	DET
fcis-30019	79	21	dynamic	dynamic	ADJ
fcis-30019	79	22	filter	filter	NOUN
fcis-30019	79	23	.	.	PUNCT
fcis-30019	80	1	a	a	PRON
fcis-30019	80	2	is	be	AUX
fcis-30019	80	3	the	the	DET
fcis-30019	80	4	continuous	continuous	ADJ
fcis-30019	80	5	real	real	ADJ
fcis-30019	80	6	map	map	NOUN
fcis-30019	80	7	,	,	PUNCT
fcis-30019	80	8	including	include	VERB
fcis-30019	80	9	point	point	NOUN
fcis-30019	80	10	convolution	convolution	NOUN
fcis-30019	80	11	and	and	CCONJ
fcis-30019	80	12	identity	identity	NOUN
fcis-30019	80	13	map	map	NOUN
fcis-30019	80	14	,	,	PUNCT
fcis-30019	80	15	⊙	⊙	PROPN
fcis-30019	80	16	denotes	denote	NOUN
fcis-30019	80	17	matrix	matrix	NOUN
fcis-30019	80	18	dot	dot	NOUN
fcis-30019	80	19	multiplication	multiplication	NOUN
fcis-30019	80	20	,	,	PUNCT
fcis-30019	80	21	∘	∘	PROPN
fcis-30019	80	22	denotes	denote	VERB
fcis-30019	80	23	composite	composite	ADJ
fcis-30019	80	24	map	map	NOUN
fcis-30019	80	25	.	.	PUNCT
fcis-30019	81	1	3.4	3.4	NUM
fcis-30019	81	2	.	.	PUNCT
fcis-30019	81	3	fecam	fecam	NOUN
fcis-30019	81	4	in	in	ADP
fcis-30019	81	5	the	the	DET
fcis-30019	81	6	process	process	NOUN
fcis-30019	81	7	of	of	ADP
fcis-30019	81	8	frequency	frequency	NOUN
fcis-30019	81	9	domain	domain	NOUN
fcis-30019	81	10	feature	feature	NOUN
fcis-30019	81	11	extraction	extraction	NOUN
fcis-30019	81	12	,	,	PUNCT
fcis-30019	81	13	the	the	DET
fcis-30019	81	14	loss	loss	NOUN
fcis-30019	81	15	of	of	ADP
fcis-30019	81	16	key	key	ADJ
fcis-30019	81	17	information	information	NOUN
fcis-30019	81	18	is	be	AUX
fcis-30019	81	19	inevitable	inevitable	ADJ
fcis-30019	81	20	with	with	ADP
fcis-30019	81	21	the	the	DET
fcis-30019	81	22	increase	increase	NOUN
fcis-30019	81	23	of	of	ADP
fcis-30019	81	24	calculation	calculation	NOUN
fcis-30019	81	25	.	.	PUNCT
fcis-30019	82	1	this	this	PRON
fcis-30019	82	2	is	be	AUX
fcis-30019	82	3	the	the	DET
fcis-30019	82	4	problem	problem	NOUN
fcis-30019	82	5	with	with	ADP
fcis-30019	82	6	the	the	DET
fcis-30019	82	7	model	model	NOUN
fcis-30019	82	8	when	when	SCONJ
fcis-30019	82	9	it	it	PRON
fcis-30019	82	10	comes	come	VERB
fcis-30019	82	11	to	to	ADP
fcis-30019	82	12	capturing	capture	VERB
fcis-30019	82	13	frequency	frequency	NOUN
fcis-30019	82	14	information	information	NOUN
fcis-30019	82	15	.	.	PUNCT
fcis-30019	83	1	at	at	ADP
fcis-30019	83	2	present	present	ADJ
fcis-30019	83	3	,	,	PUNCT
fcis-30019	83	4	the	the	DET
fcis-30019	83	5	mainstream	mainstream	NOUN
fcis-30019	83	6	frequency	frequency	NOUN
fcis-30019	83	7	information	information	NOUN
fcis-30019	83	8	extraction	extraction	NOUN
fcis-30019	83	9	methods	method	NOUN
fcis-30019	83	10	are	be	AUX
fcis-30019	83	11	based	base	VERB
fcis-30019	83	12	on	on	ADP
fcis-30019	83	13	fourier	fourier	NOUN
fcis-30019	83	14	transform	transform	NOUN
fcis-30019	83	15	.	.	PUNCT
fcis-30019	84	1	however	however	ADV
fcis-30019	84	2	,	,	PUNCT
fcis-30019	84	3	the	the	DET
fcis-30019	84	4	use	use	NOUN
fcis-30019	84	5	of	of	ADP
fcis-30019	84	6	the	the	DET
fcis-30019	84	7	fourier	fourier	NOUN
fcis-30019	84	8	transform	transform	NOUN
fcis-30019	84	9	is	be	AUX
fcis-30019	84	10	problematic	problematic	ADJ
fcis-30019	84	11	due	due	ADP
fcis-30019	84	12	to	to	ADP
fcis-30019	84	13	the	the	DET
fcis-30019	84	14	gibbs	gibbs	PROPN
fcis-30019	84	15	phenomenon	phenomenon	NOUN
fcis-30019	84	16	,	,	PUNCT
fcis-30019	84	17	and	and	CCONJ
fcis-30019	84	18	the	the	DET
fcis-30019	84	19	discrete	discrete	ADJ
fcis-30019	84	20	fourier	fourier	NOUN
fcis-30019	84	21	transform	transform	NOUN
fcis-30019	84	22	presented	present	VERB
fcis-30019	84	23	in	in	ADP
fcis-30019	84	24	the	the	DET
fcis-30019	84	25	previous	previous	ADJ
fcis-30019	84	26	section	section	NOUN
fcis-30019	84	27	creates	create	VERB
fcis-30019	84	28	the	the	DET
fcis-30019	84	29	same	same	ADJ
fcis-30019	84	30	problem	problem	NOUN
fcis-30019	84	31	.	.	PUNCT
fcis-30019	85	1	if	if	SCONJ
fcis-30019	85	2	the	the	DET
fcis-30019	85	3	values	value	NOUN
fcis-30019	85	4	on	on	ADP
fcis-30019	85	5	the	the	DET
fcis-30019	85	6	two	two	NUM
fcis-30019	85	7	sides	side	NOUN
fcis-30019	85	8	of	of	ADP
fcis-30019	85	9	the	the	DET
fcis-30019	85	10	sequence	sequence	NOUN
fcis-30019	85	11	are	be	AUX
fcis-30019	85	12	very	very	ADV
fcis-30019	85	13	different	different	ADJ
fcis-30019	85	14	,	,	PUNCT
fcis-30019	85	15	an	an	DET
fcis-30019	85	16	oscillatory	oscillatory	ADJ
fcis-30019	85	17	approximation	approximation	NOUN
fcis-30019	85	18	is	be	AUX
fcis-30019	85	19	observed	observe	VERB
fcis-30019	85	20	around	around	ADP
fcis-30019	85	21	the	the	DET
fcis-30019	85	22	sides	side	NOUN
fcis-30019	85	23	and	and	CCONJ
fcis-30019	85	24	highfrequency	highfrequency	NOUN
fcis-30019	85	25	noise	noise	NOUN
fcis-30019	85	26	is	be	AUX
fcis-30019	85	27	introduced	introduce	VERB
fcis-30019	85	28	.	.	PUNCT
fcis-30019	86	1	therefore	therefore	ADV
fcis-30019	86	2	,	,	PUNCT
fcis-30019	86	3	this	this	DET
fcis-30019	86	4	chapter	chapter	NOUN
fcis-30019	86	5	uses	use	VERB
fcis-30019	86	6	the	the	DET
fcis-30019	86	7	fecam	fecam	PROPN
fcis-30019	86	8	frequency	frequency	NOUN
fcis-30019	86	9	enhancement	enhancement	NOUN
fcis-30019	86	10	module	module	NOUN
fcis-30019	86	11	to	to	PART
fcis-30019	86	12	ameliorate	ameliorate	VERB
fcis-30019	86	13	this	this	DET
fcis-30019	86	14	problem	problem	NOUN
fcis-30019	86	15	by	by	ADP
fcis-30019	86	16	adaptively	adaptively	ADV
fcis-30019	86	17	modeling	model	VERB
fcis-30019	86	18	the	the	DET
fcis-30019	86	19	interdependence	interdependence	NOUN
fcis-30019	86	20	between	between	ADP
fcis-30019	86	21	frequencies	frequency	NOUN
fcis-30019	86	22	based	base	VERB
fcis-30019	86	23	on	on	ADP
fcis-30019	86	24	the	the	DET
fcis-30019	86	25	discrete	discrete	ADJ
fcis-30019	86	26	cosine	cosine	NOUN
fcis-30019	86	27	transform	transform	NOUN
fcis-30019	86	28	,	,	PUNCT
fcis-30019	86	29	which	which	PRON
fcis-30019	86	30	will	will	AUX
fcis-30019	86	31	essentially	essentially	ADV
fcis-30019	86	32	avoid	avoid	VERB
fcis-30019	86	33	the	the	DET
fcis-30019	86	34	problems	problem	NOUN
fcis-30019	86	35	caused	cause	VERB
fcis-30019	86	36	by	by	ADP
fcis-30019	86	37	highfrequency	highfrequency	NOUN
fcis-30019	86	38	noise	noise	NOUN
fcis-30019	86	39	during	during	ADP
fcis-30019	86	40	the	the	DET
fcis-30019	86	41	fourier	fourier	NOUN
fcis-30019	86	42	transform	transform	NOUN
fcis-30019	86	43	.	.	PUNCT
fcis-30019	86	44	fecam	fecam	PROPN
fcis-30019	87	1	[	[	X
fcis-30019	87	2	13]works	13]works	NUM
fcis-30019	87	3	by	by	ADP
fcis-30019	87	4	first	first	ADV
fcis-30019	87	5	applying	apply	VERB
fcis-30019	87	6	dct	dct	PROPN
fcis-30019	87	7	to	to	ADP
fcis-30019	87	8	the	the	DET
fcis-30019	87	9	time	time	NOUN
fcis-30019	87	10	series	series	PROPN
fcis-30019	87	11	data	datum	NOUN
fcis-30019	87	12	of	of	ADP
fcis-30019	87	13	each	each	DET
fcis-30019	87	14	channel	channel	NOUN
fcis-30019	87	15	to	to	PART
fcis-30019	87	16	convert	convert	VERB
fcis-30019	87	17	it	it	PRON
fcis-30019	87	18	into	into	ADP
fcis-30019	87	19	a	a	DET
fcis-30019	87	20	frequency	frequency	NOUN
fcis-30019	87	21	domain	domain	NOUN
fcis-30019	87	22	representation	representation	NOUN
fcis-30019	87	23	.	.	PUNCT
fcis-30019	88	1	then	then	ADV
fcis-30019	88	2	,	,	PUNCT
fcis-30019	88	3	through	through	ADP
fcis-30019	88	4	the	the	DET
fcis-30019	88	5	channel	channel	NOUN
fcis-30019	88	6	attention	attention	NOUN
fcis-30019	88	7	mechanism	mechanism	NOUN
fcis-30019	88	8	,	,	PUNCT
fcis-30019	88	9	the	the	DET
fcis-30019	88	10	frequency	frequency	NOUN
fcis-30019	88	11	information	information	NOUN
fcis-30019	88	12	is	be	AUX
fcis-30019	88	13	weighted	weight	VERB
fcis-30019	88	14	,	,	PUNCT
fcis-30019	88	15	focusing	focus	VERB
fcis-30019	88	16	57	57	NUM
fcis-30019	88	17	on	on	ADP
fcis-30019	88	18	those	those	DET
fcis-30019	88	19	frequency	frequency	NOUN
fcis-30019	88	20	components	component	NOUN
fcis-30019	88	21	that	that	PRON
fcis-30019	88	22	contribute	contribute	VERB
fcis-30019	88	23	most	most	ADJ
fcis-30019	88	24	to	to	ADP
fcis-30019	88	25	the	the	DET
fcis-30019	88	26	prediction	prediction	NOUN
fcis-30019	88	27	task	task	NOUN
fcis-30019	88	28	.	.	PUNCT
fcis-30019	89	1	this	this	DET
fcis-30019	89	2	enhanced	enhance	VERB
fcis-30019	89	3	frequency	frequency	NOUN
fcis-30019	89	4	information	information	NOUN
fcis-30019	89	5	is	be	AUX
fcis-30019	89	6	then	then	ADV
fcis-30019	89	7	incorporated	incorporate	VERB
fcis-30019	89	8	into	into	ADP
fcis-30019	89	9	the	the	DET
fcis-30019	89	10	model	model	NOUN
fcis-30019	89	11	to	to	PART
fcis-30019	89	12	generate	generate	VERB
fcis-30019	89	13	more	more	ADV
fcis-30019	89	14	accurate	accurate	ADJ
fcis-30019	89	15	predictions	prediction	NOUN
fcis-30019	89	16	.	.	PUNCT
fcis-30019	90	1	as	as	SCONJ
fcis-30019	90	2	shown	show	VERB
fcis-30019	90	3	in	in	ADP
fcis-30019	90	4	figure	figure	NOUN
fcis-30019	90	5	4	4	NUM
fcis-30019	90	6	,	,	PUNCT
fcis-30019	90	7	fecam	fecam	PROPN
fcis-30019	90	8	divides	divide	VERB
fcis-30019	90	9	the	the	DET
fcis-30019	90	10	input	input	NOUN
fcis-30019	90	11	feature	feature	NOUN
fcis-30019	90	12	map	map	NOUN
fcis-30019	90	13	into	into	ADP
fcis-30019	90	14	n	n	DET
fcis-30019	90	15	groups	group	NOUN
fcis-30019	90	16	along	along	ADP
fcis-30019	90	17	the	the	DET
fcis-30019	90	18	channel	channel	NOUN
fcis-30019	90	19	dimension	dimension	NOUN
fcis-30019	90	20	.	.	PUNCT
fcis-30019	91	1	subsequently	subsequently	ADV
fcis-30019	91	2	,	,	PUNCT
fcis-30019	91	3	the	the	DET
fcis-30019	91	4	groups	group	NOUN
fcis-30019	91	5	are	be	AUX
fcis-30019	91	6	processed	process	VERB
fcis-30019	91	7	by	by	ADP
fcis-30019	91	8	the	the	DET
fcis-30019	91	9	corresponding	corresponding	ADJ
fcis-30019	91	10	dct	dct	PROPN
fcis-30019	91	11	frequency	frequency	NOUN
fcis-30019	91	12	components	component	NOUN
fcis-30019	91	13	from	from	ADP
fcis-30019	91	14	low	low	ADJ
fcis-30019	91	15	frequency	frequency	NOUN
fcis-30019	91	16	to	to	ADP
fcis-30019	91	17	high	high	ADJ
fcis-30019	91	18	frequency	frequency	NOUN
fcis-30019	91	19	,	,	PUNCT
fcis-30019	91	20	and	and	CCONJ
fcis-30019	91	21	each	each	DET
fcis-30019	91	22	individual	individual	ADJ
fcis-30019	91	23	channel	channel	NOUN
fcis-30019	91	24	is	be	AUX
fcis-30019	91	25	processed	process	VERB
fcis-30019	91	26	by	by	ADP
fcis-30019	91	27	the	the	DET
fcis-30019	91	28	same	same	ADJ
fcis-30019	91	29	frequency	frequency	NOUN
fcis-30019	91	30	components	component	NOUN
fcis-30019	91	31	to	to	PART
fcis-30019	91	32	obtain	obtain	VERB
fcis-30019	91	33	the	the	DET
fcis-30019	91	34	freq	freq	ADJ
fcis-30019	91	35	sequence	sequence	NOUN
fcis-30019	91	36	,	,	PUNCT
fcis-30019	91	37	which	which	PRON
fcis-30019	91	38	is	be	AUX
fcis-30019	91	39	defined	define	VERB
fcis-30019	91	40	as	as	ADP
fcis-30019	91	41	:	:	PUNCT
fcis-30019	91	42	fig	fig	NOUN
fcis-30019	91	43	4	4	NUM
fcis-30019	91	44	.	.	PUNCT
fcis-30019	91	45	overall	overall	ADJ
fcis-30019	91	46	structure	structure	NOUN
fcis-30019	91	47	of	of	ADP
fcis-30019	91	48	asff[13	asff[13	PROPN
fcis-30019	91	49	]	]	X
fcis-30019	91	50	⊙	⊙	PROPN
fcis-30019	91	51	6	6	NUM
fcis-30019	91	52	,	,	PUNCT
fcis-30019	91	53	,	,	PUNCT
fcis-30019	91	54	⋯	⋯	PROPN
fcis-30019	91	55	,	,	PUNCT
fcis-30019	91	56	7	7	NUM
fcis-30019	91	57	where	where	SCONJ
fcis-30019	91	58	is	be	AUX
fcis-30019	91	59	the	the	DET
fcis-30019	91	60	group	group	NOUN
fcis-30019	91	61	number	number	NOUN
fcis-30019	91	62	from	from	ADP
fcis-30019	91	63	0	0	NUM
fcis-30019	91	64	to	to	ADP
fcis-30019	91	65	n-1	n-1	PROPN
fcis-30019	91	66	,	,	PUNCT
fcis-30019	91	67	is	be	AUX
fcis-30019	91	68	the	the	DET
fcis-30019	91	69	transform	transform	NOUN
fcis-30019	91	70	applied	apply	VERB
fcis-30019	91	71	,	,	PUNCT
fcis-30019	91	72	is	be	AUX
fcis-30019	91	73	the	the	DET
fcis-30019	91	74	feature	feature	NOUN
fcis-30019	91	75	map	map	NOUN
fcis-30019	91	76	group	group	NOUN
fcis-30019	91	77	,	,	PUNCT
fcis-30019	91	78	and	and	CCONJ
fcis-30019	91	79	then	then	ADV
fcis-30019	91	80	can	can	AUX
fcis-30019	91	81	be	be	AUX
fcis-30019	91	82	fused	fuse	VERB
fcis-30019	91	83	by	by	ADP
fcis-30019	91	84	stack	stack	NOUN
fcis-30019	91	85	stacking.in	stacking.in	NUM
fcis-30019	91	86	this	this	DET
fcis-30019	91	87	way	way	NOUN
fcis-30019	91	88	,	,	PUNCT
fcis-30019	91	89	each	each	DET
fcis-30019	91	90	channel	channel	NOUN
fcis-30019	91	91	feature	feature	NOUN
fcis-30019	91	92	interacts	interact	VERB
fcis-30019	91	93	with	with	ADP
fcis-30019	91	94	each	each	DET
fcis-30019	91	95	frequency	frequency	NOUN
fcis-30019	91	96	component	component	NOUN
fcis-30019	91	97	to	to	PART
fcis-30019	91	98	comprehensively	comprehensively	ADV
fcis-30019	91	99	obtain	obtain	VERB
fcis-30019	91	100	important	important	ADJ
fcis-30019	91	101	time	time	NOUN
fcis-30019	91	102	domain	domain	NOUN
fcis-30019	91	103	information	information	NOUN
fcis-30019	91	104	from	from	ADP
fcis-30019	91	105	the	the	DET
fcis-30019	91	106	frequency	frequency	NOUN
fcis-30019	91	107	domain	domain	NOUN
fcis-30019	91	108	,	,	PUNCT
fcis-30019	91	109	which	which	PRON
fcis-30019	91	110	will	will	AUX
fcis-30019	91	111	prompt	prompt	VERB
fcis-30019	91	112	the	the	DET
fcis-30019	91	113	network	network	NOUN
fcis-30019	91	114	to	to	PART
fcis-30019	91	115	enhance	enhance	VERB
fcis-30019	91	116	the	the	DET
fcis-30019	91	117	diversity	diversity	NOUN
fcis-30019	91	118	of	of	ADP
fcis-30019	91	119	extracted	extract	VERB
fcis-30019	91	120	features	feature	NOUN
fcis-30019	91	121	and	and	CCONJ
fcis-30019	91	122	define	define	VERB
fcis-30019	91	123	the	the	DET
fcis-30019	91	124	output	output	NOUN
fcis-30019	91	125	frequency	frequency	NOUN
fcis-30019	91	126	domain	domain	NOUN
fcis-30019	91	127	feature	feature	NOUN
fcis-30019	91	128	as	as	ADP
fcis-30019	91	129	.	.	PUNCT
fcis-30019	91	130	4	4	X
fcis-30019	91	131	.	.	NOUN
fcis-30019	91	132	result	result	NOUN
fcis-30019	91	133	and	and	CCONJ
fcis-30019	91	134	conclusion	conclusion	NOUN
fcis-30019	91	135	4.1	4.1	NUM
fcis-30019	91	136	.	.	PUNCT
fcis-30019	92	1	dataset	dataset	VERB
fcis-30019	92	2	in	in	ADP
fcis-30019	92	3	order	order	NOUN
fcis-30019	92	4	to	to	PART
fcis-30019	92	5	effectively	effectively	ADV
fcis-30019	92	6	prove	prove	VERB
fcis-30019	92	7	the	the	DET
fcis-30019	92	8	robustness	robustness	NOUN
fcis-30019	92	9	of	of	ADP
fcis-30019	92	10	the	the	DET
fcis-30019	92	11	model	model	NOUN
fcis-30019	92	12	proposed	propose	VERB
fcis-30019	92	13	in	in	ADP
fcis-30019	92	14	this	this	DET
fcis-30019	92	15	chapter	chapter	NOUN
fcis-30019	92	16	,	,	PUNCT
fcis-30019	92	17	multiple	multiple	ADJ
fcis-30019	92	18	data	datum	NOUN
fcis-30019	92	19	sets	set	NOUN
fcis-30019	92	20	are	be	AUX
fcis-30019	92	21	selected	select	VERB
fcis-30019	92	22	for	for	ADP
fcis-30019	92	23	testing	testing	NOUN
fcis-30019	92	24	.	.	PUNCT
fcis-30019	93	1	the	the	DET
fcis-30019	93	2	datasets	dataset	NOUN
fcis-30019	93	3	selected	select	VERB
fcis-30019	93	4	for	for	ADP
fcis-30019	93	5	this	this	DET
fcis-30019	93	6	chapter	chapter	NOUN
fcis-30019	93	7	are	be	AUX
fcis-30019	93	8	ff++	ff++	NOUN
fcis-30019	93	9	,	,	PUNCT
fcis-30019	93	10	celeb	celeb	NOUN
fcis-30019	93	11	-	-	PUNCT
fcis-30019	93	12	dfv2	dfv2	PROPN
fcis-30019	93	13	,	,	PUNCT
fcis-30019	93	14	biodeepav	biodeepav	NOUN
fcis-30019	93	15	,	,	PUNCT
fcis-30019	93	16	deepforensic	deepforensic	ADJ
fcis-30019	93	17	,	,	PUNCT
fcis-30019	93	18	and	and	CCONJ
fcis-30019	93	19	faceshifter	faceshifter	NOUN
fcis-30019	93	20	datasets[14	datasets[14	PROPN
fcis-30019	93	21	]	]	PUNCT
fcis-30019	93	22	.	.	PUNCT
fcis-30019	93	23	ff++	ff++	PROPN
fcis-30019	94	1	[	[	X
fcis-30019	94	2	15	15	NUM
fcis-30019	94	3	]	]	PUNCT
fcis-30019	94	4	is	be	AUX
fcis-30019	94	5	a	a	DET
fcis-30019	94	6	large	large	ADJ
fcis-30019	94	7	-	-	PUNCT
fcis-30019	94	8	scale	scale	NOUN
fcis-30019	94	9	and	and	CCONJ
fcis-30019	94	10	diverse	diverse	ADJ
fcis-30019	94	11	fake	fake	ADJ
fcis-30019	94	12	video	video	NOUN
fcis-30019	94	13	dataset	dataset	NOUN
fcis-30019	94	14	,	,	PUNCT
fcis-30019	94	15	which	which	PRON
fcis-30019	94	16	contains	contain	VERB
fcis-30019	94	17	a	a	DET
fcis-30019	94	18	variety	variety	NOUN
fcis-30019	94	19	of	of	ADP
fcis-30019	94	20	attack	attack	NOUN
fcis-30019	94	21	methods	method	NOUN
fcis-30019	94	22	,	,	PUNCT
fcis-30019	94	23	such	such	ADJ
fcis-30019	94	24	as	as	ADP
fcis-30019	94	25	deepfake	deepfake	NOUN
fcis-30019	94	26	,	,	PUNCT
fcis-30019	94	27	facial	facial	ADJ
fcis-30019	94	28	replacement	replacement	NOUN
fcis-30019	94	29	and	and	CCONJ
fcis-30019	94	30	expression	expression	NOUN
fcis-30019	94	31	imitation	imitation	NOUN
fcis-30019	94	32	.	.	PUNCT
fcis-30019	95	1	the	the	DET
fcis-30019	95	2	dataset	dataset	NOUN
fcis-30019	95	3	covers	cover	VERB
fcis-30019	95	4	real	real	ADJ
fcis-30019	95	5	videos	video	NOUN
fcis-30019	95	6	of	of	ADP
fcis-30019	95	7	different	different	ADJ
fcis-30019	95	8	races	race	NOUN
fcis-30019	95	9	,	,	PUNCT
fcis-30019	95	10	genders	gender	NOUN
fcis-30019	95	11	and	and	CCONJ
fcis-30019	95	12	ages	age	NOUN
fcis-30019	95	13	,	,	PUNCT
fcis-30019	95	14	as	as	ADV
fcis-30019	95	15	well	well	ADV
fcis-30019	95	16	as	as	ADP
fcis-30019	95	17	synthetic	synthetic	ADJ
fcis-30019	95	18	fake	fake	ADJ
fcis-30019	95	19	videos	video	NOUN
fcis-30019	95	20	.	.	PUNCT
fcis-30019	96	1	in	in	ADP
fcis-30019	96	2	addition	addition	NOUN
fcis-30019	96	3	,	,	PUNCT
fcis-30019	96	4	the	the	DET
fcis-30019	96	5	ff++	ff++	NOUN
fcis-30019	96	6	dataset	dataset	NOUN
fcis-30019	96	7	provides	provide	VERB
fcis-30019	96	8	a	a	DET
fcis-30019	96	9	variety	variety	NOUN
fcis-30019	96	10	of	of	ADP
fcis-30019	96	11	scene	scene	NOUN
fcis-30019	96	12	,	,	PUNCT
fcis-30019	96	13	lighting	lighting	NOUN
fcis-30019	96	14	,	,	PUNCT
fcis-30019	96	15	and	and	CCONJ
fcis-30019	96	16	camera	camera	NOUN
fcis-30019	96	17	conditions	condition	NOUN
fcis-30019	96	18	to	to	PART
fcis-30019	96	19	enhance	enhance	VERB
fcis-30019	96	20	its	its	PRON
fcis-30019	96	21	generalization	generalization	NOUN
fcis-30019	96	22	ability	ability	NOUN
fcis-30019	96	23	in	in	ADP
fcis-30019	96	24	practical	practical	ADJ
fcis-30019	96	25	applications	application	NOUN
fcis-30019	96	26	.	.	PUNCT
fcis-30019	97	1	celeb	celeb	NOUN
fcis-30019	97	2	-	-	PUNCT
fcis-30019	97	3	dfv2	dfv2	PROPN
fcis-30019	98	1	[	[	X
fcis-30019	98	2	16	16	NUM
fcis-30019	98	3	]	]	PUNCT
fcis-30019	98	4	is	be	AUX
fcis-30019	98	5	a	a	DET
fcis-30019	98	6	high	high	ADJ
fcis-30019	98	7	-	-	PUNCT
fcis-30019	98	8	quality	quality	NOUN
fcis-30019	98	9	dataset	dataset	NOUN
fcis-30019	98	10	specifically	specifically	ADV
fcis-30019	98	11	for	for	ADP
fcis-30019	98	12	deepfake	deepfake	NOUN
fcis-30019	98	13	detection	detection	NOUN
fcis-30019	98	14	,	,	PUNCT
fcis-30019	98	15	containing	contain	VERB
fcis-30019	98	16	590	590	NUM
fcis-30019	98	17	real	real	ADJ
fcis-30019	98	18	videos	video	NOUN
fcis-30019	98	19	and	and	CCONJ
fcis-30019	98	20	5639	5639	NUM
fcis-30019	98	21	fake	fake	ADJ
fcis-30019	98	22	videos	video	NOUN
fcis-30019	98	23	.	.	PUNCT
fcis-30019	99	1	these	these	DET
fcis-30019	99	2	videos	video	NOUN
fcis-30019	99	3	are	be	AUX
fcis-30019	99	4	produced	produce	VERB
fcis-30019	99	5	using	use	VERB
fcis-30019	99	6	advanced	advanced	ADJ
fcis-30019	99	7	generative	generative	ADJ
fcis-30019	99	8	adversarial	adversarial	ADJ
fcis-30019	99	9	network	network	NOUN
fcis-30019	99	10	(	(	PUNCT
fcis-30019	99	11	gan	gan	NOUN
fcis-30019	99	12	)	)	PUNCT
fcis-30019	99	13	technology	technology	NOUN
fcis-30019	99	14	and	and	CCONJ
fcis-30019	99	15	have	have	VERB
fcis-30019	99	16	high	high	ADJ
fcis-30019	99	17	visual	visual	ADJ
fcis-30019	99	18	quality	quality	NOUN
fcis-30019	99	19	and	and	CCONJ
fcis-30019	99	20	deception	deception	NOUN
fcis-30019	99	21	.	.	PUNCT
fcis-30019	100	1	the	the	DET
fcis-30019	100	2	celeb	celeb	PROPN
fcis-30019	100	3	-	-	PUNCT
fcis-30019	100	4	dfv2	dfv2	PROPN
fcis-30019	100	5	dataset	dataset	NOUN
fcis-30019	100	6	covers	cover	VERB
fcis-30019	100	7	a	a	DET
fcis-30019	100	8	variety	variety	NOUN
fcis-30019	100	9	of	of	ADP
fcis-30019	100	10	expressions	expression	NOUN
fcis-30019	100	11	,	,	PUNCT
fcis-30019	100	12	poses	pose	NOUN
fcis-30019	100	13	and	and	CCONJ
fcis-30019	100	14	lighting	lighting	NOUN
fcis-30019	100	15	conditions	condition	NOUN
fcis-30019	100	16	,	,	PUNCT
fcis-30019	100	17	which	which	PRON
fcis-30019	100	18	provides	provide	VERB
fcis-30019	100	19	rich	rich	ADJ
fcis-30019	100	20	training	training	NOUN
fcis-30019	100	21	and	and	CCONJ
fcis-30019	100	22	testing	testing	NOUN
fcis-30019	100	23	samples	sample	NOUN
fcis-30019	100	24	for	for	ADP
fcis-30019	100	25	deepfake	deepfake	NOUN
fcis-30019	100	26	detection	detection	NOUN
fcis-30019	100	27	tasks	task	NOUN
fcis-30019	100	28	.	.	PUNCT
fcis-30019	101	1	deeperforensics	deeperforensic	NOUN
fcis-30019	102	1	[	[	X
fcis-30019	102	2	17	17	NUM
fcis-30019	102	3	]	]	PUNCT
fcis-30019	102	4	represents	represent	VERB
fcis-30019	102	5	the	the	DET
fcis-30019	102	6	largest	large	ADJ
fcis-30019	102	7	face	face	NOUN
fcis-30019	102	8	forgery	forgery	NOUN
fcis-30019	102	9	detection	detection	NOUN
fcis-30019	102	10	dataset	dataset	VERB
fcis-30019	102	11	to	to	ADP
fcis-30019	102	12	date	date	NOUN
fcis-30019	102	13	,	,	PUNCT
fcis-30019	102	14	containing	contain	VERB
fcis-30019	102	15	60,000	60,000	NUM
fcis-30019	102	16	videos	video	NOUN
fcis-30019	102	17	consisting	consist	VERB
fcis-30019	102	18	of	of	ADP
fcis-30019	102	19	17.6	17.6	NUM
fcis-30019	102	20	million	million	NUM
fcis-30019	102	21	frames	frame	NOUN
fcis-30019	102	22	.	.	PUNCT
fcis-30019	103	1	more	more	ADV
fcis-30019	103	2	challenging	challenging	ADJ
fcis-30019	103	3	benchmarks	benchmark	NOUN
fcis-30019	103	4	that	that	PRON
fcis-30019	103	5	apply	apply	VERB
fcis-30019	103	6	a	a	DET
fcis-30019	103	7	wide	wide	ADJ
fcis-30019	103	8	range	range	NOUN
fcis-30019	103	9	of	of	ADP
fcis-30019	103	10	real	real	ADJ
fcis-30019	103	11	-	-	PUNCT
fcis-30019	103	12	world	world	NOUN
fcis-30019	103	13	perturbations	perturbation	NOUN
fcis-30019	103	14	to	to	PART
fcis-30019	103	15	obtain	obtain	VERB
fcis-30019	103	16	larger	large	ADJ
fcis-30019	103	17	scale	scale	NOUN
fcis-30019	103	18	and	and	CCONJ
fcis-30019	103	19	higher	high	ADJ
fcis-30019	103	20	diversity	diversity	NOUN
fcis-30019	103	21	.	.	PUNCT
fcis-30019	104	1	all	all	DET
fcis-30019	104	2	source	source	NOUN
fcis-30019	104	3	videos	video	NOUN
fcis-30019	104	4	in	in	ADP
fcis-30019	104	5	the	the	DET
fcis-30019	104	6	dataset	dataset	NOUN
fcis-30019	104	7	are	be	AUX
fcis-30019	104	8	carefully	carefully	ADV
fcis-30019	104	9	collected	collect	VERB
fcis-30019	104	10	and	and	CCONJ
fcis-30019	104	11	fake	fake	ADJ
fcis-30019	104	12	videos	video	NOUN
fcis-30019	104	13	are	be	AUX
fcis-30019	104	14	generated	generate	VERB
fcis-30019	104	15	by	by	ADP
fcis-30019	104	16	the	the	DET
fcis-30019	104	17	newly	newly	ADV
fcis-30019	104	18	proposed	propose	VERB
fcis-30019	104	19	end	end	NOUN
fcis-30019	104	20	-	-	PUNCT
fcis-30019	104	21	to	to	ADP
fcis-30019	104	22	-	-	PUNCT
fcis-30019	104	23	end	end	NOUN
fcis-30019	104	24	face	face	NOUN
fcis-30019	104	25	exchange	exchange	NOUN
fcis-30019	104	26	framework	framework	NOUN
fcis-30019	104	27	.	.	PUNCT
fcis-30019	105	1	as	as	SCONJ
fcis-30019	105	2	verified	verify	VERB
fcis-30019	105	3	by	by	ADP
fcis-30019	105	4	the	the	DET
fcis-30019	105	5	user	user	NOUN
fcis-30019	105	6	study	study	NOUN
fcis-30019	105	7	,	,	PUNCT
fcis-30019	105	8	the	the	DET
fcis-30019	105	9	quality	quality	NOUN
fcis-30019	105	10	of	of	ADP
fcis-30019	105	11	the	the	DET
fcis-30019	105	12	generated	generate	VERB
fcis-30019	105	13	videos	video	NOUN
fcis-30019	105	14	is	be	AUX
fcis-30019	105	15	better	well	ADJ
fcis-30019	105	16	than	than	ADP
fcis-30019	105	17	that	that	PRON
fcis-30019	105	18	in	in	ADP
fcis-30019	105	19	existing	exist	VERB
fcis-30019	105	20	datasets	dataset	NOUN
fcis-30019	105	21	.	.	PUNCT
fcis-30019	106	1	faceshifter	faceshifter	NOUN
fcis-30019	107	1	[	[	X
fcis-30019	107	2	18	18	NUM
fcis-30019	107	3	]	]	PUNCT
fcis-30019	107	4	is	be	AUX
fcis-30019	107	5	a	a	DET
fcis-30019	107	6	fake	fake	ADJ
fcis-30019	107	7	video	video	NOUN
fcis-30019	107	8	dataset	dataset	NOUN
fcis-30019	107	9	based	base	VERB
fcis-30019	107	10	on	on	ADP
fcis-30019	107	11	face	face	NOUN
fcis-30019	107	12	swapping	swap	VERB
fcis-30019	107	13	technology	technology	NOUN
fcis-30019	107	14	,	,	PUNCT
fcis-30019	107	15	which	which	PRON
fcis-30019	107	16	consists	consist	VERB
fcis-30019	107	17	of	of	ADP
fcis-30019	107	18	two	two	NUM
fcis-30019	107	19	networks	network	NOUN
fcis-30019	107	20	,	,	PUNCT
fcis-30019	107	21	aeinet	aeinet	PROPN
fcis-30019	107	22	and	and	CCONJ
fcis-30019	107	23	hear	hear	NOUN
fcis-30019	107	24	-	-	PUNCT
fcis-30019	107	25	net	net	NOUN
fcis-30019	107	26	.	.	PUNCT
fcis-30019	108	1	aei	aei	PROPN
fcis-30019	108	2	-	-	PUNCT
fcis-30019	108	3	net	net	NOUN
fcis-30019	108	4	produces	produce	VERB
fcis-30019	108	5	preliminary	preliminary	ADJ
fcis-30019	108	6	face	face	NOUN
fcis-30019	108	7	swapping	swap	VERB
fcis-30019	108	8	results	result	NOUN
fcis-30019	108	9	,	,	PUNCT
fcis-30019	108	10	while	while	SCONJ
fcis-30019	108	11	hear	hear	NOUN
fcis-30019	108	12	-	-	PUNCT
fcis-30019	108	13	net	net	NOUN
fcis-30019	108	14	refines	refine	NOUN
fcis-30019	108	15	this	this	DET
fcis-30019	108	16	output	output	NOUN
fcis-30019	108	17	.	.	PUNCT
fcis-30019	109	1	it	it	PRON
fcis-30019	109	2	contains	contain	VERB
fcis-30019	109	3	1000	1000	NUM
fcis-30019	109	4	real	real	ADJ
fcis-30019	109	5	videos	video	NOUN
fcis-30019	109	6	and	and	CCONJ
fcis-30019	109	7	1000	1000	NUM
fcis-30019	109	8	fake	fake	ADJ
fcis-30019	109	9	videos	video	NOUN
fcis-30019	109	10	.	.	PUNCT
fcis-30019	110	1	these	these	DET
fcis-30019	110	2	fake	fake	ADJ
fcis-30019	110	3	videos	video	NOUN
fcis-30019	110	4	are	be	AUX
fcis-30019	110	5	generated	generate	VERB
fcis-30019	110	6	by	by	ADP
fcis-30019	110	7	face	face	NOUN
fcis-30019	110	8	swapping	swap	VERB
fcis-30019	110	9	technique	technique	NOUN
fcis-30019	110	10	and	and	CCONJ
fcis-30019	110	11	have	have	VERB
fcis-30019	110	12	high	high	ADJ
fcis-30019	110	13	visual	visual	ADJ
fcis-30019	110	14	quality	quality	NOUN
fcis-30019	110	15	and	and	CCONJ
fcis-30019	110	16	deception	deception	NOUN
fcis-30019	110	17	.	.	PUNCT
fcis-30019	111	1	the	the	DET
fcis-30019	111	2	faceshifter	faceshifter	NOUN
fcis-30019	111	3	dataset	dataset	NOUN
fcis-30019	111	4	covers	cover	VERB
fcis-30019	111	5	people	people	NOUN
fcis-30019	111	6	of	of	ADP
fcis-30019	111	7	multiple	multiple	ADJ
fcis-30019	111	8	races	race	NOUN
fcis-30019	111	9	,	,	PUNCT
fcis-30019	111	10	genders	gender	NOUN
fcis-30019	111	11	,	,	PUNCT
fcis-30019	111	12	and	and	CCONJ
fcis-30019	111	13	ages	age	NOUN
fcis-30019	111	14	,	,	PUNCT
fcis-30019	111	15	as	as	ADV
fcis-30019	111	16	well	well	ADV
fcis-30019	111	17	as	as	ADP
fcis-30019	111	18	multiple	multiple	ADJ
fcis-30019	111	19	expressions	expression	NOUN
fcis-30019	111	20	,	,	PUNCT
fcis-30019	111	21	poses	pose	NOUN
fcis-30019	111	22	,	,	PUNCT
fcis-30019	111	23	and	and	CCONJ
fcis-30019	111	24	lighting	lighting	NOUN
fcis-30019	111	25	conditions	condition	NOUN
fcis-30019	111	26	,	,	PUNCT
fcis-30019	111	27	which	which	PRON
fcis-30019	111	28	provides	provide	VERB
fcis-30019	111	29	important	important	ADJ
fcis-30019	111	30	data	datum	NOUN
fcis-30019	111	31	support	support	NOUN
fcis-30019	111	32	for	for	ADP
fcis-30019	111	33	studying	study	VERB
fcis-30019	111	34	the	the	DET
fcis-30019	111	35	application	application	NOUN
fcis-30019	111	36	of	of	ADP
fcis-30019	111	37	faceshifter	faceshifter	NOUN
fcis-30019	111	38	technology	technology	NOUN
fcis-30019	111	39	in	in	ADP
fcis-30019	111	40	forgery	forgery	NOUN
fcis-30019	111	41	video	video	NOUN
fcis-30019	111	42	detection	detection	NOUN
fcis-30019	111	43	.	.	PUNCT
fcis-30019	112	1	biodeepav	biodeepav	PROPN
fcis-30019	113	1	[	[	X
fcis-30019	113	2	19	19	NUM
fcis-30019	113	3	]	]	PUNCT
fcis-30019	113	4	is	be	AUX
fcis-30019	113	5	a	a	DET
fcis-30019	113	6	multimodal	multimodal	ADJ
fcis-30019	113	7	benchmark	benchmark	NOUN
fcis-30019	113	8	dataset	dataset	NOUN
fcis-30019	113	9	to	to	PART
fcis-30019	113	10	evaluate	evaluate	VERB
fcis-30019	113	11	the	the	DET
fcis-30019	113	12	performance	performance	NOUN
fcis-30019	113	13	of	of	ADP
fcis-30019	113	14	deepfake	deepfake	NOUN
fcis-30019	113	15	detectors	detector	NOUN
fcis-30019	113	16	in	in	ADP
fcis-30019	113	17	the	the	DET
fcis-30019	113	18	face	face	NOUN
fcis-30019	113	19	of	of	ADP
fcis-30019	113	20	unknown	unknown	ADJ
fcis-30019	113	21	generators	generator	NOUN
fcis-30019	113	22	.	.	PUNCT
fcis-30019	114	1	contains	contain	VERB
fcis-30019	114	2	over	over	ADP
fcis-30019	114	3	1600	1600	NUM
fcis-30019	114	4	deepfake	deepfake	NOUN
fcis-30019	114	5	videos	video	NOUN
fcis-30019	114	6	,	,	PUNCT
fcis-30019	114	7	and	and	CCONJ
fcis-30019	114	8	the	the	DET
fcis-30019	114	9	videos	video	NOUN
fcis-30019	114	10	are	be	AUX
fcis-30019	114	11	generated	generate	VERB
fcis-30019	114	12	using	use	VERB
fcis-30019	114	13	four	four	NUM
fcis-30019	114	14	state	state	NOUN
fcis-30019	114	15	-	-	PUNCT
fcis-30019	114	16	of	of	ADP
fcis-30019	114	17	-	-	PUNCT
fcis-30019	114	18	the	the	DET
fcis-30019	114	19	-	-	PUNCT
fcis-30019	114	20	art	art	NOUN
fcis-30019	114	21	methods	method	NOUN
fcis-30019	114	22	specifically	specifically	ADV
fcis-30019	114	23	for	for	ADP
fcis-30019	114	24	video	video	NOUN
fcis-30019	114	25	synthesis	synthesis	NOUN
fcis-30019	114	26	.	.	PUNCT
fcis-30019	115	1	the	the	DET
fcis-30019	115	2	dataset	dataset	NOUN
fcis-30019	115	3	covers	cover	VERB
fcis-30019	115	4	multiple	multiple	ADJ
fcis-30019	115	5	identities	identity	NOUN
fcis-30019	115	6	and	and	CCONJ
fcis-30019	115	7	expressions	expression	NOUN
fcis-30019	115	8	,	,	PUNCT
fcis-30019	115	9	sampling	sample	VERB
fcis-30019	115	10	real	real	ADJ
fcis-30019	115	11	videos	video	NOUN
fcis-30019	115	12	from	from	ADP
fcis-30019	115	13	hdtf	hdtf	NOUN
fcis-30019	115	14	and	and	CCONJ
fcis-30019	115	15	talkinghead-1kh	talkinghead-1kh	ADP
fcis-30019	115	16	datasets	dataset	NOUN
fcis-30019	115	17	,	,	PUNCT
fcis-30019	115	18	and	and	CCONJ
fcis-30019	115	19	using	use	VERB
fcis-30019	115	20	face	face	NOUN
fcis-30019	115	21	images	image	NOUN
fcis-30019	115	22	from	from	ADP
fcis-30019	115	23	sources	source	NOUN
fcis-30019	115	24	such	such	ADJ
fcis-30019	115	25	as	as	ADP
fcis-30019	115	26	realvisxl	realvisxl	ADJ
fcis-30019	115	27	,	,	PUNCT
fcis-30019	115	28	laion	laion	NOUN
fcis-30019	115	29	-	-	PUNCT
fcis-30019	115	30	face	face	NOUN
fcis-30019	115	31	and	and	CCONJ
fcis-30019	115	32	hdtf	hdtf	NOUN
fcis-30019	115	33	,	,	PUNCT
fcis-30019	115	34	as	as	ADV
fcis-30019	115	35	well	well	ADV
fcis-30019	115	36	as	as	ADP
fcis-30019	115	37	english	english	ADJ
fcis-30019	115	38	dialects	dialect	NOUN
fcis-30019	115	39	,	,	PUNCT
fcis-30019	115	40	hdtf	hdtf	NOUN
fcis-30019	115	41	datasets	dataset	NOUN
fcis-30019	115	42	.	.	PUNCT
fcis-30019	116	1	to	to	PART
fcis-30019	116	2	ensure	ensure	VERB
fcis-30019	116	3	the	the	DET
fcis-30019	116	4	effectiveness	effectiveness	NOUN
fcis-30019	116	5	of	of	ADP
fcis-30019	116	6	the	the	DET
fcis-30019	116	7	amount	amount	NOUN
fcis-30019	116	8	of	of	ADP
fcis-30019	116	9	experimental	experimental	ADJ
fcis-30019	116	10	data	datum	NOUN
fcis-30019	116	11	,	,	PUNCT
fcis-30019	116	12	1000	1000	NUM
fcis-30019	116	13	real	real	ADJ
fcis-30019	116	14	and	and	CCONJ
fcis-30019	116	15	fake	fake	ADJ
fcis-30019	116	16	videos	video	NOUN
fcis-30019	116	17	are	be	AUX
fcis-30019	116	18	randomly	randomly	ADV
fcis-30019	116	19	selected	select	VERB
fcis-30019	116	20	for	for	ADP
fcis-30019	116	21	the	the	DET
fcis-30019	116	22	above	above	ADJ
fcis-30019	116	23	five	five	NUM
fcis-30019	116	24	data	datum	NOUN
fcis-30019	116	25	sets	set	NOUN
fcis-30019	116	26	,	,	PUNCT
fcis-30019	116	27	and	and	CCONJ
fcis-30019	116	28	32	32	NUM
fcis-30019	116	29	frames	frame	NOUN
fcis-30019	116	30	are	be	AUX
fcis-30019	116	31	selected	select	VERB
fcis-30019	116	32	at	at	ADP
fcis-30019	116	33	equal	equal	ADJ
fcis-30019	116	34	intervals	interval	NOUN
fcis-30019	116	35	.	.	PUNCT
fcis-30019	117	1	among	among	ADP
fcis-30019	117	2	them	they	PRON
fcis-30019	117	3	,	,	PUNCT
fcis-30019	117	4	celeb	celeb	NOUN
fcis-30019	117	5	-	-	PUNCT
fcis-30019	117	6	dfv2	dfv2	PROPN
fcis-30019	117	7	selects	select	VERB
fcis-30019	117	8	590	590	NUM
fcis-30019	117	9	real	real	ADJ
fcis-30019	117	10	videos	video	NOUN
fcis-30019	117	11	and	and	CCONJ
fcis-30019	117	12	divides	divide	VERB
fcis-30019	117	13	the	the	DET
fcis-30019	117	14	training	training	NOUN
fcis-30019	117	15	set	set	NOUN
fcis-30019	117	16	and	and	CCONJ
fcis-30019	117	17	the	the	DET
fcis-30019	117	18	test	test	NOUN
fcis-30019	117	19	set	set	VERB
fcis-30019	117	20	with	with	ADP
fcis-30019	117	21	the	the	DET
fcis-30019	117	22	ratio	ratio	NOUN
fcis-30019	117	23	of	of	ADP
fcis-30019	117	24	7:3	7:3	NUM
fcis-30019	117	25	.	.	PUNCT
fcis-30019	118	1	the	the	DET
fcis-30019	118	2	training	training	NOUN
fcis-30019	118	3	set	set	NOUN
fcis-30019	118	4	of	of	ADP
fcis-30019	118	5	each	each	DET
fcis-30019	118	6	dataset	dataset	NOUN
fcis-30019	118	7	contains	contain	VERB
fcis-30019	118	8	22,400	22,400	NUM
fcis-30019	118	9	real	real	ADJ
fcis-30019	118	10	video	video	NOUN
fcis-30019	118	11	frame	frame	NOUN
fcis-30019	118	12	images	image	NOUN
fcis-30019	118	13	and	and	CCONJ
fcis-30019	118	14	22,400	22,400	NUM
fcis-30019	118	15	fake	fake	ADJ
fcis-30019	118	16	video	video	NOUN
fcis-30019	118	17	frame	frame	NOUN
fcis-30019	118	18	images	image	NOUN
fcis-30019	118	19	.	.	PUNCT
fcis-30019	119	1	the	the	DET
fcis-30019	119	2	test	test	NOUN
fcis-30019	119	3	set	set	NOUN
fcis-30019	119	4	contains	contain	VERB
fcis-30019	119	5	9600	9600	NUM
fcis-30019	119	6	real	real	ADJ
fcis-30019	119	7	video	video	NOUN
fcis-30019	119	8	frame	frame	NOUN
fcis-30019	119	9	images	image	NOUN
fcis-30019	119	10	and	and	CCONJ
fcis-30019	119	11	9600	9600	NUM
fcis-30019	119	12	fake	fake	ADJ
fcis-30019	119	13	video	video	NOUN
fcis-30019	119	14	frame	frame	NOUN
fcis-30019	119	15	images	image	NOUN
fcis-30019	119	16	,	,	PUNCT
fcis-30019	119	17	and	and	CCONJ
fcis-30019	119	18	the	the	DET
fcis-30019	119	19	experimental	experimental	ADJ
fcis-30019	119	20	data	datum	NOUN
fcis-30019	119	21	has	have	VERB
fcis-30019	119	22	320,000	320,000	NUM
fcis-30019	119	23	face	face	NOUN
fcis-30019	119	24	images	image	NOUN
fcis-30019	119	25	.	.	PUNCT
fcis-30019	120	1	4.2	4.2	NUM
fcis-30019	120	2	.	.	PUNCT
fcis-30019	120	3	evaluation	evaluation	NOUN
fcis-30019	120	4	metrics	metric	NOUN
fcis-30019	120	5	and	and	CCONJ
fcis-30019	120	6	experimental	experimental	ADJ
fcis-30019	120	7	setup	setup	NOUN
fcis-30019	120	8	the	the	DET
fcis-30019	120	9	selected	select	VERB
fcis-30019	120	10	evaluation	evaluation	NOUN
fcis-30019	120	11	metrics	metric	NOUN
fcis-30019	120	12	in	in	ADP
fcis-30019	120	13	the	the	DET
fcis-30019	120	14	experiment	experiment	NOUN
fcis-30019	120	15	are	be	AUX
fcis-30019	120	16	accuracy	accuracy	NOUN
fcis-30019	120	17	(	(	PUNCT
fcis-30019	120	18	acc	acc	PROPN
fcis-30019	120	19	)	)	PUNCT
fcis-30019	120	20	and	and	CCONJ
fcis-30019	120	21	area	area	NOUN
fcis-30019	120	22	under	under	ADP
fcis-30019	120	23	the	the	DET
fcis-30019	120	24	curve	curve	NOUN
fcis-30019	120	25	(	(	PUNCT
fcis-30019	120	26	auc	auc	NOUN
fcis-30019	120	27	)	)	PUNCT
fcis-30019	120	28	as	as	SCONJ
fcis-30019	120	29	the	the	DET
fcis-30019	120	30	key	key	ADJ
fcis-30019	120	31	metrics	metric	NOUN
fcis-30019	120	32	to	to	PART
fcis-30019	120	33	evaluate	evaluate	VERB
fcis-30019	120	34	the	the	DET
fcis-30019	120	35	performance	performance	NOUN
fcis-30019	120	36	of	of	ADP
fcis-30019	120	37	the	the	DET
fcis-30019	120	38	model	model	NOUN
fcis-30019	120	39	,	,	PUNCT
fcis-30019	120	40	and	and	CCONJ
fcis-30019	120	41	the	the	DET
fcis-30019	120	42	specific	specific	ADJ
fcis-30019	120	43	calculation	calculation	NOUN
fcis-30019	120	44	formulas	formula	NOUN
fcis-30019	120	45	of	of	ADP
fcis-30019	120	46	these	these	DET
fcis-30019	120	47	two	two	NUM
fcis-30019	120	48	-	-	PUNCT
fcis-30019	120	49	evaluation	evaluation	NOUN
fcis-30019	120	50	metrics	metric	NOUN
fcis-30019	120	51	have	have	AUX
fcis-30019	120	52	been	be	AUX
fcis-30019	120	53	elaborated	elaborate	VERB
fcis-30019	120	54	in	in	ADP
fcis-30019	120	55	the	the	DET
fcis-30019	120	56	third	third	ADJ
fcis-30019	120	57	chapter	chapter	NOUN
fcis-30019	120	58	of	of	ADP
fcis-30019	120	59	the	the	DET
fcis-30019	120	60	report	report	NOUN
fcis-30019	120	61	.	.	PUNCT
fcis-30019	121	1	acc	acc	PROPN
fcis-30019	121	2	measures	measure	VERB
fcis-30019	121	3	how	how	SCONJ
fcis-30019	121	4	well	well	ADV
fcis-30019	121	5	the	the	DET
fcis-30019	121	6	model	model	NOUN
fcis-30019	121	7	classifies	classify	VERB
fcis-30019	121	8	the	the	DET
fcis-30019	121	9	positive	positive	ADJ
fcis-30019	121	10	and	and	CCONJ
fcis-30019	121	11	negative	negative	ADJ
fcis-30019	121	12	samples	sample	NOUN
fcis-30019	121	13	,	,	PUNCT
fcis-30019	121	14	while	while	SCONJ
fcis-30019	121	15	auc	auc	VERB
fcis-30019	121	16	measures	measure	NOUN
fcis-30019	121	17	how	how	SCONJ
fcis-30019	121	18	well	well	ADV
fcis-30019	121	19	the	the	DET
fcis-30019	121	20	model	model	NOUN
fcis-30019	121	21	classifies	classify	VERB
fcis-30019	121	22	the	the	DET
fcis-30019	121	23	positive	positive	ADJ
fcis-30019	121	24	and	and	CCONJ
fcis-30019	121	25	negative	negative	ADJ
fcis-30019	121	26	samples	sample	NOUN
fcis-30019	121	27	.	.	PUNCT
fcis-30019	122	1	58	58	NUM
fcis-30019	122	2	fig	fig	NOUN
fcis-30019	122	3	5	5	NUM
fcis-30019	122	4	.	.	PUNCT
fcis-30019	123	1	examples	example	NOUN
fcis-30019	123	2	of	of	ADP
fcis-30019	123	3	datasets	dataset	NOUN
fcis-30019	123	4	used	use	VERB
fcis-30019	123	5	for	for	ADP
fcis-30019	123	6	experiments	experiment	NOUN
fcis-30019	123	7	the	the	DET
fcis-30019	123	8	learning	learning	NOUN
fcis-30019	123	9	rate	rate	NOUN
fcis-30019	123	10	determines	determine	VERB
fcis-30019	123	11	the	the	DET
fcis-30019	123	12	step	step	NOUN
fcis-30019	123	13	size	size	NOUN
fcis-30019	123	14	at	at	ADP
fcis-30019	123	15	which	which	PRON
fcis-30019	123	16	the	the	DET
fcis-30019	123	17	model	model	NOUN
fcis-30019	123	18	updates	update	VERB
fcis-30019	123	19	the	the	DET
fcis-30019	123	20	weights	weight	NOUN
fcis-30019	123	21	during	during	ADP
fcis-30019	123	22	the	the	DET
fcis-30019	123	23	optimization	optimization	NOUN
fcis-30019	123	24	process	process	NOUN
fcis-30019	123	25	.	.	PUNCT
fcis-30019	124	1	an	an	DET
fcis-30019	124	2	appropriate	appropriate	ADJ
fcis-30019	124	3	learning	learning	NOUN
fcis-30019	124	4	rate	rate	NOUN
fcis-30019	124	5	can	can	AUX
fcis-30019	124	6	speed	speed	VERB
fcis-30019	124	7	up	up	ADP
fcis-30019	124	8	the	the	DET
fcis-30019	124	9	convergence	convergence	NOUN
fcis-30019	124	10	of	of	ADP
fcis-30019	124	11	the	the	DET
fcis-30019	124	12	model	model	NOUN
fcis-30019	124	13	while	while	SCONJ
fcis-30019	124	14	avoiding	avoid	VERB
fcis-30019	124	15	oscillations	oscillation	NOUN
fcis-30019	124	16	or	or	CCONJ
fcis-30019	124	17	overstepping	overstep	VERB
fcis-30019	124	18	the	the	DET
fcis-30019	124	19	optimal	optimal	ADJ
fcis-30019	124	20	solution	solution	NOUN
fcis-30019	124	21	during	during	ADP
fcis-30019	124	22	the	the	DET
fcis-30019	124	23	training	training	NOUN
fcis-30019	124	24	process	process	NOUN
fcis-30019	124	25	.	.	PUNCT
fcis-30019	125	1	in	in	ADP
fcis-30019	125	2	this	this	DET
fcis-30019	125	3	experiment	experiment	NOUN
fcis-30019	125	4	,	,	PUNCT
fcis-30019	125	5	we	we	PRON
fcis-30019	125	6	set	set	VERB
fcis-30019	125	7	the	the	DET
fcis-30019	125	8	learning	learning	NOUN
fcis-30019	125	9	rate	rate	NOUN
fcis-30019	125	10	to	to	ADP
fcis-30019	125	11	0.001	0.001	NUM
fcis-30019	125	12	,	,	PUNCT
fcis-30019	125	13	which	which	PRON
fcis-30019	125	14	is	be	AUX
fcis-30019	125	15	a	a	DET
fcis-30019	125	16	starting	starting	NOUN
fcis-30019	125	17	value	value	NOUN
fcis-30019	125	18	that	that	PRON
fcis-30019	125	19	is	be	AUX
fcis-30019	125	20	often	often	ADV
fcis-30019	125	21	used	use	VERB
fcis-30019	125	22	in	in	ADP
fcis-30019	125	23	practice	practice	NOUN
fcis-30019	125	24	.	.	PUNCT
fcis-30019	126	1	the	the	DET
fcis-30019	126	2	batch	batch	NOUN
fcis-30019	126	3	size	size	NOUN
fcis-30019	126	4	defines	define	VERB
fcis-30019	126	5	the	the	DET
fcis-30019	126	6	number	number	NOUN
fcis-30019	126	7	of	of	ADP
fcis-30019	126	8	samples	sample	NOUN
fcis-30019	126	9	used	use	VERB
fcis-30019	126	10	in	in	ADP
fcis-30019	126	11	each	each	DET
fcis-30019	126	12	training	training	NOUN
fcis-30019	126	13	iteration	iteration	NOUN
fcis-30019	126	14	.	.	PUNCT
fcis-30019	127	1	a	a	DET
fcis-30019	127	2	large	large	ADJ
fcis-30019	127	3	batch	batch	NOUN
fcis-30019	127	4	size	size	NOUN
fcis-30019	127	5	can	can	AUX
fcis-30019	127	6	reduce	reduce	VERB
fcis-30019	127	7	the	the	DET
fcis-30019	127	8	variance	variance	NOUN
fcis-30019	127	9	in	in	ADP
fcis-30019	127	10	the	the	DET
fcis-30019	127	11	training	training	NOUN
fcis-30019	127	12	process	process	NOUN
fcis-30019	127	13	and	and	CCONJ
fcis-30019	127	14	improve	improve	VERB
fcis-30019	127	15	the	the	DET
fcis-30019	127	16	efficiency	efficiency	NOUN
fcis-30019	127	17	of	of	ADP
fcis-30019	127	18	memory	memory	NOUN
fcis-30019	127	19	utilization	utilization	NOUN
fcis-30019	127	20	,	,	PUNCT
fcis-30019	127	21	but	but	CCONJ
fcis-30019	127	22	it	it	PRON
fcis-30019	127	23	may	may	AUX
fcis-30019	127	24	reduce	reduce	VERB
fcis-30019	127	25	the	the	DET
fcis-30019	127	26	generalization	generalization	NOUN
fcis-30019	127	27	ability	ability	NOUN
fcis-30019	127	28	of	of	ADP
fcis-30019	127	29	the	the	DET
fcis-30019	127	30	model	model	NOUN
fcis-30019	127	31	.	.	PUNCT
fcis-30019	128	1	a	a	DET
fcis-30019	128	2	smaller	small	ADJ
fcis-30019	128	3	batch	batch	NOUN
fcis-30019	128	4	size	size	NOUN
fcis-30019	128	5	may	may	AUX
fcis-30019	128	6	increase	increase	VERB
fcis-30019	128	7	the	the	DET
fcis-30019	128	8	training	training	NOUN
fcis-30019	128	9	noise	noise	NOUN
fcis-30019	128	10	,	,	PUNCT
fcis-30019	128	11	but	but	CCONJ
fcis-30019	128	12	it	it	PRON
fcis-30019	128	13	helps	help	VERB
fcis-30019	128	14	the	the	DET
fcis-30019	128	15	model	model	NOUN
fcis-30019	128	16	capture	capture	VERB
fcis-30019	128	17	more	more	ADJ
fcis-30019	128	18	feature	feature	NOUN
fcis-30019	128	19	details	detail	NOUN
fcis-30019	128	20	.	.	PUNCT
fcis-30019	129	1	in	in	ADP
fcis-30019	129	2	this	this	DET
fcis-30019	129	3	experiment	experiment	NOUN
fcis-30019	129	4	,	,	PUNCT
fcis-30019	129	5	we	we	PRON
fcis-30019	129	6	chose	choose	VERB
fcis-30019	129	7	128	128	NUM
fcis-30019	129	8	as	as	ADP
fcis-30019	129	9	the	the	DET
fcis-30019	129	10	value	value	NOUN
fcis-30019	129	11	of	of	ADP
fcis-30019	129	12	batch	batch	NOUN
fcis-30019	129	13	size	size	NOUN
fcis-30019	129	14	.	.	PUNCT
fcis-30019	130	1	the	the	DET
fcis-30019	130	2	number	number	NOUN
fcis-30019	130	3	of	of	ADP
fcis-30019	130	4	epochs	epoch	NOUN
fcis-30019	130	5	represents	represent	VERB
fcis-30019	130	6	the	the	DET
fcis-30019	130	7	number	number	NOUN
fcis-30019	130	8	of	of	ADP
fcis-30019	130	9	times	time	NOUN
fcis-30019	130	10	the	the	DET
fcis-30019	130	11	model	model	NOUN
fcis-30019	130	12	iterates	iterate	VERB
fcis-30019	130	13	over	over	ADP
fcis-30019	130	14	the	the	DET
fcis-30019	130	15	entire	entire	ADJ
fcis-30019	130	16	training	training	NOUN
fcis-30019	130	17	set	set	NOUN
fcis-30019	130	18	.	.	PUNCT
fcis-30019	131	1	a	a	DET
fcis-30019	131	2	sufficient	sufficient	ADJ
fcis-30019	131	3	number	number	NOUN
fcis-30019	131	4	of	of	ADP
fcis-30019	131	5	iterations	iteration	NOUN
fcis-30019	131	6	ensures	ensure	VERB
fcis-30019	131	7	that	that	SCONJ
fcis-30019	131	8	the	the	DET
fcis-30019	131	9	model	model	NOUN
fcis-30019	131	10	sufficiently	sufficiently	ADV
fcis-30019	131	11	learns	learn	VERB
fcis-30019	131	12	the	the	DET
fcis-30019	131	13	patterns	pattern	NOUN
fcis-30019	131	14	in	in	ADP
fcis-30019	131	15	the	the	DET
fcis-30019	131	16	training	training	NOUN
fcis-30019	131	17	data	datum	NOUN
fcis-30019	131	18	,	,	PUNCT
fcis-30019	131	19	but	but	CCONJ
fcis-30019	131	20	too	too	ADV
fcis-30019	131	21	many	many	ADJ
fcis-30019	131	22	iterations	iteration	NOUN
fcis-30019	131	23	may	may	AUX
fcis-30019	131	24	cause	cause	VERB
fcis-30019	131	25	the	the	DET
fcis-30019	131	26	model	model	NOUN
fcis-30019	131	27	to	to	PART
fcis-30019	131	28	overfit	overfit	VERB
fcis-30019	131	29	.	.	PUNCT
fcis-30019	132	1	in	in	ADP
fcis-30019	132	2	the	the	DET
fcis-30019	132	3	experiments	experiment	NOUN
fcis-30019	132	4	,	,	PUNCT
fcis-30019	132	5	we	we	PRON
fcis-30019	132	6	set	set	VERB
fcis-30019	132	7	the	the	DET
fcis-30019	132	8	number	number	NOUN
fcis-30019	132	9	of	of	ADP
fcis-30019	132	10	iterations	iteration	NOUN
fcis-30019	132	11	as	as	ADP
fcis-30019	132	12	30	30	NUM
fcis-30019	132	13	,	,	PUNCT
fcis-30019	132	14	which	which	PRON
fcis-30019	132	15	is	be	AUX
fcis-30019	132	16	an	an	DET
fcis-30019	132	17	empirically	empirically	ADV
fcis-30019	132	18	based	base	VERB
fcis-30019	132	19	choice	choice	NOUN
fcis-30019	132	20	designed	design	VERB
fcis-30019	132	21	to	to	PART
fcis-30019	132	22	balance	balance	VERB
fcis-30019	132	23	training	training	NOUN
fcis-30019	132	24	time	time	NOUN
fcis-30019	132	25	and	and	CCONJ
fcis-30019	132	26	model	model	NOUN
fcis-30019	132	27	performance	performance	NOUN
fcis-30019	132	28	.	.	PUNCT
fcis-30019	133	1	the	the	DET
fcis-30019	133	2	optimizer	optimizer	NOUN
fcis-30019	133	3	determines	determine	VERB
fcis-30019	133	4	the	the	DET
fcis-30019	133	5	strategy	strategy	NOUN
fcis-30019	133	6	for	for	ADP
fcis-30019	133	7	updating	update	VERB
fcis-30019	133	8	the	the	DET
fcis-30019	133	9	model	model	NOUN
fcis-30019	133	10	weights	weight	NOUN
fcis-30019	133	11	.	.	PUNCT
fcis-30019	134	1	in	in	ADP
fcis-30019	134	2	this	this	DET
fcis-30019	134	3	experiment	experiment	NOUN
fcis-30019	134	4	,	,	PUNCT
fcis-30019	134	5	we	we	PRON
fcis-30019	134	6	adopted	adopt	VERB
fcis-30019	134	7	the	the	DET
fcis-30019	134	8	adam	adam	PROPN
fcis-30019	134	9	optimization	optimization	NOUN
fcis-30019	134	10	algorithm	algorithm	NOUN
fcis-30019	134	11	because	because	SCONJ
fcis-30019	134	12	it	it	PRON
fcis-30019	134	13	combines	combine	VERB
fcis-30019	134	14	the	the	DET
fcis-30019	134	15	momentum	momentum	NOUN
fcis-30019	134	16	method	method	NOUN
fcis-30019	134	17	and	and	CCONJ
fcis-30019	134	18	the	the	DET
fcis-30019	134	19	first	first	ADJ
fcis-30019	134	20	moment	moment	NOUN
fcis-30019	134	21	estimation	estimation	NOUN
fcis-30019	134	22	of	of	ADP
fcis-30019	134	23	the	the	DET
fcis-30019	134	24	gradient	gradient	NOUN
fcis-30019	134	25	,	,	PUNCT
fcis-30019	134	26	which	which	PRON
fcis-30019	134	27	is	be	AUX
fcis-30019	134	28	generally	generally	ADV
fcis-30019	134	29	able	able	ADJ
fcis-30019	134	30	to	to	PART
fcis-30019	134	31	provide	provide	VERB
fcis-30019	134	32	good	good	ADJ
fcis-30019	134	33	performance	performance	NOUN
fcis-30019	134	34	in	in	ADP
fcis-30019	134	35	a	a	DET
fcis-30019	134	36	variety	variety	NOUN
fcis-30019	134	37	of	of	ADP
fcis-30019	134	38	different	different	ADJ
fcis-30019	134	39	optimization	optimization	NOUN
fcis-30019	134	40	problems	problem	NOUN
fcis-30019	134	41	.	.	PUNCT
fcis-30019	135	1	in	in	ADP
fcis-30019	135	2	addition	addition	NOUN
fcis-30019	135	3	,	,	PUNCT
fcis-30019	135	4	to	to	PART
fcis-30019	135	5	mitigate	mitigate	VERB
fcis-30019	135	6	the	the	DET
fcis-30019	135	7	possible	possible	ADJ
fcis-30019	135	8	overfitting	overfitting	NOUN
fcis-30019	135	9	of	of	ADP
fcis-30019	135	10	the	the	DET
fcis-30019	135	11	model	model	NOUN
fcis-30019	135	12	,	,	PUNCT
fcis-30019	135	13	we	we	PRON
fcis-30019	135	14	employ	employ	VERB
fcis-30019	135	15	dropout	dropout	NOUN
fcis-30019	135	16	technique	technique	NOUN
fcis-30019	135	17	.	.	PUNCT
fcis-30019	136	1	by	by	ADP
fcis-30019	136	2	randomly	randomly	ADV
fcis-30019	136	3	dropping	drop	VERB
fcis-30019	136	4	a	a	DET
fcis-30019	136	5	part	part	NOUN
fcis-30019	136	6	of	of	ADP
fcis-30019	136	7	the	the	DET
fcis-30019	136	8	neurons	neuron	NOUN
fcis-30019	136	9	in	in	ADP
fcis-30019	136	10	the	the	DET
fcis-30019	136	11	network	network	NOUN
fcis-30019	136	12	during	during	ADP
fcis-30019	136	13	training	training	NOUN
fcis-30019	136	14	,	,	PUNCT
fcis-30019	136	15	dropout	dropout	NOUN
fcis-30019	136	16	effectively	effectively	ADV
fcis-30019	136	17	reduces	reduce	VERB
fcis-30019	136	18	the	the	DET
fcis-30019	136	19	dependence	dependence	NOUN
fcis-30019	136	20	of	of	ADP
fcis-30019	136	21	the	the	DET
fcis-30019	136	22	model	model	NOUN
fcis-30019	136	23	on	on	ADP
fcis-30019	136	24	specific	specific	ADJ
fcis-30019	136	25	training	training	NOUN
fcis-30019	136	26	samples	sample	NOUN
fcis-30019	136	27	,	,	PUNCT
fcis-30019	136	28	thereby	thereby	ADV
fcis-30019	136	29	improving	improve	VERB
fcis-30019	136	30	the	the	DET
fcis-30019	136	31	generalization	generalization	NOUN
fcis-30019	136	32	ability	ability	NOUN
fcis-30019	136	33	of	of	ADP
fcis-30019	136	34	the	the	DET
fcis-30019	136	35	model	model	NOUN
fcis-30019	136	36	.	.	PUNCT
fcis-30019	137	1	in	in	ADP
fcis-30019	137	2	this	this	DET
fcis-30019	137	3	experiment	experiment	NOUN
fcis-30019	137	4	,	,	PUNCT
fcis-30019	137	5	we	we	PRON
fcis-30019	137	6	apply	apply	VERB
fcis-30019	137	7	dropout	dropout	NOUN
fcis-30019	137	8	appropriately	appropriately	ADV
fcis-30019	137	9	to	to	PART
fcis-30019	137	10	ensure	ensure	VERB
fcis-30019	137	11	that	that	SCONJ
fcis-30019	137	12	the	the	DET
fcis-30019	137	13	model	model	NOUN
fcis-30019	137	14	can	can	AUX
fcis-30019	137	15	maintain	maintain	VERB
fcis-30019	137	16	good	good	ADJ
fcis-30019	137	17	prediction	prediction	NOUN
fcis-30019	137	18	performance	performance	NOUN
fcis-30019	137	19	even	even	ADV
fcis-30019	137	20	on	on	ADP
fcis-30019	137	21	unseen	unseen	ADJ
fcis-30019	137	22	data	datum	NOUN
fcis-30019	137	23	.	.	PUNCT
fcis-30019	138	1	with	with	ADP
fcis-30019	138	2	these	these	DET
fcis-30019	138	3	meticulous	meticulous	ADJ
fcis-30019	138	4	hyperparameter	hyperparameter	NOUN
fcis-30019	138	5	tuning	tuning	NOUN
fcis-30019	138	6	and	and	CCONJ
fcis-30019	138	7	policy	policy	NOUN
fcis-30019	138	8	application	application	NOUN
fcis-30019	138	9	,	,	PUNCT
fcis-30019	138	10	we	we	PRON
fcis-30019	138	11	expect	expect	VERB
fcis-30019	138	12	to	to	PART
fcis-30019	138	13	be	be	AUX
fcis-30019	138	14	able	able	ADJ
fcis-30019	138	15	to	to	PART
fcis-30019	138	16	train	train	VERB
fcis-30019	138	17	a	a	DET
fcis-30019	138	18	model	model	NOUN
fcis-30019	138	19	that	that	PRON
fcis-30019	138	20	is	be	AUX
fcis-30019	138	21	both	both	CCONJ
fcis-30019	138	22	accurate	accurate	ADJ
fcis-30019	138	23	and	and	CCONJ
fcis-30019	138	24	robust	robust	ADJ
fcis-30019	138	25	.	.	PUNCT
fcis-30019	139	1	table	table	NOUN
fcis-30019	139	2	1	1	NUM
fcis-30019	139	3	.	.	PUNCT
fcis-30019	140	1	setting	set	VERB
fcis-30019	140	2	of	of	ADP
fcis-30019	140	3	experimental	experimental	ADJ
fcis-30019	140	4	hyperparameters	hyperparameter	NOUN
fcis-30019	140	5	parameter	parameter	NOUN
fcis-30019	140	6	value	value	NOUN
fcis-30019	140	7	learning	learn	VERB
fcis-30019	140	8	rate	rate	NOUN
fcis-30019	140	9	0.001	0.001	NUM
fcis-30019	140	10	batch	batch	NOUN
fcis-30019	140	11	size	size	NOUN
fcis-30019	140	12	128	128	NUM
fcis-30019	140	13	epochs	epoch	NOUN
fcis-30019	140	14	30	30	NUM
fcis-30019	140	15	optimizer	optimizer	NOUN
fcis-30019	140	16	adam	adam	PROPN
fcis-30019	140	17	dropout	dropout	PROPN
fcis-30019	140	18	0.5	0.5	NUM
fcis-30019	140	19	4.3	4.3	NUM
fcis-30019	140	20	.	.	PUNCT
fcis-30019	140	21	ablation	ablation	NOUN
fcis-30019	140	22	experiment	experiment	NOUN
fcis-30019	140	23	in	in	ADP
fcis-30019	140	24	order	order	NOUN
fcis-30019	140	25	to	to	PART
fcis-30019	140	26	analyze	analyze	VERB
fcis-30019	140	27	the	the	DET
fcis-30019	140	28	gain	gain	NOUN
fcis-30019	140	29	of	of	ADP
fcis-30019	140	30	different	different	ADJ
fcis-30019	140	31	network	network	NOUN
fcis-30019	140	32	modules	module	NOUN
fcis-30019	140	33	for	for	ADP
fcis-30019	140	34	model	model	NOUN
fcis-30019	140	35	checking	checking	NOUN
fcis-30019	140	36	,	,	PUNCT
fcis-30019	140	37	ablation	ablation	NOUN
fcis-30019	140	38	experiments	experiment	NOUN
fcis-30019	140	39	are	be	AUX
fcis-30019	140	40	carried	carry	VERB
fcis-30019	140	41	out	out	ADP
fcis-30019	140	42	for	for	ADP
fcis-30019	140	43	verification	verification	NOUN
fcis-30019	140	44	.	.	PUNCT
fcis-30019	141	1	firstly	firstly	ADV
fcis-30019	141	2	,	,	PUNCT
fcis-30019	141	3	the	the	DET
fcis-30019	141	4	baseline	baseline	NOUN
fcis-30019	141	5	models	model	NOUN
fcis-30019	141	6	mvit	mvit	ADJ
fcis-30019	141	7	and	and	CCONJ
fcis-30019	141	8	mvit_r	mvit_r	NOUN
fcis-30019	141	9	are	be	AUX
fcis-30019	141	10	used	use	VERB
fcis-30019	141	11	as	as	ADP
fcis-30019	141	12	the	the	DET
fcis-30019	141	13	baseline	baseline	NOUN
fcis-30019	141	14	,	,	PUNCT
fcis-30019	141	15	and	and	CCONJ
fcis-30019	141	16	then	then	ADV
fcis-30019	141	17	the	the	DET
fcis-30019	141	18	asff	asff	PROPN
fcis-30019	141	19	adaptive	adaptive	PROPN
fcis-30019	141	20	module	module	NOUN
fcis-30019	141	21	is	be	AUX
fcis-30019	141	22	introduced	introduce	VERB
fcis-30019	141	23	in	in	ADP
fcis-30019	141	24	the	the	DET
fcis-30019	141	25	spatial	spatial	ADJ
fcis-30019	141	26	domain	domain	NOUN
fcis-30019	141	27	and	and	CCONJ
fcis-30019	141	28	the	the	DET
fcis-30019	141	29	fecam	fecam	PROPN
fcis-30019	141	30	frequency	frequency	NOUN
fcis-30019	141	31	enhancement	enhancement	NOUN
fcis-30019	141	32	module	module	NOUN
fcis-30019	141	33	is	be	AUX
fcis-30019	141	34	introduced	introduce	VERB
fcis-30019	141	35	in	in	ADP
fcis-30019	141	36	the	the	DET
fcis-30019	141	37	frequency	frequency	NOUN
fcis-30019	141	38	domain	domain	NOUN
fcis-30019	141	39	to	to	PART
fcis-30019	141	40	discuss	discuss	VERB
fcis-30019	141	41	the	the	DET
fcis-30019	141	42	gain	gain	NOUN
fcis-30019	141	43	of	of	ADP
fcis-30019	141	44	each	each	DET
fcis-30019	141	45	part	part	NOUN
fcis-30019	141	46	on	on	ADP
fcis-30019	141	47	the	the	DET
fcis-30019	141	48	accuracy	accuracy	NOUN
fcis-30019	141	49	,	,	PUNCT
fcis-30019	141	50	and	and	CCONJ
fcis-30019	141	51	the	the	DET
fcis-30019	141	52	influence	influence	NOUN
fcis-30019	141	53	of	of	ADP
fcis-30019	141	54	different	different	ADJ
fcis-30019	141	55	frameworks	framework	NOUN
fcis-30019	141	56	is	be	AUX
fcis-30019	141	57	explored	explore	VERB
fcis-30019	141	58	.	.	PUNCT
fcis-30019	142	1	keep	keep	VERB
fcis-30019	142	2	the	the	DET
fcis-30019	142	3	parameters	parameter	NOUN
fcis-30019	142	4	and	and	CCONJ
fcis-30019	142	5	other	other	ADJ
fcis-30019	142	6	conditions	condition	NOUN
fcis-30019	142	7	consistent	consistent	ADJ
fcis-30019	142	8	for	for	ADP
fcis-30019	142	9	each	each	DET
fcis-30019	142	10	experiment	experiment	NOUN
fcis-30019	142	11	.	.	PUNCT
fcis-30019	143	1	the	the	DET
fcis-30019	143	2	model	model	NOUN
fcis-30019	143	3	was	be	AUX
fcis-30019	143	4	trained	train	VERB
fcis-30019	143	5	and	and	CCONJ
fcis-30019	143	6	tested	test	VERB
fcis-30019	143	7	on	on	ADP
fcis-30019	143	8	the	the	DET
fcis-30019	143	9	ff++	ff++	PROPN
fcis-30019	143	10	,	,	PUNCT
fcis-30019	143	11	celeb	celeb	NOUN
fcis-30019	143	12	-	-	PUNCT
fcis-30019	143	13	dfv2	dfv2	PROPN
fcis-30019	143	14	and	and	CCONJ
fcis-30019	143	15	biodeepav	biodeepav	VERB
fcis-30019	143	16	datasets	dataset	NOUN
fcis-30019	143	17	respectively	respectively	ADV
fcis-30019	143	18	,	,	PUNCT
fcis-30019	143	19	and	and	CCONJ
fcis-30019	143	20	acc	acc	PROPN
fcis-30019	143	21	(	(	PUNCT
fcis-30019	143	22	video	video	NOUN
fcis-30019	143	23	level	level	NOUN
fcis-30019	143	24	)	)	PUNCT
fcis-30019	143	25	was	be	AUX
fcis-30019	143	26	used	use	VERB
fcis-30019	143	27	as	as	ADP
fcis-30019	143	28	the	the	DET
fcis-30019	143	29	evaluation	evaluation	NOUN
fcis-30019	143	30	index	index	NOUN
fcis-30019	143	31	.	.	PUNCT
fcis-30019	144	1	the	the	DET
fcis-30019	144	2	final	final	ADJ
fcis-30019	144	3	experimental	experimental	ADJ
fcis-30019	144	4	results	result	NOUN
fcis-30019	144	5	are	be	AUX
fcis-30019	144	6	shown	show	VERB
fcis-30019	144	7	in	in	ADP
fcis-30019	144	8	the	the	DET
fcis-30019	144	9	table	table	NOUN
fcis-30019	144	10	.	.	PUNCT
fcis-30019	145	1	as	as	SCONJ
fcis-30019	145	2	can	can	AUX
fcis-30019	145	3	be	be	AUX
fcis-30019	145	4	seen	see	VERB
fcis-30019	145	5	from	from	ADP
fcis-30019	145	6	the	the	DET
fcis-30019	145	7	table	table	NOUN
fcis-30019	145	8	,	,	PUNCT
fcis-30019	145	9	the	the	DET
fcis-30019	145	10	accuracy	accuracy	NOUN
fcis-30019	145	11	of	of	ADP
fcis-30019	145	12	detection	detection	NOUN
fcis-30019	145	13	generally	generally	ADV
fcis-30019	145	14	keeps	keep	VERB
fcis-30019	145	15	increasing	increase	VERB
fcis-30019	145	16	with	with	ADP
fcis-30019	145	17	the	the	DET
fcis-30019	145	18	combination	combination	NOUN
fcis-30019	145	19	of	of	ADP
fcis-30019	145	20	modules	module	NOUN
fcis-30019	145	21	.	.	PUNCT
fcis-30019	146	1	whether	whether	SCONJ
fcis-30019	146	2	mvit	mvit	ADJ
fcis-30019	146	3	or	or	CCONJ
fcis-30019	146	4	mvit_r	mvit_r	NOUN
fcis-30019	146	5	is	be	AUX
fcis-30019	146	6	used	use	VERB
fcis-30019	146	7	as	as	ADP
fcis-30019	146	8	the	the	DET
fcis-30019	146	9	skeleton	skeleton	NOUN
fcis-30019	146	10	,	,	PUNCT
fcis-30019	146	11	their	their	PRON
fcis-30019	146	12	detection	detection	NOUN
fcis-30019	146	13	accuracy	accuracy	NOUN
fcis-30019	146	14	is	be	AUX
fcis-30019	146	15	generally	generally	ADV
fcis-30019	146	16	improved	improve	VERB
fcis-30019	146	17	after	after	ADP
fcis-30019	146	18	adding	add	VERB
fcis-30019	146	19	different	different	ADJ
fcis-30019	146	20	modules	module	NOUN
fcis-30019	146	21	.	.	PUNCT
fcis-30019	147	1	this	this	PRON
fcis-30019	147	2	indicates	indicate	VERB
fcis-30019	147	3	that	that	SCONJ
fcis-30019	147	4	the	the	DET
fcis-30019	147	5	module	module	NOUN
fcis-30019	147	6	of	of	ADP
fcis-30019	147	7	this	this	DET
fcis-30019	147	8	chapter	chapter	NOUN
fcis-30019	147	9	is	be	AUX
fcis-30019	147	10	effective	effective	ADJ
fcis-30019	147	11	for	for	ADP
fcis-30019	147	12	enhancing	enhance	VERB
fcis-30019	147	13	the	the	DET
fcis-30019	147	14	accuracy	accuracy	NOUN
fcis-30019	147	15	of	of	ADP
fcis-30019	147	16	forgery	forgery	NOUN
fcis-30019	147	17	detection	detection	NOUN
fcis-30019	147	18	.	.	PUNCT
fcis-30019	148	1	on	on	ADP
fcis-30019	148	2	the	the	DET
fcis-30019	148	3	mvit	mvit	ADJ
fcis-30019	148	4	architecture	architecture	NOUN
fcis-30019	148	5	,	,	PUNCT
fcis-30019	148	6	after	after	ADP
fcis-30019	148	7	adding	add	VERB
fcis-30019	148	8	the	the	DET
fcis-30019	148	9	asff	asff	NOUN
fcis-30019	148	10	module	module	NOUN
fcis-30019	148	11	,	,	PUNCT
fcis-30019	148	12	the	the	DET
fcis-30019	148	13	accuracy	accuracy	NOUN
fcis-30019	148	14	on	on	ADP
fcis-30019	148	15	deepfakes	deepfake	NOUN
fcis-30019	148	16	dataset	dataset	VERB
fcis-30019	148	17	is	be	AUX
fcis-30019	148	18	improved	improve	VERB
fcis-30019	148	19	from	from	ADP
fcis-30019	148	20	90.94	90.94	NUM
fcis-30019	148	21	%	%	NOUN
fcis-30019	148	22	to	to	ADP
fcis-30019	148	23	91.55	91.55	NUM
fcis-30019	148	24	%	%	NOUN
fcis-30019	148	25	,	,	PUNCT
fcis-30019	148	26	and	and	CCONJ
fcis-30019	148	27	faceswap	faceswap	NOUN
fcis-30019	148	28	is	be	AUX
fcis-30019	148	29	improved	improve	VERB
fcis-30019	148	30	from	from	ADP
fcis-30019	148	31	90.29	90.29	NUM
fcis-30019	148	32	%	%	NOUN
fcis-30019	148	33	to	to	ADP
fcis-30019	148	34	91.82	91.82	NUM
fcis-30019	148	35	%	%	NOUN
fcis-30019	148	36	.	.	PUNCT
fcis-30019	149	1	the	the	DET
fcis-30019	149	2	accuracy	accuracy	NOUN
fcis-30019	149	3	on	on	ADP
fcis-30019	149	4	celeb	celeb	NOUN
fcis-30019	149	5	-	-	PUNCT
fcis-30019	149	6	dfv2	dfv2	PROPN
fcis-30019	149	7	and	and	CCONJ
fcis-30019	149	8	biodeepav	biodeepav	VERB
fcis-30019	149	9	datasets	dataset	NOUN
fcis-30019	149	10	is	be	AUX
fcis-30019	149	11	improved	improve	VERB
fcis-30019	149	12	from	from	ADP
fcis-30019	149	13	92.19	92.19	NUM
fcis-30019	149	14	%	%	NOUN
fcis-30019	149	15	to	to	ADP
fcis-30019	149	16	95.98	95.98	NUM
fcis-30019	149	17	%	%	NOUN
fcis-30019	149	18	and	and	CCONJ
fcis-30019	149	19	75.92	75.92	NUM
fcis-30019	149	20	%	%	NOUN
fcis-30019	149	21	to	to	ADP
fcis-30019	149	22	77.73	77.73	NUM
fcis-30019	149	23	%	%	NOUN
fcis-30019	149	24	,	,	PUNCT
fcis-30019	149	25	respectively	respectively	ADV
fcis-30019	149	26	.	.	PUNCT
fcis-30019	150	1	when	when	SCONJ
fcis-30019	150	2	the	the	DET
fcis-30019	150	3	fecam	fecam	NOUN
fcis-30019	150	4	module	module	NOUN
fcis-30019	150	5	is	be	AUX
fcis-30019	150	6	added	add	VERB
fcis-30019	150	7	alone	alone	ADV
fcis-30019	150	8	,	,	PUNCT
fcis-30019	150	9	the	the	DET
fcis-30019	150	10	accuracy	accuracy	NOUN
fcis-30019	150	11	on	on	ADP
fcis-30019	150	12	deepfakes	deepfake	NOUN
fcis-30019	150	13	dataset	dataset	VERB
fcis-30019	150	14	is	be	AUX
fcis-30019	150	15	significantly	significantly	ADV
fcis-30019	150	16	improved	improve	VERB
fcis-30019	150	17	to	to	ADP
fcis-30019	150	18	95.46	95.46	NUM
fcis-30019	150	19	%	%	NOUN
fcis-30019	150	20	,	,	PUNCT
fcis-30019	150	21	faceswap	faceswap	PROPN
fcis-30019	150	22	is	be	AUX
fcis-30019	150	23	improved	improve	VERB
fcis-30019	150	24	from	from	ADP
fcis-30019	150	25	92.32	92.32	NUM
fcis-30019	150	26	%	%	NOUN
fcis-30019	150	27	to	to	ADP
fcis-30019	150	28	94.19	94.19	NUM
fcis-30019	150	29	%	%	NOUN
fcis-30019	150	30	,	,	PUNCT
fcis-30019	150	31	and	and	CCONJ
fcis-30019	150	32	the	the	DET
fcis-30019	150	33	accuracy	accuracy	NOUN
fcis-30019	150	34	on	on	ADP
fcis-30019	150	35	celeb	celeb	NOUN
fcis-30019	150	36	-	-	PUNCT
fcis-30019	150	37	dfv2	dfv2	PROPN
fcis-30019	150	38	and	and	CCONJ
fcis-30019	150	39	biodeepav	biodeepav	VERB
fcis-30019	150	40	datasets	dataset	NOUN
fcis-30019	150	41	is	be	AUX
fcis-30019	150	42	improved	improve	VERB
fcis-30019	150	43	to	to	ADP
fcis-30019	150	44	95.89	95.89	NUM
fcis-30019	150	45	%	%	NOUN
fcis-30019	150	46	and	and	CCONJ
fcis-30019	150	47	80.78	80.78	NUM
fcis-30019	150	48	%	%	NOUN
fcis-30019	150	49	,	,	PUNCT
fcis-30019	150	50	respectively	respectively	ADV
fcis-30019	150	51	.	.	PUNCT
fcis-30019	151	1	for	for	ADP
fcis-30019	151	2	59	59	NUM
fcis-30019	151	3	the	the	DET
fcis-30019	151	4	mvit_r	mvit_r	NOUN
fcis-30019	151	5	architecture	architecture	NOUN
fcis-30019	151	6	acting	act	VERB
fcis-30019	151	7	on	on	ADP
fcis-30019	151	8	modules	module	NOUN
fcis-30019	151	9	alone	alone	ADV
fcis-30019	151	10	,	,	PUNCT
fcis-30019	151	11	we	we	PRON
fcis-30019	151	12	can	can	AUX
fcis-30019	151	13	observe	observe	VERB
fcis-30019	151	14	a	a	DET
fcis-30019	151	15	similar	similar	ADJ
fcis-30019	151	16	trend	trend	NOUN
fcis-30019	151	17	.	.	PUNCT
fcis-30019	152	1	it	it	PRON
fcis-30019	152	2	is	be	AUX
fcis-30019	152	3	particularly	particularly	ADV
fcis-30019	152	4	worth	worth	ADJ
fcis-30019	152	5	noting	note	VERB
fcis-30019	152	6	that	that	SCONJ
fcis-30019	152	7	when	when	SCONJ
fcis-30019	152	8	the	the	DET
fcis-30019	152	9	mvit_r	mvit_r	NOUN
fcis-30019	152	10	skeleton	skeleton	NOUN
fcis-30019	152	11	combines	combine	VERB
fcis-30019	152	12	the	the	DET
fcis-30019	152	13	two	two	NUM
fcis-30019	152	14	modules	module	NOUN
fcis-30019	152	15	asff	asff	NOUN
fcis-30019	152	16	and	and	CCONJ
fcis-30019	152	17	fecam	fecam	PROPN
fcis-30019	152	18	,	,	PUNCT
fcis-30019	152	19	the	the	DET
fcis-30019	152	20	accuracy	accuracy	NOUN
fcis-30019	152	21	on	on	ADP
fcis-30019	152	22	deepfakes	deepfake	NOUN
fcis-30019	152	23	,	,	PUNCT
fcis-30019	152	24	faceswap	faceswap	NOUN
fcis-30019	152	25	and	and	CCONJ
fcis-30019	152	26	celeb	celeb	NOUN
fcis-30019	152	27	-	-	PUNCT
fcis-30019	152	28	dfv2	dfv2	PROPN
fcis-30019	152	29	datasets	dataset	NOUN
fcis-30019	152	30	reaches	reach	VERB
fcis-30019	152	31	the	the	DET
fcis-30019	152	32	highest	high	ADJ
fcis-30019	152	33	97.89	97.89	NUM
fcis-30019	152	34	%	%	NOUN
fcis-30019	152	35	,	,	PUNCT
fcis-30019	152	36	98.11	98.11	NUM
fcis-30019	152	37	%	%	NOUN
fcis-30019	152	38	and	and	CCONJ
fcis-30019	152	39	96.23	96.23	NUM
fcis-30019	152	40	%	%	NOUN
fcis-30019	152	41	.	.	PUNCT
fcis-30019	153	1	the	the	DET
fcis-30019	153	2	accuracy	accuracy	NOUN
fcis-30019	153	3	on	on	ADP
fcis-30019	153	4	biodeepav	biodeepav	PROPN
fcis-30019	153	5	dataset	dataset	PROPN
fcis-30019	153	6	reaches	reach	VERB
fcis-30019	153	7	83.89	83.89	NUM
fcis-30019	153	8	%	%	NOUN
fcis-30019	153	9	.	.	PUNCT
fcis-30019	154	1	this	this	PRON
fcis-30019	154	2	indicates	indicate	VERB
fcis-30019	154	3	that	that	SCONJ
fcis-30019	154	4	the	the	DET
fcis-30019	154	5	fusion	fusion	NOUN
fcis-30019	154	6	of	of	ADP
fcis-30019	154	7	multiple	multiple	ADJ
fcis-30019	154	8	modules	module	NOUN
fcis-30019	154	9	can	can	AUX
fcis-30019	154	10	produce	produce	VERB
fcis-30019	154	11	a	a	DET
fcis-30019	154	12	synergistic	synergistic	ADJ
fcis-30019	154	13	effect	effect	NOUN
fcis-30019	154	14	to	to	PART
fcis-30019	154	15	achieve	achieve	VERB
fcis-30019	154	16	better	well	ADJ
fcis-30019	154	17	detection	detection	NOUN
fcis-30019	154	18	performance	performance	NOUN
fcis-30019	154	19	.	.	PUNCT
fcis-30019	155	1	table	table	NOUN
fcis-30019	155	2	2	2	NUM
fcis-30019	155	3	.	.	PUNCT
fcis-30019	155	4	ablation	ablation	NOUN
fcis-30019	155	5	experiment	experiment	NOUN
fcis-30019	155	6	result	result	NOUN
fcis-30019	155	7	method	method	NOUN
fcis-30019	155	8	deepfakes	deepfake	VERB
fcis-30019	155	9	faceswap	faceswap	ADJ
fcis-30019	155	10	celeb	celeb	PROPN
fcis-30019	155	11	-	-	PUNCT
fcis-30019	155	12	dfv2	dfv2	PROPN
fcis-30019	155	13	biodeepav	biodeepav	PROPN
fcis-30019	155	14	mvit	mvit	PROPN
fcis-30019	155	15	mvit_r	mvit_r	PROPN
fcis-30019	155	16	asff	asff	PROPN
fcis-30019	155	17	fecam	fecam	PROPN
fcis-30019	155	18	acc	acc	PROPN
fcis-30019	155	19	acc	acc	PROPN
fcis-30019	155	20	acc	acc	PROPN
fcis-30019	155	21	acc	acc	PROPN
fcis-30019	156	1	√	√	ADV
fcis-30019	156	2	90.94	90.94	NUM
fcis-30019	156	3	%	%	NOUN
fcis-30019	156	4	90.29	90.29	NUM
fcis-30019	156	5	%	%	NOUN
fcis-30019	156	6	92.19	92.19	NUM
fcis-30019	156	7	%	%	NOUN
fcis-30019	156	8	75.92	75.92	NUM
fcis-30019	156	9	%	%	NOUN
fcis-30019	156	10	√	√	NOUN
fcis-30019	156	11	√	√	PROPN
fcis-30019	156	12	91.55	91.55	NUM
fcis-30019	156	13	%	%	NOUN
fcis-30019	156	14	92.32	92.32	NUM
fcis-30019	156	15	%	%	NOUN
fcis-30019	156	16	95.98	95.98	NUM
fcis-30019	156	17	%	%	NOUN
fcis-30019	156	18	77.73	77.73	NUM
fcis-30019	156	19	%	%	NOUN
fcis-30019	156	20	√	√	NOUN
fcis-30019	157	1	√	√	NUM
fcis-30019	157	2	95.46	95.46	NUM
fcis-30019	157	3	%	%	NOUN
fcis-30019	157	4	94.19	94.19	NUM
fcis-30019	157	5	%	%	NOUN
fcis-30019	157	6	95.89	95.89	NUM
fcis-30019	157	7	%	%	NOUN
fcis-30019	157	8	80.78	80.78	NUM
fcis-30019	157	9	%	%	NOUN
fcis-30019	157	10	√	√	NUM
fcis-30019	157	11	√	√	NUM
fcis-30019	158	1	√	√	NUM
fcis-30019	158	2	97.32	97.32	NUM
fcis-30019	158	3	%	%	NOUN
fcis-30019	158	4	96.88	96.88	NUM
fcis-30019	158	5	%	%	NOUN
fcis-30019	158	6	96.02	96.02	NUM
fcis-30019	158	7	%	%	NOUN
fcis-30019	158	8	84.63	84.63	NUM
fcis-30019	158	9	%	%	NOUN
fcis-30019	158	10	√	√	NOUN
fcis-30019	158	11	93.54	93.54	NUM
fcis-30019	158	12	%	%	NOUN
fcis-30019	158	13	93.86	93.86	NUM
fcis-30019	158	14	%	%	NOUN
fcis-30019	158	15	91.26	91.26	NUM
fcis-30019	158	16	%	%	NOUN
fcis-30019	158	17	77.05	77.05	NUM
fcis-30019	158	18	%	%	NOUN
fcis-30019	158	19	√	√	NOUN
fcis-30019	158	20	√	√	NUM
fcis-30019	158	21	91.25	91.25	NUM
fcis-30019	158	22	%	%	NOUN
fcis-30019	158	23	94.28	94.28	NUM
fcis-30019	158	24	%	%	NOUN
fcis-30019	158	25	94.23	94.23	NUM
fcis-30019	158	26	%	%	NOUN
fcis-30019	158	27	79.91	79.91	NUM
fcis-30019	158	28	%	%	NOUN
fcis-30019	158	29	√	√	NOUN
fcis-30019	159	1	√	√	NUM
fcis-30019	159	2	96.46	96.46	NUM
fcis-30019	159	3	%	%	NOUN
fcis-30019	159	4	96.48	96.48	NUM
fcis-30019	159	5	%	%	NOUN
fcis-30019	159	6	95.51	95.51	NUM
fcis-30019	159	7	%	%	NOUN
fcis-30019	159	8	78.59	78.59	NUM
fcis-30019	159	9	%	%	NOUN
fcis-30019	159	10	√	√	NUM
fcis-30019	159	11	√	√	NUM
fcis-30019	160	1	√	√	NUM
fcis-30019	160	2	97.89	97.89	NUM
fcis-30019	160	3	%	%	NOUN
fcis-30019	160	4	98.11	98.11	NUM
fcis-30019	160	5	%	%	NOUN
fcis-30019	160	6	96.23	96.23	NUM
fcis-30019	160	7	%	%	NOUN
fcis-30019	160	8	83.89	83.89	NUM
fcis-30019	160	9	%	%	NOUN
fcis-30019	160	10	the	the	DET
fcis-30019	160	11	results	result	NOUN
fcis-30019	160	12	show	show	VERB
fcis-30019	160	13	that	that	DET
fcis-30019	160	14	mvit	mvit	NOUN
fcis-30019	160	15	and	and	CCONJ
fcis-30019	160	16	mvit_r	mvit_r	NOUN
fcis-30019	160	17	,	,	PUNCT
fcis-30019	160	18	as	as	ADP
fcis-30019	160	19	two	two	NUM
fcis-30019	160	20	different	different	ADJ
fcis-30019	160	21	frameworks	framework	NOUN
fcis-30019	160	22	,	,	PUNCT
fcis-30019	160	23	can	can	AUX
fcis-30019	160	24	significantly	significantly	ADV
fcis-30019	160	25	improve	improve	VERB
fcis-30019	160	26	the	the	DET
fcis-30019	160	27	detection	detection	NOUN
fcis-30019	160	28	accuracy	accuracy	NOUN
fcis-30019	160	29	of	of	ADP
fcis-30019	160	30	deepfake	deepfake	NOUN
fcis-30019	160	31	videos	video	NOUN
fcis-30019	160	32	after	after	ADP
fcis-30019	160	33	combining	combine	VERB
fcis-30019	160	34	asff	asff	NOUN
fcis-30019	160	35	and	and	CCONJ
fcis-30019	160	36	fecam	fecam	PROPN
fcis-30019	160	37	modules	module	NOUN
fcis-30019	160	38	.	.	PUNCT
fcis-30019	161	1	4.4	4.4	NUM
fcis-30019	161	2	.	.	PUNCT
fcis-30019	162	1	cross	cross	NOUN
fcis-30019	162	2	compression	compression	NOUN
fcis-30019	162	3	experiment	experiment	NOUN
fcis-30019	162	4	most	most	ADJ
fcis-30019	162	5	of	of	ADP
fcis-30019	162	6	the	the	DET
fcis-30019	162	7	existing	exist	VERB
fcis-30019	162	8	deepfake	deepfake	NOUN
fcis-30019	162	9	detection	detection	NOUN
fcis-30019	162	10	models	model	NOUN
fcis-30019	162	11	are	be	AUX
fcis-30019	162	12	trained	train	VERB
fcis-30019	162	13	on	on	ADP
fcis-30019	162	14	finely	finely	ADV
fcis-30019	162	15	constructed	construct	VERB
fcis-30019	162	16	high	high	ADJ
fcis-30019	162	17	-	-	PUNCT
fcis-30019	162	18	quality	quality	NOUN
fcis-30019	162	19	datasets	dataset	NOUN
fcis-30019	162	20	.	.	PUNCT
fcis-30019	163	1	however	however	ADV
fcis-30019	163	2	,	,	PUNCT
fcis-30019	163	3	in	in	ADP
fcis-30019	163	4	practical	practical	ADJ
fcis-30019	163	5	applications	application	NOUN
fcis-30019	163	6	,	,	PUNCT
fcis-30019	163	7	fake	fake	ADJ
fcis-30019	163	8	face	face	NOUN
fcis-30019	163	9	videos	video	NOUN
fcis-30019	163	10	are	be	AUX
fcis-30019	163	11	often	often	ADV
fcis-30019	163	12	compressed	compress	VERB
fcis-30019	163	13	in	in	ADP
fcis-30019	163	14	the	the	DET
fcis-30019	163	15	process	process	NOUN
fcis-30019	163	16	of	of	ADP
fcis-30019	163	17	generation	generation	NOUN
fcis-30019	163	18	and	and	CCONJ
fcis-30019	163	19	dissemination	dissemination	NOUN
fcis-30019	163	20	on	on	ADP
fcis-30019	163	21	social	social	ADJ
fcis-30019	163	22	media	medium	NOUN
fcis-30019	163	23	.	.	PUNCT
fcis-30019	164	1	this	this	DET
fcis-30019	164	2	process	process	NOUN
fcis-30019	164	3	will	will	AUX
fcis-30019	164	4	lead	lead	VERB
fcis-30019	164	5	to	to	ADP
fcis-30019	164	6	the	the	DET
fcis-30019	164	7	loss	loss	NOUN
fcis-30019	164	8	of	of	ADP
fcis-30019	164	9	a	a	DET
fcis-30019	164	10	large	large	ADJ
fcis-30019	164	11	number	number	NOUN
fcis-30019	164	12	of	of	ADP
fcis-30019	164	13	details	detail	NOUN
fcis-30019	164	14	during	during	ADP
fcis-30019	164	15	the	the	DET
fcis-30019	164	16	coding	coding	NOUN
fcis-30019	164	17	and	and	CCONJ
fcis-30019	164	18	compression	compression	NOUN
fcis-30019	164	19	of	of	ADP
fcis-30019	164	20	the	the	DET
fcis-30019	164	21	image	image	NOUN
fcis-30019	164	22	,	,	PUNCT
fcis-30019	164	23	making	make	VERB
fcis-30019	164	24	the	the	DET
fcis-30019	164	25	forged	forge	VERB
fcis-30019	164	26	traces	trace	NOUN
fcis-30019	164	27	become	become	VERB
fcis-30019	164	28	unclear	unclear	ADJ
fcis-30019	164	29	.	.	PUNCT
fcis-30019	165	1	therefore	therefore	ADV
fcis-30019	165	2	,	,	PUNCT
fcis-30019	165	3	when	when	SCONJ
fcis-30019	165	4	these	these	DET
fcis-30019	165	5	models	model	NOUN
fcis-30019	165	6	trained	train	VERB
fcis-30019	165	7	on	on	ADP
fcis-30019	165	8	high	high	ADJ
fcis-30019	165	9	-	-	PUNCT
fcis-30019	165	10	quality	quality	NOUN
fcis-30019	165	11	datasets	dataset	NOUN
fcis-30019	165	12	are	be	AUX
fcis-30019	165	13	applied	apply	VERB
fcis-30019	165	14	to	to	PART
fcis-30019	165	15	detect	detect	VERB
fcis-30019	165	16	low	low	ADJ
fcis-30019	165	17	-	-	PUNCT
fcis-30019	165	18	quality	quality	NOUN
fcis-30019	165	19	data	datum	NOUN
fcis-30019	165	20	,	,	PUNCT
fcis-30019	165	21	their	their	PRON
fcis-30019	165	22	accuracy	accuracy	NOUN
fcis-30019	165	23	often	often	ADV
fcis-30019	165	24	degrades	degrade	VERB
fcis-30019	165	25	significantly	significantly	ADV
fcis-30019	165	26	.	.	PUNCT
fcis-30019	166	1	cross	cross	ADJ
fcis-30019	166	2	-	-	ADJ
fcis-30019	166	3	compression	compression	ADJ
fcis-30019	166	4	ratio	ratio	NOUN
fcis-30019	166	5	test	test	NOUN
fcis-30019	166	6	is	be	AUX
fcis-30019	166	7	a	a	DET
fcis-30019	166	8	challenge	challenge	NOUN
fcis-30019	166	9	often	often	ADV
fcis-30019	166	10	encountered	encounter	VERB
fcis-30019	166	11	in	in	ADP
fcis-30019	166	12	practical	practical	ADJ
fcis-30019	166	13	detection	detection	NOUN
fcis-30019	166	14	environments	environment	NOUN
fcis-30019	166	15	,	,	PUNCT
fcis-30019	166	16	and	and	CCONJ
fcis-30019	166	17	it	it	PRON
fcis-30019	166	18	is	be	AUX
fcis-30019	166	19	also	also	ADV
fcis-30019	166	20	a	a	DET
fcis-30019	166	21	key	key	ADJ
fcis-30019	166	22	indicator	indicator	NOUN
fcis-30019	166	23	to	to	PART
fcis-30019	166	24	test	test	VERB
fcis-30019	166	25	the	the	DET
fcis-30019	166	26	generalization	generalization	NOUN
fcis-30019	166	27	ability	ability	NOUN
fcis-30019	166	28	of	of	ADP
fcis-30019	166	29	the	the	DET
fcis-30019	166	30	algorithm	algorithm	NOUN
fcis-30019	166	31	.	.	PUNCT
fcis-30019	167	1	to	to	PART
fcis-30019	167	2	evaluate	evaluate	VERB
fcis-30019	167	3	the	the	DET
fcis-30019	167	4	performance	performance	NOUN
fcis-30019	167	5	of	of	ADP
fcis-30019	167	6	our	our	PRON
fcis-30019	167	7	model	model	NOUN
fcis-30019	167	8	in	in	ADP
fcis-30019	167	9	dealing	deal	VERB
fcis-30019	167	10	with	with	ADP
fcis-30019	167	11	compression	compression	NOUN
fcis-30019	167	12	,	,	PUNCT
fcis-30019	167	13	we	we	PRON
fcis-30019	167	14	conduct	conduct	VERB
fcis-30019	167	15	a	a	DET
fcis-30019	167	16	series	series	NOUN
fcis-30019	167	17	of	of	ADP
fcis-30019	167	18	crosscompression	crosscompression	NOUN
fcis-30019	167	19	experiments	experiment	NOUN
fcis-30019	167	20	and	and	CCONJ
fcis-30019	167	21	compare	compare	VERB
fcis-30019	167	22	it	it	PRON
fcis-30019	167	23	with	with	ADP
fcis-30019	167	24	the	the	DET
fcis-30019	167	25	state	state	NOUN
fcis-30019	167	26	of	of	ADP
fcis-30019	167	27	the	the	DET
fcis-30019	167	28	art	art	NOUN
fcis-30019	167	29	deepfake	deepfake	NOUN
fcis-30019	167	30	detection	detection	NOUN
fcis-30019	167	31	techniques	technique	NOUN
fcis-30019	167	32	.	.	PUNCT
fcis-30019	168	1	in	in	ADP
fcis-30019	168	2	terms	term	NOUN
fcis-30019	168	3	of	of	ADP
fcis-30019	168	4	experimental	experimental	ADJ
fcis-30019	168	5	data	datum	NOUN
fcis-30019	168	6	selection	selection	NOUN
fcis-30019	168	7	,	,	PUNCT
fcis-30019	168	8	we	we	PRON
fcis-30019	168	9	chose	choose	VERB
fcis-30019	168	10	to	to	PART
fcis-30019	168	11	train	train	VERB
fcis-30019	168	12	the	the	DET
fcis-30019	168	13	model	model	NOUN
fcis-30019	168	14	on	on	ADP
fcis-30019	168	15	the	the	DET
fcis-30019	168	16	high	high	ADJ
fcis-30019	168	17	-	-	PUNCT
fcis-30019	168	18	quality	quality	NOUN
fcis-30019	168	19	c23	c23	NOUN
fcis-30019	168	20	compression	compression	NOUN
fcis-30019	168	21	rate	rate	NOUN
fcis-30019	168	22	ff++	ff++	NOUN
fcis-30019	168	23	dataset	dataset	NOUN
fcis-30019	168	24	.	.	PUNCT
fcis-30019	169	1	the	the	DET
fcis-30019	169	2	trained	train	VERB
fcis-30019	169	3	model	model	NOUN
fcis-30019	169	4	is	be	AUX
fcis-30019	169	5	used	use	VERB
fcis-30019	169	6	to	to	PART
fcis-30019	169	7	test	test	VERB
fcis-30019	169	8	deepfakes	deepfake	NOUN
fcis-30019	169	9	(	(	PUNCT
fcis-30019	169	10	df	df	PROPN
fcis-30019	169	11	)	)	PUNCT
fcis-30019	169	12	,	,	PUNCT
fcis-30019	169	13	faceswap	faceswap	NOUN
fcis-30019	169	14	(	(	PUNCT
fcis-30019	169	15	fs	fs	PROPN
fcis-30019	169	16	)	)	PUNCT
fcis-30019	169	17	,	,	PUNCT
fcis-30019	169	18	face2face	face2face	NOUN
fcis-30019	169	19	(	(	PUNCT
fcis-30019	169	20	f2f	f2f	NOUN
fcis-30019	169	21	)	)	PUNCT
fcis-30019	169	22	and	and	CCONJ
fcis-30019	169	23	neuraltextures	neuraltexture	NOUN
fcis-30019	169	24	(	(	PUNCT
fcis-30019	169	25	nt	not	PART
fcis-30019	169	26	)	)	PUNCT
fcis-30019	169	27	on	on	ADP
fcis-30019	169	28	the	the	DET
fcis-30019	169	29	ff++	ff++	NOUN
fcis-30019	169	30	dataset	dataset	VERB
fcis-30019	169	31	with	with	ADP
fcis-30019	169	32	low	low	ADJ
fcis-30019	169	33	quality	quality	NOUN
fcis-30019	169	34	c40	c40	NOUN
fcis-30019	169	35	compression	compression	NOUN
fcis-30019	169	36	rate	rate	NOUN
fcis-30019	169	37	.	.	PUNCT
fcis-30019	170	1	in	in	ADP
fcis-30019	170	2	this	this	DET
fcis-30019	170	3	section	section	NOUN
fcis-30019	170	4	,	,	PUNCT
fcis-30019	170	5	the	the	DET
fcis-30019	170	6	video	video	NOUN
fcis-30019	170	7	-	-	PUNCT
fcis-30019	170	8	level	level	NOUN
fcis-30019	170	9	accuracy	accuracy	NOUN
fcis-30019	170	10	acc	acc	NOUN
fcis-30019	170	11	is	be	AUX
fcis-30019	170	12	used	use	VERB
fcis-30019	170	13	as	as	ADP
fcis-30019	170	14	the	the	DET
fcis-30019	170	15	standard	standard	NOUN
fcis-30019	170	16	to	to	PART
fcis-30019	170	17	measure	measure	VERB
fcis-30019	170	18	the	the	DET
fcis-30019	170	19	performance	performance	NOUN
fcis-30019	170	20	,	,	PUNCT
fcis-30019	170	21	and	and	CCONJ
fcis-30019	170	22	the	the	DET
fcis-30019	170	23	specific	specific	ADJ
fcis-30019	170	24	experimental	experimental	ADJ
fcis-30019	170	25	results	result	NOUN
fcis-30019	170	26	are	be	AUX
fcis-30019	170	27	shown	show	VERB
fcis-30019	170	28	in	in	ADP
fcis-30019	170	29	table	table	NOUN
fcis-30019	171	1	3	3	NUM
fcis-30019	171	2	.	.	PUNCT
fcis-30019	171	3	table	table	NOUN
fcis-30019	171	4	3	3	NUM
fcis-30019	171	5	.	.	PUNCT
fcis-30019	171	6	results	result	NOUN
fcis-30019	171	7	of	of	ADP
fcis-30019	171	8	the	the	DET
fcis-30019	171	9	cross	cross	NOUN
fcis-30019	171	10	-	-	ADJ
fcis-30019	171	11	compression	compression	NOUN
fcis-30019	171	12	experiment[20	experiment[20	NOUN
fcis-30019	171	13	]	]	X
fcis-30019	171	14	pretrain	pretrain	NOUN
fcis-30019	171	15	on	on	ADP
fcis-30019	171	16	ff++	ff++	PROPN
fcis-30019	171	17	c23	c23	PROPN
fcis-30019	171	18	model	model	PROPN
fcis-30019	171	19	df	df	PROPN
fcis-30019	171	20	(	(	PUNCT
fcis-30019	171	21	c40	c40	NOUN
fcis-30019	171	22	)	)	PUNCT
fcis-30019	171	23	fs(c40	fs(c40	NOUN
fcis-30019	171	24	)	)	PUNCT
fcis-30019	171	25	f2f(c40	f2f(c40	NOUN
fcis-30019	171	26	)	)	PUNCT
fcis-30019	171	27	nt(c40	nt(c40	NOUN
fcis-30019	171	28	)	)	PUNCT
fcis-30019	171	29	avg	avg	PROPN
fcis-30019	171	30	xception	xception	PROPN
fcis-30019	171	31	63.60	63.60	NUM
fcis-30019	171	32	%	%	NOUN
fcis-30019	171	33	55.83	55.83	NUM
fcis-30019	171	34	%	%	NOUN
fcis-30019	171	35	50.05	50.05	NUM
fcis-30019	171	36	%	%	NOUN
fcis-30019	171	37	55.22	55.22	NUM
fcis-30019	171	38	%	%	NOUN
fcis-30019	171	39	56.18	56.18	NUM
fcis-30019	171	40	%	%	NOUN
fcis-30019	171	41	nguyen	nguyen	NOUN
fcis-30019	171	42	67.46	67.46	NUM
fcis-30019	171	43	%	%	NOUN
fcis-30019	171	44	54.23	54.23	NUM
fcis-30019	171	45	%	%	NOUN
fcis-30019	171	46	55.90	55.90	NUM
fcis-30019	171	47	%	%	NOUN
fcis-30019	171	48	53.85	53.85	NUM
fcis-30019	171	49	%	%	NOUN
fcis-30019	171	50	57.86	57.86	NUM
fcis-30019	171	51	%	%	NOUN
fcis-30019	171	52	mesonet	mesonet	NOUN
fcis-30019	171	53	78.39	78.39	NUM
fcis-30019	171	54	%	%	NOUN
fcis-30019	171	55	59.89	59.89	NUM
fcis-30019	171	56	%	%	NOUN
fcis-30019	171	57	69.47	69.47	NUM
fcis-30019	171	58	%	%	NOUN
fcis-30019	171	59	53.94	53.94	NUM
fcis-30019	171	60	%	%	NOUN
fcis-30019	171	61	65.42	65.42	NUM
fcis-30019	171	62	%	%	NOUN
fcis-30019	171	63	cn	cn	PROPN
fcis-30019	171	64	-	-	PROPN
fcis-30019	171	65	rn	rn	PROPN
fcis-30019	171	66	71.97	71.97	NUM
fcis-30019	171	67	%	%	NOUN
fcis-30019	171	68	52.29	52.29	NUM
fcis-30019	171	69	%	%	NOUN
fcis-30019	171	70	50.26	50.26	NUM
fcis-30019	171	71	%	%	NOUN
fcis-30019	171	72	51.50	51.50	NUM
fcis-30019	171	73	%	%	NOUN
fcis-30019	171	74	56.51	56.51	NUM
fcis-30019	171	75	%	%	NOUN
fcis-30019	171	76	3d	3d	NOUN
fcis-30019	171	77	-	-	PUNCT
fcis-30019	171	78	cnn	cnn	PROPN
fcis-30019	171	79	79.99	79.99	NUM
fcis-30019	171	80	%	%	NOUN
fcis-30019	171	81	54.65	54.65	NUM
fcis-30019	171	82	%	%	NOUN
fcis-30019	171	83	69.69	69.69	NUM
fcis-30019	171	84	%	%	NOUN
fcis-30019	171	85	59.33	59.33	NUM
fcis-30019	171	86	%	%	NOUN
fcis-30019	171	87	65.92	65.92	NUM
fcis-30019	171	88	%	%	NOUN
fcis-30019	171	89	crossv	crossv	VERB
fcis-30019	171	90	63.50	63.50	NUM
fcis-30019	171	91	%	%	NOUN
fcis-30019	171	92	67.53	67.53	NUM
fcis-30019	171	93	%	%	NOUN
fcis-30019	171	94	50.85	50.85	NUM
fcis-30019	171	95	%	%	NOUN
fcis-30019	171	96	66.24	66.24	NUM
fcis-30019	171	97	%	%	NOUN
fcis-30019	171	98	62.03	62.03	NUM
fcis-30019	171	99	%	%	NOUN
fcis-30019	171	100	madd	madd	VERB
fcis-30019	171	101	88.96	88.96	NUM
fcis-30019	171	102	%	%	NOUN
fcis-30019	171	103	83.88	83.88	NUM
fcis-30019	171	104	%	%	NOUN
fcis-30019	171	105	62.04	62.04	NUM
fcis-30019	171	106	%	%	NOUN
fcis-30019	171	107	52.87	52.87	NUM
fcis-30019	171	108	%	%	NOUN
fcis-30019	171	109	71.94	71.94	NUM
fcis-30019	171	110	%	%	NOUN
fcis-30019	171	111	sfdt	sfdt	NOUN
fcis-30019	171	112	93.47	93.47	NUM
fcis-30019	171	113	%	%	NOUN
fcis-30019	171	114	86.01	86.01	NUM
fcis-30019	171	115	%	%	NOUN
fcis-30019	171	116	75.11	75.11	NUM
fcis-30019	171	117	%	%	NOUN
fcis-30019	171	118	63.68	63.68	NUM
fcis-30019	171	119	%	%	NOUN
fcis-30019	171	120	79.57	79.57	NUM
fcis-30019	171	121	%	%	NOUN
fcis-30019	171	122	according	accord	VERB
fcis-30019	171	123	to	to	ADP
fcis-30019	171	124	the	the	DET
fcis-30019	171	125	data	datum	NOUN
fcis-30019	171	126	in	in	ADP
fcis-30019	171	127	the	the	DET
fcis-30019	171	128	table	table	NOUN
fcis-30019	171	129	,	,	PUNCT
fcis-30019	171	130	the	the	DET
fcis-30019	171	131	sfdt	sfdt	NOUN
fcis-30019	171	132	method	method	NOUN
fcis-30019	171	133	has	have	AUX
fcis-30019	171	134	shown	show	VERB
fcis-30019	171	135	significant	significant	ADJ
fcis-30019	171	136	advantages	advantage	NOUN
fcis-30019	171	137	in	in	ADP
fcis-30019	171	138	cross	cross	NOUN
fcis-30019	171	139	-	-	NOUN
fcis-30019	171	140	compression	compression	NOUN
fcis-30019	171	141	(	(	PUNCT
fcis-30019	171	142	c23	c23	NOUN
fcis-30019	171	143	-	-	PUNCT
fcis-30019	171	144	c40	c40	NOUN
fcis-30019	171	145	)	)	PUNCT
fcis-30019	171	146	detection	detection	NOUN
fcis-30019	171	147	experiments	experiment	NOUN
fcis-30019	171	148	,	,	PUNCT
fcis-30019	171	149	and	and	CCONJ
fcis-30019	171	150	its	its	PRON
fcis-30019	171	151	accuracy	accuracy	NOUN
fcis-30019	171	152	has	have	AUX
fcis-30019	171	153	been	be	AUX
fcis-30019	171	154	greatly	greatly	ADV
fcis-30019	171	155	improved	improve	VERB
fcis-30019	171	156	in	in	ADP
fcis-30019	171	157	four	four	NUM
fcis-30019	171	158	different	different	ADJ
fcis-30019	171	159	ff++	ff++	NOUN
fcis-30019	171	160	datasets	dataset	NOUN
fcis-30019	171	161	.	.	PUNCT
fcis-30019	172	1	compared	compare	VERB
fcis-30019	172	2	with	with	ADP
fcis-30019	172	3	other	other	ADJ
fcis-30019	172	4	methods	method	NOUN
fcis-30019	172	5	in	in	ADP
fcis-30019	172	6	the	the	DET
fcis-30019	172	7	table	table	NOUN
fcis-30019	172	8	,	,	PUNCT
fcis-30019	172	9	sfdt	sfdt	PROPN
fcis-30019	172	10	achieves	achieve	VERB
fcis-30019	172	11	93.47	93.47	NUM
fcis-30019	172	12	%	%	NOUN
fcis-30019	172	13	,	,	PUNCT
fcis-30019	172	14	86.01	86.01	NUM
fcis-30019	172	15	%	%	NOUN
fcis-30019	172	16	,	,	PUNCT
fcis-30019	172	17	and	and	CCONJ
fcis-30019	172	18	75.11	75.11	NUM
fcis-30019	172	19	%	%	NOUN
fcis-30019	172	20	cross	cross	ADJ
fcis-30019	172	21	-	-	ADJ
fcis-30019	172	22	compression	compression	ADJ
fcis-30019	172	23	accuracy	accuracy	NOUN
fcis-30019	172	24	on	on	ADP
fcis-30019	172	25	deepfakes	deepfake	NOUN
fcis-30019	172	26	,	,	PUNCT
fcis-30019	172	27	faceswap	faceswap	NOUN
fcis-30019	172	28	,	,	PUNCT
fcis-30019	172	29	and	and	CCONJ
fcis-30019	172	30	face2face	face2face	VERB
fcis-30019	172	31	,	,	PUNCT
fcis-30019	172	32	respectively	respectively	ADV
fcis-30019	172	33	,	,	PUNCT
fcis-30019	172	34	which	which	PRON
fcis-30019	172	35	are	be	AUX
fcis-30019	172	36	the	the	DET
fcis-30019	172	37	highest	high	ADJ
fcis-30019	172	38	values	value	NOUN
fcis-30019	172	39	.	.	PUNCT
fcis-30019	173	1	compared	compare	VERB
fcis-30019	173	2	with	with	ADP
fcis-30019	173	3	the	the	DET
fcis-30019	173	4	second	second	ADV
fcis-30019	173	5	-	-	PUNCT
fcis-30019	173	6	best	good	ADJ
fcis-30019	173	7	method	method	NOUN
fcis-30019	173	8	madd	madd	PROPN
fcis-30019	173	9	,	,	PUNCT
fcis-30019	173	10	sfdt	sfdt	PROPN
fcis-30019	173	11	improves	improve	VERB
fcis-30019	173	12	the	the	DET
fcis-30019	173	13	first	first	ADJ
fcis-30019	173	14	three	three	NUM
fcis-30019	173	15	tasks	task	NOUN
fcis-30019	173	16	by	by	ADP
fcis-30019	173	17	4.51	4.51	NUM
fcis-30019	173	18	percentage	percentage	NOUN
fcis-30019	173	19	points	point	NOUN
fcis-30019	173	20	,	,	PUNCT
fcis-30019	173	21	2.13	2.13	NUM
fcis-30019	173	22	percentage	percentage	NOUN
fcis-30019	173	23	points	point	NOUN
fcis-30019	173	24	,	,	PUNCT
fcis-30019	173	25	and	and	CCONJ
fcis-30019	173	26	13.07	13.07	NUM
fcis-30019	173	27	percentage	percentage	NOUN
fcis-30019	173	28	points	point	NOUN
fcis-30019	173	29	,	,	PUNCT
fcis-30019	173	30	respectively	respectively	ADV
fcis-30019	173	31	.	.	PUNCT
fcis-30019	174	1	on	on	ADP
fcis-30019	174	2	average	average	ADJ
fcis-30019	174	3	,	,	PUNCT
fcis-30019	174	4	sfdt	sfdt	PROPN
fcis-30019	174	5	achieves	achieve	VERB
fcis-30019	174	6	an	an	DET
fcis-30019	174	7	average	average	ADJ
fcis-30019	174	8	improvement	improvement	NOUN
fcis-30019	174	9	of	of	ADP
fcis-30019	174	10	8.52	8.52	NUM
fcis-30019	174	11	percentage	percentage	NOUN
fcis-30019	174	12	points	point	NOUN
fcis-30019	174	13	on	on	ADP
fcis-30019	174	14	the	the	DET
fcis-30019	174	15	four	four	NUM
fcis-30019	174	16	tasks	task	NOUN
fcis-30019	174	17	.	.	PUNCT
fcis-30019	175	1	this	this	DET
fcis-30019	175	2	result	result	NOUN
fcis-30019	175	3	shows	show	VERB
fcis-30019	175	4	that	that	SCONJ
fcis-30019	175	5	sfdt	sfdt	NOUN
fcis-30019	175	6	has	have	VERB
fcis-30019	175	7	higher	high	ADJ
fcis-30019	175	8	cross	cross	ADJ
fcis-30019	175	9	-	-	ADJ
fcis-30019	175	10	compression	compression	ADJ
fcis-30019	175	11	detection	detection	NOUN
fcis-30019	175	12	performance	performance	NOUN
fcis-30019	175	13	when	when	SCONJ
fcis-30019	175	14	dealing	deal	VERB
fcis-30019	175	15	with	with	ADP
fcis-30019	175	16	different	different	ADJ
fcis-30019	175	17	types	type	NOUN
fcis-30019	175	18	of	of	ADP
fcis-30019	175	19	deepfake	deepfake	NOUN
fcis-30019	175	20	technologies	technology	NOUN
fcis-30019	175	21	,	,	PUNCT
fcis-30019	175	22	benefiting	benefit	VERB
fcis-30019	175	23	from	from	ADP
fcis-30019	175	24	the	the	DET
fcis-30019	175	25	blessing	blessing	NOUN
fcis-30019	175	26	of	of	ADP
fcis-30019	175	27	fusing	fuse	VERB
fcis-30019	175	28	spatial	spatial	ADJ
fcis-30019	175	29	-	-	PUNCT
fcis-30019	175	30	frequency	frequency	NOUN
fcis-30019	175	31	domain	domain	NOUN
fcis-30019	175	32	features	feature	NOUN
fcis-30019	175	33	.	.	PUNCT
fcis-30019	176	1	4.5	4.5	NUM
fcis-30019	176	2	.	.	PUNCT
fcis-30019	177	1	robustness	robustness	NOUN
fcis-30019	177	2	experiments	experiment	NOUN
fcis-30019	177	3	since	since	SCONJ
fcis-30019	177	4	deepfake	deepfake	NOUN
fcis-30019	177	5	videos	video	NOUN
fcis-30019	177	6	can	can	AUX
fcis-30019	177	7	be	be	AUX
fcis-30019	177	8	easily	easily	ADV
fcis-30019	177	9	modified	modify	VERB
fcis-30019	177	10	on	on	ADP
fcis-30019	177	11	various	various	ADJ
fcis-30019	177	12	media	medium	NOUN
fcis-30019	177	13	platforms	platform	NOUN
fcis-30019	177	14	,	,	PUNCT
fcis-30019	177	15	and	and	CCONJ
fcis-30019	177	16	deepfake	deepfake	NOUN
fcis-30019	177	17	detectors	detector	NOUN
fcis-30019	177	18	are	be	AUX
fcis-30019	177	19	highly	highly	ADV
fcis-30019	177	20	vulnerable	vulnerable	ADJ
fcis-30019	177	21	to	to	ADP
fcis-30019	177	22	interference	interference	VERB
fcis-30019	177	23	from	from	ADP
fcis-30019	177	24	adversarial	adversarial	ADJ
fcis-30019	177	25	attacks	attack	NOUN
fcis-30019	177	26	(	(	PUNCT
fcis-30019	177	27	deepfake	deepfake	VERB
fcis-30019	177	28	detection	detection	NOUN
fcis-30019	177	29	adversarial	adversarial	ADJ
fcis-30019	177	30	robustness	robustness	NOUN
fcis-30019	177	31	enhancement	enhancement	NOUN
fcis-30019	177	32	algorithm	algorithm	NOUN
fcis-30019	177	33	)	)	PUNCT
fcis-30019	177	34	,	,	PUNCT
fcis-30019	177	35	therefore	therefore	ADV
fcis-30019	177	36	,	,	PUNCT
fcis-30019	177	37	the	the	DET
fcis-30019	177	38	robustness	robustness	NOUN
fcis-30019	177	39	of	of	ADP
fcis-30019	177	40	the	the	DET
fcis-30019	177	41	model	model	NOUN
fcis-30019	177	42	is	be	AUX
fcis-30019	177	43	detected	detect	VERB
fcis-30019	177	44	by	by	ADP
fcis-30019	177	45	adding	add	VERB
fcis-30019	177	46	saturation	saturation	NOUN
fcis-30019	177	47	,	,	PUNCT
fcis-30019	177	48	contrast	contrast	NOUN
fcis-30019	177	49	,	,	PUNCT
fcis-30019	177	50	noise	noise	NOUN
fcis-30019	177	51	,	,	PUNCT
fcis-30019	177	52	blur	blur	NOUN
fcis-30019	177	53	,	,	PUNCT
fcis-30019	177	54	pixel	pixel	NOUN
fcis-30019	177	55	and	and	CCONJ
fcis-30019	177	56	so	so	ADV
fcis-30019	177	57	on	on	ADV
fcis-30019	177	58	to	to	ADP
fcis-30019	177	59	the	the	DET
fcis-30019	177	60	deepfakes	deepfake	NOUN
fcis-30019	177	61	fake	fake	ADJ
fcis-30019	177	62	video	video	NOUN
fcis-30019	177	63	data	datum	NOUN
fcis-30019	177	64	set	set	VERB
fcis-30019	177	65	in	in	ADP
fcis-30019	177	66	f++	f++	NOUN
fcis-30019	177	67	.	.	PUNCT
fcis-30019	178	1	these	these	DET
fcis-30019	178	2	jamming	jamming	NOUN
fcis-30019	178	3	methods	method	NOUN
fcis-30019	178	4	can	can	AUX
fcis-30019	178	5	effectively	effectively	ADV
fcis-30019	178	6	simulate	simulate	VERB
fcis-30019	178	7	the	the	DET
fcis-30019	178	8	natural	natural	ADJ
fcis-30019	178	9	interference	interference	NOUN
fcis-30019	178	10	or	or	CCONJ
fcis-30019	178	11	human	human	ADJ
fcis-30019	178	12	damage	damage	NOUN
fcis-30019	178	13	that	that	PRON
fcis-30019	178	14	face	face	VERB
fcis-30019	178	15	forgery	forgery	NOUN
fcis-30019	178	16	images	image	NOUN
fcis-30019	178	17	may	may	AUX
fcis-30019	178	18	encounter	encounter	VERB
fcis-30019	178	19	in	in	ADP
fcis-30019	178	20	real	real	ADJ
fcis-30019	178	21	detection	detection	NOUN
fcis-30019	178	22	scenarios	scenario	NOUN
fcis-30019	178	23	.	.	PUNCT
fcis-30019	179	1	the	the	DET
fcis-30019	179	2	corresponding	corresponding	ADJ
fcis-30019	179	3	interference	interference	NOUN
fcis-30019	179	4	visualization	visualization	NOUN
fcis-30019	179	5	used	use	VERB
fcis-30019	179	6	in	in	ADP
fcis-30019	179	7	the	the	DET
fcis-30019	179	8	experiment	experiment	NOUN
fcis-30019	179	9	is	be	AUX
fcis-30019	179	10	shown	show	VERB
fcis-30019	179	11	in	in	ADP
fcis-30019	179	12	the	the	DET
fcis-30019	179	13	figure	figure	NOUN
fcis-30019	179	14	:	:	PUNCT
fcis-30019	179	15	in	in	ADP
fcis-30019	179	16	order	order	NOUN
fcis-30019	179	17	to	to	PART
fcis-30019	179	18	deeply	deeply	ADV
fcis-30019	179	19	verify	verify	VERB
fcis-30019	179	20	the	the	DET
fcis-30019	179	21	advantages	advantage	NOUN
fcis-30019	179	22	of	of	ADP
fcis-30019	179	23	the	the	DET
fcis-30019	179	24	proposed	propose	VERB
fcis-30019	179	25	algorithm	algorithm	NOUN
fcis-30019	179	26	in	in	ADP
fcis-30019	179	27	this	this	DET
fcis-30019	179	28	chapter	chapter	NOUN
fcis-30019	179	29	in	in	ADP
fcis-30019	179	30	terms	term	NOUN
fcis-30019	179	31	of	of	ADP
fcis-30019	179	32	robustness	robustness	NOUN
fcis-30019	179	33	,	,	PUNCT
fcis-30019	179	34	this	this	DET
fcis-30019	179	35	section	section	NOUN
fcis-30019	179	36	designs	design	VERB
fcis-30019	179	37	a	a	DET
fcis-30019	179	38	comparison	comparison	NOUN
fcis-30019	179	39	experiment	experiment	NOUN
fcis-30019	179	40	with	with	ADP
fcis-30019	179	41	the	the	DET
fcis-30019	179	42	robustness	robustness	NOUN
fcis-30019	179	43	test	test	NOUN
fcis-30019	179	44	of	of	ADP
fcis-30019	179	45	mainstream	mainstream	ADJ
fcis-30019	179	46	algorithms	algorithm	NOUN
fcis-30019	179	47	.	.	PUNCT
fcis-30019	180	1	in	in	ADP
fcis-30019	180	2	view	view	NOUN
fcis-30019	180	3	of	of	ADP
fcis-30019	180	4	the	the	DET
fcis-30019	180	5	fact	fact	NOUN
fcis-30019	180	6	that	that	SCONJ
fcis-30019	180	7	previous	previous	ADJ
fcis-30019	180	8	deepfake	deepfake	NOUN
fcis-30019	180	9	research	research	NOUN
fcis-30019	180	10	has	have	AUX
fcis-30019	180	11	paid	pay	VERB
fcis-30019	180	12	relatively	relatively	ADV
fcis-30019	180	13	little	little	ADJ
fcis-30019	180	14	attention	attention	NOUN
fcis-30019	180	15	to	to	ADP
fcis-30019	180	16	robustness	robustness	NOUN
fcis-30019	180	17	testing	testing	NOUN
fcis-30019	180	18	and	and	CCONJ
fcis-30019	180	19	lacks	lack	VERB
fcis-30019	180	20	comparable	comparable	ADJ
fcis-30019	180	21	data	datum	NOUN
fcis-30019	180	22	,	,	PUNCT
fcis-30019	180	23	this	this	DET
fcis-30019	180	24	chapter	chapter	NOUN
fcis-30019	180	25	selected	select	VERB
fcis-30019	180	26	mainstream	mainstream	NOUN
fcis-30019	180	27	face	face	NOUN
fcis-30019	180	28	forgery	forgery	NOUN
fcis-30019	180	29	detection	detection	NOUN
fcis-30019	180	30	models	model	NOUN
fcis-30019	180	31	,	,	PUNCT
fcis-30019	180	32	including	include	VERB
fcis-30019	180	33	xception	xception	NOUN
fcis-30019	180	34	,	,	PUNCT
fcis-30019	180	35	facexray	facexray	NOUN
fcis-30019	180	36	,	,	PUNCT
fcis-30019	180	37	lipforensics	lipforensic	NOUN
fcis-30019	180	38	,	,	PUNCT
fcis-30019	180	39	caddm	caddm	NOUN
fcis-30019	180	40	,	,	PUNCT
fcis-30019	180	41	laa	laa	PROPN
fcis-30019	180	42	-	-	PUNCT
fcis-30019	180	43	net	net	NOUN
fcis-30019	180	44	,	,	PUNCT
fcis-30019	180	45	as	as	SCONJ
fcis-30019	180	46	the	the	DET
fcis-30019	180	47	comparison	comparison	NOUN
fcis-30019	180	48	object	object	VERB
fcis-30019	180	49	.	.	PUNCT
fcis-30019	181	1	these	these	DET
fcis-30019	181	2	models	model	NOUN
fcis-30019	181	3	are	be	AUX
fcis-30019	181	4	pre	pre	ADJ
fcis-30019	181	5	-	-	VERB
fcis-30019	181	6	trained	train	VERB
fcis-30019	181	7	on	on	ADP
fcis-30019	181	8	a	a	DET
fcis-30019	181	9	standard	standard	ADJ
fcis-30019	181	10	dataset	dataset	NOUN
fcis-30019	181	11	of	of	ADP
fcis-30019	181	12	deepfakes	deepfake	NOUN
fcis-30019	181	13	and	and	CCONJ
fcis-30019	181	14	subsequently	subsequently	ADV
fcis-30019	181	15	evaluated	evaluate	VERB
fcis-30019	181	16	for	for	ADP
fcis-30019	181	17	60	60	NUM
fcis-30019	181	18	performance	performance	NOUN
fcis-30019	181	19	on	on	ADP
fcis-30019	181	20	perturbed	perturb	VERB
fcis-30019	181	21	data	datum	NOUN
fcis-30019	181	22	.	.	PUNCT
fcis-30019	182	1	in	in	ADP
fcis-30019	182	2	this	this	DET
fcis-30019	182	3	chapter	chapter	NOUN
fcis-30019	182	4	,	,	PUNCT
fcis-30019	182	5	acc	acc	PROPN
fcis-30019	182	6	(	(	PUNCT
fcis-30019	182	7	video	video	NOUN
fcis-30019	182	8	level	level	NOUN
fcis-30019	182	9	)	)	PUNCT
fcis-30019	182	10	is	be	AUX
fcis-30019	182	11	used	use	VERB
fcis-30019	182	12	as	as	ADP
fcis-30019	182	13	the	the	DET
fcis-30019	182	14	evaluation	evaluation	NOUN
fcis-30019	182	15	index	index	NOUN
fcis-30019	182	16	in	in	ADP
fcis-30019	182	17	the	the	DET
fcis-30019	182	18	performance	performance	NOUN
fcis-30019	182	19	test	test	NOUN
fcis-30019	182	20	on	on	ADP
fcis-30019	182	21	the	the	DET
fcis-30019	182	22	standard	standard	ADJ
fcis-30019	182	23	data	datum	NOUN
fcis-30019	182	24	fig	fig	NOUN
fcis-30019	182	25	6	6	NUM
fcis-30019	182	26	.	.	PUNCT
fcis-30019	182	27	disturbance	disturbance	NOUN
fcis-30019	182	28	visualization	visualization	NOUN
fcis-30019	182	29	table	table	NOUN
fcis-30019	182	30	4	4	NUM
fcis-30019	182	31	.	.	PUNCT
fcis-30019	182	32	robustness	robustness	NOUN
fcis-30019	182	33	experiments	experiment	NOUN
fcis-30019	182	34	result[21	result[21	X
fcis-30019	182	35	]	]	PUNCT
fcis-30019	182	36	method	method	NOUN
fcis-30019	182	37	fake	fake	ADJ
fcis-30019	182	38	saturation	saturation	NOUN
fcis-30019	182	39	contrast	contrast	NOUN
fcis-30019	182	40	noise	noise	NOUN
fcis-30019	182	41	blur	blur	PROPN
fcis-30019	182	42	pixel	pixel	PROPN
fcis-30019	182	43	xception	xception	PROPN
fcis-30019	182	44	✓	✓	ADJ
fcis-30019	182	45	99.31	99.31	NUM
fcis-30019	182	46	%	%	NOUN
fcis-30019	182	47	98.62	98.62	NUM
fcis-30019	182	48	%	%	NOUN
fcis-30019	182	49	53.85	53.85	NUM
fcis-30019	182	50	%	%	NOUN
fcis-30019	182	51	60.22	60.22	NUM
fcis-30019	182	52	%	%	NOUN
fcis-30019	182	53	74.27	74.27	NUM
fcis-30019	182	54	%	%	NOUN
fcis-30019	182	55	facexray	facexray	NOUN
fcis-30019	182	56	✓	✓	ADJ
fcis-30019	182	57	97.62	97.62	NUM
fcis-30019	182	58	%	%	NOUN
fcis-30019	182	59	88.55	88.55	NUM
fcis-30019	182	60	%	%	NOUN
fcis-30019	182	61	49.83	49.83	NUM
fcis-30019	182	62	%	%	NOUN
fcis-30019	182	63	63.86	63.86	NUM
fcis-30019	182	64	%	%	NOUN
fcis-30019	182	65	88.62	88.62	NUM
fcis-30019	182	66	%	%	NOUN
fcis-30019	182	67	lipforensics	lipforensic	NOUN
fcis-30019	182	68	✓	✓	ADJ
fcis-30019	182	69	99.92	99.92	NUM
fcis-30019	182	70	%	%	NOUN
fcis-30019	182	71	99.61	99.61	NUM
fcis-30019	182	72	%	%	NOUN
fcis-30019	182	73	73.89	73.89	NUM
fcis-30019	182	74	%	%	NOUN
fcis-30019	182	75	96.14	96.14	NUM
fcis-30019	182	76	%	%	NOUN
fcis-30019	182	77	95.63	95.63	NUM
fcis-30019	182	78	%	%	NOUN
fcis-30019	182	79	caddm	caddm	NOUN
fcis-30019	182	80	✓	✓	ADJ
fcis-30019	182	81	99.63	99.63	NUM
fcis-30019	182	82	%	%	NOUN
fcis-30019	182	83	99.86	99.86	NUM
fcis-30019	182	84	%	%	NOUN
fcis-30019	182	85	87.41	87.41	NUM
fcis-30019	182	86	%	%	NOUN
fcis-30019	182	87	99.03	99.03	NUM
fcis-30019	182	88	%	%	NOUN
fcis-30019	182	89	98.86	98.86	NUM
fcis-30019	182	90	%	%	NOUN
fcis-30019	182	91	laa	laa	PROPN
fcis-30019	182	92	-	-	PUNCT
fcis-30019	182	93	net	net	ADJ
fcis-30019	182	94	✓	✓	ADJ
fcis-30019	182	95	99.95	99.95	NUM
fcis-30019	182	96	%	%	NOUN
fcis-30019	182	97	99.92	99.92	NUM
fcis-30019	182	98	%	%	NOUN
fcis-30019	182	99	53.96	53.96	NUM
fcis-30019	182	100	%	%	NOUN
fcis-30019	182	101	98.22	98.22	NUM
fcis-30019	182	102	%	%	NOUN
fcis-30019	182	103	99.81	99.81	NUM
fcis-30019	182	104	%	%	NOUN
fcis-30019	182	105	sfdt	sfdt	NOUN
fcis-30019	182	106	✓	✓	ADJ
fcis-30019	182	107	99.97	99.97	NUM
fcis-30019	182	108	%	%	NOUN
fcis-30019	182	109	99.87	99.87	NUM
fcis-30019	182	110	%	%	NOUN
fcis-30019	182	111	89.27	89.27	NUM
fcis-30019	182	112	%	%	NOUN
fcis-30019	182	113	98.48	98.48	NUM
fcis-30019	182	114	%	%	NOUN
fcis-30019	182	115	99.91	99.91	NUM
fcis-30019	182	116	%	%	NOUN
fcis-30019	182	117	compared	compare	VERB
fcis-30019	182	118	with	with	ADP
fcis-30019	182	119	other	other	ADJ
fcis-30019	182	120	algorithms	algorithm	NOUN
fcis-30019	182	121	,	,	PUNCT
fcis-30019	182	122	the	the	DET
fcis-30019	182	123	color	color	NOUN
fcis-30019	182	124	-	-	PUNCT
fcis-30019	182	125	related	relate	VERB
fcis-30019	182	126	changes	change	NOUN
fcis-30019	182	127	such	such	ADJ
fcis-30019	182	128	as	as	ADP
fcis-30019	182	129	saturation	saturation	NOUN
fcis-30019	182	130	and	and	CCONJ
fcis-30019	182	131	contrast	contrast	NOUN
fcis-30019	182	132	have	have	VERB
fcis-30019	182	133	little	little	ADJ
fcis-30019	182	134	impact	impact	NOUN
fcis-30019	182	135	on	on	ADP
fcis-30019	182	136	the	the	DET
fcis-30019	182	137	overall	overall	ADJ
fcis-30019	182	138	detection	detection	NOUN
fcis-30019	182	139	results	result	NOUN
fcis-30019	182	140	,	,	PUNCT
fcis-30019	182	141	and	and	CCONJ
fcis-30019	182	142	the	the	DET
fcis-30019	182	143	proposed	propose	VERB
fcis-30019	182	144	sfdt	sfdt	NOUN
fcis-30019	182	145	algorithm	algorithm	PROPN
fcis-30019	182	146	achieves	achieve	VERB
fcis-30019	182	147	99.9	99.9	NUM
fcis-30019	182	148	%	%	NOUN
fcis-30019	182	149	and	and	CCONJ
fcis-30019	182	150	99.8	99.8	NUM
fcis-30019	182	151	%	%	NOUN
fcis-30019	182	152	accuracy	accuracy	NOUN
fcis-30019	182	153	.	.	PUNCT
fcis-30019	183	1	compared	compare	VERB
fcis-30019	183	2	with	with	ADP
fcis-30019	183	3	caddm	caddm	ADJ
fcis-30019	183	4	method	method	NOUN
fcis-30019	183	5	,	,	PUNCT
fcis-30019	183	6	it	it	PRON
fcis-30019	183	7	improves	improve	VERB
fcis-30019	183	8	the	the	DET
fcis-30019	183	9	noise	noise	NOUN
fcis-30019	183	10	processing	processing	NOUN
fcis-30019	183	11	by	by	ADP
fcis-30019	183	12	1.8	1.8	NUM
fcis-30019	183	13	%	%	NOUN
fcis-30019	183	14	.	.	PUNCT
fcis-30019	184	1	it	it	PRON
fcis-30019	184	2	achieves	achieve	VERB
fcis-30019	184	3	98.4	98.4	NUM
fcis-30019	184	4	%	%	NOUN
fcis-30019	184	5	accuracy	accuracy	NOUN
fcis-30019	184	6	in	in	ADP
fcis-30019	184	7	fuzzy	fuzzy	ADJ
fcis-30019	184	8	processing	processing	NOUN
fcis-30019	184	9	,	,	PUNCT
fcis-30019	184	10	which	which	PRON
fcis-30019	184	11	still	still	ADV
fcis-30019	184	12	has	have	VERB
fcis-30019	184	13	room	room	NOUN
fcis-30019	184	14	for	for	ADP
fcis-30019	184	15	improvement	improvement	NOUN
fcis-30019	184	16	compared	compare	VERB
fcis-30019	184	17	with	with	ADP
fcis-30019	184	18	the	the	DET
fcis-30019	184	19	mainstream	mainstream	NOUN
fcis-30019	184	20	algorithms	algorithm	NOUN
fcis-30019	184	21	.	.	PUNCT
fcis-30019	185	1	the	the	DET
fcis-30019	185	2	pixel	pixel	PROPN
fcis-30019	185	3	perturbation	perturbation	NOUN
fcis-30019	185	4	has	have	VERB
fcis-30019	185	5	a	a	DET
fcis-30019	185	6	weak	weak	ADJ
fcis-30019	185	7	impact	impact	NOUN
fcis-30019	185	8	on	on	ADP
fcis-30019	185	9	the	the	DET
fcis-30019	185	10	detection	detection	NOUN
fcis-30019	185	11	as	as	ADP
fcis-30019	185	12	a	a	DET
fcis-30019	185	13	whole	whole	NOUN
fcis-30019	185	14	,	,	PUNCT
fcis-30019	185	15	and	and	CCONJ
fcis-30019	185	16	the	the	DET
fcis-30019	185	17	proposed	propose	VERB
fcis-30019	185	18	method	method	NOUN
fcis-30019	185	19	achieves	achieve	VERB
fcis-30019	185	20	an	an	DET
fcis-30019	185	21	accuracy	accuracy	NOUN
fcis-30019	185	22	of	of	ADP
fcis-30019	185	23	99.9	99.9	NUM
fcis-30019	185	24	%	%	NOUN
fcis-30019	185	25	,	,	PUNCT
fcis-30019	185	26	which	which	PRON
fcis-30019	185	27	proves	prove	VERB
fcis-30019	185	28	that	that	SCONJ
fcis-30019	185	29	the	the	DET
fcis-30019	185	30	method	method	NOUN
fcis-30019	185	31	in	in	ADP
fcis-30019	185	32	this	this	DET
fcis-30019	185	33	chapter	chapter	NOUN
fcis-30019	185	34	has	have	VERB
fcis-30019	185	35	certain	certain	ADJ
fcis-30019	185	36	advantages	advantage	NOUN
fcis-30019	185	37	in	in	ADP
fcis-30019	185	38	the	the	DET
fcis-30019	185	39	robustness	robustness	NOUN
fcis-30019	185	40	test	test	NOUN
fcis-30019	185	41	of	of	ADP
fcis-30019	185	42	forged	forge	VERB
fcis-30019	185	43	videos	video	NOUN
fcis-30019	185	44	.	.	PUNCT
fcis-30019	186	1	4.6	4.6	NUM
fcis-30019	186	2	.	.	PUNCT
fcis-30019	187	1	visual	visual	ADJ
fcis-30019	187	2	analysis	analysis	NOUN
fcis-30019	187	3	in	in	ADP
fcis-30019	187	4	order	order	NOUN
fcis-30019	187	5	to	to	PART
fcis-30019	187	6	visually	visually	ADV
fcis-30019	187	7	show	show	VERB
fcis-30019	187	8	how	how	SCONJ
fcis-30019	187	9	the	the	DET
fcis-30019	187	10	proposed	propose	VERB
fcis-30019	187	11	detection	detection	NOUN
fcis-30019	187	12	model	model	NOUN
fcis-30019	187	13	makes	make	VERB
fcis-30019	187	14	decisions	decision	NOUN
fcis-30019	187	15	,	,	PUNCT
fcis-30019	187	16	this	this	DET
fcis-30019	187	17	chapter	chapter	NOUN
fcis-30019	187	18	visualizes	visualize	VERB
fcis-30019	187	19	the	the	DET
fcis-30019	187	20	classification	classification	NOUN
fcis-30019	187	21	results	result	NOUN
fcis-30019	187	22	of	of	ADP
fcis-30019	187	23	the	the	DET
fcis-30019	187	24	input	input	NOUN
fcis-30019	187	25	face	face	NOUN
fcis-30019	187	26	image	image	NOUN
fcis-30019	187	27	by	by	ADP
fcis-30019	187	28	the	the	DET
fcis-30019	187	29	detection	detection	NOUN
fcis-30019	187	30	model	model	NOUN
fcis-30019	187	31	and	and	CCONJ
fcis-30019	187	32	the	the	DET
fcis-30019	187	33	final	final	ADJ
fcis-30019	187	34	feature	feature	NOUN
fcis-30019	187	35	vector	vector	NOUN
fcis-30019	187	36	input	input	NOUN
fcis-30019	187	37	to	to	ADP
fcis-30019	187	38	the	the	DET
fcis-30019	187	39	classifier	classifier	NOUN
fcis-30019	187	40	,	,	PUNCT
fcis-30019	187	41	revealing	reveal	VERB
fcis-30019	187	42	the	the	DET
fcis-30019	187	43	basis	basis	NOUN
fcis-30019	187	44	for	for	ADP
fcis-30019	187	45	the	the	DET
fcis-30019	187	46	detection	detection	NOUN
fcis-30019	187	47	model	model	NOUN
fcis-30019	187	48	to	to	PART
fcis-30019	187	49	determine	determine	VERB
fcis-30019	187	50	the	the	DET
fcis-30019	187	51	truth	truth	NOUN
fcis-30019	187	52	or	or	CCONJ
fcis-30019	187	53	falsehood	falsehood	NOUN
fcis-30019	187	54	.	.	PUNCT
fcis-30019	188	1	this	this	PRON
fcis-30019	188	2	is	be	AUX
fcis-30019	188	3	exactly	exactly	ADV
fcis-30019	188	4	the	the	DET
fcis-30019	188	5	face	face	NOUN
fcis-30019	188	6	region	region	NOUN
fcis-30019	188	7	that	that	SCONJ
fcis-30019	188	8	the	the	DET
fcis-30019	188	9	detection	detection	NOUN
fcis-30019	188	10	model	model	NOUN
fcis-30019	188	11	focuses	focus	VERB
fcis-30019	188	12	on	on	ADP
fcis-30019	188	13	when	when	SCONJ
fcis-30019	188	14	processing	process	VERB
fcis-30019	188	15	the	the	DET
fcis-30019	188	16	input	input	NOUN
fcis-30019	188	17	samples	sample	NOUN
fcis-30019	188	18	,	,	PUNCT
fcis-30019	188	19	and	and	CCONJ
fcis-30019	188	20	it	it	PRON
fcis-30019	188	21	is	be	AUX
fcis-30019	188	22	also	also	ADV
fcis-30019	188	23	the	the	DET
fcis-30019	188	24	region	region	NOUN
fcis-30019	188	25	that	that	PRON
fcis-30019	188	26	the	the	DET
fcis-30019	188	27	model	model	NOUN
fcis-30019	188	28	thinks	think	VERB
fcis-30019	188	29	may	may	AUX
fcis-30019	188	30	have	have	VERB
fcis-30019	188	31	deepfake	deepfake	VERB
fcis-30019	188	32	traces	trace	NOUN
fcis-30019	188	33	.	.	PUNCT
fcis-30019	189	1	in	in	ADP
fcis-30019	189	2	this	this	DET
fcis-30019	189	3	chapter	chapter	NOUN
fcis-30019	189	4	,	,	PUNCT
fcis-30019	189	5	the	the	DET
fcis-30019	189	6	model	model	NOUN
fcis-30019	189	7	visualization	visualization	NOUN
fcis-30019	189	8	analysis	analysis	NOUN
fcis-30019	189	9	is	be	AUX
fcis-30019	189	10	carried	carry	VERB
fcis-30019	189	11	out	out	ADP
fcis-30019	189	12	by	by	ADP
fcis-30019	189	13	grad	grad	ADJ
fcis-30019	189	14	cam[22	cam[22	PROPN
fcis-30019	189	15	]	]	PUNCT
fcis-30019	189	16	.	.	PUNCT
fcis-30019	190	1	celebdf	celebdf	NOUN
fcis-30019	190	2	-	-	PUNCT
fcis-30019	190	3	v2	v2	PROPN
fcis-30019	190	4	has	have	VERB
fcis-30019	190	5	superior	superior	ADJ
fcis-30019	190	6	forgery	forgery	NOUN
fcis-30019	190	7	methods	method	NOUN
fcis-30019	190	8	compared	compare	VERB
fcis-30019	190	9	with	with	ADP
fcis-30019	190	10	other	other	ADJ
fcis-30019	190	11	datasets	dataset	NOUN
fcis-30019	190	12	.	.	PUNCT
fcis-30019	191	1	it	it	PRON
fcis-30019	191	2	uses	use	VERB
fcis-30019	191	3	advanced	advanced	ADJ
fcis-30019	191	4	generative	generative	ADJ
fcis-30019	191	5	adversarial	adversarial	ADJ
fcis-30019	191	6	network	network	NOUN
fcis-30019	191	7	(	(	PUNCT
fcis-30019	191	8	gan	gan	NOUN
fcis-30019	191	9	)	)	PUNCT
fcis-30019	191	10	technology	technology	NOUN
fcis-30019	191	11	to	to	PART
fcis-30019	191	12	generate	generate	VERB
fcis-30019	191	13	high	high	ADJ
fcis-30019	191	14	-	-	PUNCT
fcis-30019	191	15	quality	quality	NOUN
fcis-30019	191	16	fake	fake	ADJ
fcis-30019	191	17	videos	video	NOUN
fcis-30019	191	18	covering	cover	VERB
fcis-30019	191	19	multiple	multiple	ADJ
fcis-30019	191	20	well	well	ADV
fcis-30019	191	21	-	-	PUNCT
fcis-30019	191	22	known	know	VERB
fcis-30019	191	23	public	public	ADJ
fcis-30019	191	24	figures	figure	NOUN
fcis-30019	191	25	,	,	PUNCT
fcis-30019	191	26	which	which	PRON
fcis-30019	191	27	ensures	ensure	VERB
fcis-30019	191	28	the	the	DET
fcis-30019	191	29	diversity	diversity	NOUN
fcis-30019	191	30	and	and	CCONJ
fcis-30019	191	31	representation	representation	NOUN
fcis-30019	191	32	of	of	ADP
fcis-30019	191	33	the	the	DET
fcis-30019	191	34	dataset	dataset	NOUN
fcis-30019	191	35	and	and	CCONJ
fcis-30019	191	36	is	be	AUX
fcis-30019	191	37	more	more	ADV
fcis-30019	191	38	difficult	difficult	ADJ
fcis-30019	191	39	to	to	PART
fcis-30019	191	40	be	be	AUX
fcis-30019	191	41	detected	detect	VERB
fcis-30019	191	42	by	by	ADP
fcis-30019	191	43	the	the	DET
fcis-30019	191	44	naked	naked	ADJ
fcis-30019	191	45	eyes	eye	NOUN
fcis-30019	191	46	.	.	PUNCT
fcis-30019	192	1	therefore	therefore	ADV
fcis-30019	192	2	,	,	PUNCT
fcis-30019	192	3	the	the	DET
fcis-30019	192	4	forged	forge	VERB
fcis-30019	192	5	data	datum	NOUN
fcis-30019	192	6	in	in	ADP
fcis-30019	192	7	the	the	DET
fcis-30019	192	8	celebdf	celebdf	NOUN
fcis-30019	192	9	-	-	PUNCT
fcis-30019	192	10	v2	v2	NOUN
fcis-30019	192	11	dataset	dataset	NOUN
fcis-30019	192	12	are	be	AUX
fcis-30019	192	13	selected	select	VERB
fcis-30019	192	14	for	for	ADP
fcis-30019	192	15	testing	testing	NOUN
fcis-30019	192	16	.	.	PUNCT
fcis-30019	193	1	by	by	ADP
fcis-30019	193	2	showing	show	VERB
fcis-30019	193	3	the	the	DET
fcis-30019	193	4	activation	activation	NOUN
fcis-30019	193	5	of	of	ADP
fcis-30019	193	6	different	different	ADJ
fcis-30019	193	7	channels	channel	NOUN
fcis-30019	193	8	downstream	downstream	ADV
fcis-30019	193	9	of	of	ADP
fcis-30019	193	10	the	the	DET
fcis-30019	193	11	model	model	NOUN
fcis-30019	193	12	,	,	PUNCT
fcis-30019	193	13	we	we	PRON
fcis-30019	193	14	can	can	AUX
fcis-30019	193	15	observe	observe	VERB
fcis-30019	193	16	that	that	SCONJ
fcis-30019	193	17	the	the	DET
fcis-30019	193	18	model	model	NOUN
fcis-30019	193	19	smooths	smooth	VERB
fcis-30019	193	20	the	the	DET
fcis-30019	193	21	skin	skin	NOUN
fcis-30019	193	22	texture	texture	NOUN
fcis-30019	193	23	,	,	PUNCT
fcis-30019	193	24	and	and	CCONJ
fcis-30019	193	25	pays	pay	VERB
fcis-30019	193	26	more	more	ADJ
fcis-30019	193	27	attention	attention	NOUN
fcis-30019	193	28	to	to	ADP
fcis-30019	193	29	the	the	DET
fcis-30019	193	30	key	key	ADJ
fcis-30019	193	31	edge	edge	NOUN
fcis-30019	193	32	parts	part	NOUN
fcis-30019	193	33	of	of	ADP
fcis-30019	193	34	the	the	DET
fcis-30019	193	35	face	face	NOUN
fcis-30019	193	36	downstream	downstream	ADJ
fcis-30019	193	37	,	,	PUNCT
fcis-30019	193	38	revealing	reveal	VERB
fcis-30019	193	39	possible	possible	ADJ
fcis-30019	193	40	regions	region	NOUN
fcis-30019	193	41	of	of	ADP
fcis-30019	193	42	forgery	forgery	NOUN
fcis-30019	193	43	traces	trace	NOUN
fcis-30019	193	44	.	.	PUNCT
fcis-30019	194	1	as	as	SCONJ
fcis-30019	194	2	shown	show	VERB
fcis-30019	194	3	in	in	ADP
fcis-30019	194	4	figure	figure	NOUN
fcis-30019	194	5	7	7	NUM
fcis-30019	194	6	:	:	PUNCT
fcis-30019	194	7	fig	fig	NOUN
fcis-30019	194	8	7	7	NUM
fcis-30019	194	9	.	.	PUNCT
fcis-30019	194	10	activation	activation	NOUN
fcis-30019	194	11	map	map	NOUN
fcis-30019	194	12	visualization	visualization	NOUN
fcis-30019	194	13	downstream	downstream	NOUN
fcis-30019	194	14	of	of	ADP
fcis-30019	194	15	the	the	DET
fcis-30019	194	16	model	model	NOUN
fcis-30019	194	17	61	61	NUM
fcis-30019	194	18	grad	grad	ADJ
fcis-30019	194	19	cam	cam	NOUN
fcis-30019	194	20	is	be	AUX
fcis-30019	194	21	usually	usually	ADV
fcis-30019	194	22	used	use	VERB
fcis-30019	194	23	to	to	PART
fcis-30019	194	24	explain	explain	VERB
fcis-30019	194	25	the	the	DET
fcis-30019	194	26	partial	partial	ADJ
fcis-30019	194	27	feature	feature	NOUN
fcis-30019	194	28	contribution	contribution	NOUN
fcis-30019	194	29	of	of	ADP
fcis-30019	194	30	a	a	DET
fcis-30019	194	31	single	single	ADJ
fcis-30019	194	32	instance	instance	NOUN
fcis-30019	194	33	,	,	PUNCT
fcis-30019	194	34	and	and	CCONJ
fcis-30019	194	35	it	it	PRON
fcis-30019	194	36	can	can	AUX
fcis-30019	194	37	show	show	VERB
fcis-30019	194	38	the	the	DET
fcis-30019	194	39	contribution	contribution	NOUN
fcis-30019	194	40	of	of	ADP
fcis-30019	194	41	the	the	DET
fcis-30019	194	42	features	feature	NOUN
fcis-30019	194	43	in	in	ADP
fcis-30019	194	44	the	the	DET
fcis-30019	194	45	space	space	NOUN
fcis-30019	194	46	to	to	ADP
fcis-30019	194	47	the	the	DET
fcis-30019	194	48	prediction	prediction	NOUN
fcis-30019	194	49	of	of	ADP
fcis-30019	194	50	the	the	DET
fcis-30019	194	51	current	current	ADJ
fcis-30019	194	52	sample	sample	NOUN
fcis-30019	194	53	through	through	ADP
fcis-30019	194	54	the	the	DET
fcis-30019	194	55	heat	heat	NOUN
fcis-30019	194	56	map	map	NOUN
fcis-30019	194	57	.	.	PUNCT
fcis-30019	195	1	with	with	ADP
fcis-30019	195	2	this	this	DET
fcis-30019	195	3	partial	partial	ADJ
fcis-30019	195	4	visualization	visualization	NOUN
fcis-30019	195	5	,	,	PUNCT
fcis-30019	195	6	it	it	PRON
fcis-30019	195	7	's	be	AUX
fcis-30019	195	8	possible	possible	ADJ
fcis-30019	195	9	to	to	PART
fcis-30019	195	10	see	see	VERB
fcis-30019	195	11	at	at	ADP
fcis-30019	195	12	a	a	DET
fcis-30019	195	13	glance	glance	NOUN
fcis-30019	195	14	which	which	PRON
fcis-30019	195	15	features	feature	VERB
fcis-30019	195	16	play	play	VERB
fcis-30019	195	17	a	a	DET
fcis-30019	195	18	key	key	ADJ
fcis-30019	195	19	role	role	NOUN
fcis-30019	195	20	in	in	ADP
fcis-30019	195	21	individual	individual	ADJ
fcis-30019	195	22	predictions	prediction	NOUN
fcis-30019	195	23	.	.	PUNCT
fcis-30019	196	1	fig	fig	NOUN
fcis-30019	196	2	8	8	NUM
fcis-30019	196	3	.	.	PUNCT
fcis-30019	197	1	falsified	falsified	ADJ
fcis-30019	197	2	data	datum	NOUN
fcis-30019	197	3	with	with	ADP
fcis-30019	197	4	sfdt	sfdt	NOUN
fcis-30019	197	5	area	area	NOUN
fcis-30019	197	6	of	of	ADP
fcis-30019	197	7	concern	concern	NOUN
fcis-30019	197	8	visualization	visualization	NOUN
fcis-30019	197	9	as	as	SCONJ
fcis-30019	197	10	can	can	AUX
fcis-30019	197	11	be	be	AUX
fcis-30019	197	12	observed	observe	VERB
fcis-30019	197	13	from	from	ADP
fcis-30019	197	14	the	the	DET
fcis-30019	197	15	figure	figure	NOUN
fcis-30019	197	16	,	,	PUNCT
fcis-30019	197	17	for	for	ADP
fcis-30019	197	18	different	different	ADJ
fcis-30019	197	19	samples	sample	NOUN
fcis-30019	197	20	,	,	PUNCT
fcis-30019	197	21	the	the	DET
fcis-30019	197	22	sfdt	sfdt	NOUN
fcis-30019	197	23	detection	detection	NOUN
fcis-30019	197	24	model	model	NOUN
fcis-30019	197	25	proposed	propose	VERB
fcis-30019	197	26	in	in	ADP
fcis-30019	197	27	this	this	DET
fcis-30019	197	28	chapter	chapter	NOUN
fcis-30019	197	29	accurately	accurately	ADV
fcis-30019	197	30	focuses	focus	VERB
fcis-30019	197	31	on	on	ADP
fcis-30019	197	32	the	the	DET
fcis-30019	197	33	fake	fake	ADJ
fcis-30019	197	34	facial	facial	ADJ
fcis-30019	197	35	regions	region	NOUN
fcis-30019	197	36	,	,	PUNCT
fcis-30019	197	37	such	such	ADJ
fcis-30019	197	38	as	as	ADP
fcis-30019	197	39	the	the	DET
fcis-30019	197	40	lower	low	ADJ
fcis-30019	197	41	jaw	jaw	NOUN
fcis-30019	197	42	,	,	PUNCT
fcis-30019	197	43	the	the	DET
fcis-30019	197	44	forehead	forehead	NOUN
fcis-30019	197	45	,	,	PUNCT
fcis-30019	197	46	the	the	DET
fcis-30019	197	47	nose	nose	NOUN
fcis-30019	197	48	,	,	PUNCT
fcis-30019	197	49	the	the	DET
fcis-30019	197	50	border	border	NOUN
fcis-30019	197	51	after	after	ADP
fcis-30019	197	52	the	the	DET
fcis-30019	197	53	face	face	NOUN
fcis-30019	197	54	change	change	NOUN
fcis-30019	197	55	,	,	PUNCT
fcis-30019	197	56	etc	etc	X
fcis-30019	197	57	.	.	X
fcis-30019	198	1	it	it	PRON
fcis-30019	198	2	is	be	AUX
fcis-30019	198	3	easy	easy	ADJ
fcis-30019	198	4	to	to	PART
fcis-30019	198	5	see	see	VERB
fcis-30019	198	6	that	that	SCONJ
fcis-30019	198	7	the	the	DET
fcis-30019	198	8	forgery	forgery	NOUN
fcis-30019	198	9	regions	region	NOUN
fcis-30019	198	10	focused	focus	VERB
fcis-30019	198	11	on	on	ADP
fcis-30019	198	12	by	by	ADP
fcis-30019	198	13	the	the	DET
fcis-30019	198	14	proposed	propose	VERB
fcis-30019	198	15	model	model	NOUN
fcis-30019	198	16	in	in	ADP
fcis-30019	198	17	deepfake	deepfake	NOUN
fcis-30019	198	18	detection	detection	NOUN
fcis-30019	198	19	are	be	AUX
fcis-30019	198	20	more	more	ADV
fcis-30019	198	21	detailed	detailed	ADJ
fcis-30019	198	22	and	and	CCONJ
fcis-30019	198	23	accurate	accurate	ADJ
fcis-30019	198	24	,	,	PUNCT
fcis-30019	198	25	and	and	CCONJ
fcis-30019	198	26	can	can	AUX
fcis-30019	198	27	almost	almost	ADV
fcis-30019	198	28	perfectly	perfectly	ADV
fcis-30019	198	29	correspond	correspond	VERB
fcis-30019	198	30	to	to	ADP
fcis-30019	198	31	the	the	DET
fcis-30019	198	32	information	information	NOUN
fcis-30019	198	33	abnormal	abnormal	ADJ
fcis-30019	198	34	regions	region	NOUN
fcis-30019	198	35	in	in	ADP
fcis-30019	198	36	the	the	DET
fcis-30019	198	37	input	input	NOUN
fcis-30019	198	38	samples	sample	NOUN
fcis-30019	198	39	.	.	PUNCT
fcis-30019	199	1	the	the	DET
fcis-30019	199	2	visual	visual	ADJ
fcis-30019	199	3	experimental	experimental	ADJ
fcis-30019	199	4	results	result	NOUN
fcis-30019	199	5	fully	fully	ADV
fcis-30019	199	6	prove	prove	VERB
fcis-30019	199	7	the	the	DET
fcis-30019	199	8	reliability	reliability	NOUN
fcis-30019	199	9	and	and	CCONJ
fcis-30019	199	10	persuadability	persuadability	NOUN
fcis-30019	199	11	of	of	ADP
fcis-30019	199	12	the	the	DET
fcis-30019	199	13	detection	detection	NOUN
fcis-30019	199	14	model	model	NOUN
fcis-30019	199	15	.	.	PUNCT
fcis-30019	200	1	5	5	X
fcis-30019	200	2	.	.	X
fcis-30019	200	3	conclusion	conclusion	NOUN
fcis-30019	200	4	in	in	ADP
fcis-30019	200	5	this	this	DET
fcis-30019	200	6	paper	paper	NOUN
fcis-30019	200	7	,	,	PUNCT
fcis-30019	200	8	to	to	PART
fcis-30019	200	9	enhance	enhance	VERB
fcis-30019	200	10	the	the	DET
fcis-30019	200	11	robustness	robustness	NOUN
fcis-30019	200	12	of	of	ADP
fcis-30019	200	13	deepfake	deepfake	NOUN
fcis-30019	200	14	detection	detection	NOUN
fcis-30019	200	15	model	model	NOUN
fcis-30019	200	16	,	,	PUNCT
fcis-30019	200	17	this	this	DET
fcis-30019	200	18	paper	paper	NOUN
fcis-30019	200	19	proposes	propose	VERB
fcis-30019	200	20	a	a	DET
fcis-30019	200	21	detection	detection	NOUN
fcis-30019	200	22	algorithm	algorithm	NOUN
fcis-30019	200	23	that	that	PRON
fcis-30019	200	24	fuses	fuse	VERB
fcis-30019	200	25	spatial	spatial	ADJ
fcis-30019	200	26	-	-	PUNCT
fcis-30019	200	27	frequency	frequency	NOUN
fcis-30019	200	28	domain	domain	NOUN
fcis-30019	200	29	features	feature	NOUN
fcis-30019	200	30	.	.	PUNCT
fcis-30019	201	1	the	the	DET
fcis-30019	201	2	algorithm	algorithm	NOUN
fcis-30019	201	3	combines	combine	VERB
fcis-30019	201	4	spatial	spatial	ADJ
fcis-30019	201	5	domain	domain	NOUN
fcis-30019	201	6	and	and	CCONJ
fcis-30019	201	7	frequency	frequency	NOUN
fcis-30019	201	8	domain	domain	NOUN
fcis-30019	201	9	information	information	NOUN
fcis-30019	201	10	to	to	PART
fcis-30019	201	11	improve	improve	VERB
fcis-30019	201	12	the	the	DET
fcis-30019	201	13	detection	detection	NOUN
fcis-30019	201	14	effect	effect	NOUN
fcis-30019	201	15	of	of	ADP
fcis-30019	201	16	cross	cross	NOUN
fcis-30019	201	17	-	-	NOUN
fcis-30019	201	18	compression	compression	NOUN
fcis-30019	201	19	and	and	CCONJ
fcis-30019	201	20	antiinterference	antiinterference	NOUN
fcis-30019	201	21	.	.	PUNCT
fcis-30019	202	1	firstly	firstly	ADV
fcis-30019	202	2	,	,	PUNCT
fcis-30019	202	3	this	this	DET
fcis-30019	202	4	paper	paper	NOUN
fcis-30019	202	5	expounds	expound	VERB
fcis-30019	202	6	the	the	DET
fcis-30019	202	7	research	research	NOUN
fcis-30019	202	8	motivation	motivation	NOUN
fcis-30019	202	9	of	of	ADP
fcis-30019	202	10	the	the	DET
fcis-30019	202	11	algorithm	algorithm	NOUN
fcis-30019	202	12	,	,	PUNCT
fcis-30019	202	13	and	and	CCONJ
fcis-30019	202	14	describes	describe	VERB
fcis-30019	202	15	the	the	DET
fcis-30019	202	16	overall	overall	ADJ
fcis-30019	202	17	process	process	NOUN
fcis-30019	202	18	of	of	ADP
fcis-30019	202	19	the	the	DET
fcis-30019	202	20	model	model	NOUN
fcis-30019	202	21	,	,	PUNCT
fcis-30019	202	22	the	the	DET
fcis-30019	202	23	spatial	spatial	ADJ
fcis-30019	202	24	domain	domain	NOUN
fcis-30019	202	25	feature	feature	NOUN
fcis-30019	202	26	extraction	extraction	NOUN
fcis-30019	202	27	part	part	NOUN
fcis-30019	202	28	,	,	PUNCT
fcis-30019	202	29	the	the	DET
fcis-30019	202	30	frequency	frequency	NOUN
fcis-30019	202	31	domain	domain	NOUN
fcis-30019	202	32	information	information	NOUN
fcis-30019	202	33	extraction	extraction	NOUN
fcis-30019	202	34	part	part	NOUN
fcis-30019	202	35	,	,	PUNCT
fcis-30019	202	36	and	and	CCONJ
fcis-30019	202	37	the	the	DET
fcis-30019	202	38	cmf	cmf	PROPN
fcis-30019	202	39	feature	feature	NOUN
fcis-30019	202	40	fusion	fusion	NOUN
fcis-30019	202	41	module	module	NOUN
fcis-30019	202	42	in	in	ADP
fcis-30019	202	43	detail	detail	NOUN
fcis-30019	202	44	in	in	ADP
fcis-30019	202	45	the	the	DET
fcis-30019	202	46	overall	overall	ADJ
fcis-30019	202	47	algorithm	algorithm	NOUN
fcis-30019	202	48	architecture	architecture	NOUN
fcis-30019	202	49	.	.	PUNCT
fcis-30019	203	1	through	through	ADP
fcis-30019	203	2	a	a	DET
fcis-30019	203	3	series	series	NOUN
fcis-30019	203	4	of	of	ADP
fcis-30019	203	5	experiments	experiment	NOUN
fcis-30019	203	6	to	to	PART
fcis-30019	203	7	verify	verify	VERB
fcis-30019	203	8	the	the	DET
fcis-30019	203	9	performance	performance	NOUN
fcis-30019	203	10	of	of	ADP
fcis-30019	203	11	the	the	DET
fcis-30019	203	12	algorithm	algorithm	NOUN
fcis-30019	203	13	in	in	ADP
fcis-30019	203	14	this	this	DET
fcis-30019	203	15	chapter	chapter	NOUN
fcis-30019	203	16	,	,	PUNCT
fcis-30019	203	17	the	the	DET
fcis-30019	203	18	influence	influence	NOUN
fcis-30019	203	19	of	of	ADP
fcis-30019	203	20	each	each	DET
fcis-30019	203	21	module	module	NOUN
fcis-30019	203	22	is	be	AUX
fcis-30019	203	23	analyzed	analyze	VERB
fcis-30019	203	24	through	through	ADP
fcis-30019	203	25	ablation	ablation	NOUN
fcis-30019	203	26	experiments	experiment	NOUN
fcis-30019	203	27	,	,	PUNCT
fcis-30019	203	28	the	the	DET
fcis-30019	203	29	transferability	transferability	NOUN
fcis-30019	203	30	evaluation	evaluation	NOUN
fcis-30019	203	31	on	on	ADP
fcis-30019	203	32	cross	cross	ADJ
fcis-30019	203	33	-	-	ADJ
fcis-30019	203	34	data	data	ADJ
fcis-30019	203	35	sets	set	NOUN
fcis-30019	203	36	,	,	PUNCT
fcis-30019	203	37	the	the	DET
fcis-30019	203	38	crosscompression	crosscompression	NOUN
fcis-30019	203	39	ability	ability	NOUN
fcis-30019	203	40	test	test	NOUN
fcis-30019	203	41	under	under	ADP
fcis-30019	203	42	different	different	ADJ
fcis-30019	203	43	compression	compression	NOUN
fcis-30019	203	44	rates	rate	NOUN
fcis-30019	203	45	,	,	PUNCT
fcis-30019	203	46	the	the	DET
fcis-30019	203	47	model	model	NOUN
fcis-30019	203	48	is	be	AUX
fcis-30019	203	49	presented	present	VERB
fcis-30019	203	50	in	in	ADP
fcis-30019	203	51	the	the	DET
fcis-30019	203	52	form	form	NOUN
fcis-30019	203	53	of	of	ADP
fcis-30019	203	54	visualization	visualization	NOUN
fcis-30019	203	55	,	,	PUNCT
fcis-30019	203	56	which	which	PRON
fcis-30019	203	57	shows	show	VERB
fcis-30019	203	58	the	the	DET
fcis-30019	203	59	superiority	superiority	NOUN
fcis-30019	203	60	of	of	ADP
fcis-30019	203	61	the	the	DET
fcis-30019	203	62	algorithm	algorithm	NOUN
fcis-30019	203	63	in	in	ADP
fcis-30019	203	64	the	the	DET
fcis-30019	203	65	generalization	generalization	NOUN
fcis-30019	203	66	ability	ability	NOUN
fcis-30019	203	67	.	.	PUNCT
fcis-30019	204	1	robustness	robustness	NOUN
fcis-30019	204	2	test	test	NOUN
fcis-30019	204	3	of	of	ADP
fcis-30019	204	4	anti	anti	ADJ
fcis-30019	204	5	-	-	NOUN
fcis-30019	204	6	interference	interference	NOUN
fcis-30019	204	7	,	,	PUNCT
fcis-30019	204	8	and	and	CCONJ
fcis-30019	204	9	the	the	DET
fcis-30019	204	10	attention	attention	NOUN
fcis-30019	204	11	to	to	ADP
fcis-30019	204	12	the	the	DET
fcis-30019	204	13	references	reference	NOUN
fcis-30019	204	14	[	[	X
fcis-30019	204	15	1	1	NUM
fcis-30019	204	16	]	]	X
fcis-30019	204	17	mohammad	mohammad	PROPN
fcis-30019	204	18	rami	rami	PROPN
fcis-30019	204	19	koujan	koujan	PROPN
fcis-30019	204	20	,	,	PUNCT
fcis-30019	204	21	michail	michail	NOUN
fcis-30019	204	22	christos	christo	NOUN
fcis-30019	204	23	doukas	doukas	NOUN
fcis-30019	204	24	,	,	PUNCT
fcis-30019	204	25	anastasios	anastasio	NOUN
fcis-30019	204	26	roussos	rousso	NOUN
fcis-30019	204	27	,	,	PUNCT
fcis-30019	204	28	and	and	CCONJ
fcis-30019	204	29	stefanos	stefanos	PROPN
fcis-30019	204	30	zafeiriou	zafeiriou	NOUN
fcis-30019	204	31	.	.	PUNCT
fcis-30019	205	1	2020	2020	NUM
fcis-30019	205	2	.	.	PUNCT
fcis-30019	206	1	head2head	head2head	NOUN
fcis-30019	206	2	:	:	PUNCT
fcis-30019	206	3	video	video	NOUN
fcis-30019	206	4	-	-	PUNCT
fcis-30019	206	5	based	base	VERB
fcis-30019	206	6	neural	neural	ADJ
fcis-30019	206	7	head	head	NOUN
fcis-30019	206	8	synthesis	synthesis	NOUN
fcis-30019	206	9	.	.	PUNCT
fcis-30019	207	1	arxiv	arxiv	PROPN
fcis-30019	207	2	preprint	preprint	PROPN
fcis-30019	207	3	arxiv	arxiv	PROPN
fcis-30019	207	4	:	:	PUNCT
fcis-30019	207	5	2005	2005	NUM
fcis-30019	207	6	.	.	PUNCT
fcis-30019	208	1	10954	10954	NUM
fcis-30019	208	2	(	(	PUNCT
fcis-30019	208	3	2020	2020	NUM
fcis-30019	208	4	)	)	PUNCT
fcis-30019	208	5	.	.	PUNCT
fcis-30019	209	1	mohammad	mohammad	PROPN
fcis-30019	209	2	rami	rami	PROPN
fcis-30019	209	3	koujan	koujan	PROPN
fcis-30019	209	4	,	,	PUNCT
fcis-30019	209	5	michail	michail	NOUN
fcis-30019	209	6	christos	christo	NOUN
fcis-30019	209	7	doukas	doukas	NOUN
fcis-30019	209	8	,	,	PUNCT
fcis-30019	209	9	anastasios	anastasio	NOUN
fcis-30019	209	10	roussos	rousso	NOUN
fcis-30019	209	11	,	,	PUNCT
fcis-30019	209	12	and	and	CCONJ
fcis-30019	209	13	stefanos	stefanos	PROPN
fcis-30019	209	14	zafeiriou	zafeiriou	NOUN
fcis-30019	209	15	.	.	PUNCT
fcis-30019	210	1	2020	2020	NUM
fcis-30019	210	2	.	.	PUNCT
fcis-30019	211	1	head2head	head2head	NOUN
fcis-30019	211	2	:	:	PUNCT
fcis-30019	211	3	video	video	NOUN
fcis-30019	211	4	-	-	PUNCT
fcis-30019	211	5	based	base	VERB
fcis-30019	211	6	neural	neural	ADJ
fcis-30019	211	7	head	head	NOUN
fcis-30019	211	8	synthesis	synthesis	NOUN
fcis-30019	211	9	.	.	PUNCT
fcis-30019	212	1	arxiv	arxiv	PROPN
fcis-30019	212	2	preprint	preprint	NOUN
fcis-30019	212	3	arxiv:2005.10954	arxiv:2005.10954	NOUN
fcis-30019	212	4	.	.	PUNCT
fcis-30019	213	1	[	[	X
fcis-30019	213	2	2	2	NUM
fcis-30019	213	3	]	]	X
fcis-30019	213	4	goodfellow	goodfellow	PROPN
fcis-30019	213	5	i	i	PROPN
fcis-30019	213	6	,	,	PUNCT
fcis-30019	213	7	pouget	pouget	NOUN
fcis-30019	213	8	-	-	PUNCT
fcis-30019	213	9	abadie	abadie	NOUN
fcis-30019	213	10	j	j	PROPN
fcis-30019	213	11	,	,	PUNCT
fcis-30019	213	12	mirza	mirza	PROPN
fcis-30019	213	13	m	m	PROPN
fcis-30019	213	14	,	,	PUNCT
fcis-30019	213	15	et	et	PROPN
fcis-30019	213	16	al	al	PROPN
fcis-30019	213	17	.	.	PROPN
fcis-30019	213	18	generative	generative	PROPN
fcis-30019	213	19	adversarial	adversarial	ADJ
fcis-30019	213	20	networks[j	networks[j	PROPN
fcis-30019	213	21	]	]	PUNCT
fcis-30019	213	22	.	.	PUNCT
fcis-30019	214	1	communications	communication	NOUN
fcis-30019	214	2	of	of	ADP
fcis-30019	214	3	the	the	DET
fcis-30019	214	4	acm	acm	NOUN
fcis-30019	214	5	,	,	PUNCT
fcis-30019	214	6	2020	2020	NUM
fcis-30019	214	7	,	,	PUNCT
fcis-30019	214	8	63(11	63(11	NUM
fcis-30019	214	9	):	):	PUNCT
fcis-30019	214	10	139	139	NUM
fcis-30019	214	11	-	-	SYM
fcis-30019	214	12	144	144	NUM
fcis-30019	214	13	.	.	PUNCT
fcis-30019	215	1	[	[	X
fcis-30019	215	2	3	3	X
fcis-30019	215	3	]	]	X
fcis-30019	215	4	rana	rana	PROPN
fcis-30019	215	5	m	m	PROPN
fcis-30019	215	6	s	s	PROPN
fcis-30019	215	7	,	,	PUNCT
fcis-30019	215	8	nobi	nobi	PROPN
fcis-30019	215	9	m	m	PROPN
fcis-30019	215	10	n	n	CCONJ
fcis-30019	215	11	,	,	PUNCT
fcis-30019	215	12	murali	murali	PROPN
fcis-30019	215	13	b	b	PROPN
fcis-30019	215	14	,	,	PUNCT
fcis-30019	215	15	et	et	PROPN
fcis-30019	215	16	al	al	PROPN
fcis-30019	215	17	.	.	PROPN
fcis-30019	215	18	deepfake	deepfake	PROPN
fcis-30019	215	19	detection	detection	NOUN
fcis-30019	215	20	:	:	PUNCT
fcis-30019	215	21	a	a	DET
fcis-30019	215	22	systematic	systematic	ADJ
fcis-30019	215	23	literature	literature	NOUN
fcis-30019	215	24	review[j	review[j	PROPN
fcis-30019	215	25	]	]	PUNCT
fcis-30019	215	26	.	.	PUNCT
fcis-30019	216	1	ieee	ieee	NOUN
fcis-30019	216	2	access	access	NOUN
fcis-30019	216	3	,	,	PUNCT
fcis-30019	216	4	2022	2022	NUM
fcis-30019	216	5	,	,	PUNCT
fcis-30019	216	6	10	10	NUM
fcis-30019	216	7	:	:	SYM
fcis-30019	216	8	2549425513	2549425513	NUM
fcis-30019	216	9	.	.	PUNCT
fcis-30019	217	1	[	[	X
fcis-30019	217	2	4	4	X
fcis-30019	217	3	]	]	PUNCT
fcis-30019	217	4	suthaharan	suthaharan	VERB
fcis-30019	217	5	s	s	PROPN
fcis-30019	217	6	,	,	PUNCT
fcis-30019	217	7	suthaharan	suthaharan	VERB
fcis-30019	217	8	s.	s.	PROPN
fcis-30019	217	9	support	support	VERB
fcis-30019	217	10	vector	vector	NOUN
fcis-30019	217	11	machine[j	machine[j	PROPN
fcis-30019	217	12	]	]	PUNCT
fcis-30019	217	13	.	.	PUNCT
fcis-30019	218	1	machine	machine	NOUN
fcis-30019	218	2	learning	learning	NOUN
fcis-30019	218	3	models	model	NOUN
fcis-30019	218	4	and	and	CCONJ
fcis-30019	218	5	algorithms	algorithm	NOUN
fcis-30019	218	6	for	for	ADP
fcis-30019	218	7	big	big	ADJ
fcis-30019	218	8	data	datum	NOUN
fcis-30019	218	9	classification	classification	NOUN
fcis-30019	218	10	:	:	PUNCT
fcis-30019	218	11	thinking	think	VERB
fcis-30019	218	12	with	with	ADP
fcis-30019	218	13	examples	example	NOUN
fcis-30019	218	14	for	for	ADP
fcis-30019	218	15	effective	effective	ADJ
fcis-30019	218	16	learning	learning	NOUN
fcis-30019	218	17	,	,	PUNCT
fcis-30019	218	18	2016	2016	NUM
fcis-30019	218	19	:	:	PUNCT
fcis-30019	218	20	207	207	NUM
fcis-30019	218	21	-	-	SYM
fcis-30019	218	22	235	235	NUM
fcis-30019	218	23	.	.	PUNCT
fcis-30019	219	1	[	[	X
fcis-30019	219	2	5	5	X
fcis-30019	219	3	]	]	X
fcis-30019	219	4	afchar	afchar	PROPN
fcis-30019	219	5	d	d	NOUN
fcis-30019	219	6	,	,	PUNCT
fcis-30019	219	7	nozick	nozick	PROPN
fcis-30019	219	8	v	v	NOUN
fcis-30019	219	9	,	,	PUNCT
fcis-30019	219	10	yamagishi	yamagishi	PROPN
fcis-30019	219	11	j	j	PROPN
fcis-30019	219	12	,	,	PUNCT
fcis-30019	220	1	et	et	PROPN
fcis-30019	220	2	al	al	PROPN
fcis-30019	220	3	.	.	PUNCT
fcis-30019	220	4	mesonet	mesonet	PROPN
fcis-30019	220	5	:	:	PUNCT
fcis-30019	220	6	a	a	DET
fcis-30019	220	7	compact	compact	ADJ
fcis-30019	220	8	facial	facial	ADJ
fcis-30019	220	9	video	video	NOUN
fcis-30019	220	10	forgery	forgery	NOUN
fcis-30019	220	11	detection	detection	NOUN
fcis-30019	220	12	network[c]//2018	network[c]//2018	SYM
fcis-30019	220	13	ieee	ieee	NOUN
fcis-30019	220	14	international	international	ADJ
fcis-30019	220	15	workshop	workshop	NOUN
fcis-30019	220	16	on	on	ADP
fcis-30019	220	17	information	information	NOUN
fcis-30019	220	18	forensics	forensic	NOUN
fcis-30019	220	19	and	and	CCONJ
fcis-30019	220	20	security	security	NOUN
fcis-30019	220	21	(	(	PUNCT
fcis-30019	220	22	wifs	wifs	PROPN
fcis-30019	220	23	)	)	PUNCT
fcis-30019	220	24	.	.	PUNCT
fcis-30019	221	1	ieee	ieee	NOUN
fcis-30019	221	2	,	,	PUNCT
fcis-30019	221	3	2018	2018	NUM
fcis-30019	221	4	:	:	PUNCT
fcis-30019	221	5	1	1	NUM
fcis-30019	221	6	-	-	SYM
fcis-30019	221	7	7	7	NUM
fcis-30019	221	8	.	.	PUNCT
fcis-30019	222	1	[	[	X
fcis-30019	222	2	6	6	NUM
fcis-30019	222	3	]	]	X
fcis-30019	222	4	lu	lu	PROPN
fcis-30019	222	5	x	x	NOUN
fcis-30019	222	6	,	,	PUNCT
fcis-30019	222	7	firoozeh	firoozeh	PROPN
fcis-30019	222	8	abolhasani	abolhasani	PROPN
fcis-30019	222	9	zadeh	zadeh	PROPN
fcis-30019	222	10	y	y	PROPN
fcis-30019	222	11	a.	a.	PROPN
fcis-30019	222	12	deep	deep	ADV
fcis-30019	222	13	learning‐based	learning‐base	VERB
fcis-30019	222	14	classification	classification	NOUN
fcis-30019	222	15	for	for	ADP
fcis-30019	222	16	melanoma	melanoma	NOUN
fcis-30019	222	17	detection	detection	NOUN
fcis-30019	222	18	using	use	VERB
fcis-30019	222	19	xceptionnet	xceptionnet	PROPN
fcis-30019	223	1	[	[	X
fcis-30019	223	2	j	j	X
fcis-30019	223	3	]	]	X
fcis-30019	223	4	.	.	PUNCT
fcis-30019	224	1	journal	journal	PROPN
fcis-30019	224	2	of	of	ADP
fcis-30019	224	3	healthcare	healthcare	PROPN
fcis-30019	224	4	engineering	engineering	NOUN
fcis-30019	224	5	,	,	PUNCT
fcis-30019	224	6	2022	2022	NUM
fcis-30019	224	7	,	,	PUNCT
fcis-30019	224	8	2022(1	2022(1	NUM
fcis-30019	224	9	):	):	PUNCT
fcis-30019	224	10	2196096	2196096	NUM
fcis-30019	224	11	.	.	PUNCT
fcis-30019	225	1	[	[	X
fcis-30019	225	2	7	7	X
fcis-30019	225	3	]	]	X
fcis-30019	225	4	sun	sun	PROPN
fcis-30019	225	5	b	b	PROPN
fcis-30019	225	6	,	,	PUNCT
fcis-30019	225	7	liu	liu	PROPN
fcis-30019	225	8	g	g	PROPN
fcis-30019	225	9	,	,	PUNCT
fcis-30019	225	10	yuan	yuan	PROPN
fcis-30019	225	11	y.	y.	PROPN
fcis-30019	225	12	f3	f3	PROPN
fcis-30019	225	13	-	-	PUNCT
fcis-30019	225	14	net	net	NOUN
fcis-30019	225	15	:	:	PUNCT
fcis-30019	225	16	multiview	multiview	NOUN
fcis-30019	225	17	scene	scene	NOUN
fcis-30019	225	18	matching	match	VERB
fcis-30019	225	19	for	for	ADP
fcis-30019	225	20	drone	drone	NOUN
fcis-30019	225	21	-	-	PUNCT
fcis-30019	225	22	based	base	VERB
fcis-30019	225	23	geo	geo	PROPN
fcis-30019	225	24	-	-	PUNCT
fcis-30019	225	25	localization[j	localization[j	PROPN
fcis-30019	225	26	]	]	PUNCT
fcis-30019	225	27	.	.	PUNCT
fcis-30019	226	1	ieee	ieee	NOUN
fcis-30019	226	2	transactions	transaction	NOUN
fcis-30019	226	3	on	on	ADP
fcis-30019	226	4	geoscience	geoscience	NOUN
fcis-30019	226	5	and	and	CCONJ
fcis-30019	226	6	remote	remote	ADJ
fcis-30019	226	7	sensing	sensing	NOUN
fcis-30019	226	8	,	,	PUNCT
fcis-30019	226	9	2023	2023	NUM
fcis-30019	226	10	,	,	PUNCT
fcis-30019	226	11	61	61	NUM
fcis-30019	226	12	:	:	SYM
fcis-30019	226	13	1	1	NUM
fcis-30019	226	14	-	-	SYM
fcis-30019	226	15	11	11	NUM
fcis-30019	226	16	.	.	PUNCT
fcis-30019	227	1	[	[	X
fcis-30019	227	2	8	8	NUM
fcis-30019	227	3	]	]	PUNCT
fcis-30019	227	4	yu	yu	PROPN
fcis-30019	227	5	y	y	PROPN
fcis-30019	227	6	,	,	PUNCT
fcis-30019	227	7	si	si	PROPN
fcis-30019	227	8	x	x	PROPN
fcis-30019	227	9	,	,	PUNCT
fcis-30019	227	10	hu	hu	PROPN
fcis-30019	227	11	c	c	X
fcis-30019	227	12	,	,	PUNCT
fcis-30019	227	13	et	et	PROPN
fcis-30019	227	14	al	al	PROPN
fcis-30019	227	15	.	.	PUNCT
fcis-30019	228	1	a	a	DET
fcis-30019	228	2	review	review	NOUN
fcis-30019	228	3	of	of	ADP
fcis-30019	228	4	recurrent	recurrent	ADJ
fcis-30019	228	5	neural	neural	ADJ
fcis-30019	228	6	networks	network	NOUN
fcis-30019	228	7	:	:	PUNCT
fcis-30019	228	8	lstm	lstm	NOUN
fcis-30019	228	9	cells	cell	NOUN
fcis-30019	228	10	and	and	CCONJ
fcis-30019	228	11	network	network	NOUN
fcis-30019	228	12	architectures[j	architectures[j	PROPN
fcis-30019	228	13	]	]	PUNCT
fcis-30019	228	14	.	.	PUNCT
fcis-30019	229	1	neural	neural	ADJ
fcis-30019	229	2	computation	computation	NOUN
fcis-30019	229	3	,	,	PUNCT
fcis-30019	229	4	2019	2019	NUM
fcis-30019	229	5	,	,	PUNCT
fcis-30019	229	6	31(7	31(7	NUM
fcis-30019	229	7	):	):	PUNCT
fcis-30019	229	8	1235	1235	NUM
fcis-30019	229	9	-	-	SYM
fcis-30019	229	10	1270	1270	NUM
fcis-30019	229	11	.	.	PUNCT
fcis-30019	230	1	[	[	X
fcis-30019	230	2	9	9	NUM
fcis-30019	230	3	]	]	PUNCT
fcis-30019	230	4	fan	fan	NOUN
fcis-30019	230	5	h	h	PROPN
fcis-30019	230	6	,	,	PUNCT
fcis-30019	230	7	xiong	xiong	PROPN
fcis-30019	230	8	b	b	PROPN
fcis-30019	230	9	,	,	PUNCT
fcis-30019	230	10	mangalam	mangalam	PROPN
fcis-30019	230	11	k	k	PROPN
fcis-30019	230	12	,	,	PUNCT
fcis-30019	230	13	et	et	PROPN
fcis-30019	231	1	al	al	PROPN
fcis-30019	231	2	.	.	PROPN
fcis-30019	231	3	multiscale	multiscale	PROPN
fcis-30019	231	4	vision	vision	NOUN
fcis-30019	231	5	transformers[c]//proceedings	transformers[c]//proceeding	NOUN
fcis-30019	231	6	of	of	ADP
fcis-30019	231	7	the	the	DET
fcis-30019	231	8	ieee	ieee	NOUN
fcis-30019	231	9	/	/	SYM
fcis-30019	231	10	cvf	cvf	NOUN
fcis-30019	231	11	international	international	ADJ
fcis-30019	231	12	conference	conference	NOUN
fcis-30019	231	13	on	on	ADP
fcis-30019	231	14	computer	computer	NOUN
fcis-30019	231	15	vision	vision	NOUN
fcis-30019	231	16	.	.	PUNCT
fcis-30019	232	1	2021	2021	NUM
fcis-30019	232	2	:	:	PUNCT
fcis-30019	232	3	6824	6824	NUM
fcis-30019	232	4	-	-	SYM
fcis-30019	232	5	6835	6835	NUM
fcis-30019	232	6	.	.	PUNCT
fcis-30019	233	1	[	[	X
fcis-30019	233	2	10	10	NUM
fcis-30019	233	3	]	]	X
fcis-30019	233	4	chen	chen	PROPN
fcis-30019	233	5	m	m	PROPN
fcis-30019	233	6	,	,	PUNCT
fcis-30019	233	7	peng	peng	PROPN
fcis-30019	233	8	h	h	PROPN
fcis-30019	233	9	,	,	PUNCT
fcis-30019	233	10	fu	fu	PROPN
fcis-30019	233	11	j	j	PROPN
fcis-30019	233	12	,	,	PUNCT
fcis-30019	233	13	et	et	PROPN
fcis-30019	233	14	al	al	PROPN
fcis-30019	233	15	.	.	PUNCT
fcis-30019	233	16	autoformer	autoformer	PROPN
fcis-30019	233	17	:	:	PUNCT
fcis-30019	233	18	searching	search	VERB
fcis-30019	233	19	transformers	transformer	NOUN
fcis-30019	233	20	for	for	ADP
fcis-30019	233	21	visual	visual	ADJ
fcis-30019	233	22	recognition[c]//proceedings	recognition[c]//proceeding	NOUN
fcis-30019	233	23	of	of	ADP
fcis-30019	233	24	the	the	DET
fcis-30019	233	25	ieee	ieee	NOUN
fcis-30019	233	26	/	/	SYM
fcis-30019	233	27	cvf	cvf	NOUN
fcis-30019	233	28	international	international	ADJ
fcis-30019	233	29	conference	conference	NOUN
fcis-30019	233	30	on	on	ADP
fcis-30019	233	31	computer	computer	NOUN
fcis-30019	233	32	vision	vision	NOUN
fcis-30019	233	33	.	.	PUNCT
fcis-30019	234	1	2021	2021	NUM
fcis-30019	234	2	:	:	PUNCT
fcis-30019	234	3	12270	12270	NUM
fcis-30019	234	4	-	-	SYM
fcis-30019	234	5	12280	12280	NUM
fcis-30019	234	6	.	.	PUNCT
fcis-30019	235	1	[	[	X
fcis-30019	235	2	11	11	NUM
fcis-30019	235	3	]	]	X
fcis-30019	235	4	wen	wen	PROPN
fcis-30019	235	5	g	g	PROPN
fcis-30019	235	6	,	,	PUNCT
fcis-30019	235	7	li	li	PROPN
fcis-30019	235	8	z	z	PROPN
fcis-30019	235	9	,	,	PUNCT
fcis-30019	235	10	azizzadenesheli	azizzadenesheli	PROPN
fcis-30019	235	11	k	k	PROPN
fcis-30019	235	12	,	,	PUNCT
fcis-30019	235	13	et	et	PROPN
fcis-30019	235	14	al	al	PROPN
fcis-30019	235	15	.	.	PUNCT
fcis-30019	236	1	u	u	PROPN
fcis-30019	236	2	-	-	PROPN
fcis-30019	236	3	fno	fno	PROPN
fcis-30019	236	4	—	—	PUNCT
fcis-30019	236	5	an	an	DET
fcis-30019	236	6	enhanced	enhanced	ADJ
fcis-30019	236	7	fourier	fourier	NOUN
fcis-30019	236	8	neural	neural	ADJ
fcis-30019	236	9	operator	operator	NOUN
fcis-30019	236	10	-	-	PUNCT
fcis-30019	236	11	based	base	VERB
fcis-30019	236	12	deep	deep	ADJ
fcis-30019	236	13	-	-	PUNCT
fcis-30019	236	14	learning	learn	VERB
fcis-30019	236	15	model	model	NOUN
fcis-30019	236	16	for	for	ADP
fcis-30019	236	17	multiphase	multiphase	PROPN
fcis-30019	236	18	flow[j	flow[j	PROPN
fcis-30019	236	19	]	]	PUNCT
fcis-30019	236	20	.	.	PUNCT
fcis-30019	237	1	advances	advance	NOUN
fcis-30019	237	2	in	in	ADP
fcis-30019	237	3	water	water	NOUN
fcis-30019	237	4	resources	resource	NOUN
fcis-30019	237	5	,	,	PUNCT
fcis-30019	237	6	2022	2022	NUM
fcis-30019	237	7	,	,	PUNCT
fcis-30019	237	8	163	163	NUM
fcis-30019	237	9	:	:	PUNCT
fcis-30019	237	10	104180	104180	NUM
fcis-30019	237	11	.	.	PUNCT
fcis-30019	238	1	[	[	X
fcis-30019	238	2	12	12	NUM
fcis-30019	238	3	]	]	PUNCT
fcis-30019	238	4	men	man	NOUN
fcis-30019	238	5	y	y	PROPN
fcis-30019	238	6	,	,	PUNCT
fcis-30019	238	7	yao	yao	PROPN
fcis-30019	238	8	y	y	PROPN
fcis-30019	238	9	,	,	PUNCT
fcis-30019	238	10	cui	cui	PROPN
fcis-30019	238	11	m	m	PROPN
fcis-30019	238	12	,	,	PUNCT
fcis-30019	238	13	et	et	PROPN
fcis-30019	238	14	al	al	PROPN
fcis-30019	238	15	.	.	PUNCT
fcis-30019	238	16	dct	dct	PROPN
fcis-30019	238	17	-	-	PUNCT
fcis-30019	238	18	net	net	NOUN
fcis-30019	238	19	:	:	PUNCT
fcis-30019	238	20	domain	domain	NOUN
fcis-30019	238	21	-	-	PUNCT
fcis-30019	238	22	calibrated	calibrate	VERB
fcis-30019	238	23	translation	translation	NOUN
fcis-30019	238	24	for	for	ADP
fcis-30019	238	25	portrait	portrait	NOUN
fcis-30019	238	26	stylization[j	stylization[j	PROPN
fcis-30019	238	27	]	]	PUNCT
fcis-30019	238	28	.	.	PUNCT
fcis-30019	238	29	acm	acm	PROPN
fcis-30019	238	30	transactions	transaction	NOUN
fcis-30019	238	31	on	on	ADP
fcis-30019	238	32	graphics	graphic	NOUN
fcis-30019	238	33	(	(	PUNCT
fcis-30019	238	34	tog	tog	PROPN
fcis-30019	238	35	)	)	PUNCT
fcis-30019	238	36	,	,	PUNCT
fcis-30019	238	37	2022	2022	NUM
fcis-30019	238	38	,	,	PUNCT
fcis-30019	238	39	41(4	41(4	NUM
fcis-30019	238	40	):	):	PUNCT
fcis-30019	238	41	1	1	NUM
fcis-30019	238	42	-	-	SYM
fcis-30019	238	43	9	9	NUM
fcis-30019	238	44	.	.	PUNCT
fcis-30019	239	1	[	[	X
fcis-30019	239	2	13	13	NUM
fcis-30019	239	3	]	]	X
fcis-30019	239	4	jiang	jiang	PROPN
fcis-30019	239	5	m	m	PROPN
fcis-30019	239	6	,	,	PUNCT
fcis-30019	239	7	zeng	zeng	PROPN
fcis-30019	239	8	p	p	PROPN
fcis-30019	239	9	,	,	PUNCT
fcis-30019	239	10	wang	wang	PROPN
fcis-30019	239	11	k	k	PROPN
fcis-30019	239	12	,	,	PUNCT
fcis-30019	239	13	et	et	PROPN
fcis-30019	239	14	al	al	PROPN
fcis-30019	239	15	.	.	PROPN
fcis-30019	239	16	fecam	fecam	PROPN
fcis-30019	239	17	:	:	PUNCT
fcis-30019	239	18	frequency	frequency	NOUN
fcis-30019	239	19	enhanced	enhance	VERB
fcis-30019	239	20	channel	channel	NOUN
fcis-30019	239	21	attention	attention	NOUN
fcis-30019	239	22	mechanism	mechanism	NOUN
fcis-30019	239	23	for	for	ADP
fcis-30019	239	24	time	time	NOUN
fcis-30019	239	25	series	series	PROPN
fcis-30019	239	26	forecasting[j	forecasting[j	PROPN
fcis-30019	239	27	]	]	PUNCT
fcis-30019	239	28	.	.	PUNCT
fcis-30019	240	1	advanced	advanced	ADJ
fcis-30019	240	2	engineering	engineering	NOUN
fcis-30019	240	3	informatics	informatic	NOUN
fcis-30019	240	4	,	,	PUNCT
fcis-30019	240	5	2023	2023	NUM
fcis-30019	240	6	,	,	PUNCT
fcis-30019	240	7	58	58	NUM
fcis-30019	240	8	:	:	SYM
fcis-30019	240	9	102158	102158	NUM
fcis-30019	240	10	.	.	PUNCT
fcis-30019	241	1	[	[	X
fcis-30019	241	2	14	14	NUM
fcis-30019	241	3	]	]	X
fcis-30019	241	4	testa	testa	PROPN
fcis-30019	241	5	r	r	PROPN
fcis-30019	241	6	l	l	NOUN
fcis-30019	241	7	,	,	PUNCT
fcis-30019	241	8	machado	machado	PROPN
fcis-30019	241	9	-	-	PUNCT
fcis-30019	241	10	lima	lima	PROPN
fcis-30019	241	11	a	a	PROPN
fcis-30019	241	12	,	,	PUNCT
fcis-30019	241	13	nunes	nunes	PROPN
fcis-30019	241	14	f	f	PROPN
fcis-30019	241	15	l	l	PROPN
fcis-30019	241	16	s.	s.	PROPN
fcis-30019	241	17	deepfake	deepfake	PROPN
fcis-30019	241	18	detection	detection	NOUN
fcis-30019	241	19	on	on	ADP
fcis-30019	241	20	videos	video	NOUN
fcis-30019	241	21	based	base	VERB
fcis-30019	241	22	on	on	ADP
fcis-30019	241	23	ratio	ratio	NOUN
fcis-30019	241	24	images[c]//2022	images[c]//2022	PROPN
fcis-30019	241	25	12th	12th	PROPN
fcis-30019	241	26	international	international	ADJ
fcis-30019	241	27	congress	congress	PROPN
fcis-30019	241	28	on	on	ADP
fcis-30019	241	29	advanced	advanced	ADJ
fcis-30019	241	30	applied	apply	VERB
fcis-30019	241	31	informatics	informatic	NOUN
fcis-30019	241	32	(	(	PUNCT
fcis-30019	241	33	iiai	iiai	NOUN
fcis-30019	241	34	-	-	PUNCT
fcis-30019	241	35	aai	aai	PROPN
fcis-30019	241	36	)	)	PUNCT
fcis-30019	241	37	.	.	PUNCT
fcis-30019	242	1	ieee	ieee	NOUN
fcis-30019	242	2	,	,	PUNCT
fcis-30019	242	3	2022	2022	NUM
fcis-30019	242	4	:	:	PUNCT
fcis-30019	242	5	403	403	NUM
fcis-30019	242	6	-	-	SYM
fcis-30019	242	7	408	408	NUM
fcis-30019	242	8	.	.	PUNCT
fcis-30019	243	1	[	[	X
fcis-30019	243	2	15	15	NUM
fcis-30019	243	3	]	]	X
fcis-30019	243	4	rossler	rossler	NOUN
fcis-30019	243	5	a	a	PROPN
fcis-30019	243	6	,	,	PUNCT
fcis-30019	243	7	cozzolino	cozzolino	PROPN
fcis-30019	243	8	d	d	PROPN
fcis-30019	243	9	,	,	PUNCT
fcis-30019	243	10	verdoliva	verdoliva	PROPN
fcis-30019	243	11	l	l	NOUN
fcis-30019	243	12	,	,	PUNCT
fcis-30019	243	13	et	et	PROPN
fcis-30019	243	14	al	al	PROPN
fcis-30019	243	15	.	.	PUNCT
fcis-30019	244	1	faceforensics++	faceforensics++	ADV
fcis-30019	244	2	:	:	PUNCT
fcis-30019	244	3	learning	learn	VERB
fcis-30019	244	4	to	to	PART
fcis-30019	244	5	detect	detect	VERB
fcis-30019	244	6	manipulated	manipulate	VERB
fcis-30019	244	7	facial	facial	ADJ
fcis-30019	244	8	images[c]//	images[c]//	ADJ
fcis-30019	244	9	proceedings	proceeding	NOUN
fcis-30019	244	10	62	62	NUM
fcis-30019	244	11	of	of	ADP
fcis-30019	244	12	the	the	DET
fcis-30019	244	13	ieee	ieee	NOUN
fcis-30019	244	14	/	/	SYM
fcis-30019	244	15	cvf	cvf	NOUN
fcis-30019	244	16	international	international	ADJ
fcis-30019	244	17	conference	conference	NOUN
fcis-30019	244	18	on	on	ADP
fcis-30019	244	19	computer	computer	NOUN
fcis-30019	244	20	vision	vision	NOUN
fcis-30019	244	21	.	.	PUNCT
fcis-30019	245	1	2019	2019	NUM
fcis-30019	245	2	:	:	PUNCT
fcis-30019	245	3	1	1	NUM
fcis-30019	245	4	-	-	SYM
fcis-30019	245	5	11	11	NUM
fcis-30019	245	6	.	.	PUNCT
fcis-30019	246	1	[	[	X
fcis-30019	246	2	16	16	NUM
fcis-30019	246	3	]	]	SYM
fcis-30019	246	4	li	li	PROPN
fcis-30019	246	5	,	,	PUNCT
fcis-30019	246	6	yuezun	yuezun	PROPN
fcis-30019	246	7	,	,	PUNCT
fcis-30019	246	8	et	et	PROPN
fcis-30019	246	9	al	al	PROPN
fcis-30019	246	10	.	.	PUNCT
fcis-30019	247	1	"	"	PUNCT
fcis-30019	247	2	celeb	celeb	NOUN
fcis-30019	247	3	-	-	PUNCT
fcis-30019	247	4	df	df	PROPN
fcis-30019	247	5	:	:	PUNCT
fcis-30019	247	6	a	a	DET
fcis-30019	247	7	large	large	ADJ
fcis-30019	247	8	-	-	PUNCT
fcis-30019	247	9	scale	scale	NOUN
fcis-30019	247	10	challenging	challenging	ADJ
fcis-30019	247	11	dataset	dataset	NOUN
fcis-30019	247	12	for	for	ADP
fcis-30019	247	13	deepfake	deepfake	NOUN
fcis-30019	247	14	forensics	forensic	NOUN
fcis-30019	247	15	.	.	PUNCT
fcis-30019	247	16	"	"	PUNCT
fcis-30019	248	1	proceedings	proceeding	NOUN
fcis-30019	248	2	of	of	ADP
fcis-30019	248	3	the	the	DET
fcis-30019	248	4	ieee	ieee	NOUN
fcis-30019	248	5	/	/	SYM
fcis-30019	248	6	cvf	cvf	NOUN
fcis-30019	248	7	conference	conference	NOUN
fcis-30019	248	8	on	on	ADP
fcis-30019	248	9	computer	computer	NOUN
fcis-30019	248	10	vision	vision	NOUN
fcis-30019	248	11	and	and	CCONJ
fcis-30019	248	12	pattern	pattern	NOUN
fcis-30019	248	13	recognition	recognition	NOUN
fcis-30019	248	14	.	.	PUNCT
fcis-30019	249	1	2020	2020	NUM
fcis-30019	249	2	.	.	PUNCT
fcis-30019	250	1	[	[	X
fcis-30019	250	2	17	17	NUM
fcis-30019	250	3	]	]	X
fcis-30019	250	4	jiang	jiang	PROPN
fcis-30019	250	5	,	,	PUNCT
fcis-30019	250	6	liming	liming	NOUN
fcis-30019	250	7	,	,	PUNCT
fcis-30019	250	8	et	et	PROPN
fcis-30019	250	9	al	al	PROPN
fcis-30019	250	10	.	.	PUNCT
fcis-30019	251	1	"	"	PUNCT
fcis-30019	251	2	deeperforensics-1.0	deeperforensics-1.0	NOUN
fcis-30019	251	3	:	:	PUNCT
fcis-30019	251	4	a	a	DET
fcis-30019	251	5	large	large	ADJ
fcis-30019	251	6	-	-	PUNCT
fcis-30019	251	7	scale	scale	NOUN
fcis-30019	251	8	dataset	dataset	NOUN
fcis-30019	251	9	for	for	ADP
fcis-30019	251	10	real	real	ADJ
fcis-30019	251	11	-	-	PUNCT
fcis-30019	251	12	world	world	NOUN
fcis-30019	251	13	face	face	NOUN
fcis-30019	251	14	forgery	forgery	NOUN
fcis-30019	251	15	detection	detection	NOUN
fcis-30019	251	16	.	.	PUNCT
fcis-30019	251	17	"	"	PUNCT
fcis-30019	252	1	proceedings	proceeding	NOUN
fcis-30019	252	2	of	of	ADP
fcis-30019	252	3	the	the	DET
fcis-30019	252	4	ieee	ieee	NOUN
fcis-30019	252	5	/	/	SYM
fcis-30019	252	6	cvf	cvf	NOUN
fcis-30019	252	7	conference	conference	NOUN
fcis-30019	252	8	on	on	ADP
fcis-30019	252	9	computer	computer	NOUN
fcis-30019	252	10	vision	vision	NOUN
fcis-30019	252	11	and	and	CCONJ
fcis-30019	252	12	pattern	pattern	NOUN
fcis-30019	252	13	recognition	recognition	NOUN
fcis-30019	252	14	.	.	PUNCT
fcis-30019	253	1	2020	2020	NUM
fcis-30019	253	2	.	.	PUNCT
fcis-30019	254	1	[	[	X
fcis-30019	254	2	18	18	NUM
fcis-30019	254	3	]	]	SYM
fcis-30019	254	4	li	li	PROPN
fcis-30019	254	5	,	,	PUNCT
fcis-30019	254	6	lingzhi	lingzhi	PROPN
fcis-30019	254	7	,	,	PUNCT
fcis-30019	254	8	et	et	PROPN
fcis-30019	254	9	al	al	PROPN
fcis-30019	254	10	.	.	PUNCT
fcis-30019	254	11	"	"	PUNCT
fcis-30019	254	12	faceshifter	faceshifter	NOUN
fcis-30019	254	13	:	:	PUNCT
fcis-30019	254	14	towards	towards	ADP
fcis-30019	254	15	high	high	ADJ
fcis-30019	254	16	fidelity	fidelity	NOUN
fcis-30019	254	17	and	and	CCONJ
fcis-30019	254	18	occlusion	occlusion	NOUN
fcis-30019	254	19	aware	aware	ADJ
fcis-30019	254	20	face	face	NOUN
fcis-30019	254	21	swapping	swap	VERB
fcis-30019	254	22	.	.	PUNCT
fcis-30019	254	23	"	"	PUNCT
fcis-30019	255	1	arxiv	arxiv	PROPN
fcis-30019	255	2	preprint	preprint	NOUN
fcis-30019	255	3	arxiv	arxiv	PROPN
fcis-30019	255	4	:	:	PUNCT
fcis-30019	255	5	1912	1912	NUM
fcis-30019	255	6	.	.	PUNCT
fcis-30019	256	1	13457	13457	NUM
fcis-30019	256	2	(	(	PUNCT
fcis-30019	256	3	2019	2019	NUM
fcis-30019	256	4	)	)	PUNCT
fcis-30019	256	5	.	.	PUNCT
fcis-30019	257	1	[	[	X
fcis-30019	257	2	19	19	NUM
fcis-30019	257	3	]	]	X
fcis-30019	257	4	croitoru	croitoru	PROPN
fcis-30019	257	5	f	f	PROPN
fcis-30019	257	6	a	a	PROPN
fcis-30019	257	7	,	,	PUNCT
fcis-30019	257	8	hiji	hiji	VERB
fcis-30019	257	9	a	a	DET
fcis-30019	257	10	i	i	PROPN
fcis-30019	257	11	,	,	PUNCT
fcis-30019	257	12	hondru	hondru	NOUN
fcis-30019	257	13	v	v	NOUN
fcis-30019	257	14	,	,	PUNCT
fcis-30019	257	15	et	et	PROPN
fcis-30019	257	16	al	al	PROPN
fcis-30019	257	17	.	.	PROPN
fcis-30019	258	1	deepfake	deepfake	PROPN
fcis-30019	258	2	media	medium	NOUN
fcis-30019	258	3	generation	generation	NOUN
fcis-30019	258	4	and	and	CCONJ
fcis-30019	258	5	detection	detection	NOUN
fcis-30019	258	6	in	in	ADP
fcis-30019	258	7	the	the	DET
fcis-30019	258	8	generative	generative	NOUN
fcis-30019	258	9	ai	ai	VERB
fcis-30019	258	10	era	era	NOUN
fcis-30019	258	11	:	:	PUNCT
fcis-30019	258	12	a	a	DET
fcis-30019	258	13	survey	survey	NOUN
fcis-30019	258	14	and	and	CCONJ
fcis-30019	258	15	outlook[j	outlook[j	NOUN
fcis-30019	258	16	]	]	PUNCT
fcis-30019	258	17	.	.	PUNCT
fcis-30019	259	1	arxiv	arxiv	PROPN
fcis-30019	259	2	preprint	preprint	NOUN
fcis-30019	259	3	arxiv:2411.19537	arxiv:2411.19537	PROPN
fcis-30019	259	4	,	,	PUNCT
fcis-30019	259	5	2024	2024	NUM
fcis-30019	259	6	.	.	PUNCT
fcis-30019	260	1	[	[	X
fcis-30019	260	2	20	20	NUM
fcis-30019	260	3	]	]	X
fcis-30019	260	4	wan	wan	PROPN
fcis-30019	260	5	da	da	PROPN
fcis-30019	260	6	.	.	PUNCT
fcis-30019	260	7	research	research	NOUN
fcis-30019	260	8	on	on	ADP
fcis-30019	260	9	face	face	NOUN
fcis-30019	260	10	forgery	forgery	NOUN
fcis-30019	260	11	image	image	NOUN
fcis-30019	260	12	detection	detection	NOUN
fcis-30019	260	13	method	method	NOUN
fcis-30019	260	14	based	base	VERB
fcis-30019	260	15	on	on	ADP
fcis-30019	260	16	deep	deep	ADJ
fcis-30019	260	17	learning	learning	NOUN
fcis-30019	260	18	[	[	X
fcis-30019	260	19	d	d	X
fcis-30019	260	20	]	]	X
fcis-30019	260	21	.	.	PUNCT
fcis-30019	261	1	people	people	NOUN
fcis-30019	261	2	's	's	PART
fcis-30019	261	3	public	public	ADJ
fcis-30019	261	4	security	security	NOUN
fcis-30019	261	5	university	university	PROPN
fcis-30019	261	6	of	of	ADP
fcis-30019	261	7	china,2024.doi:10	china,2024.doi:10	PROPN
fcis-30019	261	8	.	.	PROPN
fcis-30019	261	9	27634	27634	NUM
fcis-30019	261	10	/	/	SYM
fcis-30019	261	11	d.	d.	PROPN
fcis-30019	261	12	cnki	cnki	PROPN
fcis-30019	261	13	.	.	PUNCT
fcis-30019	262	1	gzrgu	gzrgu	PROPN
fcis-30019	262	2	.	.	PUNCT
fcis-30019	263	1	2024	2024	NUM
fcis-30019	263	2	.	.	PUNCT
fcis-30019	264	1	000268	000268	NUM
fcis-30019	264	2	.	.	PUNCT
fcis-30019	265	1	[	[	X
fcis-30019	265	2	21	21	NUM
fcis-30019	265	3	]	]	X
fcis-30019	265	4	nguyen	nguyen	NOUN
fcis-30019	265	5	d	d	NOUN
fcis-30019	265	6	,	,	PUNCT
fcis-30019	265	7	mejri	mejri	ADJ
fcis-30019	265	8	n	n	CCONJ
fcis-30019	265	9	,	,	PUNCT
fcis-30019	265	10	singh	singh	PROPN
fcis-30019	265	11	i	i	PRON
fcis-30019	265	12	p	p	PROPN
fcis-30019	265	13	,	,	PUNCT
fcis-30019	265	14	et	et	PROPN
fcis-30019	265	15	al	al	PROPN
fcis-30019	265	16	.	.	PUNCT
fcis-30019	266	1	laa	laa	PROPN
fcis-30019	266	2	-	-	PUNCT
fcis-30019	266	3	net	net	NOUN
fcis-30019	266	4	:	:	PUNCT
fcis-30019	266	5	localized	localized	ADJ
fcis-30019	266	6	artifact	artifact	ADJ
fcis-30019	266	7	attention	attention	NOUN
fcis-30019	266	8	network	network	NOUN
fcis-30019	266	9	for	for	ADP
fcis-30019	266	10	quality	quality	NOUN
fcis-30019	266	11	-	-	PUNCT
fcis-30019	266	12	agnostic	agnostic	ADJ
fcis-30019	266	13	and	and	CCONJ
fcis-30019	266	14	generalizable	generalizable	ADJ
fcis-30019	266	15	deepfake	deepfake	NOUN
fcis-30019	266	16	detection[c]//proceedings	detection[c]//proceeding	NOUN
fcis-30019	266	17	of	of	ADP
fcis-30019	266	18	the	the	DET
fcis-30019	266	19	ieee	ieee	NOUN
fcis-30019	266	20	/	/	SYM
fcis-30019	266	21	cvf	cvf	NOUN
fcis-30019	266	22	conference	conference	NOUN
fcis-30019	266	23	on	on	ADP
fcis-30019	266	24	computer	computer	NOUN
fcis-30019	266	25	vision	vision	NOUN
fcis-30019	266	26	and	and	CCONJ
fcis-30019	266	27	pattern	pattern	NOUN
fcis-30019	266	28	recognition	recognition	NOUN
fcis-30019	266	29	.	.	PUNCT
fcis-30019	267	1	2024	2024	NUM
fcis-30019	267	2	:	:	PUNCT
fcis-30019	267	3	17395	17395	NUM
fcis-30019	267	4	-	-	SYM
fcis-30019	267	5	17405	17405	NUM
fcis-30019	267	6	.	.	PUNCT
fcis-30019	268	1	[	[	X
fcis-30019	268	2	22	22	NUM
fcis-30019	268	3	]	]	PUNCT
fcis-30019	268	4	selvaraju	selvaraju	VERB
fcis-30019	268	5	r	r	NOUN
fcis-30019	268	6	r	r	PROPN
fcis-30019	268	7	,	,	PUNCT
fcis-30019	268	8	cogswell	cogswell	PROPN
fcis-30019	268	9	m	m	PROPN
fcis-30019	268	10	,	,	PUNCT
fcis-30019	268	11	das	das	PROPN
fcis-30019	268	12	a	a	PRON
fcis-30019	268	13	,	,	PUNCT
fcis-30019	268	14	et	et	PROPN
fcis-30019	268	15	al	al	PROPN
fcis-30019	268	16	.	.	PROPN
fcis-30019	268	17	grad	grad	PROPN
fcis-30019	268	18	-	-	PUNCT
fcis-30019	268	19	cam	cam	PROPN
fcis-30019	268	20	:	:	PUNCT
fcis-30019	268	21	visual	visual	ADJ
fcis-30019	268	22	explanations	explanation	NOUN
fcis-30019	268	23	from	from	ADP
fcis-30019	268	24	deep	deep	ADJ
fcis-30019	268	25	networks	network	NOUN
fcis-30019	268	26	via	via	ADP
fcis-30019	268	27	gradient	gradient	NOUN
fcis-30019	268	28	-	-	PUNCT
fcis-30019	268	29	based	base	VERB
fcis-30019	268	30	localization[c]//proceedings	localization[c]//proceeding	NOUN
fcis-30019	268	31	of	of	ADP
fcis-30019	268	32	the	the	DET
fcis-30019	268	33	ieee	ieee	NOUN
fcis-30019	268	34	international	international	PROPN
fcis-30019	268	35	conference	conference	NOUN
fcis-30019	268	36	on	on	ADP
fcis-30019	268	37	computer	computer	NOUN
fcis-30019	268	38	vision	vision	NOUN
fcis-30019	268	39	.	.	PUNCT
fcis-30019	269	1	2017	2017	NUM
fcis-30019	269	2	:	:	PUNCT
fcis-30019	269	3	618	618	NUM
fcis-30019	269	4	-	-	SYM
fcis-30019	269	5	626	626	NUM
fcis-30019	269	6	.	.	PUNCT