id	sid	tid	token	lemma	pos
fcis-16180	1	1	frontiers	frontier	NOUN
fcis-16180	1	2	in	in	ADP
fcis-16180	1	3	computing	computing	NOUN
fcis-16180	1	4	and	and	CCONJ
fcis-16180	1	5	intelligent	intelligent	ADJ
fcis-16180	1	6	systems	system	NOUN
fcis-16180	1	7	issn	issn	VERB
fcis-16180	1	8	:	:	PUNCT
fcis-16180	1	9	2832	2832	NUM
fcis-16180	1	10	-	-	SYM
fcis-16180	1	11	6024	6024	NUM
fcis-16180	1	12	|	|	NOUN
fcis-16180	1	13	vol	vol	NOUN
fcis-16180	1	14	.	.	PROPN
fcis-16180	2	1	7	7	NUM
fcis-16180	2	2	,	,	PUNCT
fcis-16180	2	3	no	no	INTJ
fcis-16180	2	4	.	.	NOUN
fcis-16180	2	5	1	1	NUM
fcis-16180	2	6	,	,	PUNCT
fcis-16180	2	7	2024	2024	NUM
fcis-16180	2	8	18	18	NUM
fcis-16180	2	9	ppbrfl	ppbrfl	NOUN
fcis-16180	2	10	:	:	PUNCT
fcis-16180	2	11	privacy‐preserving	privacy‐preserving	NOUN
fcis-16180	2	12	byzantine‐robust	byzantine‐robust	ADV
fcis-16180	3	1	federated	federated	ADJ
fcis-16180	3	2	learning	learning	NOUN
fcis-16180	3	3	qun	qun	PROPN
fcis-16180	3	4	zhou	zhou	PROPN
fcis-16180	3	5	*	*	PUNCT
fcis-16180	3	6	college	college	PROPN
fcis-16180	3	7	of	of	ADP
fcis-16180	3	8	computer	computer	NOUN
fcis-16180	3	9	science	science	NOUN
fcis-16180	3	10	and	and	CCONJ
fcis-16180	3	11	technology	technology	NOUN
fcis-16180	3	12	,	,	PUNCT
fcis-16180	3	13	qingdao	qingdao	PROPN
fcis-16180	3	14	university	university	PROPN
fcis-16180	3	15	,	,	PUNCT
fcis-16180	3	16	qingdao	qingdao	PROPN
fcis-16180	3	17	,	,	PUNCT
fcis-16180	3	18	shandong	shandong	PROPN
fcis-16180	3	19	,	,	PUNCT
fcis-16180	3	20	china	china	PROPN
fcis-16180	3	21	*	*	PUNCT
fcis-16180	3	22	corresponding	correspond	VERB
fcis-16180	3	23	author	author	NOUN
fcis-16180	3	24	email	email	NOUN
fcis-16180	3	25	:	:	PUNCT
fcis-16180	3	26	zhouqun1999@163.com	zhouqun1999@163.com	X
fcis-16180	3	27	abstract	abstract	NOUN
fcis-16180	3	28	:	:	PUNCT
fcis-16180	3	29	federated	federated	ADJ
fcis-16180	3	30	learning	learning	NOUN
fcis-16180	3	31	is	be	AUX
fcis-16180	3	32	a	a	DET
fcis-16180	3	33	distributed	distribute	VERB
fcis-16180	3	34	machine	machine	NOUN
fcis-16180	3	35	learning	learn	VERB
fcis-16180	3	36	approach	approach	NOUN
fcis-16180	3	37	that	that	PRON
fcis-16180	3	38	allows	allow	VERB
fcis-16180	3	39	the	the	DET
fcis-16180	3	40	neural	neural	ADJ
fcis-16180	3	41	network	network	NOUN
fcis-16180	3	42	to	to	PART
fcis-16180	3	43	be	be	AUX
fcis-16180	3	44	trained	train	VERB
fcis-16180	3	45	without	without	ADP
fcis-16180	3	46	exposing	expose	VERB
fcis-16180	3	47	private	private	ADJ
fcis-16180	3	48	user	user	NOUN
fcis-16180	3	49	data	datum	NOUN
fcis-16180	3	50	.	.	PUNCT
fcis-16180	4	1	despite	despite	SCONJ
fcis-16180	4	2	its	its	PRON
fcis-16180	4	3	advantages	advantage	NOUN
fcis-16180	4	4	,	,	PUNCT
fcis-16180	4	5	federated	federate	VERB
fcis-16180	4	6	learning	learning	NOUN
fcis-16180	4	7	schemes	scheme	NOUN
fcis-16180	4	8	still	still	ADV
fcis-16180	4	9	face	face	VERB
fcis-16180	4	10	two	two	NUM
fcis-16180	4	11	critical	critical	ADJ
fcis-16180	4	12	security	security	NOUN
fcis-16180	4	13	challenges	challenge	NOUN
fcis-16180	4	14	:	:	PUNCT
fcis-16180	4	15	user	user	NOUN
fcis-16180	4	16	privacy	privacy	NOUN
fcis-16180	4	17	disclosure	disclosure	NOUN
fcis-16180	4	18	and	and	CCONJ
fcis-16180	4	19	byzantine	byzantine	ADJ
fcis-16180	4	20	robustness	robustness	NOUN
fcis-16180	4	21	.	.	PUNCT
fcis-16180	5	1	the	the	DET
fcis-16180	5	2	adversary	adversary	NOUN
fcis-16180	5	3	may	may	AUX
fcis-16180	5	4	try	try	VERB
fcis-16180	5	5	to	to	PART
fcis-16180	5	6	infer	infer	VERB
fcis-16180	5	7	the	the	DET
fcis-16180	5	8	private	private	ADJ
fcis-16180	5	9	data	datum	NOUN
fcis-16180	5	10	from	from	ADP
fcis-16180	5	11	the	the	DET
fcis-16180	5	12	trained	train	VERB
fcis-16180	5	13	local	local	ADJ
fcis-16180	5	14	gradients	gradient	NOUN
fcis-16180	5	15	or	or	CCONJ
fcis-16180	5	16	compromise	compromise	VERB
fcis-16180	5	17	the	the	DET
fcis-16180	5	18	global	global	ADJ
fcis-16180	5	19	model	model	NOUN
fcis-16180	5	20	update	update	NOUN
fcis-16180	5	21	.	.	PUNCT
fcis-16180	6	1	to	to	PART
fcis-16180	6	2	tackle	tackle	VERB
fcis-16180	6	3	the	the	DET
fcis-16180	6	4	above	above	ADJ
fcis-16180	6	5	challenges	challenge	NOUN
fcis-16180	6	6	,	,	PUNCT
fcis-16180	6	7	we	we	PRON
fcis-16180	6	8	propose	propose	VERB
fcis-16180	6	9	ppbrfl	ppbrfl	NOUN
fcis-16180	6	10	,	,	PUNCT
fcis-16180	6	11	a	a	DET
fcis-16180	6	12	privacy	privacy	NOUN
fcis-16180	6	13	-	-	PUNCT
fcis-16180	6	14	preserving	preserve	VERB
fcis-16180	6	15	byzantinerobust	byzantinerobust	ADV
fcis-16180	6	16	federated	federated	ADJ
fcis-16180	6	17	learning	learning	NOUN
fcis-16180	6	18	scheme	scheme	NOUN
fcis-16180	6	19	.	.	PUNCT
fcis-16180	7	1	to	to	PART
fcis-16180	7	2	resist	resist	VERB
fcis-16180	7	3	byzantine	byzantine	ADJ
fcis-16180	7	4	attacks	attack	NOUN
fcis-16180	7	5	,	,	PUNCT
fcis-16180	7	6	we	we	PRON
fcis-16180	7	7	design	design	VERB
fcis-16180	7	8	a	a	DET
fcis-16180	7	9	novel	novel	ADJ
fcis-16180	7	10	byzantine	byzantine	ADJ
fcis-16180	7	11	-	-	PUNCT
fcis-16180	7	12	robust	robust	ADJ
fcis-16180	7	13	aggregation	aggregation	NOUN
fcis-16180	7	14	method	method	NOUN
fcis-16180	7	15	based	base	VERB
fcis-16180	7	16	on	on	ADP
fcis-16180	7	17	cosine	cosine	NOUN
fcis-16180	7	18	similarity	similarity	NOUN
fcis-16180	7	19	,	,	PUNCT
fcis-16180	7	20	which	which	PRON
fcis-16180	7	21	can	can	AUX
fcis-16180	7	22	guarantee	guarantee	VERB
fcis-16180	7	23	the	the	DET
fcis-16180	7	24	global	global	ADJ
fcis-16180	7	25	model	model	NOUN
fcis-16180	7	26	update	update	NOUN
fcis-16180	7	27	and	and	CCONJ
fcis-16180	7	28	improve	improve	VERB
fcis-16180	7	29	the	the	DET
fcis-16180	7	30	model	model	NOUN
fcis-16180	7	31	’s	’s	PART
fcis-16180	7	32	classification	classification	NOUN
fcis-16180	7	33	accuracy	accuracy	NOUN
fcis-16180	7	34	.	.	PUNCT
fcis-16180	8	1	furthermore	furthermore	ADV
fcis-16180	8	2	,	,	PUNCT
fcis-16180	8	3	we	we	PRON
fcis-16180	8	4	introduce	introduce	VERB
fcis-16180	8	5	a	a	DET
fcis-16180	8	6	reward	reward	NOUN
fcis-16180	8	7	and	and	CCONJ
fcis-16180	8	8	penalty	penalty	NOUN
fcis-16180	8	9	mechanism	mechanism	NOUN
fcis-16180	8	10	that	that	PRON
fcis-16180	8	11	considers	consider	VERB
fcis-16180	8	12	users	user	NOUN
fcis-16180	8	13	’	'	PUNCT
fcis-16180	8	14	behavior	behavior	NOUN
fcis-16180	8	15	to	to	PART
fcis-16180	8	16	mitigate	mitigate	VERB
fcis-16180	8	17	the	the	DET
fcis-16180	8	18	impact	impact	NOUN
fcis-16180	8	19	of	of	ADP
fcis-16180	8	20	byzantine	byzantine	ADJ
fcis-16180	8	21	users	user	NOUN
fcis-16180	8	22	on	on	ADP
fcis-16180	8	23	the	the	DET
fcis-16180	8	24	global	global	ADJ
fcis-16180	8	25	model	model	NOUN
fcis-16180	8	26	.	.	PUNCT
fcis-16180	9	1	to	to	PART
fcis-16180	9	2	protect	protect	VERB
fcis-16180	9	3	user	user	NOUN
fcis-16180	9	4	privacy	privacy	NOUN
fcis-16180	9	5	,	,	PUNCT
fcis-16180	9	6	we	we	PRON
fcis-16180	9	7	utilize	utilize	VERB
fcis-16180	9	8	symmetric	symmetric	ADJ
fcis-16180	9	9	homomorphic	homomorphic	ADJ
fcis-16180	9	10	encryption	encryption	NOUN
fcis-16180	9	11	to	to	PART
fcis-16180	9	12	encrypt	encrypt	VERB
fcis-16180	9	13	the	the	DET
fcis-16180	9	14	users	user	NOUN
fcis-16180	9	15	’	'	PUNCT
fcis-16180	9	16	trained	train	VERB
fcis-16180	9	17	local	local	ADJ
fcis-16180	9	18	models	model	NOUN
fcis-16180	9	19	,	,	PUNCT
fcis-16180	9	20	which	which	PRON
fcis-16180	9	21	requires	require	VERB
fcis-16180	9	22	low	low	ADJ
fcis-16180	9	23	computation	computation	NOUN
fcis-16180	9	24	cost	cost	NOUN
fcis-16180	9	25	while	while	SCONJ
fcis-16180	9	26	maintaining	maintain	VERB
fcis-16180	9	27	model	model	NOUN
fcis-16180	9	28	accuracy	accuracy	NOUN
fcis-16180	9	29	.	.	PUNCT
fcis-16180	10	1	we	we	PRON
fcis-16180	10	2	conduct	conduct	VERB
fcis-16180	10	3	the	the	DET
fcis-16180	10	4	experimental	experimental	ADJ
fcis-16180	10	5	assessment	assessment	NOUN
fcis-16180	10	6	of	of	ADP
fcis-16180	10	7	the	the	DET
fcis-16180	10	8	performance	performance	NOUN
fcis-16180	10	9	of	of	ADP
fcis-16180	10	10	ppbrfl	ppbrfl	NOUN
fcis-16180	10	11	.	.	PUNCT
fcis-16180	11	1	the	the	DET
fcis-16180	11	2	experimental	experimental	ADJ
fcis-16180	11	3	results	result	NOUN
fcis-16180	11	4	show	show	VERB
fcis-16180	11	5	that	that	SCONJ
fcis-16180	11	6	ppbrfl	ppbrfl	NOUN
fcis-16180	11	7	maintains	maintain	VERB
fcis-16180	11	8	model	model	NOUN
fcis-16180	11	9	classification	classification	NOUN
fcis-16180	11	10	accuracy	accuracy	NOUN
fcis-16180	11	11	while	while	SCONJ
fcis-16180	11	12	ensuring	ensure	VERB
fcis-16180	11	13	privacy	privacy	NOUN
fcis-16180	11	14	preservation	preservation	NOUN
fcis-16180	11	15	and	and	CCONJ
fcis-16180	11	16	byzantine	byzantine	ADJ
fcis-16180	11	17	robustness	robustness	NOUN
fcis-16180	11	18	compared	compare	VERB
fcis-16180	11	19	to	to	ADP
fcis-16180	11	20	traditional	traditional	ADJ
fcis-16180	11	21	federated	federated	ADJ
fcis-16180	11	22	learning	learning	NOUN
fcis-16180	11	23	scheme	scheme	NOUN
fcis-16180	11	24	.	.	PUNCT
fcis-16180	12	1	keywords	keyword	NOUN
fcis-16180	12	2	:	:	PUNCT
fcis-16180	12	3	federated	federated	ADJ
fcis-16180	12	4	learning	learning	NOUN
fcis-16180	12	5	;	;	PUNCT
fcis-16180	12	6	byzantine	byzantine	ADJ
fcis-16180	12	7	robustness	robustness	NOUN
fcis-16180	12	8	;	;	PUNCT
fcis-16180	12	9	privacy	privacy	NOUN
fcis-16180	12	10	-	-	PUNCT
fcis-16180	12	11	preserving	preserve	VERB
fcis-16180	12	12	;	;	PUNCT
fcis-16180	12	13	data	data	NOUN
fcis-16180	12	14	privacy	privacy	NOUN
fcis-16180	12	15	.	.	PUNCT
fcis-16180	13	1	1	1	X
fcis-16180	13	2	.	.	X
fcis-16180	13	3	introduction	introduction	NOUN
fcis-16180	13	4	machine	machine	NOUN
fcis-16180	13	5	learning	learning	NOUN
fcis-16180	13	6	has	have	AUX
fcis-16180	13	7	demonstrated	demonstrate	VERB
fcis-16180	13	8	exceptional	exceptional	ADJ
fcis-16180	13	9	performance	performance	NOUN
fcis-16180	13	10	in	in	ADP
fcis-16180	13	11	various	various	ADJ
fcis-16180	13	12	data	data	NOUN
fcis-16180	13	13	-	-	PUNCT
fcis-16180	13	14	driven	drive	VERB
fcis-16180	13	15	applications	application	NOUN
fcis-16180	13	16	,	,	PUNCT
fcis-16180	13	17	such	such	ADJ
fcis-16180	13	18	as	as	ADP
fcis-16180	13	19	financial	financial	ADJ
fcis-16180	13	20	analysis	analysis	NOUN
fcis-16180	13	21	[	[	X
fcis-16180	13	22	1	1	NUM
fcis-16180	13	23	]	]	PUNCT
fcis-16180	13	24	,	,	PUNCT
fcis-16180	13	25	medical	medical	ADJ
fcis-16180	13	26	diagnosis	diagnosis	NOUN
fcis-16180	13	27	[	[	X
fcis-16180	13	28	2	2	NUM
fcis-16180	13	29	]	]	PUNCT
fcis-16180	13	30	and	and	CCONJ
fcis-16180	13	31	speech	speech	NOUN
fcis-16180	13	32	recognition	recognition	NOUN
fcis-16180	14	1	[	[	X
fcis-16180	14	2	3	3	NUM
fcis-16180	14	3	]	]	PUNCT
fcis-16180	14	4	.	.	PUNCT
fcis-16180	15	1	in	in	ADP
fcis-16180	15	2	the	the	DET
fcis-16180	15	3	field	field	NOUN
fcis-16180	15	4	of	of	ADP
fcis-16180	15	5	finance	finance	NOUN
fcis-16180	15	6	,	,	PUNCT
fcis-16180	15	7	machine	machine	NOUN
fcis-16180	15	8	learning	learning	NOUN
fcis-16180	15	9	techniques	technique	NOUN
fcis-16180	15	10	are	be	AUX
fcis-16180	15	11	used	use	VERB
fcis-16180	15	12	to	to	PART
fcis-16180	15	13	predict	predict	VERB
fcis-16180	15	14	stock	stock	NOUN
fcis-16180	15	15	market	market	NOUN
fcis-16180	15	16	trends	trend	NOUN
fcis-16180	15	17	,	,	PUNCT
fcis-16180	15	18	while	while	SCONJ
fcis-16180	15	19	in	in	ADP
fcis-16180	15	20	the	the	DET
fcis-16180	15	21	medical	medical	ADJ
fcis-16180	15	22	field	field	NOUN
fcis-16180	15	23	,	,	PUNCT
fcis-16180	15	24	they	they	PRON
fcis-16180	15	25	are	be	AUX
fcis-16180	15	26	employed	employ	VERB
fcis-16180	15	27	for	for	ADP
fcis-16180	15	28	disease	disease	NOUN
fcis-16180	15	29	diagnosis	diagnosis	NOUN
fcis-16180	15	30	.	.	PUNCT
fcis-16180	16	1	the	the	DET
fcis-16180	16	2	development	development	NOUN
fcis-16180	16	3	of	of	ADP
fcis-16180	16	4	machine	machine	NOUN
fcis-16180	16	5	learning	learning	NOUN
fcis-16180	16	6	is	be	AUX
fcis-16180	16	7	predominantly	predominantly	ADV
fcis-16180	16	8	attributed	attribute	VERB
fcis-16180	16	9	to	to	ADP
fcis-16180	16	10	the	the	DET
fcis-16180	16	11	abundance	abundance	NOUN
fcis-16180	16	12	of	of	ADP
fcis-16180	16	13	extensive	extensive	ADJ
fcis-16180	16	14	datasets	dataset	NOUN
fcis-16180	16	15	.	.	PUNCT
fcis-16180	17	1	in	in	ADP
fcis-16180	17	2	machine	machine	NOUN
fcis-16180	17	3	learning	learning	NOUN
fcis-16180	17	4	,	,	PUNCT
fcis-16180	17	5	training	train	VERB
fcis-16180	17	6	a	a	DET
fcis-16180	17	7	powerful	powerful	ADJ
fcis-16180	17	8	and	and	CCONJ
fcis-16180	17	9	accurate	accurate	ADJ
fcis-16180	17	10	neural	neural	ADJ
fcis-16180	17	11	network	network	NOUN
fcis-16180	17	12	model	model	NOUN
fcis-16180	17	13	requires	require	VERB
fcis-16180	17	14	a	a	DET
fcis-16180	17	15	large	large	ADJ
fcis-16180	17	16	amount	amount	NOUN
fcis-16180	17	17	of	of	ADP
fcis-16180	17	18	user	user	NOUN
fcis-16180	17	19	data	datum	NOUN
fcis-16180	17	20	.	.	PUNCT
fcis-16180	18	1	however	however	ADV
fcis-16180	18	2	,	,	PUNCT
fcis-16180	18	3	utilizing	utilize	VERB
fcis-16180	18	4	traditional	traditional	ADJ
fcis-16180	18	5	centralized	centralized	ADJ
fcis-16180	18	6	machine	machine	NOUN
fcis-16180	18	7	learning	learning	NOUN
fcis-16180	18	8	for	for	ADP
fcis-16180	18	9	training	training	NOUN
fcis-16180	18	10	can	can	AUX
fcis-16180	18	11	lead	lead	VERB
fcis-16180	18	12	to	to	ADP
fcis-16180	18	13	the	the	DET
fcis-16180	18	14	leakage	leakage	NOUN
fcis-16180	18	15	of	of	ADP
fcis-16180	18	16	user	user	NOUN
fcis-16180	18	17	data	data	PROPN
fcis-16180	18	18	privacy	privacy	NOUN
fcis-16180	18	19	[	[	X
fcis-16180	18	20	4	4	NUM
fcis-16180	18	21	]	]	PUNCT
fcis-16180	18	22	.	.	PUNCT
fcis-16180	19	1	in	in	ADP
fcis-16180	19	2	2016	2016	NUM
fcis-16180	19	3	,	,	PUNCT
fcis-16180	19	4	google	google	PROPN
fcis-16180	19	5	introduced	introduce	VERB
fcis-16180	19	6	the	the	DET
fcis-16180	19	7	concept	concept	NOUN
fcis-16180	19	8	of	of	ADP
fcis-16180	19	9	federated	federated	ADJ
fcis-16180	19	10	learning	learning	NOUN
fcis-16180	19	11	[	[	X
fcis-16180	19	12	5	5	NUM
fcis-16180	19	13	]	]	PUNCT
fcis-16180	19	14	.	.	PUNCT
fcis-16180	20	1	federated	federated	ADJ
fcis-16180	20	2	learning	learning	NOUN
fcis-16180	20	3	is	be	AUX
fcis-16180	20	4	a	a	DET
fcis-16180	20	5	distributed	distribute	VERB
fcis-16180	20	6	machine	machine	NOUN
fcis-16180	20	7	learning	learn	VERB
fcis-16180	20	8	training	training	NOUN
fcis-16180	20	9	technique	technique	NOUN
fcis-16180	20	10	that	that	PRON
fcis-16180	20	11	can	can	AUX
fcis-16180	20	12	effectively	effectively	ADV
fcis-16180	20	13	tackle	tackle	VERB
fcis-16180	20	14	user	user	NOUN
fcis-16180	20	15	data	data	PROPN
fcis-16180	20	16	privacy	privacy	NOUN
fcis-16180	20	17	concerns	concern	NOUN
fcis-16180	20	18	[	[	X
fcis-16180	20	19	6	6	NUM
fcis-16180	20	20	]	]	PUNCT
fcis-16180	20	21	.	.	PUNCT
fcis-16180	21	1	in	in	ADP
fcis-16180	21	2	the	the	DET
fcis-16180	21	3	training	training	NOUN
fcis-16180	21	4	process	process	NOUN
fcis-16180	21	5	of	of	ADP
fcis-16180	21	6	federated	federated	ADJ
fcis-16180	21	7	learning	learning	NOUN
fcis-16180	21	8	,	,	PUNCT
fcis-16180	21	9	users	user	NOUN
fcis-16180	21	10	locally	locally	ADV
fcis-16180	21	11	train	train	VERB
fcis-16180	21	12	the	the	DET
fcis-16180	21	13	neural	neural	ADJ
fcis-16180	21	14	network	network	NOUN
fcis-16180	21	15	with	with	ADP
fcis-16180	21	16	their	their	PRON
fcis-16180	21	17	respective	respective	ADJ
fcis-16180	21	18	datasets	dataset	NOUN
fcis-16180	21	19	.	.	PUNCT
fcis-16180	22	1	after	after	SCONJ
fcis-16180	22	2	the	the	DET
fcis-16180	22	3	training	training	NOUN
fcis-16180	22	4	is	be	AUX
fcis-16180	22	5	completed	complete	VERB
fcis-16180	22	6	,	,	PUNCT
fcis-16180	22	7	the	the	DET
fcis-16180	22	8	trained	train	VERB
fcis-16180	22	9	local	local	ADJ
fcis-16180	22	10	gradients	gradient	NOUN
fcis-16180	22	11	are	be	AUX
fcis-16180	22	12	sent	send	VERB
fcis-16180	22	13	to	to	ADP
fcis-16180	22	14	the	the	DET
fcis-16180	22	15	cloud	cloud	NOUN
fcis-16180	22	16	server	server	NOUN
fcis-16180	22	17	for	for	ADP
fcis-16180	22	18	aggregation	aggregation	NOUN
fcis-16180	22	19	.	.	PUNCT
fcis-16180	23	1	in	in	ADP
fcis-16180	23	2	this	this	DET
fcis-16180	23	3	process	process	NOUN
fcis-16180	23	4	,	,	PUNCT
fcis-16180	23	5	the	the	DET
fcis-16180	23	6	user	user	NOUN
fcis-16180	23	7	data	data	PROPN
fcis-16180	23	8	privacy	privacy	NOUN
fcis-16180	23	9	can	can	AUX
fcis-16180	23	10	be	be	AUX
fcis-16180	23	11	protected	protect	VERB
fcis-16180	23	12	since	since	SCONJ
fcis-16180	23	13	the	the	DET
fcis-16180	23	14	user	user	NOUN
fcis-16180	23	15	’s	’s	PART
fcis-16180	23	16	private	private	ADJ
fcis-16180	23	17	data	datum	NOUN
fcis-16180	23	18	does	do	AUX
fcis-16180	23	19	not	not	PART
fcis-16180	23	20	require	require	VERB
fcis-16180	23	21	to	to	PART
fcis-16180	23	22	be	be	AUX
fcis-16180	23	23	uploaded	upload	VERB
fcis-16180	23	24	to	to	ADP
fcis-16180	23	25	the	the	DET
fcis-16180	23	26	cloud	cloud	NOUN
fcis-16180	23	27	server	server	NOUN
fcis-16180	23	28	.	.	PUNCT
fcis-16180	24	1	however	however	ADV
fcis-16180	24	2	,	,	PUNCT
fcis-16180	24	3	some	some	DET
fcis-16180	24	4	subsequent	subsequent	ADJ
fcis-16180	24	5	studies	study	NOUN
fcis-16180	24	6	[	[	X
fcis-16180	24	7	7	7	NUM
fcis-16180	24	8	]	]	PUNCT
fcis-16180	24	9	,	,	PUNCT
fcis-16180	24	10	[	[	X
fcis-16180	24	11	8	8	NUM
fcis-16180	24	12	]	]	PUNCT
fcis-16180	24	13	have	have	AUX
fcis-16180	24	14	demonstrated	demonstrate	VERB
fcis-16180	24	15	that	that	SCONJ
fcis-16180	24	16	the	the	DET
fcis-16180	24	17	cloud	cloud	NOUN
fcis-16180	24	18	server	server	NOUN
fcis-16180	24	19	can	can	AUX
fcis-16180	24	20	still	still	ADV
fcis-16180	24	21	infer	infer	VERB
fcis-16180	24	22	the	the	DET
fcis-16180	24	23	users	user	NOUN
fcis-16180	24	24	’	’	PART
fcis-16180	24	25	private	private	ADJ
fcis-16180	24	26	training	training	NOUN
fcis-16180	24	27	data	datum	NOUN
fcis-16180	24	28	from	from	ADP
fcis-16180	24	29	the	the	DET
fcis-16180	24	30	trained	train	VERB
fcis-16180	24	31	local	local	ADJ
fcis-16180	24	32	gradients	gradient	NOUN
fcis-16180	24	33	sent	send	VERB
fcis-16180	24	34	by	by	ADP
fcis-16180	24	35	the	the	DET
fcis-16180	24	36	users	user	NOUN
fcis-16180	24	37	.	.	PUNCT
fcis-16180	25	1	subsequently	subsequently	ADV
fcis-16180	25	2	,	,	PUNCT
fcis-16180	25	3	many	many	ADJ
fcis-16180	25	4	privacypreserving	privacypreserve	VERB
fcis-16180	25	5	federated	federated	ADJ
fcis-16180	25	6	learning	learning	NOUN
fcis-16180	25	7	schemes	scheme	NOUN
fcis-16180	25	8	[	[	X
fcis-16180	25	9	8	8	NUM
fcis-16180	25	10	-	-	SYM
fcis-16180	25	11	14	14	NUM
fcis-16180	25	12	]	]	PUNCT
fcis-16180	25	13	have	have	AUX
fcis-16180	25	14	been	be	AUX
fcis-16180	25	15	proposed	propose	VERB
fcis-16180	25	16	to	to	PART
fcis-16180	25	17	address	address	VERB
fcis-16180	25	18	the	the	DET
fcis-16180	25	19	privacy	privacy	NOUN
fcis-16180	25	20	implications	implication	NOUN
fcis-16180	25	21	in	in	ADP
fcis-16180	25	22	federated	federated	ADJ
fcis-16180	25	23	learning	learning	NOUN
fcis-16180	25	24	.	.	PUNCT
fcis-16180	26	1	these	these	DET
fcis-16180	26	2	schemes	scheme	NOUN
fcis-16180	26	3	encrypt	encrypt	VERB
fcis-16180	26	4	the	the	DET
fcis-16180	26	5	users	user	NOUN
fcis-16180	26	6	’	'	PUNCT
fcis-16180	26	7	trained	train	VERB
fcis-16180	26	8	local	local	ADJ
fcis-16180	26	9	gradients	gradient	NOUN
fcis-16180	26	10	by	by	ADP
fcis-16180	26	11	using	use	VERB
fcis-16180	26	12	cryptographic	cryptographic	ADJ
fcis-16180	26	13	primitives	primitive	NOUN
fcis-16180	26	14	such	such	ADJ
fcis-16180	26	15	as	as	ADP
fcis-16180	26	16	homomorphic	homomorphic	ADJ
fcis-16180	26	17	encryption	encryption	NOUN
fcis-16180	26	18	[	[	X
fcis-16180	26	19	7	7	NUM
fcis-16180	26	20	]	]	PUNCT
fcis-16180	26	21	,	,	PUNCT
fcis-16180	26	22	[	[	X
fcis-16180	26	23	10	10	NUM
fcis-16180	26	24	-	-	SYM
fcis-16180	26	25	11	11	NUM
fcis-16180	26	26	]	]	PUNCT
fcis-16180	26	27	,	,	PUNCT
fcis-16180	26	28	differential	differential	ADJ
fcis-16180	26	29	privacy	privacy	NOUN
fcis-16180	26	30	[	[	X
fcis-16180	26	31	14	14	NUM
fcis-16180	26	32	]	]	PUNCT
fcis-16180	26	33	and	and	CCONJ
fcis-16180	26	34	secure	secure	VERB
fcis-16180	26	35	multi	multi	ADJ
fcis-16180	26	36	-	-	ADJ
fcis-16180	26	37	party	party	ADJ
fcis-16180	26	38	computation	computation	NOUN
fcis-16180	27	1	[	[	X
fcis-16180	27	2	9	9	NUM
fcis-16180	27	3	]	]	PUNCT
fcis-16180	27	4	,	,	PUNCT
fcis-16180	28	1	[	[	X
fcis-16180	28	2	12	12	NUM
fcis-16180	28	3	-	-	SYM
fcis-16180	28	4	13	13	NUM
fcis-16180	28	5	]	]	PUNCT
fcis-16180	28	6	.	.	PUNCT
fcis-16180	29	1	furthermore	furthermore	ADV
fcis-16180	29	2	,	,	PUNCT
fcis-16180	29	3	federated	federated	ADJ
fcis-16180	29	4	learning	learning	NOUN
fcis-16180	29	5	may	may	AUX
fcis-16180	29	6	be	be	AUX
fcis-16180	29	7	vulnerable	vulnerable	ADJ
fcis-16180	29	8	to	to	ADP
fcis-16180	29	9	byzantine	byzantine	ADJ
fcis-16180	29	10	attacks	attack	NOUN
fcis-16180	29	11	[	[	X
fcis-16180	29	12	15	15	NUM
fcis-16180	29	13	]	]	PUNCT
fcis-16180	29	14	.	.	PUNCT
fcis-16180	30	1	byzantine	byzantine	ADJ
fcis-16180	30	2	users	user	NOUN
fcis-16180	30	3	can	can	AUX
fcis-16180	30	4	upload	upload	VERB
fcis-16180	30	5	false	false	ADJ
fcis-16180	30	6	parameters	parameter	NOUN
fcis-16180	30	7	to	to	ADP
fcis-16180	30	8	the	the	DET
fcis-16180	30	9	cloud	cloud	NOUN
fcis-16180	30	10	server	server	NOUN
fcis-16180	30	11	,	,	PUNCT
fcis-16180	30	12	thereby	thereby	ADV
fcis-16180	30	13	compromising	compromise	VERB
fcis-16180	30	14	global	global	ADJ
fcis-16180	30	15	model	model	NOUN
fcis-16180	30	16	’s	’s	PART
fcis-16180	30	17	accuracy	accuracy	NOUN
fcis-16180	30	18	or	or	CCONJ
fcis-16180	30	19	manipulating	manipulate	VERB
fcis-16180	30	20	the	the	DET
fcis-16180	30	21	global	global	ADJ
fcis-16180	30	22	model	model	NOUN
fcis-16180	30	23	to	to	PART
fcis-16180	30	24	produce	produce	VERB
fcis-16180	30	25	specific	specific	ADJ
fcis-16180	30	26	outputs	output	NOUN
fcis-16180	30	27	through	through	ADP
fcis-16180	30	28	poisoning	poisoning	NOUN
fcis-16180	30	29	attacks	attack	NOUN
fcis-16180	30	30	[	[	X
fcis-16180	30	31	10	10	NUM
fcis-16180	30	32	]	]	PUNCT
fcis-16180	30	33	.	.	PUNCT
fcis-16180	31	1	such	such	ADJ
fcis-16180	31	2	actions	action	NOUN
fcis-16180	31	3	can	can	AUX
fcis-16180	31	4	severely	severely	ADV
fcis-16180	31	5	undermine	undermine	VERB
fcis-16180	31	6	the	the	DET
fcis-16180	31	7	robustness	robustness	NOUN
fcis-16180	31	8	of	of	ADP
fcis-16180	31	9	the	the	DET
fcis-16180	31	10	federated	federated	ADJ
fcis-16180	31	11	learning	learning	NOUN
fcis-16180	31	12	.	.	PUNCT
fcis-16180	32	1	to	to	PART
fcis-16180	32	2	resist	resist	VERB
fcis-16180	32	3	byzantine	byzantine	ADJ
fcis-16180	32	4	attacks	attack	NOUN
fcis-16180	32	5	,	,	PUNCT
fcis-16180	32	6	a	a	DET
fcis-16180	32	7	lot	lot	NOUN
fcis-16180	32	8	of	of	ADP
fcis-16180	32	9	byzantine	byzantine	ADJ
fcis-16180	32	10	-	-	ADJ
fcis-16180	32	11	robust	robust	ADJ
fcis-16180	32	12	federated	federated	ADJ
fcis-16180	32	13	learning	learning	NOUN
fcis-16180	32	14	schemes	scheme	NOUN
fcis-16180	32	15	have	have	AUX
fcis-16180	32	16	been	be	AUX
fcis-16180	32	17	constructed	construct	VERB
fcis-16180	32	18	.	.	PUNCT
fcis-16180	33	1	blanchard	blanchard	PROPN
fcis-16180	33	2	et	et	PROPN
fcis-16180	33	3	al	al	PROPN
fcis-16180	33	4	.	.	PUNCT
fcis-16180	34	1	[	[	X
fcis-16180	34	2	16	16	NUM
fcis-16180	34	3	]	]	PUNCT
fcis-16180	34	4	calculated	calculate	VERB
fcis-16180	34	5	the	the	DET
fcis-16180	34	6	euclidean	euclidean	ADJ
fcis-16180	34	7	distance	distance	NOUN
fcis-16180	34	8	between	between	ADP
fcis-16180	34	9	two	two	NUM
fcis-16180	34	10	users	user	NOUN
fcis-16180	34	11	to	to	PART
fcis-16180	34	12	filter	filter	VERB
fcis-16180	34	13	out	out	ADP
fcis-16180	34	14	the	the	DET
fcis-16180	34	15	aggregated	aggregate	VERB
fcis-16180	34	16	users	user	NOUN
fcis-16180	34	17	.	.	PUNCT
fcis-16180	35	1	yin	yin	PROPN
fcis-16180	35	2	et	et	PROPN
fcis-16180	35	3	al	al	PROPN
fcis-16180	35	4	.	.	PUNCT
fcis-16180	36	1	[	[	X
fcis-16180	36	2	17	17	NUM
fcis-16180	36	3	]	]	PUNCT
fcis-16180	36	4	utilized	utilize	VERB
fcis-16180	36	5	the	the	DET
fcis-16180	36	6	mean	mean	NOUN
fcis-16180	36	7	and	and	CCONJ
fcis-16180	36	8	median	median	NOUN
fcis-16180	36	9	of	of	ADP
fcis-16180	36	10	user	user	NOUN
fcis-16180	36	11	models	model	NOUN
fcis-16180	36	12	as	as	ADP
fcis-16180	36	13	the	the	DET
fcis-16180	36	14	global	global	ADJ
fcis-16180	36	15	model	model	NOUN
fcis-16180	36	16	for	for	ADP
fcis-16180	36	17	updating	update	VERB
fcis-16180	36	18	.	.	PUNCT
fcis-16180	37	1	cao	cao	PROPN
fcis-16180	37	2	et	et	PROPN
fcis-16180	37	3	al	al	PROPN
fcis-16180	37	4	.	.	PUNCT
fcis-16180	38	1	[	[	X
fcis-16180	38	2	18	18	NUM
fcis-16180	38	3	]	]	PUNCT
fcis-16180	38	4	combined	combined	ADJ
fcis-16180	38	5	cosine	cosine	NOUN
fcis-16180	38	6	similarity	similarity	NOUN
fcis-16180	38	7	with	with	ADP
fcis-16180	38	8	the	the	DET
fcis-16180	38	9	relu	relu	NOUN
fcis-16180	38	10	function	function	NOUN
fcis-16180	38	11	to	to	PART
fcis-16180	38	12	assign	assign	VERB
fcis-16180	38	13	model	model	NOUN
fcis-16180	38	14	update	update	NOUN
fcis-16180	38	15	weight	weight	NOUN
fcis-16180	38	16	to	to	ADP
fcis-16180	38	17	each	each	DET
fcis-16180	38	18	user	user	NOUN
fcis-16180	38	19	.	.	PUNCT
fcis-16180	39	1	however	however	ADV
fcis-16180	39	2	,	,	PUNCT
fcis-16180	39	3	most	most	ADJ
fcis-16180	39	4	of	of	ADP
fcis-16180	39	5	the	the	DET
fcis-16180	39	6	existing	exist	VERB
fcis-16180	39	7	byzantine	byzantine	ADJ
fcis-16180	39	8	-	-	PUNCT
fcis-16180	39	9	robust	robust	ADJ
fcis-16180	39	10	federated	federated	ADJ
fcis-16180	39	11	learning	learn	VERB
fcis-16180	39	12	schemes	scheme	NOUN
fcis-16180	39	13	do	do	AUX
fcis-16180	39	14	not	not	PART
fcis-16180	39	15	consider	consider	VERB
fcis-16180	39	16	the	the	DET
fcis-16180	39	17	issue	issue	NOUN
fcis-16180	39	18	of	of	ADP
fcis-16180	39	19	privacy	privacy	NOUN
fcis-16180	39	20	,	,	PUNCT
fcis-16180	39	21	which	which	PRON
fcis-16180	39	22	leads	lead	VERB
fcis-16180	39	23	to	to	ADP
fcis-16180	39	24	privacy	privacy	NOUN
fcis-16180	39	25	leakage	leakage	NOUN
fcis-16180	39	26	.	.	PUNCT
fcis-16180	40	1	robustness	robustness	NOUN
fcis-16180	40	2	and	and	CCONJ
fcis-16180	40	3	privacy	privacy	NOUN
fcis-16180	40	4	are	be	AUX
fcis-16180	40	5	two	two	NUM
fcis-16180	40	6	critical	critical	ADJ
fcis-16180	40	7	issues	issue	NOUN
fcis-16180	40	8	that	that	PRON
fcis-16180	40	9	need	need	VERB
fcis-16180	40	10	to	to	PART
fcis-16180	40	11	be	be	AUX
fcis-16180	40	12	solved	solve	VERB
fcis-16180	40	13	in	in	ADP
fcis-16180	40	14	federated	federated	ADJ
fcis-16180	40	15	learning	learning	NOUN
fcis-16180	40	16	.	.	PUNCT
fcis-16180	41	1	in	in	ADP
fcis-16180	41	2	this	this	DET
fcis-16180	41	3	paper	paper	NOUN
fcis-16180	41	4	,	,	PUNCT
fcis-16180	41	5	we	we	PRON
fcis-16180	41	6	propose	propose	VERB
fcis-16180	41	7	a	a	DET
fcis-16180	41	8	novel	novel	ADJ
fcis-16180	41	9	privacy	privacy	NOUN
fcis-16180	41	10	-	-	PUNCT
fcis-16180	41	11	preserving	preserve	VERB
fcis-16180	41	12	byzantine	byzantine	ADJ
fcis-16180	41	13	-	-	PUNCT
fcis-16180	41	14	robust	robust	ADJ
fcis-16180	41	15	federated	federated	ADJ
fcis-16180	41	16	learning	learning	NOUN
fcis-16180	41	17	scheme	scheme	NOUN
fcis-16180	41	18	(	(	PUNCT
fcis-16180	41	19	ppbrfl	ppbrfl	NOUN
fcis-16180	41	20	)	)	PUNCT
fcis-16180	41	21	,	,	PUNCT
fcis-16180	41	22	which	which	PRON
fcis-16180	41	23	can	can	AUX
fcis-16180	41	24	achieve	achieve	VERB
fcis-16180	41	25	privacy	privacy	NOUN
fcis-16180	41	26	protection	protection	NOUN
fcis-16180	41	27	and	and	CCONJ
fcis-16180	41	28	resist	resist	VERB
fcis-16180	41	29	byzantine	byzantine	ADJ
fcis-16180	41	30	attacks	attack	NOUN
fcis-16180	41	31	.	.	PUNCT
fcis-16180	42	1	furthermore	furthermore	ADV
fcis-16180	42	2	,	,	PUNCT
fcis-16180	42	3	ppbrfl	ppbrfl	NOUN
fcis-16180	42	4	allows	allow	VERB
fcis-16180	42	5	users	user	NOUN
fcis-16180	42	6	to	to	PART
fcis-16180	42	7	drop	drop	VERB
fcis-16180	42	8	out	out	ADP
fcis-16180	42	9	in	in	ADP
fcis-16180	42	10	the	the	DET
fcis-16180	42	11	training	training	NOUN
fcis-16180	42	12	phase	phase	NOUN
fcis-16180	42	13	.	.	PUNCT
fcis-16180	43	1	contribution	contribution	NOUN
fcis-16180	43	2	.	.	PUNCT
fcis-16180	44	1	the	the	DET
fcis-16180	44	2	contributions	contribution	NOUN
fcis-16180	44	3	of	of	ADP
fcis-16180	44	4	this	this	DET
fcis-16180	44	5	paper	paper	NOUN
fcis-16180	44	6	are	be	AUX
fcis-16180	44	7	summarized	summarize	VERB
fcis-16180	44	8	as	as	SCONJ
fcis-16180	44	9	follows	follow	VERB
fcis-16180	44	10	:	:	PUNCT
fcis-16180	44	11	(	(	PUNCT
fcis-16180	44	12	1	1	X
fcis-16180	44	13	)	)	PUNCT
fcis-16180	44	14	we	we	PRON
fcis-16180	44	15	construct	construct	VERB
fcis-16180	44	16	ppbrfl	ppbrfl	NOUN
fcis-16180	44	17	,	,	PUNCT
fcis-16180	44	18	a	a	DET
fcis-16180	44	19	novel	novel	ADJ
fcis-16180	44	20	scheme	scheme	NOUN
fcis-16180	44	21	to	to	PART
fcis-16180	44	22	achieve	achieve	VERB
fcis-16180	44	23	privacy	privacy	NOUN
fcis-16180	44	24	protection	protection	NOUN
fcis-16180	44	25	and	and	CCONJ
fcis-16180	44	26	byzantine	byzantine	ADJ
fcis-16180	44	27	robustness	robustness	NOUN
fcis-16180	44	28	.	.	PUNCT
fcis-16180	45	1	we	we	PRON
fcis-16180	45	2	design	design	VERB
fcis-16180	45	3	a	a	DET
fcis-16180	45	4	novel	novel	ADJ
fcis-16180	45	5	byzantine	byzantine	ADJ
fcis-16180	45	6	-	-	PUNCT
fcis-16180	45	7	robust	robust	ADJ
fcis-16180	45	8	aggregation	aggregation	NOUN
fcis-16180	45	9	method	method	NOUN
fcis-16180	45	10	based	base	VERB
fcis-16180	45	11	on	on	ADP
fcis-16180	45	12	cosine	cosine	NOUN
fcis-16180	45	13	similarity	similarity	NOUN
fcis-16180	45	14	.	.	PUNCT
fcis-16180	46	1	specifically	specifically	ADV
fcis-16180	46	2	,	,	PUNCT
fcis-16180	46	3	each	each	DET
fcis-16180	46	4	user	user	NOUN
fcis-16180	46	5	computes	compute	VERB
fcis-16180	46	6	the	the	DET
fcis-16180	46	7	cosine	cosine	NOUN
fcis-16180	46	8	similarity	similarity	NOUN
fcis-16180	46	9	of	of	ADP
fcis-16180	46	10	local	local	ADJ
fcis-16180	46	11	model	model	NOUN
fcis-16180	46	12	and	and	CCONJ
fcis-16180	46	13	trained	train	VERB
fcis-16180	46	14	model	model	NOUN
fcis-16180	46	15	.	.	PUNCT
fcis-16180	47	1	based	base	VERB
fcis-16180	47	2	on	on	ADP
fcis-16180	47	3	the	the	DET
fcis-16180	47	4	cosine	cosine	NOUN
fcis-16180	47	5	similarity	similarity	NOUN
fcis-16180	47	6	values	value	NOUN
fcis-16180	47	7	of	of	ADP
fcis-16180	47	8	users	user	NOUN
fcis-16180	47	9	,	,	PUNCT
fcis-16180	47	10	the	the	DET
fcis-16180	47	11	cloud	cloud	NOUN
fcis-16180	47	12	server	server	NOUN
fcis-16180	47	13	selects	select	VERB
fcis-16180	47	14	a	a	DET
fcis-16180	47	15	subset	subset	NOUN
fcis-16180	47	16	of	of	ADP
fcis-16180	47	17	users	user	NOUN
fcis-16180	47	18	and	and	CCONJ
fcis-16180	47	19	aggregates	aggregate	VERB
fcis-16180	47	20	these	these	DET
fcis-16180	47	21	selected	select	VERB
fcis-16180	47	22	users	user	NOUN
fcis-16180	47	23	’	'	PUNCT
fcis-16180	47	24	encrypted	encrypt	VERB
fcis-16180	47	25	trained	train	VERB
fcis-16180	47	26	models	model	NOUN
fcis-16180	47	27	instead	instead	ADV
fcis-16180	47	28	of	of	ADP
fcis-16180	47	29	aggregating	aggregate	VERB
fcis-16180	47	30	the	the	DET
fcis-16180	47	31	encrypted	encrypt	VERB
fcis-16180	47	32	trained	train	VERB
fcis-16180	47	33	models	model	NOUN
fcis-16180	47	34	of	of	ADP
fcis-16180	47	35	all	all	DET
fcis-16180	47	36	users	user	NOUN
fcis-16180	47	37	.	.	PUNCT
fcis-16180	48	1	using	use	VERB
fcis-16180	48	2	this	this	DET
fcis-16180	48	3	aggregation	aggregation	NOUN
fcis-16180	48	4	method	method	NOUN
fcis-16180	48	5	,	,	PUNCT
fcis-16180	48	6	ppbrfl	ppbrfl	NOUN
fcis-16180	48	7	is	be	AUX
fcis-16180	48	8	resistant	resistant	ADJ
fcis-16180	48	9	to	to	ADP
fcis-16180	48	10	byzantine	byzantine	ADJ
fcis-16180	48	11	attacks	attack	NOUN
fcis-16180	48	12	,	,	PUNCT
fcis-16180	48	13	thus	thus	ADV
fcis-16180	48	14	improving	improve	VERB
fcis-16180	48	15	the	the	DET
fcis-16180	48	16	model	model	NOUN
fcis-16180	48	17	’s	’s	PART
fcis-16180	48	18	classification	classification	NOUN
fcis-16180	48	19	accuracy	accuracy	NOUN
fcis-16180	48	20	.	.	PUNCT
fcis-16180	49	1	furthermore	furthermore	ADV
fcis-16180	49	2	,	,	PUNCT
fcis-16180	49	3	we	we	PRON
fcis-16180	49	4	present	present	VERB
fcis-16180	49	5	a	a	DET
fcis-16180	49	6	reward	reward	NOUN
fcis-16180	49	7	and	and	CCONJ
fcis-16180	49	8	penalty	penalty	NOUN
fcis-16180	49	9	mechanism	mechanism	NOUN
fcis-16180	49	10	that	that	PRON
fcis-16180	49	11	takes	take	VERB
fcis-16180	49	12	into	into	ADP
fcis-16180	49	13	account	account	NOUN
fcis-16180	49	14	the	the	DET
fcis-16180	49	15	behavior	behavior	NOUN
fcis-16180	49	16	of	of	ADP
fcis-16180	49	17	the	the	DET
fcis-16180	49	18	users	user	NOUN
fcis-16180	49	19	to	to	PART
fcis-16180	49	20	mitigate	mitigate	VERB
fcis-16180	49	21	the	the	DET
fcis-16180	49	22	influence	influence	NOUN
fcis-16180	49	23	of	of	ADP
fcis-16180	49	24	byzantine	byzantine	ADJ
fcis-16180	49	25	users	user	NOUN
fcis-16180	49	26	on	on	ADP
fcis-16180	49	27	the	the	DET
fcis-16180	49	28	global	global	ADJ
fcis-16180	49	29	model	model	NOUN
fcis-16180	49	30	and	and	CCONJ
fcis-16180	49	31	safeguard	safeguard	VERB
fcis-16180	49	32	the	the	DET
fcis-16180	49	33	fairness	fairness	NOUN
fcis-16180	49	34	of	of	ADP
fcis-16180	49	35	users	user	NOUN
fcis-16180	49	36	.	.	PUNCT
fcis-16180	50	1	(	(	PUNCT
fcis-16180	50	2	2	2	X
fcis-16180	50	3	)	)	PUNCT
fcis-16180	50	4	we	we	PRON
fcis-16180	50	5	use	use	VERB
fcis-16180	50	6	symmetric	symmetric	ADJ
fcis-16180	50	7	homomorphic	homomorphic	ADJ
fcis-16180	50	8	encryption	encryption	NOUN
fcis-16180	50	9	to	to	PART
fcis-16180	50	10	encrypt	encrypt	VERB
fcis-16180	50	11	the	the	DET
fcis-16180	50	12	users	user	NOUN
fcis-16180	50	13	’	'	PUNCT
fcis-16180	50	14	trained	train	VERB
fcis-16180	50	15	local	local	ADJ
fcis-16180	50	16	models	model	NOUN
fcis-16180	50	17	and	and	CCONJ
fcis-16180	50	18	achieve	achieve	VERB
fcis-16180	50	19	privacy	privacy	NOUN
fcis-16180	50	20	19	19	NUM
fcis-16180	50	21	protection	protection	NOUN
fcis-16180	50	22	of	of	ADP
fcis-16180	50	23	user	user	NOUN
fcis-16180	50	24	data	data	PROPN
fcis-16180	50	25	.	.	PUNCT
fcis-16180	51	1	compared	compare	VERB
fcis-16180	51	2	with	with	ADP
fcis-16180	51	3	traditional	traditional	ADJ
fcis-16180	51	4	homomorphic	homomorphic	ADJ
fcis-16180	51	5	encryption	encryption	NOUN
fcis-16180	51	6	,	,	PUNCT
fcis-16180	51	7	symmetric	symmetric	ADJ
fcis-16180	51	8	homomorphic	homomorphic	ADJ
fcis-16180	51	9	encryption	encryption	NOUN
fcis-16180	51	10	reduces	reduce	VERB
fcis-16180	51	11	the	the	DET
fcis-16180	51	12	computation	computation	NOUN
fcis-16180	51	13	cost	cost	NOUN
fcis-16180	51	14	while	while	SCONJ
fcis-16180	51	15	maintaining	maintain	VERB
fcis-16180	51	16	the	the	DET
fcis-16180	51	17	accuracy	accuracy	NOUN
fcis-16180	51	18	of	of	ADP
fcis-16180	51	19	the	the	DET
fcis-16180	51	20	model	model	NOUN
fcis-16180	51	21	.	.	PUNCT
fcis-16180	52	1	(	(	PUNCT
fcis-16180	52	2	3	3	X
fcis-16180	52	3	)	)	PUNCT
fcis-16180	52	4	we	we	PRON
fcis-16180	52	5	evaluate	evaluate	VERB
fcis-16180	52	6	the	the	DET
fcis-16180	52	7	performance	performance	NOUN
fcis-16180	52	8	of	of	ADP
fcis-16180	52	9	ppbrfl	ppbrfl	NOUN
fcis-16180	52	10	by	by	ADP
fcis-16180	52	11	conducting	conduct	VERB
fcis-16180	52	12	experiments	experiment	NOUN
fcis-16180	52	13	on	on	ADP
fcis-16180	52	14	two	two	NUM
fcis-16180	52	15	well	well	ADV
fcis-16180	52	16	-	-	PUNCT
fcis-16180	52	17	known	know	VERB
fcis-16180	52	18	datasets	dataset	NOUN
fcis-16180	52	19	and	and	CCONJ
fcis-16180	52	20	comparing	compare	VERB
fcis-16180	52	21	ppbrfl	ppbrfl	NOUN
fcis-16180	52	22	with	with	ADP
fcis-16180	52	23	existing	exist	VERB
fcis-16180	52	24	state	state	NOUN
fcis-16180	52	25	-	-	PUNCT
fcis-16180	52	26	of	of	ADP
fcis-16180	52	27	-	-	PUNCT
fcis-16180	52	28	the	the	DET
fcis-16180	52	29	-	-	PUNCT
fcis-16180	52	30	art	art	NOUN
fcis-16180	52	31	schemes	scheme	NOUN
fcis-16180	52	32	.	.	PUNCT
fcis-16180	53	1	the	the	DET
fcis-16180	53	2	experimental	experimental	ADJ
fcis-16180	53	3	results	result	NOUN
fcis-16180	53	4	indicate	indicate	VERB
fcis-16180	53	5	that	that	SCONJ
fcis-16180	53	6	ppbrfl	ppbrfl	NOUN
fcis-16180	53	7	exhibits	exhibit	VERB
fcis-16180	53	8	superior	superior	ADJ
fcis-16180	53	9	performance	performance	NOUN
fcis-16180	53	10	in	in	ADP
fcis-16180	53	11	terms	term	NOUN
fcis-16180	53	12	of	of	ADP
fcis-16180	53	13	both	both	DET
fcis-16180	53	14	classification	classification	NOUN
fcis-16180	53	15	accuracy	accuracy	NOUN
fcis-16180	53	16	and	and	CCONJ
fcis-16180	53	17	efficiency	efficiency	NOUN
fcis-16180	53	18	,	,	PUNCT
fcis-16180	53	19	while	while	SCONJ
fcis-16180	53	20	achieving	achieve	VERB
fcis-16180	53	21	privacy	privacy	NOUN
fcis-16180	53	22	protection	protection	NOUN
fcis-16180	53	23	and	and	CCONJ
fcis-16180	53	24	byzantine	byzantine	ADJ
fcis-16180	53	25	robustness	robustness	NOUN
fcis-16180	53	26	.	.	PUNCT
fcis-16180	54	1	2	2	X
fcis-16180	54	2	.	.	NUM
fcis-16180	54	3	related	relate	VERB
fcis-16180	54	4	works	work	NOUN
fcis-16180	54	5	currently	currently	ADV
fcis-16180	54	6	,	,	PUNCT
fcis-16180	54	7	federated	federated	ADJ
fcis-16180	54	8	learning	learning	NOUN
fcis-16180	54	9	has	have	AUX
fcis-16180	54	10	garnered	garner	VERB
fcis-16180	54	11	substantial	substantial	ADJ
fcis-16180	54	12	attention	attention	NOUN
fcis-16180	54	13	across	across	ADP
fcis-16180	54	14	various	various	ADJ
fcis-16180	54	15	industries	industry	NOUN
fcis-16180	54	16	and	and	CCONJ
fcis-16180	54	17	is	be	AUX
fcis-16180	54	18	being	be	AUX
fcis-16180	54	19	adopted	adopt	VERB
fcis-16180	54	20	in	in	ADP
fcis-16180	54	21	a	a	DET
fcis-16180	54	22	growing	grow	VERB
fcis-16180	54	23	array	array	NOUN
fcis-16180	54	24	of	of	ADP
fcis-16180	54	25	applications	application	NOUN
fcis-16180	54	26	[	[	X
fcis-16180	54	27	19	19	NUM
fcis-16180	54	28	]	]	PUNCT
fcis-16180	54	29	.	.	PUNCT
fcis-16180	55	1	the	the	DET
fcis-16180	55	2	extensive	extensive	ADJ
fcis-16180	55	3	adoption	adoption	NOUN
fcis-16180	55	4	of	of	ADP
fcis-16180	55	5	federated	federated	ADJ
fcis-16180	55	6	learning	learning	NOUN
fcis-16180	55	7	accentuates	accentuate	VERB
fcis-16180	55	8	the	the	DET
fcis-16180	55	9	significance	significance	NOUN
fcis-16180	55	10	of	of	ADP
fcis-16180	55	11	privacy	privacy	NOUN
fcis-16180	55	12	and	and	CCONJ
fcis-16180	55	13	security	security	NOUN
fcis-16180	55	14	concerns	concern	NOUN
fcis-16180	55	15	associated	associate	VERB
fcis-16180	55	16	with	with	ADP
fcis-16180	55	17	this	this	DET
fcis-16180	55	18	approach	approach	NOUN
fcis-16180	55	19	.	.	PUNCT
fcis-16180	56	1	privacy	privacy	NOUN
fcis-16180	56	2	leakage	leakage	NOUN
fcis-16180	56	3	in	in	ADP
fcis-16180	56	4	federated	federated	ADJ
fcis-16180	56	5	learning	learning	NOUN
fcis-16180	56	6	occurs	occur	VERB
fcis-16180	56	7	mainly	mainly	ADV
fcis-16180	56	8	through	through	ADP
fcis-16180	56	9	the	the	DET
fcis-16180	56	10	intermediate	intermediate	ADJ
fcis-16180	56	11	gradients	gradient	NOUN
fcis-16180	56	12	uploaded	upload	VERB
fcis-16180	56	13	by	by	ADP
fcis-16180	56	14	users	user	NOUN
fcis-16180	56	15	during	during	ADP
fcis-16180	56	16	model	model	NOUN
fcis-16180	56	17	training	training	NOUN
fcis-16180	56	18	[	[	X
fcis-16180	56	19	20	20	NUM
fcis-16180	56	20	]	]	PUNCT
fcis-16180	56	21	.	.	PUNCT
fcis-16180	57	1	therefore	therefore	ADV
fcis-16180	57	2	,	,	PUNCT
fcis-16180	57	3	the	the	DET
fcis-16180	57	4	crucial	crucial	ADJ
fcis-16180	57	5	aspect	aspect	NOUN
fcis-16180	57	6	of	of	ADP
fcis-16180	57	7	safeguarding	safeguard	VERB
fcis-16180	57	8	privacy	privacy	NOUN
fcis-16180	57	9	lies	lie	VERB
fcis-16180	57	10	in	in	ADP
fcis-16180	57	11	encrypting	encrypt	VERB
fcis-16180	57	12	the	the	DET
fcis-16180	57	13	gradients	gradient	NOUN
fcis-16180	57	14	or	or	CCONJ
fcis-16180	57	15	models	model	NOUN
fcis-16180	57	16	uploaded	upload	VERB
fcis-16180	57	17	by	by	ADP
fcis-16180	57	18	users	user	NOUN
fcis-16180	57	19	.	.	PUNCT
fcis-16180	58	1	to	to	PART
fcis-16180	58	2	achieve	achieve	VERB
fcis-16180	58	3	private	private	ADJ
fcis-16180	58	4	protection	protection	NOUN
fcis-16180	58	5	,	,	PUNCT
fcis-16180	58	6	many	many	ADJ
fcis-16180	58	7	privacypreserving	privacypreserve	VERB
fcis-16180	58	8	federated	federated	ADJ
fcis-16180	58	9	learning	learning	NOUN
fcis-16180	58	10	schemes	scheme	NOUN
fcis-16180	58	11	[	[	X
fcis-16180	58	12	21	21	NUM
fcis-16180	58	13	-	-	SYM
fcis-16180	58	14	24	24	NUM
fcis-16180	58	15	]	]	PUNCT
fcis-16180	58	16	are	be	AUX
fcis-16180	58	17	proposed	propose	VERB
fcis-16180	58	18	.	.	PUNCT
fcis-16180	59	1	bonawitz	bonawitz	NOUN
fcis-16180	59	2	et	et	PROPN
fcis-16180	59	3	al	al	PROPN
fcis-16180	59	4	.	.	PUNCT
fcis-16180	60	1	[	[	X
fcis-16180	60	2	6	6	NUM
fcis-16180	60	3	]	]	PUNCT
fcis-16180	60	4	achieve	achieve	VERB
fcis-16180	60	5	privacy	privacy	NOUN
fcis-16180	60	6	protection	protection	NOUN
fcis-16180	60	7	through	through	ADP
fcis-16180	60	8	secret	secret	ADJ
fcis-16180	60	9	sharing	sharing	NOUN
fcis-16180	60	10	techniques	technique	NOUN
fcis-16180	60	11	in	in	ADP
fcis-16180	60	12	federated	federated	ADJ
fcis-16180	60	13	learning	learning	NOUN
fcis-16180	60	14	.	.	PUNCT
fcis-16180	61	1	phong	phong	PROPN
fcis-16180	61	2	et	et	PROPN
fcis-16180	61	3	al	al	PROPN
fcis-16180	61	4	.	.	PUNCT
fcis-16180	62	1	[	[	X
fcis-16180	62	2	7	7	X
fcis-16180	62	3	]	]	PUNCT
fcis-16180	62	4	proposed	propose	VERB
fcis-16180	62	5	a	a	DET
fcis-16180	62	6	privacy	privacy	NOUN
fcis-16180	62	7	-	-	PUNCT
fcis-16180	62	8	preserving	preserve	VERB
fcis-16180	62	9	federated	federated	ADJ
fcis-16180	62	10	learning	learning	NOUN
fcis-16180	62	11	scheme	scheme	NOUN
fcis-16180	62	12	based	base	VERB
fcis-16180	62	13	on	on	ADP
fcis-16180	62	14	additive	additive	ADJ
fcis-16180	62	15	homomorphic	homomorphic	ADJ
fcis-16180	62	16	encryption	encryption	NOUN
fcis-16180	62	17	that	that	PRON
fcis-16180	62	18	is	be	AUX
fcis-16180	62	19	able	able	ADJ
fcis-16180	62	20	to	to	PART
fcis-16180	62	21	train	train	VERB
fcis-16180	62	22	the	the	DET
fcis-16180	62	23	model	model	NOUN
fcis-16180	62	24	without	without	ADP
fcis-16180	62	25	loss	loss	NOUN
fcis-16180	62	26	of	of	ADP
fcis-16180	62	27	accuracy	accuracy	NOUN
fcis-16180	62	28	.	.	PUNCT
fcis-16180	63	1	ma	ma	PROPN
fcis-16180	63	2	et	et	PROPN
fcis-16180	63	3	al	al	PROPN
fcis-16180	63	4	.	.	PUNCT
fcis-16180	64	1	[	[	X
fcis-16180	64	2	25	25	NUM
fcis-16180	64	3	]	]	PUNCT
fcis-16180	64	4	used	use	VERB
fcis-16180	64	5	two	two	NUM
fcis-16180	64	6	-	-	PUNCT
fcis-16180	64	7	trapdoor	trapdoor	NOUN
fcis-16180	64	8	homomorphic	homomorphic	ADJ
fcis-16180	64	9	encryption	encryption	NOUN
fcis-16180	64	10	to	to	PART
fcis-16180	64	11	protect	protect	VERB
fcis-16180	64	12	the	the	DET
fcis-16180	64	13	users	user	NOUN
fcis-16180	64	14	’	’	PART
fcis-16180	64	15	data	datum	NOUN
fcis-16180	64	16	privacy	privacy	NOUN
fcis-16180	64	17	.	.	PUNCT
fcis-16180	65	1	fang	fang	PROPN
fcis-16180	65	2	et	et	PROPN
fcis-16180	65	3	al	al	PROPN
fcis-16180	65	4	.	.	PUNCT
fcis-16180	66	1	[	[	X
fcis-16180	66	2	26	26	NUM
fcis-16180	66	3	]	]	X
fcis-16180	66	4	utilized	utilize	VERB
fcis-16180	66	5	elgamal	elgamal	ADJ
fcis-16180	66	6	multiplicative	multiplicative	ADJ
fcis-16180	66	7	homomorphic	homomorphic	ADJ
fcis-16180	66	8	encryption	encryption	NOUN
fcis-16180	66	9	technology	technology	NOUN
fcis-16180	66	10	to	to	PART
fcis-16180	66	11	construct	construct	VERB
fcis-16180	66	12	a	a	DET
fcis-16180	66	13	privacypreserving	privacypreserving	NOUN
fcis-16180	66	14	federated	federated	ADJ
fcis-16180	66	15	learning	learning	NOUN
fcis-16180	66	16	scheme	scheme	NOUN
fcis-16180	66	17	.	.	PUNCT
fcis-16180	67	1	differential	differential	ADJ
fcis-16180	67	2	privacy	privacy	NOUN
fcis-16180	67	3	is	be	AUX
fcis-16180	67	4	a	a	DET
fcis-16180	67	5	commonly	commonly	ADV
fcis-16180	67	6	used	use	VERB
fcis-16180	67	7	privacy	privacy	NOUN
fcis-16180	67	8	-	-	PUNCT
fcis-16180	67	9	preserving	preserve	VERB
fcis-16180	67	10	technique	technique	NOUN
fcis-16180	67	11	in	in	ADP
fcis-16180	67	12	federated	federated	ADJ
fcis-16180	67	13	learning	learning	NOUN
fcis-16180	67	14	.	.	PUNCT
fcis-16180	68	1	jia	jia	PROPN
fcis-16180	68	2	et	et	PROPN
fcis-16180	68	3	al	al	PROPN
fcis-16180	68	4	.	.	PUNCT
fcis-16180	69	1	[	[	X
fcis-16180	69	2	27	27	NUM
fcis-16180	69	3	]	]	PUNCT
fcis-16180	69	4	and	and	CCONJ
fcis-16180	69	5	zhou	zhou	PROPN
fcis-16180	69	6	et	et	PROPN
fcis-16180	69	7	al	al	PROPN
fcis-16180	69	8	.	.	PUNCT
fcis-16180	70	1	[	[	X
fcis-16180	70	2	28	28	NUM
fcis-16180	70	3	]	]	X
fcis-16180	70	4	used	use	VERB
fcis-16180	70	5	differential	differential	NOUN
fcis-16180	70	6	privacy	privacy	NOUN
fcis-16180	70	7	to	to	PART
fcis-16180	70	8	protect	protect	VERB
fcis-16180	70	9	the	the	DET
fcis-16180	70	10	user	user	NOUN
fcis-16180	70	11	’s	’s	PART
fcis-16180	70	12	data	data	PROPN
fcis-16180	70	13	privacy	privacy	PROPN
fcis-16180	70	14	.	.	PUNCT
fcis-16180	71	1	based	base	VERB
fcis-16180	71	2	on	on	ADP
fcis-16180	71	3	differential	differential	ADJ
fcis-16180	71	4	privacy	privacy	NOUN
fcis-16180	71	5	and	and	CCONJ
fcis-16180	71	6	secure	secure	VERB
fcis-16180	71	7	multi	multi	ADJ
fcis-16180	71	8	-	-	ADJ
fcis-16180	71	9	party	party	ADJ
fcis-16180	71	10	computation	computation	NOUN
fcis-16180	71	11	techniques	technique	NOUN
fcis-16180	71	12	,	,	PUNCT
fcis-16180	71	13	mugunthan	mugunthan	NOUN
fcis-16180	71	14	et	et	PROPN
fcis-16180	71	15	al	al	PROPN
fcis-16180	71	16	.	.	PUNCT
fcis-16180	72	1	[	[	X
fcis-16180	72	2	29	29	NUM
fcis-16180	72	3	]	]	PUNCT
fcis-16180	72	4	designed	design	VERB
fcis-16180	72	5	a	a	DET
fcis-16180	72	6	privacypreserving	privacypreserve	VERB
fcis-16180	72	7	federated	federated	ADJ
fcis-16180	72	8	learning	learning	NOUN
fcis-16180	72	9	scheme	scheme	NOUN
fcis-16180	72	10	by	by	ADP
fcis-16180	72	11	applying	apply	VERB
fcis-16180	72	12	a	a	DET
fcis-16180	72	13	combination	combination	NOUN
fcis-16180	72	14	of	of	ADP
fcis-16180	72	15	these	these	DET
fcis-16180	72	16	two	two	NUM
fcis-16180	72	17	techniques	technique	NOUN
fcis-16180	72	18	to	to	ADP
fcis-16180	72	19	federated	federated	ADJ
fcis-16180	72	20	learning	learning	NOUN
fcis-16180	72	21	.	.	PUNCT
fcis-16180	73	1	zhou	zhou	PROPN
fcis-16180	73	2	et	et	PROPN
fcis-16180	73	3	al	al	PROPN
fcis-16180	73	4	.	.	PUNCT
fcis-16180	74	1	[	[	X
fcis-16180	74	2	30	30	NUM
fcis-16180	74	3	]	]	PUNCT
fcis-16180	74	4	introduced	introduce	VERB
fcis-16180	74	5	a	a	DET
fcis-16180	74	6	privacypreserving	privacypreserving	NOUN
fcis-16180	74	7	federated	federated	ADJ
fcis-16180	74	8	learning	learning	NOUN
fcis-16180	74	9	approach	approach	NOUN
fcis-16180	74	10	,	,	PUNCT
fcis-16180	74	11	which	which	PRON
fcis-16180	74	12	introduces	introduce	VERB
fcis-16180	74	13	a	a	DET
fcis-16180	74	14	trusted	trust	VERB
fcis-16180	74	15	blinding	blind	VERB
fcis-16180	74	16	server	server	NOUN
fcis-16180	74	17	to	to	PART
fcis-16180	74	18	blind	blind	VERB
fcis-16180	74	19	the	the	DET
fcis-16180	74	20	gradient	gradient	ADJ
fcis-16180	74	21	ciphertexts	ciphertexts	NOUN
fcis-16180	74	22	in	in	ADP
fcis-16180	74	23	the	the	DET
fcis-16180	74	24	federated	federated	ADJ
fcis-16180	74	25	learning	learning	NOUN
fcis-16180	74	26	process	process	NOUN
fcis-16180	74	27	.	.	PUNCT
fcis-16180	75	1	moreover	moreover	ADV
fcis-16180	75	2	,	,	PUNCT
fcis-16180	75	3	this	this	DET
fcis-16180	75	4	scheme	scheme	NOUN
fcis-16180	75	5	can	can	AUX
fcis-16180	75	6	effectively	effectively	ADV
fcis-16180	75	7	solve	solve	VERB
fcis-16180	75	8	the	the	DET
fcis-16180	75	9	problem	problem	NOUN
fcis-16180	75	10	of	of	ADP
fcis-16180	75	11	collusion	collusion	NOUN
fcis-16180	75	12	between	between	ADP
fcis-16180	75	13	cloud	cloud	NOUN
fcis-16180	75	14	servers	server	NOUN
fcis-16180	75	15	and	and	CCONJ
fcis-16180	75	16	users	user	NOUN
fcis-16180	75	17	.	.	PUNCT
fcis-16180	76	1	to	to	PART
fcis-16180	76	2	resist	resist	VERB
fcis-16180	76	3	byzantine	byzantine	ADJ
fcis-16180	76	4	user	user	NOUN
fcis-16180	76	5	attacks	attack	NOUN
fcis-16180	76	6	and	and	CCONJ
fcis-16180	76	7	improve	improve	VERB
fcis-16180	76	8	byzantine	byzantine	ADJ
fcis-16180	76	9	robustness	robustness	NOUN
fcis-16180	76	10	in	in	ADP
fcis-16180	76	11	federated	federated	ADJ
fcis-16180	76	12	learning	learning	NOUN
fcis-16180	76	13	,	,	PUNCT
fcis-16180	76	14	a	a	DET
fcis-16180	76	15	number	number	NOUN
fcis-16180	76	16	of	of	ADP
fcis-16180	76	17	byzantinerobust	byzantinerobust	ADJ
fcis-16180	76	18	federated	federated	ADJ
fcis-16180	76	19	learning	learning	NOUN
fcis-16180	76	20	schemes	scheme	NOUN
fcis-16180	76	21	[	[	X
fcis-16180	76	22	15	15	NUM
fcis-16180	76	23	]	]	PUNCT
fcis-16180	76	24	,	,	PUNCT
fcis-16180	76	25	[	[	X
fcis-16180	76	26	31	31	NUM
fcis-16180	76	27	-	-	SYM
fcis-16180	76	28	35	35	NUM
fcis-16180	76	29	]	]	PUNCT
fcis-16180	76	30	have	have	AUX
fcis-16180	76	31	been	be	AUX
fcis-16180	76	32	proposed	propose	VERB
fcis-16180	76	33	.	.	PUNCT
fcis-16180	77	1	blanchard	blanchard	PROPN
fcis-16180	77	2	et	et	PROPN
fcis-16180	77	3	al	al	PROPN
fcis-16180	77	4	.	.	PUNCT
fcis-16180	78	1	[	[	X
fcis-16180	78	2	16	16	NUM
fcis-16180	78	3	]	]	PUNCT
fcis-16180	78	4	introduced	introduce	VERB
fcis-16180	78	5	the	the	DET
fcis-16180	78	6	krum	krum	PROPN
fcis-16180	78	7	scheme	scheme	PROPN
fcis-16180	78	8	,	,	PUNCT
fcis-16180	78	9	which	which	PRON
fcis-16180	78	10	selects	select	VERB
fcis-16180	78	11	specific	specific	ADJ
fcis-16180	78	12	users	user	NOUN
fcis-16180	78	13	for	for	ADP
fcis-16180	78	14	gradient	gradient	ADJ
fcis-16180	78	15	aggregation	aggregation	NOUN
fcis-16180	78	16	based	base	VERB
fcis-16180	78	17	on	on	ADP
fcis-16180	78	18	their	their	PRON
fcis-16180	78	19	euclidean	euclidean	ADJ
fcis-16180	78	20	distance	distance	NOUN
fcis-16180	78	21	,	,	PUNCT
fcis-16180	78	22	aiming	aim	VERB
fcis-16180	78	23	to	to	PART
fcis-16180	78	24	reduce	reduce	VERB
fcis-16180	78	25	the	the	DET
fcis-16180	78	26	impact	impact	NOUN
fcis-16180	78	27	of	of	ADP
fcis-16180	78	28	byzantine	byzantine	ADJ
fcis-16180	78	29	users	user	NOUN
fcis-16180	78	30	on	on	ADP
fcis-16180	78	31	the	the	DET
fcis-16180	78	32	accuracy	accuracy	NOUN
fcis-16180	78	33	of	of	ADP
fcis-16180	78	34	trained	train	VERB
fcis-16180	78	35	models	model	NOUN
fcis-16180	78	36	in	in	ADP
fcis-16180	78	37	federated	federated	ADJ
fcis-16180	78	38	learning	learning	NOUN
fcis-16180	78	39	.	.	PUNCT
fcis-16180	79	1	in	in	ADP
fcis-16180	79	2	yin	yin	PROPN
fcis-16180	79	3	et	et	PROPN
fcis-16180	79	4	al	al	PROPN
fcis-16180	79	5	.	.	PROPN
fcis-16180	79	6	’s	’s	PART
fcis-16180	79	7	scheme	scheme	NOUN
fcis-16180	79	8	[	[	X
fcis-16180	79	9	17	17	NUM
fcis-16180	79	10	]	]	PUNCT
fcis-16180	79	11	,	,	PUNCT
fcis-16180	79	12	each	each	DET
fcis-16180	79	13	component	component	NOUN
fcis-16180	79	14	of	of	ADP
fcis-16180	79	15	the	the	DET
fcis-16180	79	16	global	global	ADJ
fcis-16180	79	17	model	model	NOUN
fcis-16180	79	18	is	be	AUX
fcis-16180	79	19	updated	update	VERB
fcis-16180	79	20	individually	individually	ADV
fcis-16180	79	21	,	,	PUNCT
fcis-16180	79	22	with	with	ADP
fcis-16180	79	23	each	each	DET
fcis-16180	79	24	part	part	NOUN
fcis-16180	79	25	being	be	AUX
fcis-16180	79	26	updated	update	VERB
fcis-16180	79	27	by	by	ADP
fcis-16180	79	28	taking	take	VERB
fcis-16180	79	29	the	the	DET
fcis-16180	79	30	average	average	NOUN
fcis-16180	79	31	or	or	CCONJ
fcis-16180	79	32	median	median	NOUN
fcis-16180	79	33	of	of	ADP
fcis-16180	79	34	the	the	DET
fcis-16180	79	35	local	local	ADJ
fcis-16180	79	36	models	model	NOUN
fcis-16180	79	37	from	from	ADP
fcis-16180	79	38	all	all	DET
fcis-16180	79	39	users	user	NOUN
fcis-16180	79	40	.	.	PUNCT
fcis-16180	80	1	this	this	DET
fcis-16180	80	2	approach	approach	NOUN
fcis-16180	80	3	helps	help	VERB
fcis-16180	80	4	to	to	PART
fcis-16180	80	5	mitigate	mitigate	VERB
fcis-16180	80	6	the	the	DET
fcis-16180	80	7	impact	impact	NOUN
fcis-16180	80	8	caused	cause	VERB
fcis-16180	80	9	by	by	ADP
fcis-16180	80	10	byzantine	byzantine	ADJ
fcis-16180	80	11	users	user	NOUN
fcis-16180	80	12	.	.	PUNCT
fcis-16180	81	1	wang	wang	PROPN
fcis-16180	81	2	et	et	PROPN
fcis-16180	81	3	al	al	PROPN
fcis-16180	81	4	.	.	PUNCT
fcis-16180	82	1	[	[	X
fcis-16180	82	2	36	36	NUM
fcis-16180	82	3	]	]	PUNCT
fcis-16180	82	4	designed	design	VERB
fcis-16180	82	5	a	a	DET
fcis-16180	82	6	model	model	NOUN
fcis-16180	82	7	segmentation	segmentation	NOUN
fcis-16180	82	8	aggregation	aggregation	NOUN
fcis-16180	82	9	strategy	strategy	NOUN
fcis-16180	82	10	to	to	PART
fcis-16180	82	11	defend	defend	VERB
fcis-16180	82	12	against	against	ADP
fcis-16180	82	13	attacks	attack	NOUN
fcis-16180	82	14	by	by	ADP
fcis-16180	82	15	byzantine	byzantine	ADJ
fcis-16180	82	16	users	user	NOUN
fcis-16180	82	17	.	.	PUNCT
fcis-16180	83	1	in	in	ADP
fcis-16180	83	2	wang	wang	PROPN
fcis-16180	83	3	et	et	PROPN
fcis-16180	83	4	al	al	PROPN
fcis-16180	83	5	.	.	PUNCT
fcis-16180	84	1	[	[	X
fcis-16180	84	2	36	36	NUM
fcis-16180	84	3	]	]	PUNCT
fcis-16180	84	4	’s	’s	PART
fcis-16180	84	5	scheme	scheme	NOUN
fcis-16180	84	6	,	,	PUNCT
fcis-16180	84	7	the	the	DET
fcis-16180	84	8	classification	classification	NOUN
fcis-16180	84	9	accuracy	accuracy	NOUN
fcis-16180	84	10	can	can	AUX
fcis-16180	84	11	be	be	AUX
fcis-16180	84	12	improved	improve	VERB
fcis-16180	84	13	by	by	ADP
fcis-16180	84	14	using	use	VERB
fcis-16180	84	15	pairwise	pairwise	NOUN
fcis-16180	84	16	adjusted	adjust	VERB
fcis-16180	84	17	cosine	cosine	NOUN
fcis-16180	84	18	similarity	similarity	NOUN
fcis-16180	84	19	.	.	PUNCT
fcis-16180	85	1	li	li	PROPN
fcis-16180	85	2	et	et	PROPN
fcis-16180	85	3	al	al	PROPN
fcis-16180	85	4	.	.	PUNCT
fcis-16180	86	1	[	[	X
fcis-16180	86	2	37	37	NUM
fcis-16180	86	3	]	]	PUNCT
fcis-16180	86	4	constructed	construct	VERB
fcis-16180	86	5	a	a	DET
fcis-16180	86	6	byzantine	byzantine	ADJ
fcis-16180	86	7	-	-	PUNCT
fcis-16180	86	8	robust	robust	ADJ
fcis-16180	86	9	federated	federated	ADJ
fcis-16180	86	10	learning	learning	NOUN
fcis-16180	86	11	scheme	scheme	NOUN
fcis-16180	86	12	based	base	VERB
fcis-16180	86	13	on	on	ADP
fcis-16180	86	14	an	an	DET
fcis-16180	86	15	auto	auto	NOUN
fcis-16180	86	16	-	-	PUNCT
fcis-16180	86	17	weighted	weight	VERB
fcis-16180	86	18	geometric	geometric	ADJ
fcis-16180	86	19	median	median	ADJ
fcis-16180	86	20	method	method	NOUN
fcis-16180	86	21	.	.	PUNCT
fcis-16180	87	1	cao	cao	PROPN
fcis-16180	87	2	et	et	PROPN
fcis-16180	87	3	al	al	PROPN
fcis-16180	87	4	.	.	PUNCT
fcis-16180	88	1	[	[	X
fcis-16180	88	2	18	18	NUM
fcis-16180	88	3	]	]	PUNCT
fcis-16180	88	4	combined	combine	VERB
fcis-16180	88	5	the	the	DET
fcis-16180	88	6	relu	relu	NOUN
fcis-16180	88	7	function	function	NOUN
fcis-16180	88	8	with	with	ADP
fcis-16180	88	9	cosine	cosine	NOUN
fcis-16180	88	10	similarity	similarity	NOUN
fcis-16180	88	11	to	to	PART
fcis-16180	88	12	assign	assign	VERB
fcis-16180	88	13	model	model	NOUN
fcis-16180	88	14	update	update	NOUN
fcis-16180	88	15	weight	weight	NOUN
fcis-16180	88	16	to	to	ADP
fcis-16180	88	17	each	each	DET
fcis-16180	88	18	user	user	NOUN
fcis-16180	88	19	.	.	PUNCT
fcis-16180	89	1	in	in	ADP
fcis-16180	89	2	this	this	DET
fcis-16180	89	3	way	way	NOUN
fcis-16180	89	4	,	,	PUNCT
fcis-16180	89	5	honest	honest	ADJ
fcis-16180	89	6	users	user	NOUN
fcis-16180	89	7	are	be	AUX
fcis-16180	89	8	allocated	allocate	VERB
fcis-16180	89	9	larger	large	ADJ
fcis-16180	89	10	update	update	NOUN
fcis-16180	89	11	weights	weight	NOUN
fcis-16180	89	12	while	while	SCONJ
fcis-16180	89	13	malicious	malicious	ADJ
fcis-16180	89	14	byzantine	byzantine	ADJ
fcis-16180	89	15	users	user	NOUN
fcis-16180	89	16	are	be	AUX
fcis-16180	89	17	assigned	assign	VERB
fcis-16180	89	18	smaller	small	ADJ
fcis-16180	89	19	update	update	NOUN
fcis-16180	89	20	weights	weight	NOUN
fcis-16180	89	21	,	,	PUNCT
fcis-16180	89	22	thereby	thereby	ADV
fcis-16180	89	23	enhancing	enhance	VERB
fcis-16180	89	24	the	the	DET
fcis-16180	89	25	byzantine	byzantine	ADJ
fcis-16180	89	26	robustness	robustness	NOUN
fcis-16180	89	27	of	of	ADP
fcis-16180	89	28	this	this	DET
fcis-16180	89	29	scheme	scheme	NOUN
fcis-16180	89	30	.	.	PUNCT
fcis-16180	90	1	in	in	ADP
fcis-16180	90	2	tang	tang	PROPN
fcis-16180	90	3	et	et	PROPN
fcis-16180	90	4	al	al	PROPN
fcis-16180	90	5	.	.	PROPN
fcis-16180	90	6	’s	’s	PART
fcis-16180	90	7	scheme	scheme	NOUN
fcis-16180	91	1	[	[	X
fcis-16180	91	2	38	38	NUM
fcis-16180	91	3	]	]	PUNCT
fcis-16180	91	4	,	,	PUNCT
fcis-16180	91	5	the	the	DET
fcis-16180	91	6	server	server	NOUN
fcis-16180	91	7	can	can	AUX
fcis-16180	91	8	determine	determine	VERB
fcis-16180	91	9	whether	whether	SCONJ
fcis-16180	91	10	the	the	DET
fcis-16180	91	11	users	user	NOUN
fcis-16180	91	12	are	be	AUX
fcis-16180	91	13	honest	honest	ADJ
fcis-16180	91	14	or	or	CCONJ
fcis-16180	91	15	not	not	PART
fcis-16180	91	16	by	by	ADP
fcis-16180	91	17	verifying	verify	VERB
fcis-16180	91	18	the	the	DET
fcis-16180	91	19	gradients	gradient	NOUN
fcis-16180	91	20	uploaded	upload	VERB
fcis-16180	91	21	by	by	ADP
fcis-16180	91	22	these	these	DET
fcis-16180	91	23	users	user	NOUN
fcis-16180	91	24	.	.	PUNCT
fcis-16180	92	1	in	in	ADP
fcis-16180	92	2	this	this	DET
fcis-16180	92	3	way	way	NOUN
fcis-16180	92	4	,	,	PUNCT
fcis-16180	92	5	the	the	DET
fcis-16180	92	6	server	server	NOUN
fcis-16180	92	7	can	can	AUX
fcis-16180	92	8	avoid	avoid	VERB
fcis-16180	92	9	using	use	VERB
fcis-16180	92	10	the	the	DET
fcis-16180	92	11	wrong	wrong	ADJ
fcis-16180	92	12	gradients	gradient	NOUN
fcis-16180	92	13	sent	send	VERB
fcis-16180	92	14	by	by	ADP
fcis-16180	92	15	the	the	DET
fcis-16180	92	16	byzantine	byzantine	ADJ
fcis-16180	92	17	users	user	NOUN
fcis-16180	92	18	for	for	ADP
fcis-16180	92	19	aggregation	aggregation	NOUN
fcis-16180	92	20	,	,	PUNCT
fcis-16180	92	21	which	which	PRON
fcis-16180	92	22	prevents	prevent	VERB
fcis-16180	92	23	the	the	DET
fcis-16180	92	24	byzantine	byzantine	ADJ
fcis-16180	92	25	users	user	NOUN
fcis-16180	92	26	from	from	ADP
fcis-16180	92	27	influencing	influence	VERB
fcis-16180	92	28	the	the	DET
fcis-16180	92	29	training	training	NOUN
fcis-16180	92	30	of	of	ADP
fcis-16180	92	31	federated	federated	ADJ
fcis-16180	92	32	learning	learning	NOUN
fcis-16180	92	33	.	.	PUNCT
fcis-16180	93	1	however	however	ADV
fcis-16180	93	2	,	,	PUNCT
fcis-16180	93	3	the	the	DET
fcis-16180	93	4	above	above	ADJ
fcis-16180	93	5	byzantine	byzantine	ADJ
fcis-16180	93	6	-	-	PUNCT
fcis-16180	93	7	robust	robust	ADJ
fcis-16180	93	8	federated	federated	ADJ
fcis-16180	93	9	learning	learning	NOUN
fcis-16180	93	10	schemes	scheme	NOUN
fcis-16180	93	11	can	can	AUX
fcis-16180	93	12	not	not	PART
fcis-16180	93	13	ensure	ensure	VERB
fcis-16180	93	14	the	the	DET
fcis-16180	93	15	users	user	NOUN
fcis-16180	93	16	’	’	PART
fcis-16180	93	17	data	datum	NOUN
fcis-16180	93	18	privacy	privacy	NOUN
fcis-16180	93	19	.	.	PUNCT
fcis-16180	94	1	therefore	therefore	ADV
fcis-16180	94	2	,	,	PUNCT
fcis-16180	94	3	it	it	PRON
fcis-16180	94	4	is	be	AUX
fcis-16180	94	5	meaning	mean	VERB
fcis-16180	94	6	to	to	PART
fcis-16180	94	7	explore	explore	VERB
fcis-16180	94	8	a	a	DET
fcis-16180	94	9	federated	federated	ADJ
fcis-16180	94	10	learning	learning	NOUN
fcis-16180	94	11	scheme	scheme	NOUN
fcis-16180	94	12	that	that	PRON
fcis-16180	94	13	can	can	AUX
fcis-16180	94	14	protect	protect	VERB
fcis-16180	94	15	users	user	NOUN
fcis-16180	94	16	’	’	PART
fcis-16180	94	17	data	datum	NOUN
fcis-16180	94	18	privacy	privacy	NOUN
fcis-16180	94	19	while	while	SCONJ
fcis-16180	94	20	maintaining	maintain	VERB
fcis-16180	94	21	byzantine	byzantine	ADJ
fcis-16180	94	22	robustness	robustness	NOUN
fcis-16180	94	23	.	.	PUNCT
fcis-16180	95	1	3	3	X
fcis-16180	95	2	.	.	X
fcis-16180	95	3	the	the	DET
fcis-16180	95	4	construction	construction	NOUN
fcis-16180	95	5	of	of	ADP
fcis-16180	95	6	ppbrfl	ppbrfl	NOUN
fcis-16180	95	7	3.1	3.1	NUM
fcis-16180	95	8	.	.	PUNCT
fcis-16180	95	9	system	system	NOUN
fcis-16180	95	10	model	model	NOUN
fcis-16180	95	11	as	as	SCONJ
fcis-16180	95	12	shown	show	VERB
fcis-16180	95	13	in	in	ADP
fcis-16180	95	14	fig.1	fig.1	PROPN
fcis-16180	95	15	,	,	PUNCT
fcis-16180	95	16	ppbrfl	ppbrfl	NOUN
fcis-16180	95	17	is	be	AUX
fcis-16180	95	18	mainly	mainly	ADV
fcis-16180	95	19	composed	compose	VERB
fcis-16180	95	20	of	of	ADP
fcis-16180	95	21	three	three	NUM
fcis-16180	95	22	entities	entity	NOUN
fcis-16180	95	23	:	:	PUNCT
fcis-16180	95	24	key	key	ADJ
fcis-16180	95	25	generator	generator	NOUN
fcis-16180	95	26	center	center	NOUN
fcis-16180	95	27	(	(	PUNCT
fcis-16180	95	28	kgc	kgc	PROPN
fcis-16180	95	29	)	)	PUNCT
fcis-16180	95	30	,	,	PUNCT
fcis-16180	95	31	users	user	NOUN
fcis-16180	95	32	and	and	CCONJ
fcis-16180	95	33	cloud	cloud	PROPN
fcis-16180	95	34	server	server	NOUN
fcis-16180	95	35	.	.	PROPN
fcis-16180	96	1	•	•	PROPN
fcis-16180	96	2	kgc	kgc	PROPN
fcis-16180	96	3	:	:	PUNCT
fcis-16180	96	4	kgc	kgc	PROPN
fcis-16180	96	5	is	be	AUX
fcis-16180	96	6	an	an	DET
fcis-16180	96	7	honest	honest	ADJ
fcis-16180	96	8	entity	entity	NOUN
fcis-16180	96	9	that	that	PRON
fcis-16180	96	10	is	be	AUX
fcis-16180	96	11	responsible	responsible	ADJ
fcis-16180	96	12	for	for	ADP
fcis-16180	96	13	generating	generate	VERB
fcis-16180	96	14	the	the	DET
fcis-16180	96	15	keys	key	NOUN
fcis-16180	96	16	and	and	CCONJ
fcis-16180	96	17	the	the	DET
fcis-16180	96	18	public	public	ADJ
fcis-16180	96	19	parameters	parameter	NOUN
fcis-16180	96	20	for	for	ADP
fcis-16180	96	21	the	the	DET
fcis-16180	96	22	users	user	NOUN
fcis-16180	96	23	.	.	PUNCT
fcis-16180	97	1	•	•	NUM
fcis-16180	97	2	users	user	NOUN
fcis-16180	97	3	:	:	PUNCT
fcis-16180	97	4	the	the	DET
fcis-16180	97	5	users	user	NOUN
fcis-16180	97	6	train	train	VERB
fcis-16180	97	7	the	the	DET
fcis-16180	97	8	model	model	NOUN
fcis-16180	97	9	with	with	ADP
fcis-16180	97	10	their	their	PRON
fcis-16180	97	11	own	own	ADJ
fcis-16180	97	12	local	local	ADJ
fcis-16180	97	13	datasets	dataset	NOUN
fcis-16180	97	14	.	.	PUNCT
fcis-16180	98	1	then	then	ADV
fcis-16180	98	2	,	,	PUNCT
fcis-16180	98	3	the	the	DET
fcis-16180	98	4	user	user	NOUN
fcis-16180	98	5	calculates	calculate	VERB
fcis-16180	98	6	the	the	DET
fcis-16180	98	7	value	value	NOUN
fcis-16180	98	8	of	of	ADP
fcis-16180	98	9	cosine	cosine	NOUN
fcis-16180	98	10	similarity	similarity	NOUN
fcis-16180	98	11	between	between	ADP
fcis-16180	98	12	the	the	DET
fcis-16180	98	13	local	local	ADJ
fcis-16180	98	14	model	model	NOUN
fcis-16180	98	15	and	and	CCONJ
fcis-16180	98	16	the	the	DET
fcis-16180	98	17	trained	train	VERB
fcis-16180	98	18	model	model	NOUN
fcis-16180	98	19	,	,	PUNCT
fcis-16180	98	20	and	and	CCONJ
fcis-16180	98	21	encrypts	encrypt	VERB
fcis-16180	98	22	the	the	DET
fcis-16180	98	23	trained	train	VERB
fcis-16180	98	24	model	model	NOUN
fcis-16180	98	25	.	.	PUNCT
fcis-16180	99	1	finally	finally	ADV
fcis-16180	99	2	,	,	PUNCT
fcis-16180	99	3	the	the	DET
fcis-16180	99	4	user	user	NOUN
fcis-16180	99	5	sends	send	VERB
fcis-16180	99	6	the	the	DET
fcis-16180	99	7	encrypted	encrypt	VERB
fcis-16180	99	8	trained	train	VERB
fcis-16180	99	9	model	model	NOUN
fcis-16180	99	10	along	along	ADP
fcis-16180	99	11	with	with	ADP
fcis-16180	99	12	the	the	DET
fcis-16180	99	13	cosine	cosine	NOUN
fcis-16180	99	14	similarity	similarity	NOUN
fcis-16180	99	15	value	value	NOUN
fcis-16180	99	16	to	to	ADP
fcis-16180	99	17	the	the	DET
fcis-16180	99	18	cloud	cloud	NOUN
fcis-16180	99	19	server	server	NOUN
fcis-16180	99	20	.	.	PUNCT
fcis-16180	99	21	•	•	PROPN
fcis-16180	99	22	cloud	cloud	PROPN
fcis-16180	99	23	server	server	NOUN
fcis-16180	99	24	:	:	PUNCT
fcis-16180	99	25	the	the	DET
fcis-16180	99	26	cloud	cloud	NOUN
fcis-16180	99	27	server	server	NOUN
fcis-16180	99	28	is	be	AUX
fcis-16180	99	29	responsible	responsible	ADJ
fcis-16180	99	30	for	for	ADP
fcis-16180	99	31	selecting	select	VERB
fcis-16180	99	32	some	some	DET
fcis-16180	99	33	users	user	NOUN
fcis-16180	99	34	and	and	CCONJ
fcis-16180	99	35	aggregating	aggregate	VERB
fcis-16180	99	36	the	the	DET
fcis-16180	99	37	encrypted	encrypt	VERB
fcis-16180	99	38	trained	train	VERB
fcis-16180	99	39	models	model	NOUN
fcis-16180	99	40	of	of	ADP
fcis-16180	99	41	these	these	DET
fcis-16180	99	42	selected	select	VERB
fcis-16180	99	43	users	user	NOUN
fcis-16180	99	44	.	.	PUNCT
fcis-16180	100	1	specifically	specifically	ADV
fcis-16180	100	2	,	,	PUNCT
fcis-16180	100	3	the	the	DET
fcis-16180	100	4	cloud	cloud	NOUN
fcis-16180	100	5	server	server	NOUN
fcis-16180	100	6	divides	divide	VERB
fcis-16180	100	7	the	the	DET
fcis-16180	100	8	users	user	NOUN
fcis-16180	100	9	into	into	ADP
fcis-16180	100	10	three	three	NUM
fcis-16180	100	11	subgroups	subgroup	NOUN
fcis-16180	100	12	,	,	PUNCT
fcis-16180	100	13	then	then	ADV
fcis-16180	100	14	sorts	sort	VERB
fcis-16180	100	15	the	the	DET
fcis-16180	100	16	values	value	NOUN
fcis-16180	100	17	of	of	ADP
fcis-16180	100	18	cosine	cosine	NOUN
fcis-16180	100	19	similarity	similarity	NOUN
fcis-16180	100	20	of	of	ADP
fcis-16180	100	21	these	these	DET
fcis-16180	100	22	users	user	NOUN
fcis-16180	100	23	.	.	PUNCT
fcis-16180	101	1	then	then	ADV
fcis-16180	101	2	,	,	PUNCT
fcis-16180	101	3	the	the	DET
fcis-16180	101	4	cloud	cloud	ADJ
fcis-16180	101	5	server	server	NOUN
fcis-16180	101	6	selects	select	VERB
fcis-16180	101	7	some	some	DET
fcis-16180	101	8	users	user	NOUN
fcis-16180	101	9	from	from	ADP
fcis-16180	101	10	each	each	DET
fcis-16180	101	11	subgroup	subgroup	NOUN
fcis-16180	101	12	for	for	ADP
fcis-16180	101	13	aggregation	aggregation	NOUN
fcis-16180	101	14	.	.	PUNCT
fcis-16180	102	1	the	the	DET
fcis-16180	102	2	cloud	cloud	PROPN
fcis-16180	102	3	server	server	NOUN
fcis-16180	102	4	aggregates	aggregate	VERB
fcis-16180	102	5	the	the	DET
fcis-16180	102	6	encrypted	encrypt	VERB
fcis-16180	102	7	trained	train	VERB
fcis-16180	102	8	models	model	NOUN
fcis-16180	102	9	of	of	ADP
fcis-16180	102	10	the	the	DET
fcis-16180	102	11	selected	select	VERB
fcis-16180	102	12	users	user	NOUN
fcis-16180	102	13	to	to	PART
fcis-16180	102	14	get	get	VERB
fcis-16180	102	15	a	a	DET
fcis-16180	102	16	new	new	ADJ
fcis-16180	102	17	encrypted	encrypt	VERB
fcis-16180	102	18	global	global	ADJ
fcis-16180	102	19	model	model	NOUN
fcis-16180	102	20	.	.	PUNCT
fcis-16180	103	1	finally	finally	ADV
fcis-16180	103	2	,	,	PUNCT
fcis-16180	103	3	the	the	DET
fcis-16180	103	4	cloud	cloud	NOUN
fcis-16180	103	5	server	server	NOUN
fcis-16180	103	6	sends	send	VERB
fcis-16180	103	7	the	the	DET
fcis-16180	103	8	new	new	ADJ
fcis-16180	103	9	encrypted	encrypt	VERB
fcis-16180	103	10	global	global	ADJ
fcis-16180	103	11	model	model	NOUN
fcis-16180	103	12	to	to	ADP
fcis-16180	103	13	each	each	DET
fcis-16180	103	14	user	user	NOUN
fcis-16180	103	15	for	for	ADP
fcis-16180	103	16	the	the	DET
fcis-16180	103	17	next	next	ADJ
fcis-16180	103	18	round	round	NOUN
fcis-16180	103	19	of	of	ADP
fcis-16180	103	20	training	training	NOUN
fcis-16180	103	21	.	.	PUNCT
fcis-16180	104	1	figure	figure	NOUN
fcis-16180	104	2	1	1	NUM
fcis-16180	104	3	.	.	PUNCT
fcis-16180	104	4	system	system	NOUN
fcis-16180	104	5	model	model	NOUN
fcis-16180	104	6	of	of	ADP
fcis-16180	104	7	ppbrfl	ppbrfl	NOUN
fcis-16180	104	8	3.2	3.2	NUM
fcis-16180	104	9	.	.	PUNCT
fcis-16180	105	1	proposed	propose	VERB
fcis-16180	105	2	scheme	scheme	NOUN
fcis-16180	105	3	(	(	PUNCT
fcis-16180	105	4	1	1	NUM
fcis-16180	105	5	)	)	PUNCT
fcis-16180	105	6	initialization	initialization	NOUN
fcis-16180	105	7	phase	phase	NOUN
fcis-16180	105	8	20	20	NUM
fcis-16180	105	9	in	in	ADP
fcis-16180	105	10	ppbrfl	ppbrfl	NOUN
fcis-16180	105	11	,	,	PUNCT
fcis-16180	105	12	a	a	DET
fcis-16180	105	13	user	user	NOUN
fcis-16180	105	14	group	group	NOUN
fcis-16180	105	15	u	u	PROPN
fcis-16180	105	16	contains	contain	VERB
fcis-16180	105	17	n	n	PRON
fcis-16180	105	18	users	user	NOUN
fcis-16180	105	19	iu	iu	ADP
fcis-16180	105	20	(	(	PUNCT
fcis-16180	105	21	[	[	X
fcis-16180	105	22	0	0	NUM
fcis-16180	105	23	,	,	PUNCT
fcis-16180	105	24	1])i	1])i	PROPN
fcis-16180	105	25	n	n	PROPN
fcis-16180	105	26			PROPN
fcis-16180	105	27	with	with	ADP
fcis-16180	105	28	logical	logical	ADJ
fcis-16180	105	29	identities	identity	NOUN
fcis-16180	105	30	{	{	PUNCT
fcis-16180	105	31	0,1	0,1	NUM
fcis-16180	105	32	,	,	PUNCT
fcis-16180	105	33	,	,	PUNCT
fcis-16180	105	34	1}n	1}n	NUM
fcis-16180	105	35			NOUN
fcis-16180	105	36	.	.	PUNCT
fcis-16180	106	1	firstly	firstly	ADV
fcis-16180	106	2	,	,	PUNCT
fcis-16180	106	3	the	the	DET
fcis-16180	106	4	cloud	cloud	NOUN
fcis-16180	106	5	server	server	NOUN
fcis-16180	106	6	initializes	initialize	VERB
fcis-16180	106	7	the	the	DET
fcis-16180	106	8	global	global	ADJ
fcis-16180	106	9	model	model	NOUN
fcis-16180	106	10	gw	gw	PROPN
fcis-16180	106	11	and	and	CCONJ
fcis-16180	106	12	learning	learn	VERB
fcis-16180	106	13	rate	rate	NOUN
fcis-16180	106	14			NUM
fcis-16180	106	15	of	of	ADP
fcis-16180	106	16	federated	federated	ADJ
fcis-16180	106	17	learning	learning	NOUN
fcis-16180	106	18	,	,	PUNCT
fcis-16180	106	19	then	then	ADV
fcis-16180	106	20	sends	send	VERB
fcis-16180	106	21	the	the	DET
fcis-16180	106	22	global	global	ADJ
fcis-16180	106	23	model	model	NOUN
fcis-16180	106	24	gw	gw	PROPN
fcis-16180	106	25	and	and	CCONJ
fcis-16180	106	26	learning	learn	VERB
fcis-16180	106	27	rate	rate	NOUN
fcis-16180	106	28			PUNCT
fcis-16180	106	29	to	to	ADP
fcis-16180	106	30	each	each	DET
fcis-16180	106	31	user	user	NOUN
fcis-16180	106	32	of	of	ADP
fcis-16180	106	33	the	the	DET
fcis-16180	106	34	user	user	NOUN
fcis-16180	106	35	group	group	PROPN
fcis-16180	106	36	u	u	PROPN
fcis-16180	106	37	.	.	PUNCT
fcis-16180	107	1	kgc	kgc	PROPN
fcis-16180	107	2	selects	select	VERB
fcis-16180	107	3	two	two	NUM
fcis-16180	107	4	random	random	ADJ
fcis-16180	107	5	prime	prime	ADJ
fcis-16180	107	6	numbers	number	NOUN
fcis-16180	107	7	p	p	NOUN
fcis-16180	107	8	and	and	CCONJ
fcis-16180	107	9	q	q	NOUN
fcis-16180	107	10	,	,	PUNCT
fcis-16180	107	11	generates	generate	VERB
fcis-16180	107	12	the	the	DET
fcis-16180	107	13	key	key	NOUN
fcis-16180	107	14	(	(	PUNCT
fcis-16180	107	15	,	,	PUNCT
fcis-16180	107	16	)	)	PUNCT
fcis-16180	107	17	sk	sk	ADP
fcis-16180	107	18	p	p	NOUN
fcis-16180	107	19			NOUN
fcis-16180	107	20	for	for	ADP
fcis-16180	107	21	symmetric	symmetric	ADJ
fcis-16180	107	22	homomorphic	homomorphic	ADJ
fcis-16180	107	23	encryption	encryption	NOUN
fcis-16180	107	24	(	(	PUNCT
fcis-16180	107	25	she	she	NOUN
fcis-16180	107	26	)	)	PUNCT
fcis-16180	107	27	and	and	CCONJ
fcis-16180	107	28	sends	send	VERB
fcis-16180	107	29	it	it	PRON
fcis-16180	107	30	to	to	ADP
fcis-16180	107	31	users	user	NOUN
fcis-16180	107	32	iu	iu	ADP
fcis-16180	107	33	(	(	PUNCT
fcis-16180	107	34	[	[	X
fcis-16180	107	35	0	0	NUM
fcis-16180	107	36	,	,	PUNCT
fcis-16180	107	37	1])i	1])i	PROPN
fcis-16180	107	38	n	n	PROPN
fcis-16180	107	39			PROPN
fcis-16180	107	40	in	in	ADP
fcis-16180	107	41	the	the	DET
fcis-16180	107	42	user	user	NOUN
fcis-16180	107	43	group	group	NOUN
fcis-16180	107	44	u	u	PROPN
fcis-16180	107	45	.	.	PUNCT
fcis-16180	108	1	kgc	kgc	PROPN
fcis-16180	108	2	picks	pick	VERB
fcis-16180	108	3	four	four	NUM
fcis-16180	108	4	values	value	NOUN
fcis-16180	108	5	0	0	NUM
fcis-16180	108	6	1	1	NUM
fcis-16180	108	7	2	2	NUM
fcis-16180	108	8	,	,	PUNCT
fcis-16180	108	9	,	,	PUNCT
fcis-16180	108	10	,	,	PUNCT
fcis-16180	108	11	k	k	PROPN
fcis-16180	108	12	k	k	PROPN
fcis-16180	108	13	k	k	PROPN
fcis-16180	108	14	n	n	ADV
fcis-16180	108	15	used	use	VERB
fcis-16180	108	16	for	for	ADP
fcis-16180	108	17	she	she	PRON
fcis-16180	108	18	,	,	PUNCT
fcis-16180	108	19	where	where	SCONJ
fcis-16180	108	20	0k	0k	NOUN
fcis-16180	108	21	denotes	denote	VERB
fcis-16180	108	22	the	the	DET
fcis-16180	108	23	length	length	NOUN
fcis-16180	108	24	of	of	ADP
fcis-16180	108	25	two	two	NUM
fcis-16180	108	26	prime	prime	ADJ
fcis-16180	108	27	numbers	number	NOUN
fcis-16180	108	28	p	p	NOUN
fcis-16180	108	29	,	,	PUNCT
fcis-16180	108	30	q	q	X
fcis-16180	108	31	,	,	PUNCT
fcis-16180	108	32	1k	1k	NUM
fcis-16180	108	33	denotes	denote	NOUN
fcis-16180	108	34	length	length	NOUN
fcis-16180	108	35	of	of	ADP
fcis-16180	108	36	plaintext	plaintext	NOUN
fcis-16180	108	37	,	,	PUNCT
fcis-16180	108	38	2k	2k	PROPN
fcis-16180	108	39	denotes	denote	VERB
fcis-16180	108	40	the	the	DET
fcis-16180	108	41	length	length	NOUN
fcis-16180	108	42	of	of	ADP
fcis-16180	108	43	a	a	DET
fcis-16180	108	44	random	random	ADJ
fcis-16180	108	45	number	number	NOUN
fcis-16180	108	46			NOUN
fcis-16180	108	47	,	,	PUNCT
fcis-16180	108	48	and	and	CCONJ
fcis-16180	108	49	n	n	PRON
fcis-16180	108	50	denotes	denote	VERB
fcis-16180	108	51	the	the	DET
fcis-16180	108	52	product	product	NOUN
fcis-16180	108	53	of	of	ADP
fcis-16180	108	54	two	two	NUM
fcis-16180	108	55	prime	prime	ADJ
fcis-16180	108	56	numbers	number	NOUN
fcis-16180	108	57	p	p	NOUN
fcis-16180	108	58	and	and	CCONJ
fcis-16180	108	59	q	q	PROPN
fcis-16180	108	60	.	.	PUNCT
fcis-16180	109	1	kgc	kgc	PROPN
fcis-16180	109	2	exposes	expose	VERB
fcis-16180	109	3	the	the	DET
fcis-16180	109	4	public	public	ADJ
fcis-16180	109	5	parameters	parameter	NOUN
fcis-16180	109	6	0	0	NUM
fcis-16180	109	7	1	1	NUM
fcis-16180	109	8	2	2	NUM
fcis-16180	109	9	(	(	PUNCT
fcis-16180	109	10	,	,	PUNCT
fcis-16180	109	11	,	,	PUNCT
fcis-16180	109	12	,	,	PUNCT
fcis-16180	109	13	)	)	PUNCT
fcis-16180	109	14	k	k	PROPN
fcis-16180	109	15	k	k	PROPN
fcis-16180	109	16	k	k	PROPN
fcis-16180	109	17	n	n	PROPN
fcis-16180	109	18	.	.	PUNCT
fcis-16180	110	1	(	(	PUNCT
fcis-16180	110	2	2	2	X
fcis-16180	110	3	)	)	PUNCT
fcis-16180	110	4	training	training	NOUN
fcis-16180	110	5	phase	phase	NOUN
fcis-16180	110	6	after	after	ADP
fcis-16180	110	7	receiving	receive	VERB
fcis-16180	110	8	the	the	DET
fcis-16180	110	9	global	global	ADJ
fcis-16180	110	10	model	model	NOUN
fcis-16180	110	11	gw	gw	PROPN
fcis-16180	110	12	from	from	ADP
fcis-16180	110	13	the	the	DET
fcis-16180	110	14	cloud	cloud	NOUN
fcis-16180	110	15	server	server	NOUN
fcis-16180	110	16	,	,	PUNCT
fcis-16180	110	17	the	the	DET
fcis-16180	110	18	user	user	NOUN
fcis-16180	110	19	iu	iu	ADP
fcis-16180	110	20	(	(	PUNCT
fcis-16180	110	21	[	[	X
fcis-16180	110	22	0	0	NUM
fcis-16180	110	23	,	,	PUNCT
fcis-16180	110	24	1])i	1])i	PROPN
fcis-16180	110	25	n	n	PROPN
fcis-16180	110	26			PROPN
fcis-16180	110	27	uses	use	VERB
fcis-16180	110	28	the	the	DET
fcis-16180	110	29	received	receive	VERB
fcis-16180	110	30	global	global	ADJ
fcis-16180	110	31	model	model	NOUN
fcis-16180	110	32	gw	gw	PROPN
fcis-16180	110	33	as	as	ADP
fcis-16180	110	34	his	his	PRON
fcis-16180	110	35	/	/	SYM
fcis-16180	110	36	her	her	PRON
fcis-16180	110	37	local	local	ADJ
fcis-16180	110	38	training	training	NOUN
fcis-16180	110	39	model	model	NOUN
fcis-16180	110	40	l	l	PROPN
fcis-16180	110	41	iw	iw	PROPN
fcis-16180	110	42	.	.	PUNCT
fcis-16180	111	1	the	the	DET
fcis-16180	111	2	user	user	NOUN
fcis-16180	111	3	iu	iu	ADP
fcis-16180	111	4	(	(	PUNCT
fcis-16180	111	5	[	[	X
fcis-16180	111	6	0	0	NUM
fcis-16180	111	7	,	,	PUNCT
fcis-16180	111	8	1])i	1])i	PROPN
fcis-16180	111	9	n	n	PROPN
fcis-16180	111	10			NOUN
fcis-16180	111	11	trains	train	VERB
fcis-16180	111	12	the	the	DET
fcis-16180	111	13	local	local	ADJ
fcis-16180	111	14	model	model	NOUN
fcis-16180	111	15	l	l	PROPN
fcis-16180	111	16	iw	iw	INTJ
fcis-16180	111	17	using	use	VERB
fcis-16180	111	18	its	its	PRON
fcis-16180	111	19	own	own	ADJ
fcis-16180	111	20	local	local	ADJ
fcis-16180	111	21	dataset	dataset	NOUN
fcis-16180	111	22	[	[	X
fcis-16180	111	23	1	1	NUM
fcis-16180	111	24	,	,	PUNCT
fcis-16180	111	25	]	]	X
fcis-16180	111	26	{	{	PUNCT
fcis-16180	111	27	,	,	PUNCT
fcis-16180	111	28	}	}	PUNCT
fcis-16180	111	29	i	i	PRON
fcis-16180	111	30	l	l	NOUN
fcis-16180	111	31	l	l	X
fcis-16180	111	32	l	l	NOUN
fcis-16180	111	33	kd	kd	NOUN
fcis-16180	111	34	x	x	PROPN
fcis-16180	111	35	y	y	PROPN
fcis-16180	111	36			PROPN
fcis-16180	111	37			PROPN
fcis-16180	111	38			INTJ
fcis-16180	111	39	.	.	PUNCT
fcis-16180	112	1	the	the	DET
fcis-16180	112	2	loss	loss	NOUN
fcis-16180	112	3	function	function	NOUN
fcis-16180	112	4	calculated	calculate	VERB
fcis-16180	112	5	by	by	ADP
fcis-16180	112	6	the	the	DET
fcis-16180	112	7	user	user	NOUN
fcis-16180	112	8	iu	iu	ADP
fcis-16180	112	9	(	(	PUNCT
fcis-16180	112	10	[	[	X
fcis-16180	112	11	0	0	NUM
fcis-16180	112	12	,	,	PUNCT
fcis-16180	112	13	1])i	1])i	PROPN
fcis-16180	112	14	n	n	PROPN
fcis-16180	112	15			NOUN
fcis-16180	112	16	is	be	AUX
fcis-16180	112	17	as	as	SCONJ
fcis-16180	112	18	follows	follow	VERB
fcis-16180	112	19	:	:	PUNCT
fcis-16180	112	20	*	*	PUNCT
fcis-16180	112	21	*	*	PUNCT
fcis-16180	112	22	*	*	PUNCT
fcis-16180	112	23	(	(	PUNCT
fcis-16180	112	24	,	,	PUNCT
fcis-16180	112	25	)	)	PUNCT
fcis-16180	112	26	1	1	NUM
fcis-16180	112	27	(	(	PUNCT
fcis-16180	112	28	,	,	PUNCT
fcis-16180	112	29	)	)	PUNCT
fcis-16180	112	30	(	(	PUNCT
fcis-16180	112	31	(	(	PUNCT
fcis-16180	112	32	,	,	PUNCT
fcis-16180	112	33	)	)	PUNCT
fcis-16180	112	34	,	,	PUNCT
fcis-16180	112	35	)	)	PUNCT
fcis-16180	113	1	|	|	ADV
fcis-16180	114	1	|	|	ADV
fcis-16180	114	2	l	l	NOUN
fcis-16180	114	3	l	l	NOUN
fcis-16180	115	1	i	i	NOUN
fcis-16180	115	2	l	l	NOUN
fcis-16180	115	3	l	l	NOUN
fcis-16180	116	1	i	i	PRON
fcis-16180	116	2	i	i	VERB
fcis-16180	117	1	l	l	VERB
fcis-16180	118	1	i	i	VERB
fcis-16180	118	2	l	l	NOUN
fcis-16180	119	1	x	x	VERB
fcis-16180	119	2	y	y	PROPN
fcis-16180	119	3	di	di	X
fcis-16180	119	4	l	l	PROPN
fcis-16180	120	1	d	d	X
fcis-16180	120	2	w	w	PROPN
fcis-16180	120	3	x	x	X
fcis-16180	120	4	w	w	VERB
fcis-16180	120	5	y	y	PROPN
fcis-16180	120	6	d	d	X
fcis-16180	120	7			PROPN
fcis-16180	120	8			PROPN
fcis-16180	120	9			PROPN
fcis-16180	120	10			NOUN
fcis-16180	120	11			NUM
fcis-16180	120	12	where	where	SCONJ
fcis-16180	120	13	*	*	PUNCT
fcis-16180	120	14	i	i	PROPN
fcis-16180	120	15	d	d	PROPN
fcis-16180	120	16	is	be	AUX
fcis-16180	120	17	the	the	DET
fcis-16180	120	18	subset	subset	NOUN
fcis-16180	120	19	of	of	ADP
fcis-16180	120	20	i	i	PROPN
fcis-16180	120	21	d	d	PROPN
fcis-16180	120	22	,	,	PUNCT
fcis-16180	120	23	(	(	PUNCT
fcis-16180	120	24	)	)	PUNCT
fcis-16180	121	1			ADJ
fcis-16180	121	2	represents	represent	VERB
fcis-16180	121	3	the	the	DET
fcis-16180	121	4	neural	neural	ADJ
fcis-16180	121	5	network	network	NOUN
fcis-16180	121	6	,	,	PUNCT
fcis-16180	121	7	l	l	PROPN
fcis-16180	121	8	iw	iw	NOUN
fcis-16180	121	9	represents	represent	VERB
fcis-16180	121	10	trainable	trainable	ADJ
fcis-16180	121	11	model	model	NOUN
fcis-16180	121	12	in	in	ADP
fcis-16180	121	13	(	(	PUNCT
fcis-16180	121	14	)	)	PUNCT
fcis-16180	121	15			PROPN
fcis-16180	121	16	,	,	PUNCT
fcis-16180	121	17	and	and	CCONJ
fcis-16180	121	18	(	(	PUNCT
fcis-16180	121	19	)	)	PUNCT
fcis-16180	121	20			PROPN
fcis-16180	121	21	represents	represent	VERB
fcis-16180	121	22	the	the	DET
fcis-16180	121	23	difference	difference	NOUN
fcis-16180	121	24	between	between	ADP
fcis-16180	121	25	the	the	DET
fcis-16180	121	26	real	real	ADJ
fcis-16180	121	27	label	label	NOUN
fcis-16180	121	28	ly	ly	X
fcis-16180	121	29	and	and	CCONJ
fcis-16180	121	30	the	the	DET
fcis-16180	121	31	neural	neural	ADJ
fcis-16180	121	32	network	network	NOUN
fcis-16180	121	33	output	output	NOUN
fcis-16180	121	34	(	(	PUNCT
fcis-16180	121	35	,	,	PUNCT
fcis-16180	121	36	)	)	PUNCT
fcis-16180	121	37	ll	ll	AUX
fcis-16180	121	38	ix	ix	ADV
fcis-16180	121	39	w	w	PROPN
fcis-16180	121	40	.	.	PUNCT
fcis-16180	122	1	the	the	DET
fcis-16180	122	2	user	user	NOUN
fcis-16180	122	3	iu	iu	ADP
fcis-16180	122	4	(	(	PUNCT
fcis-16180	122	5	[	[	X
fcis-16180	122	6	0	0	NUM
fcis-16180	122	7	,	,	PUNCT
fcis-16180	122	8	1])i	1])i	PROPN
fcis-16180	122	9	n	n	PROPN
fcis-16180	122	10			PROPN
fcis-16180	122	11	calculates	calculate	VERB
fcis-16180	122	12	the	the	DET
fcis-16180	122	13	local	local	ADJ
fcis-16180	122	14	gradient	gradient	NOUN
fcis-16180	122	15	and	and	CCONJ
fcis-16180	122	16	updates	update	VERB
fcis-16180	122	17	the	the	DET
fcis-16180	122	18	local	local	ADJ
fcis-16180	122	19	model	model	NOUN
fcis-16180	122	20	l	l	PROPN
fcis-16180	122	21	iw	iw	PROPN
fcis-16180	122	22	with	with	ADP
fcis-16180	122	23	the	the	DET
fcis-16180	122	24	following	follow	VERB
fcis-16180	122	25	equations	equation	NOUN
fcis-16180	122	26	:	:	PUNCT
fcis-16180	122	27	*	*	PUNCT
fcis-16180	122	28	(	(	PUNCT
fcis-16180	122	29	,	,	PUNCT
fcis-16180	122	30	)	)	PUNCT
fcis-16180	122	31	li	li	PROPN
fcis-16180	123	1	i	i	PRON
fcis-16180	123	2	il	il	PROPN
fcis-16180	123	3	d	d	PROPN
fcis-16180	123	4	w	w	NUM
fcis-16180	123	5			NOUN
fcis-16180	123	6	and	and	CCONJ
fcis-16180	123	7	l	l	NOUN
fcis-16180	124	1	i	i	PRON
fcis-16180	124	2	i	i	PRON
fcis-16180	124	3	iw	iw	VERB
fcis-16180	124	4	w	w	NOUN
fcis-16180	124	5			PROPN
fcis-16180	124	6			PROPN
fcis-16180	124	7	.	.	PUNCT
fcis-16180	125	1	after	after	ADP
fcis-16180	125	2	training	training	NOUN
fcis-16180	125	3	,	,	PUNCT
fcis-16180	125	4	the	the	DET
fcis-16180	125	5	trained	train	VERB
fcis-16180	125	6	model	model	NOUN
fcis-16180	125	7	iw	iw	PROPN
fcis-16180	125	8	is	be	AUX
fcis-16180	125	9	encrypted	encrypt	VERB
fcis-16180	125	10	with	with	ADP
fcis-16180	125	11	symmetric	symmetric	ADJ
fcis-16180	125	12	homomorphic	homomorphic	ADJ
fcis-16180	125	13	encryption	encryption	NOUN
fcis-16180	125	14	.	.	PUNCT
fcis-16180	126	1	the	the	DET
fcis-16180	126	2	formula	formula	NOUN
fcis-16180	126	3	for	for	ADP
fcis-16180	126	4	encryption	encryption	NOUN
fcis-16180	126	5	is	be	AUX
fcis-16180	126	6	as	as	SCONJ
fcis-16180	126	7	follows	follow	VERB
fcis-16180	126	8	:	:	PUNCT
fcis-16180	127	1	[	[	X
fcis-16180	127	2	[	[	X
fcis-16180	127	3	]	]	X
fcis-16180	127	4	]	]	PUNCT
fcis-16180	127	5	.	.	PUNCT
fcis-16180	128	1	(	(	PUNCT
fcis-16180	128	2	,	,	PUNCT
fcis-16180	128	3	,	,	PUNCT
fcis-16180	128	4	)	)	PUNCT
fcis-16180	128	5	i	i	PRON
fcis-16180	128	6	iw	iw	VERB
fcis-16180	128	7	she	she	PRON
fcis-16180	128	8	enc	enc	PROPN
fcis-16180	128	9	w	w	PROPN
fcis-16180	128	10	p	p	PROPN
fcis-16180	128	11			VERB
fcis-16180	128	12	.to	.to	PUNCT
fcis-16180	128	13	exclude	exclude	VERB
fcis-16180	128	14	malicious	malicious	ADJ
fcis-16180	128	15	users	user	NOUN
fcis-16180	128	16	,	,	PUNCT
fcis-16180	128	17	the	the	DET
fcis-16180	128	18	user	user	NOUN
fcis-16180	128	19	calculates	calculate	VERB
fcis-16180	128	20	the	the	DET
fcis-16180	128	21	cosine	cosine	NOUN
fcis-16180	128	22	similarity	similarity	NOUN
fcis-16180	128	23	between	between	ADP
fcis-16180	128	24	the	the	DET
fcis-16180	128	25	trained	train	VERB
fcis-16180	128	26	model	model	NOUN
fcis-16180	128	27	iw	iw	PROPN
fcis-16180	128	28	and	and	CCONJ
fcis-16180	128	29	the	the	DET
fcis-16180	128	30	local	local	ADJ
fcis-16180	128	31	model	model	NOUN
fcis-16180	128	32	l	l	PROPN
fcis-16180	128	33	iw	iw	PROPN
fcis-16180	128	34	.	.	PUNCT
fcis-16180	129	1	the	the	DET
fcis-16180	129	2	cosine	cosine	NOUN
fcis-16180	129	3	similarity	similarity	NOUN
fcis-16180	129	4	is	be	AUX
fcis-16180	129	5	calculated	calculate	VERB
fcis-16180	129	6	as	as	SCONJ
fcis-16180	129	7	follows	follow	VERB
fcis-16180	129	8	:	:	PUNCT
fcis-16180	129	9	(	(	PUNCT
fcis-16180	129	10	,	,	PUNCT
fcis-16180	129	11	)	)	PUNCT
fcis-16180	130	1	l	l	NOUN
fcis-16180	130	2	l	l	NOUN
fcis-16180	131	1	i	i	PRON
fcis-16180	131	2	i	i	PRON
fcis-16180	132	1	i	i	PRON
fcis-16180	133	1	i	i	VERB
fcis-16180	133	2	l	l	VERB
fcis-16180	134	1	i	i	PRON
fcis-16180	134	2	i	i	INTJ
fcis-16180	135	1	w	w	VERB
fcis-16180	135	2	w	w	PROPN
fcis-16180	135	3	sim	sim	PROPN
fcis-16180	135	4	w	w	PROPN
fcis-16180	135	5	w	w	PROPN
fcis-16180	135	6	w	w	PROPN
fcis-16180	135	7	w	w	PROPN
fcis-16180	135	8			PROPN
fcis-16180	135	9			PRON
fcis-16180	135	10	the	the	DET
fcis-16180	135	11	user	user	NOUN
fcis-16180	135	12	iu	iu	ADP
fcis-16180	135	13	(	(	PUNCT
fcis-16180	135	14	[	[	X
fcis-16180	135	15	0	0	NUM
fcis-16180	135	16	,	,	PUNCT
fcis-16180	135	17	1])i	1])i	PROPN
fcis-16180	135	18	n	n	PROPN
fcis-16180	135	19			NOUN
fcis-16180	135	20	sends	send	VERB
fcis-16180	135	21	the	the	DET
fcis-16180	135	22	encrypted	encrypt	VERB
fcis-16180	135	23	trained	train	VERB
fcis-16180	135	24	model	model	NOUN
fcis-16180	136	1	[	[	X
fcis-16180	136	2	[	[	X
fcis-16180	136	3	]	]	X
fcis-16180	136	4	]	]	X
fcis-16180	136	5	iw	iw	X
fcis-16180	136	6	along	along	ADP
fcis-16180	136	7	with	with	ADP
fcis-16180	136	8	the	the	DET
fcis-16180	136	9	cosine	cosine	NOUN
fcis-16180	136	10	similarity	similarity	NOUN
fcis-16180	136	11	value	value	NOUN
fcis-16180	136	12	(	(	PUNCT
fcis-16180	136	13	,	,	PUNCT
fcis-16180	136	14	)	)	PUNCT
fcis-16180	136	15	li	li	PROPN
fcis-16180	136	16	isim	isim	PROPN
fcis-16180	136	17	w	w	PROPN
fcis-16180	137	1	w	w	NOUN
fcis-16180	137	2	to	to	ADP
fcis-16180	137	3	the	the	DET
fcis-16180	137	4	cloud	cloud	NOUN
fcis-16180	137	5	server	server	NOUN
fcis-16180	137	6	.	.	PUNCT
fcis-16180	138	1	(	(	PUNCT
fcis-16180	138	2	3	3	X
fcis-16180	138	3	)	)	PUNCT
fcis-16180	138	4	verification	verification	NOUN
fcis-16180	138	5	and	and	CCONJ
fcis-16180	138	6	aggregation	aggregation	NOUN
fcis-16180	138	7	phase	phase	NOUN
fcis-16180	138	8	the	the	DET
fcis-16180	138	9	cloud	cloud	NOUN
fcis-16180	138	10	server	server	NOUN
fcis-16180	138	11	receives	receive	VERB
fcis-16180	138	12	the	the	DET
fcis-16180	138	13	encrypted	encrypt	VERB
fcis-16180	138	14	trained	train	VERB
fcis-16180	138	15	model	model	NOUN
fcis-16180	138	16	[	[	X
fcis-16180	138	17	[	[	X
fcis-16180	138	18	]	]	X
fcis-16180	138	19	]	]	X
fcis-16180	138	20	iw	iw	INTJ
fcis-16180	138	21	and	and	CCONJ
fcis-16180	138	22	cosine	cosine	NOUN
fcis-16180	138	23	similarity	similarity	NOUN
fcis-16180	138	24	(	(	PUNCT
fcis-16180	138	25	,	,	PUNCT
fcis-16180	138	26	)	)	PUNCT
fcis-16180	138	27	li	li	PROPN
fcis-16180	138	28	isim	isim	PROPN
fcis-16180	138	29	w	w	AUX
fcis-16180	138	30	w	w	AUX
fcis-16180	138	31	sent	send	VERB
fcis-16180	138	32	by	by	ADP
fcis-16180	138	33	user	user	NOUN
fcis-16180	138	34	iu	iu	ADP
fcis-16180	138	35	(	(	PUNCT
fcis-16180	138	36	[	[	X
fcis-16180	138	37	0	0	NUM
fcis-16180	138	38	,	,	PUNCT
fcis-16180	138	39	1])i	1])i	PROPN
fcis-16180	138	40	n	n	PROPN
fcis-16180	138	41			PROPN
fcis-16180	138	42	.	.	PUNCT
fcis-16180	139	1	the	the	DET
fcis-16180	139	2	cloud	cloud	PROPN
fcis-16180	139	3	server	server	NOUN
fcis-16180	139	4	computes	compute	VERB
fcis-16180	139	5	a	a	DET
fcis-16180	139	6	new	new	ADJ
fcis-16180	139	7	logical	logical	ADJ
fcis-16180	139	8	identifier	identifier	NOUN
fcis-16180	139	9	(	(	PUNCT
fcis-16180	139	10	)	)	PUNCT
fcis-16180	139	11	j	j	NOUN
fcis-16180	139	12	i	i	NOUN
fcis-16180	139	13	for	for	ADP
fcis-16180	139	14	each	each	DET
fcis-16180	139	15	user	user	NOUN
fcis-16180	139	16	(	(	PUNCT
fcis-16180	139	17	[	[	X
fcis-16180	139	18	0	0	NUM
fcis-16180	139	19	,	,	PUNCT
fcis-16180	139	20	1])i	1])i	PROPN
fcis-16180	139	21	n	n	PROPN
fcis-16180	139	22			NOUN
fcis-16180	139	23	by	by	ADP
fcis-16180	139	24	using	use	VERB
fcis-16180	139	25	the	the	DET
fcis-16180	139	26	pseudo	pseudo	NOUN
fcis-16180	139	27	-	-	ADJ
fcis-16180	139	28	random	random	ADJ
fcis-16180	139	29	permutation	permutation	NOUN
fcis-16180	139	30	(	(	PUNCT
fcis-16180	139	31	)	)	PUNCT
fcis-16180	139	32	:	:	PUNCT
fcis-16180	139	33	{	{	PUNCT
fcis-16180	139	34	0,1	0,1	NUM
fcis-16180	139	35	,	,	PUNCT
fcis-16180	139	36	,	,	PUNCT
fcis-16180	139	37	1	1	X
fcis-16180	139	38	}	}	PUNCT
fcis-16180	139	39	{	{	PUNCT
fcis-16180	139	40	0,1	0,1	NOUN
fcis-16180	139	41	,	,	PUNCT
fcis-16180	139	42	,	,	PUNCT
fcis-16180	139	43	1}n	1}n	NUM
fcis-16180	139	44	n	n	VERB
fcis-16180	139	45			NOUN
fcis-16180	139	46			NOUN
fcis-16180	139	47			SYM
fcis-16180	139	48			NOUN
fcis-16180	139	49	,	,	PUNCT
fcis-16180	139	50	where	where	SCONJ
fcis-16180	139	51	i	i	PRON
fcis-16180	139	52	is	be	AUX
fcis-16180	139	53	original	original	ADJ
fcis-16180	139	54	logical	logical	ADJ
fcis-16180	139	55	identifier	identifier	NOUN
fcis-16180	139	56	.	.	PUNCT
fcis-16180	140	1	the	the	DET
fcis-16180	140	2	new	new	ADJ
fcis-16180	140	3	user	user	NOUN
fcis-16180	140	4	group	group	NOUN
fcis-16180	140	5	is	be	AUX
fcis-16180	140	6	*	*	PUNCT
fcis-16180	140	7	{	{	PUNCT
fcis-16180	141	1	|	|	ADV
fcis-16180	141	2	[	[	X
fcis-16180	141	3	0	0	NUM
fcis-16180	141	4	,	,	PUNCT
fcis-16180	141	5	1]}ju	1]}ju	NUM
fcis-16180	141	6	u	u	NOUN
fcis-16180	141	7	j	j	PROPN
fcis-16180	141	8	n	n	PROPN
fcis-16180	141	9			PROPN
fcis-16180	141	10			PROPN
fcis-16180	141	11	.	.	PUNCT
fcis-16180	142	1	the	the	DET
fcis-16180	142	2	cloud	cloud	NOUN
fcis-16180	142	3	server	server	NOUN
fcis-16180	142	4	divides	divide	VERB
fcis-16180	142	5	the	the	DET
fcis-16180	142	6	new	new	ADJ
fcis-16180	142	7	user	user	NOUN
fcis-16180	142	8	group	group	NOUN
fcis-16180	142	9	*	*	PUNCT
fcis-16180	142	10	u	u	NOUN
fcis-16180	142	11	into	into	ADP
fcis-16180	142	12	three	three	NUM
fcis-16180	142	13	subgroups	subgroup	NOUN
fcis-16180	142	14	:	:	PUNCT
fcis-16180	142	15	1	1	NUM
fcis-16180	142	16	{	{	PUNCT
fcis-16180	142	17	|	|	ADV
fcis-16180	142	18	0	0	NUM
fcis-16180	142	19	mod	mod	PROPN
fcis-16180	143	1	3}ju	3}ju	NOUN
fcis-16180	144	1	j	j	NUM
fcis-16180	144	2			PROPN
fcis-16180	145	1			PROPN
fcis-16180	145	2	,	,	PUNCT
fcis-16180	145	3	2	2	NUM
fcis-16180	145	4	{	{	PUNCT
fcis-16180	145	5	|	|	ADV
fcis-16180	146	1	1mod	1mod	NUM
fcis-16180	146	2	3}ju	3}ju	NOUN
fcis-16180	146	3	j	j	NUM
fcis-16180	147	1			PROPN
fcis-16180	147	2			PROPN
fcis-16180	147	3	and	and	CCONJ
fcis-16180	147	4	3	3	NUM
fcis-16180	147	5	{	{	PUNCT
fcis-16180	147	6	|	|	ADV
fcis-16180	147	7	2	2	NUM
fcis-16180	147	8	mod	mod	NOUN
fcis-16180	147	9	3}ju	3}ju	NOUN
fcis-16180	147	10	j	j	PUNCT
fcis-16180	148	1			PROPN
fcis-16180	148	2			PROPN
fcis-16180	148	3	.	.	PUNCT
fcis-16180	149	1	the	the	DET
fcis-16180	149	2	cloud	cloud	PROPN
fcis-16180	149	3	server	server	NOUN
fcis-16180	149	4	sorts	sort	VERB
fcis-16180	149	5	the	the	DET
fcis-16180	149	6	cosine	cosine	NOUN
fcis-16180	149	7	similarity	similarity	NOUN
fcis-16180	149	8	values	value	NOUN
fcis-16180	149	9	of	of	ADP
fcis-16180	149	10	the	the	DET
fcis-16180	149	11	three	three	NUM
fcis-16180	149	12	user	user	NOUN
fcis-16180	149	13	subgroups	subgroup	NOUN
fcis-16180	149	14	in	in	ADP
fcis-16180	149	15	descending	descend	VERB
fcis-16180	149	16	order	order	NOUN
fcis-16180	149	17	,	,	PUNCT
fcis-16180	149	18	then	then	ADV
fcis-16180	149	19	removes	remove	VERB
fcis-16180	149	20	the	the	DET
fcis-16180	149	21	users	user	NOUN
fcis-16180	149	22	whose	whose	DET
fcis-16180	149	23	cosine	cosine	NOUN
fcis-16180	149	24	similarity	similarity	NOUN
fcis-16180	149	25	values	value	NOUN
fcis-16180	149	26	are	be	AUX
fcis-16180	149	27	in	in	ADP
fcis-16180	149	28	the	the	DET
fcis-16180	149	29	top	top	ADJ
fcis-16180	149	30	0	0	NUM
fcis-16180	149	31	%	%	NOUN
fcis-16180	149	32			NOUN
fcis-16180	149	33	of	of	ADP
fcis-16180	149	34	the	the	DET
fcis-16180	149	35	three	three	NUM
fcis-16180	149	36	subgroups	subgroup	NOUN
fcis-16180	149	37	.	.	PUNCT
fcis-16180	150	1	in	in	ADP
fcis-16180	150	2	this	this	DET
fcis-16180	150	3	way	way	NOUN
fcis-16180	150	4	,	,	PUNCT
fcis-16180	150	5	the	the	DET
fcis-16180	150	6	byzantine	byzantine	ADJ
fcis-16180	150	7	users	user	NOUN
fcis-16180	150	8	with	with	ADP
fcis-16180	150	9	excessively	excessively	ADV
fcis-16180	150	10	large	large	ADJ
fcis-16180	150	11	cosine	cosine	NOUN
fcis-16180	150	12	similarity	similarity	NOUN
fcis-16180	150	13	values	value	NOUN
fcis-16180	150	14	can	can	AUX
fcis-16180	150	15	be	be	AUX
fcis-16180	150	16	removed	remove	VERB
fcis-16180	150	17	.	.	PUNCT
fcis-16180	151	1	generally	generally	ADV
fcis-16180	151	2	,	,	PUNCT
fcis-16180	151	3	these	these	DET
fcis-16180	151	4	users	user	NOUN
fcis-16180	151	5	can	can	AUX
fcis-16180	151	6	directly	directly	ADV
fcis-16180	151	7	upload	upload	VERB
fcis-16180	151	8	untrained	untrained	ADJ
fcis-16180	151	9	local	local	ADJ
fcis-16180	151	10	model	model	NOUN
fcis-16180	151	11	,	,	PUNCT
fcis-16180	151	12	do	do	AUX
fcis-16180	151	13	not	not	PART
fcis-16180	151	14	contribution	contribution	VERB
fcis-16180	151	15	to	to	ADP
fcis-16180	151	16	the	the	DET
fcis-16180	151	17	global	global	ADJ
fcis-16180	151	18	model	model	NOUN
fcis-16180	151	19	during	during	ADP
fcis-16180	151	20	federated	federated	ADJ
fcis-16180	151	21	learning	learning	NOUN
fcis-16180	151	22	,	,	PUNCT
fcis-16180	151	23	but	but	CCONJ
fcis-16180	151	24	still	still	ADV
fcis-16180	151	25	obtain	obtain	VERB
fcis-16180	151	26	a	a	DET
fcis-16180	151	27	new	new	ADJ
fcis-16180	151	28	global	global	ADJ
fcis-16180	151	29	model	model	NOUN
fcis-16180	151	30	after	after	ADP
fcis-16180	151	31	each	each	DET
fcis-16180	151	32	training	training	NOUN
fcis-16180	151	33	.	.	PUNCT
fcis-16180	152	1	the	the	DET
fcis-16180	152	2	cloud	cloud	ADJ
fcis-16180	152	3	server	server	NOUN
fcis-16180	152	4	selects	select	VERB
fcis-16180	152	5	users	user	NOUN
fcis-16180	152	6	from	from	ADP
fcis-16180	152	7	three	three	NUM
fcis-16180	152	8	user	user	NOUN
fcis-16180	152	9	subgroups	subgroup	NOUN
fcis-16180	152	10	1	1	NUM
fcis-16180	152	11	,	,	PUNCT
fcis-16180	152	12	2	2	NUM
fcis-16180	152	13	and	and	CCONJ
fcis-16180	152	14	3	3	NUM
fcis-16180	152	15	based	base	VERB
fcis-16180	152	16	on	on	ADP
fcis-16180	152	17	different	different	ADJ
fcis-16180	152	18	ranges	range	NOUN
fcis-16180	152	19	.	.	PUNCT
fcis-16180	153	1	1	1	X
fcis-16180	153	2	)	)	PUNCT
fcis-16180	153	3	for	for	ADP
fcis-16180	153	4	the	the	DET
fcis-16180	153	5	user	user	NOUN
fcis-16180	153	6	subgroup	subgroup	NOUN
fcis-16180	153	7	1	1	NUM
fcis-16180	153	8	:	:	PUNCT
fcis-16180	153	9	the	the	DET
fcis-16180	153	10	cloud	cloud	ADJ
fcis-16180	153	11	server	server	NOUN
fcis-16180	153	12	selects	select	VERB
fcis-16180	153	13	users	user	NOUN
fcis-16180	153	14	in	in	ADP
fcis-16180	153	15	user	user	NOUN
fcis-16180	153	16	subgroup	subgroup	PROPN
fcis-16180	153	17	1	1	NUM
fcis-16180	153	18	whose	whose	DET
fcis-16180	153	19	cosine	cosine	NOUN
fcis-16180	153	20	similarity	similarity	NOUN
fcis-16180	153	21	values	value	NOUN
fcis-16180	153	22	are	be	AUX
fcis-16180	153	23	ranked	rank	VERB
fcis-16180	153	24	between	between	ADP
fcis-16180	153	25	0	0	NUM
fcis-16180	153	26	1%	1%	NUM
fcis-16180	153	27			ADV
fcis-16180	153	28	%	%	NOUN
fcis-16180	153	29	and	and	CCONJ
fcis-16180	153	30	1%	1%	NUM
fcis-16180	153	31	in	in	ADP
fcis-16180	153	32	descending	descend	VERB
fcis-16180	153	33	order	order	NOUN
fcis-16180	153	34	.	.	PUNCT
fcis-16180	154	1	the	the	DET
fcis-16180	154	2	set	set	NOUN
fcis-16180	154	3	of	of	ADP
fcis-16180	154	4	selected	select	VERB
fcis-16180	154	5	users	user	NOUN
fcis-16180	154	6	is	be	AUX
fcis-16180	154	7	denoted	denote	VERB
fcis-16180	154	8	as	as	ADP
fcis-16180	154	9	*	*	X
fcis-16180	154	10	1	1	NUM
fcis-16180	154	11	.	.	NOUN
fcis-16180	155	1	2	2	NUM
fcis-16180	155	2	)	)	PUNCT
fcis-16180	155	3	for	for	ADP
fcis-16180	155	4	the	the	DET
fcis-16180	155	5	user	user	NOUN
fcis-16180	155	6	subgroup	subgroup	NOUN
fcis-16180	155	7	2	2	NUM
fcis-16180	155	8	:	:	PUNCT
fcis-16180	155	9	the	the	DET
fcis-16180	155	10	cloud	cloud	ADJ
fcis-16180	155	11	server	server	NOUN
fcis-16180	155	12	selects	select	VERB
fcis-16180	155	13	users	user	NOUN
fcis-16180	155	14	in	in	ADP
fcis-16180	155	15	user	user	NOUN
fcis-16180	155	16	subgroup	subgroup	NOUN
fcis-16180	155	17	2	2	NUM
fcis-16180	155	18	whose	whose	DET
fcis-16180	155	19	cosine	cosine	NOUN
fcis-16180	155	20	similarity	similarity	NOUN
fcis-16180	155	21	values	value	NOUN
fcis-16180	155	22	are	be	AUX
fcis-16180	155	23	ranked	rank	VERB
fcis-16180	155	24	between	between	ADP
fcis-16180	155	25	1	1	NUM
fcis-16180	155	26	1%	1%	NUM
fcis-16180	155	27			ADV
fcis-16180	155	28	and	and	CCONJ
fcis-16180	155	29	2	2	NUM
fcis-16180	155	30	%	%	NOUN
fcis-16180	155	31			NOUN
fcis-16180	155	32	in	in	ADP
fcis-16180	155	33	descending	descend	VERB
fcis-16180	155	34	order	order	NOUN
fcis-16180	155	35	.	.	PUNCT
fcis-16180	156	1	the	the	DET
fcis-16180	156	2	set	set	NOUN
fcis-16180	156	3	of	of	ADP
fcis-16180	156	4	selected	select	VERB
fcis-16180	156	5	users	user	NOUN
fcis-16180	156	6	is	be	AUX
fcis-16180	156	7	denoted	denote	VERB
fcis-16180	156	8	as	as	ADP
fcis-16180	156	9	*	*	PUNCT
fcis-16180	156	10	2	2	NUM
fcis-16180	156	11	.	.	PUNCT
fcis-16180	157	1	3	3	X
fcis-16180	157	2	)	)	PUNCT
fcis-16180	157	3	for	for	ADP
fcis-16180	157	4	the	the	DET
fcis-16180	157	5	user	user	NOUN
fcis-16180	157	6	subgroup	subgroup	PROPN
fcis-16180	157	7	3	3	NUM
fcis-16180	157	8	:	:	PUNCT
fcis-16180	157	9	the	the	DET
fcis-16180	157	10	cloud	cloud	ADJ
fcis-16180	157	11	server	server	NOUN
fcis-16180	157	12	selects	select	VERB
fcis-16180	157	13	users	user	NOUN
fcis-16180	157	14	in	in	ADP
fcis-16180	157	15	user	user	NOUN
fcis-16180	157	16	subgroup	subgroup	PROPN
fcis-16180	157	17	3	3	NUM
fcis-16180	157	18	whose	whose	DET
fcis-16180	157	19	cosine	cosine	NOUN
fcis-16180	157	20	similarity	similarity	NOUN
fcis-16180	157	21	values	value	NOUN
fcis-16180	157	22	are	be	AUX
fcis-16180	157	23	ranked	rank	VERB
fcis-16180	157	24	between	between	ADP
fcis-16180	157	25	2	2	NUM
fcis-16180	157	26	1%	1%	NUM
fcis-16180	157	27			PUNCT
fcis-16180	157	28	and	and	CCONJ
fcis-16180	157	29	3%	3%	NUM
fcis-16180	157	30	in	in	ADP
fcis-16180	157	31	descending	descend	VERB
fcis-16180	157	32	order	order	NOUN
fcis-16180	157	33	.	.	PUNCT
fcis-16180	158	1	the	the	DET
fcis-16180	158	2	set	set	NOUN
fcis-16180	158	3	of	of	ADP
fcis-16180	158	4	selected	select	VERB
fcis-16180	158	5	users	user	NOUN
fcis-16180	158	6	is	be	AUX
fcis-16180	158	7	denoted	denote	VERB
fcis-16180	158	8	as	as	ADP
fcis-16180	158	9	*	*	X
fcis-16180	158	10	3	3	NUM
fcis-16180	158	11	.	.	PUNCT
fcis-16180	159	1	the	the	DET
fcis-16180	159	2	users	user	NOUN
fcis-16180	159	3	whose	whose	DET
fcis-16180	159	4	cosine	cosine	NOUN
fcis-16180	159	5	similarity	similarity	NOUN
fcis-16180	159	6	values	value	NOUN
fcis-16180	159	7	rank	rank	VERB
fcis-16180	159	8	in	in	ADP
fcis-16180	159	9	the	the	DET
fcis-16180	159	10	bottom	bottom	NOUN
fcis-16180	159	11	3(100	3(100	NUM
fcis-16180	159	12	)	)	PUNCT
fcis-16180	160	1	%	%	NOUN
fcis-16180	161	1			NOUN
fcis-16180	161	2	of	of	ADP
fcis-16180	161	3	the	the	DET
fcis-16180	161	4	three	three	NUM
fcis-16180	161	5	subgroups	subgroup	NOUN
fcis-16180	161	6	are	be	AUX
fcis-16180	161	7	identified	identify	VERB
fcis-16180	161	8	as	as	ADP
fcis-16180	161	9	byzantine	byzantine	ADJ
fcis-16180	161	10	users	user	NOUN
fcis-16180	161	11	since	since	SCONJ
fcis-16180	161	12	their	their	PRON
fcis-16180	161	13	cosine	cosine	NOUN
fcis-16180	161	14	similarity	similarity	NOUN
fcis-16180	161	15	values	value	NOUN
fcis-16180	161	16	are	be	AUX
fcis-16180	161	17	significantly	significantly	ADV
fcis-16180	161	18	small	small	ADJ
fcis-16180	161	19	.	.	PUNCT
fcis-16180	162	1	generally	generally	ADV
fcis-16180	162	2	,	,	PUNCT
fcis-16180	162	3	these	these	DET
fcis-16180	162	4	users	user	NOUN
fcis-16180	162	5	with	with	ADP
fcis-16180	162	6	excessively	excessively	ADV
fcis-16180	162	7	small	small	ADJ
fcis-16180	162	8	cosine	cosine	NOUN
fcis-16180	162	9	similarity	similarity	NOUN
fcis-16180	162	10	can	can	AUX
fcis-16180	162	11	upload	upload	VERB
fcis-16180	162	12	random	random	ADJ
fcis-16180	162	13	local	local	ADJ
fcis-16180	162	14	model	model	NOUN
fcis-16180	162	15	,	,	PUNCT
fcis-16180	162	16	or	or	CCONJ
fcis-16180	162	17	flip	flip	VERB
fcis-16180	162	18	the	the	DET
fcis-16180	162	19	element	element	NOUN
fcis-16180	162	20	-	-	PUNCT
fcis-16180	162	21	wise	wise	ADJ
fcis-16180	162	22	signs	sign	NOUN
fcis-16180	162	23	of	of	ADP
fcis-16180	162	24	the	the	DET
fcis-16180	162	25	local	local	ADJ
fcis-16180	162	26	model	model	NOUN
fcis-16180	162	27	and	and	CCONJ
fcis-16180	162	28	upload	upload	VERB
fcis-16180	162	29	the	the	DET
fcis-16180	162	30	flipped	flip	VERB
fcis-16180	162	31	local	local	ADJ
fcis-16180	162	32	model	model	NOUN
fcis-16180	162	33	.	.	PUNCT
fcis-16180	163	1	these	these	DET
fcis-16180	163	2	users	user	NOUN
fcis-16180	163	3	also	also	ADV
fcis-16180	163	4	do	do	AUX
fcis-16180	163	5	not	not	PART
fcis-16180	163	6	contribution	contribution	VERB
fcis-16180	163	7	to	to	ADP
fcis-16180	163	8	the	the	DET
fcis-16180	163	9	global	global	ADJ
fcis-16180	163	10	model	model	NOUN
fcis-16180	163	11	during	during	ADP
fcis-16180	163	12	federated	federated	ADJ
fcis-16180	163	13	learning	learning	NOUN
fcis-16180	163	14	,	,	PUNCT
fcis-16180	163	15	but	but	CCONJ
fcis-16180	163	16	still	still	ADV
fcis-16180	163	17	obtain	obtain	VERB
fcis-16180	163	18	a	a	DET
fcis-16180	163	19	new	new	ADJ
fcis-16180	163	20	global	global	ADJ
fcis-16180	163	21	model	model	NOUN
fcis-16180	163	22	after	after	ADP
fcis-16180	163	23	each	each	DET
fcis-16180	163	24	training	training	NOUN
fcis-16180	163	25	.	.	PUNCT
fcis-16180	164	1	the	the	DET
fcis-16180	164	2	cloud	cloud	PROPN
fcis-16180	164	3	server	server	NOUN
fcis-16180	164	4	aggregates	aggregate	VERB
fcis-16180	164	5	the	the	DET
fcis-16180	164	6	encrypted	encrypt	VERB
fcis-16180	164	7	trained	train	VERB
fcis-16180	164	8	models	model	NOUN
fcis-16180	164	9	of	of	ADP
fcis-16180	164	10	the	the	DET
fcis-16180	164	11	above	above	ADV
fcis-16180	164	12	selected	select	VERB
fcis-16180	164	13	users	user	NOUN
fcis-16180	164	14	to	to	PART
fcis-16180	164	15	obtain	obtain	VERB
fcis-16180	164	16	a	a	DET
fcis-16180	164	17	new	new	ADJ
fcis-16180	164	18	encrypted	encrypt	VERB
fcis-16180	164	19	global	global	ADJ
fcis-16180	164	20	model	model	NOUN
fcis-16180	164	21	*	*	PUNCT
fcis-16180	165	1	[	[	X
fcis-16180	165	2	[	[	X
fcis-16180	165	3	]	]	X
fcis-16180	165	4	]	]	X
fcis-16180	165	5	gw	gw	NOUN
fcis-16180	165	6	as	as	SCONJ
fcis-16180	165	7	follows	follow	VERB
fcis-16180	165	8	:	:	PUNCT
fcis-16180	165	9	21	21	NUM
fcis-16180	165	10	*	*	PUNCT
fcis-16180	165	11	*	*	PUNCT
fcis-16180	165	12	21	21	NUM
fcis-16180	165	13	2	2	NUM
fcis-16180	165	14	*	*	PUNCT
fcis-16180	165	15	*	*	PUNCT
fcis-16180	166	1	*	*	PUNCT
fcis-16180	166	2	*	*	PUNCT
fcis-16180	166	3	1	1	NUM
fcis-16180	166	4	2	2	NUM
fcis-16180	166	5	3	3	NUM
fcis-16180	166	6	(	(	PUNCT
fcis-16180	166	7	[	[	X
fcis-16180	166	8	[	[	X
fcis-16180	166	9	]	]	X
fcis-16180	166	10	]	]	X
fcis-16180	167	1	[	[	X
fcis-16180	167	2	[	[	X
fcis-16180	167	3	]	]	X
fcis-16180	167	4	]	]	X
fcis-16180	167	5	[	[	X
fcis-16180	167	6	[	[	X
fcis-16180	167	7	]	]	X
fcis-16180	167	8	]	]	PUNCT
fcis-16180	167	9	)	)	PUNCT
fcis-16180	168	1	[	[	X
fcis-16180	168	2	[	[	X
fcis-16180	168	3	]	]	X
fcis-16180	168	4	]	]	X
fcis-16180	168	5	|	|	ADV
fcis-16180	168	6	|	|	ADV
fcis-16180	168	7	|	|	INTJ
fcis-16180	169	1	|	|	ADV
fcis-16180	170	1	|	|	ADV
fcis-16180	170	2	j	j	PROPN
fcis-16180	171	1	j	j	PROPN
fcis-16180	171	2	j	j	PROPN
fcis-16180	172	1	jj	jj	PROPN
fcis-16180	172	2	j	j	PROPN
fcis-16180	172	3	g	g	PROPN
fcis-16180	172	4	w	w	PROPN
fcis-16180	172	5	w	w	PROPN
fcis-16180	172	6	w	w	PROPN
fcis-16180	172	7	w	w	PROPN
fcis-16180	172	8			NOUN
fcis-16180	172	9			NOUN
fcis-16180	172	10			NOUN
fcis-16180	172	11			NOUN
fcis-16180	172	12			NOUN
fcis-16180	172	13			ADP
fcis-16180	172	14			NOUN
fcis-16180	172	15			PROPN
fcis-16180	172	16			PROPN
fcis-16180	172	17			PROPN
fcis-16180	172	18			ADV
fcis-16180	172	19			PUNCT
fcis-16180	172	20			X
fcis-16180	172	21			X
fcis-16180	172	22			X
fcis-16180	172	23	the	the	DET
fcis-16180	172	24	cloud	cloud	NOUN
fcis-16180	172	25	server	server	NOUN
fcis-16180	172	26	sends	send	VERB
fcis-16180	172	27	the	the	DET
fcis-16180	172	28	new	new	ADJ
fcis-16180	172	29	encrypted	encrypt	VERB
fcis-16180	172	30	global	global	ADJ
fcis-16180	172	31	model	model	NOUN
fcis-16180	173	1	*	*	PUNCT
fcis-16180	174	1	[	[	X
fcis-16180	174	2	[	[	X
fcis-16180	174	3	]	]	X
fcis-16180	174	4	]	]	X
fcis-16180	174	5	gw	gw	NOUN
fcis-16180	174	6	to	to	ADP
fcis-16180	174	7	each	each	DET
fcis-16180	174	8	user	user	NOUN
fcis-16180	174	9	.	.	PUNCT
fcis-16180	175	1	after	after	ADP
fcis-16180	175	2	receiving	receive	VERB
fcis-16180	175	3	the	the	DET
fcis-16180	175	4	new	new	ADJ
fcis-16180	175	5	encrypted	encrypt	VERB
fcis-16180	175	6	global	global	ADJ
fcis-16180	175	7	model	model	NOUN
fcis-16180	175	8	*	*	PUNCT
fcis-16180	175	9	[	[	X
fcis-16180	175	10	[	[	X
fcis-16180	175	11	]	]	X
fcis-16180	175	12	]	]	X
fcis-16180	175	13	gw	gw	PROPN
fcis-16180	175	14	,	,	PUNCT
fcis-16180	175	15	the	the	DET
fcis-16180	175	16	user	user	NOUN
fcis-16180	175	17	first	first	ADV
fcis-16180	175	18	decrypts	decrypt	VERB
fcis-16180	175	19	the	the	DET
fcis-16180	175	20	encrypted	encrypt	VERB
fcis-16180	175	21	global	global	ADJ
fcis-16180	175	22	model	model	NOUN
fcis-16180	175	23	to	to	PART
fcis-16180	175	24	get	get	VERB
fcis-16180	175	25	the	the	DET
fcis-16180	175	26	global	global	ADJ
fcis-16180	175	27	model	model	NOUN
fcis-16180	175	28	*	*	PUNCT
fcis-16180	176	1	*	*	PUNCT
fcis-16180	176	2	.	.	PUNCT
fcis-16180	177	1	(	(	PUNCT
fcis-16180	177	2	[	[	X
fcis-16180	177	3	[	[	X
fcis-16180	177	4	]	]	X
fcis-16180	177	5	]	]	X
fcis-16180	177	6	,	,	PUNCT
fcis-16180	177	7	,	,	PUNCT
fcis-16180	177	8	)	)	PUNCT
fcis-16180	177	9	g	g	NOUN
fcis-16180	177	10	gw	gw	PROPN
fcis-16180	178	1	she	she	PRON
fcis-16180	178	2	dec	dec	PROPN
fcis-16180	178	3	w	w	PROPN
fcis-16180	178	4	p	p	NOUN
fcis-16180	178	5			X
fcis-16180	178	6	,	,	PUNCT
fcis-16180	178	7	and	and	CCONJ
fcis-16180	178	8	then	then	ADV
fcis-16180	178	9	proceeds	proceed	VERB
fcis-16180	178	10	to	to	ADP
fcis-16180	178	11	the	the	DET
fcis-16180	178	12	next	next	ADJ
fcis-16180	178	13	round	round	NOUN
fcis-16180	178	14	of	of	ADP
fcis-16180	178	15	training	training	NOUN
fcis-16180	178	16	.	.	PUNCT
fcis-16180	179	1	note	note	NOUN
fcis-16180	179	2	:	:	PUNCT
fcis-16180	179	3	the	the	DET
fcis-16180	179	4	reason	reason	NOUN
fcis-16180	179	5	for	for	ADP
fcis-16180	179	6	selecting	select	VERB
fcis-16180	179	7	users	user	NOUN
fcis-16180	179	8	from	from	ADP
fcis-16180	179	9	the	the	DET
fcis-16180	179	10	three	three	NUM
fcis-16180	179	11	disjoint	disjoint	NOUN
fcis-16180	179	12	intervals	interval	NOUN
fcis-16180	179	13	of	of	ADP
fcis-16180	179	14	the	the	DET
fcis-16180	179	15	three	three	NUM
fcis-16180	179	16	user	user	NOUN
fcis-16180	179	17	subgroups	subgroup	NOUN
fcis-16180	179	18	is	be	AUX
fcis-16180	179	19	to	to	PART
fcis-16180	179	20	reduce	reduce	VERB
fcis-16180	179	21	the	the	DET
fcis-16180	179	22	risk	risk	NOUN
fcis-16180	179	23	of	of	ADP
fcis-16180	179	24	a	a	DET
fcis-16180	179	25	successful	successful	ADJ
fcis-16180	179	26	attack	attack	NOUN
fcis-16180	179	27	by	by	ADP
fcis-16180	179	28	byzantine	byzantine	ADJ
fcis-16180	179	29	users	user	NOUN
fcis-16180	179	30	.	.	PUNCT
fcis-16180	180	1	even	even	ADV
fcis-16180	180	2	if	if	SCONJ
fcis-16180	180	3	a	a	DET
fcis-16180	180	4	byzantine	byzantine	ADJ
fcis-16180	180	5	user	user	NOUN
fcis-16180	180	6	succeeds	succeed	VERB
fcis-16180	180	7	in	in	ADP
fcis-16180	180	8	faking	fake	VERB
fcis-16180	180	9	a	a	DET
fcis-16180	180	10	cosine	cosine	NOUN
fcis-16180	180	11	similarity	similarity	NOUN
fcis-16180	180	12	that	that	PRON
fcis-16180	180	13	closely	closely	ADV
fcis-16180	180	14	resembles	resemble	VERB
fcis-16180	180	15	that	that	PRON
fcis-16180	180	16	of	of	ADP
fcis-16180	180	17	other	other	ADJ
fcis-16180	180	18	users	user	NOUN
fcis-16180	180	19	,	,	PUNCT
fcis-16180	180	20	the	the	DET
fcis-16180	180	21	probability	probability	NOUN
fcis-16180	180	22	of	of	ADP
fcis-16180	180	23	that	that	DET
fcis-16180	180	24	byzantine	byzantine	ADJ
fcis-16180	180	25	user	user	NOUN
fcis-16180	180	26	being	be	AUX
fcis-16180	180	27	selected	select	VERB
fcis-16180	180	28	is	be	AUX
fcis-16180	180	29	greatly	greatly	ADV
fcis-16180	180	30	reduced	reduce	VERB
fcis-16180	180	31	.	.	PUNCT
fcis-16180	181	1	4	4	X
fcis-16180	181	2	.	.	X
fcis-16180	181	3	performance	performance	NOUN
fcis-16180	181	4	evaluation	evaluation	NOUN
fcis-16180	181	5	4.1	4.1	NUM
fcis-16180	181	6	.	.	PUNCT
fcis-16180	182	1	classification	classification	NOUN
fcis-16180	182	2	accuracy	accuracy	NOUN
fcis-16180	182	3	in	in	ADP
fcis-16180	182	4	this	this	DET
fcis-16180	182	5	section	section	NOUN
fcis-16180	182	6	,	,	PUNCT
fcis-16180	182	7	we	we	PRON
fcis-16180	182	8	evaluate	evaluate	VERB
fcis-16180	182	9	the	the	DET
fcis-16180	182	10	performance	performance	NOUN
fcis-16180	182	11	of	of	ADP
fcis-16180	182	12	our	our	PRON
fcis-16180	182	13	ppbrfl	ppbrfl	NOUN
fcis-16180	182	14	through	through	ADP
fcis-16180	182	15	the	the	DET
fcis-16180	182	16	following	follow	VERB
fcis-16180	182	17	experiments	experiment	NOUN
fcis-16180	182	18	.	.	PUNCT
fcis-16180	183	1	these	these	DET
fcis-16180	183	2	experiments	experiment	NOUN
fcis-16180	183	3	are	be	AUX
fcis-16180	183	4	carried	carry	VERB
fcis-16180	183	5	out	out	ADP
fcis-16180	183	6	on	on	ADP
fcis-16180	183	7	a	a	DET
fcis-16180	183	8	linux	linux	PROPN
fcis-16180	183	9	server	server	NOUN
fcis-16180	183	10	with	with	ADP
fcis-16180	183	11	intel(r	intel(r	PROPN
fcis-16180	183	12	)	)	PUNCT
fcis-16180	183	13	core	core	NOUN
fcis-16180	183	14	(	(	PUNCT
fcis-16180	183	15	tm	tm	NOUN
fcis-16180	183	16	)	)	PUNCT
fcis-16180	183	17	i9	i9	NOUN
fcis-16180	183	18	-	-	PUNCT
fcis-16180	183	19	10980xe	10980xe	NOUN
fcis-16180	183	20	,	,	PUNCT
fcis-16180	183	21	3.0ghz	3.0ghz	NUM
fcis-16180	183	22	,	,	PUNCT
fcis-16180	183	23	and	and	CCONJ
fcis-16180	183	24	32	32	NUM
fcis-16180	183	25	gb	gb	NOUN
fcis-16180	183	26	of	of	ADP
fcis-16180	183	27	ram	ram	NOUN
fcis-16180	183	28	.	.	PUNCT
fcis-16180	184	1	the	the	DET
fcis-16180	184	2	simulation	simulation	NOUN
fcis-16180	184	3	environment	environment	NOUN
fcis-16180	184	4	is	be	AUX
fcis-16180	184	5	implemented	implement	VERB
fcis-16180	184	6	using	use	VERB
fcis-16180	184	7	the	the	DET
fcis-16180	184	8	pytorch	pytorch	NOUN
fcis-16180	184	9	framework	framework	NOUN
fcis-16180	184	10	.	.	PUNCT
fcis-16180	185	1	the	the	DET
fcis-16180	185	2	evaluation	evaluation	NOUN
fcis-16180	185	3	of	of	ADP
fcis-16180	185	4	our	our	PRON
fcis-16180	185	5	simulation	simulation	NOUN
fcis-16180	185	6	experiments	experiment	NOUN
fcis-16180	185	7	focuses	focus	VERB
fcis-16180	185	8	on	on	ADP
fcis-16180	185	9	two	two	NUM
fcis-16180	185	10	aspects	aspect	NOUN
fcis-16180	185	11	:	:	PUNCT
fcis-16180	185	12	classification	classification	NOUN
fcis-16180	185	13	accuracy	accuracy	NOUN
fcis-16180	185	14	and	and	CCONJ
fcis-16180	185	15	computation	computation	NOUN
fcis-16180	185	16	efficiency	efficiency	NOUN
fcis-16180	185	17	on	on	ADP
fcis-16180	185	18	the	the	DET
fcis-16180	185	19	user	user	NOUN
fcis-16180	185	20	side	side	NOUN
fcis-16180	185	21	.	.	PUNCT
fcis-16180	186	1	we	we	PRON
fcis-16180	186	2	evaluate	evaluate	VERB
fcis-16180	186	3	ppbrfl	ppbrfl	NOUN
fcis-16180	186	4	using	use	VERB
fcis-16180	186	5	mnist	mnist	NOUN
fcis-16180	187	1	[	[	X
fcis-16180	187	2	39	39	NUM
fcis-16180	187	3	]	]	PUNCT
fcis-16180	187	4	and	and	CCONJ
fcis-16180	187	5	cifar10	cifar10	PROPN
fcis-16180	187	6	[	[	X
fcis-16180	187	7	40	40	NUM
fcis-16180	187	8	]	]	PUNCT
fcis-16180	187	9	datasets	dataset	NOUN
fcis-16180	187	10	.	.	PUNCT
fcis-16180	188	1	mnist	mnist	PROPN
fcis-16180	188	2	dataset	dataset	VERB
fcis-16180	189	1	[	[	X
fcis-16180	189	2	39	39	NUM
fcis-16180	189	3	]	]	PUNCT
fcis-16180	189	4	:	:	PUNCT
fcis-16180	189	5	the	the	DET
fcis-16180	189	6	mnist	mnist	NOUN
fcis-16180	189	7	is	be	AUX
fcis-16180	189	8	a	a	DET
fcis-16180	189	9	dataset	dataset	NOUN
fcis-16180	189	10	of	of	ADP
fcis-16180	189	11	handwritten	handwritten	ADJ
fcis-16180	189	12	numbers	number	NOUN
fcis-16180	189	13	.	.	PUNCT
fcis-16180	190	1	it	it	PRON
fcis-16180	190	2	contains	contain	VERB
fcis-16180	190	3	a	a	DET
fcis-16180	190	4	total	total	NOUN
fcis-16180	190	5	of	of	ADP
fcis-16180	190	6	70,000	70,000	NUM
fcis-16180	190	7	images	image	NOUN
fcis-16180	190	8	,	,	PUNCT
fcis-16180	190	9	with	with	ADP
fcis-16180	190	10	60,000	60,000	NUM
fcis-16180	190	11	images	image	NOUN
fcis-16180	190	12	selected	select	VERB
fcis-16180	190	13	for	for	ADP
fcis-16180	190	14	training	training	NOUN
fcis-16180	190	15	and	and	CCONJ
fcis-16180	190	16	10,000	10,000	NUM
fcis-16180	190	17	images	image	NOUN
fcis-16180	190	18	for	for	ADP
fcis-16180	190	19	testing	testing	NOUN
fcis-16180	190	20	.	.	PUNCT
fcis-16180	191	1	each	each	DET
fcis-16180	191	2	image	image	NOUN
fcis-16180	191	3	has	have	VERB
fcis-16180	191	4	a	a	DET
fcis-16180	191	5	size	size	NOUN
fcis-16180	191	6	of	of	ADP
fcis-16180	191	7	28x28	28x28	NUM
fcis-16180	191	8	pixels	pixel	NOUN
fcis-16180	191	9	and	and	CCONJ
fcis-16180	191	10	features	feature	VERB
fcis-16180	191	11	handwritten	handwritten	ADJ
fcis-16180	191	12	digits	digit	NOUN
fcis-16180	191	13	ranging	range	VERB
fcis-16180	191	14	from	from	ADP
fcis-16180	191	15	0	0	NUM
fcis-16180	191	16	to	to	ADP
fcis-16180	191	17	9	9	NUM
fcis-16180	191	18	,	,	PUNCT
fcis-16180	191	19	displayed	display	VERB
fcis-16180	191	20	in	in	ADP
fcis-16180	191	21	white	white	NOUN
fcis-16180	191	22	on	on	ADP
fcis-16180	191	23	a	a	DET
fcis-16180	191	24	black	black	ADJ
fcis-16180	191	25	background	background	NOUN
fcis-16180	191	26	.	.	PUNCT
fcis-16180	192	1	(	(	PUNCT
fcis-16180	192	2	a	a	X
fcis-16180	192	3	)	)	PUNCT
fcis-16180	192	4	(	(	PUNCT
fcis-16180	192	5	b	b	X
fcis-16180	192	6	)	)	PUNCT
fcis-16180	192	7	figure	figure	NOUN
fcis-16180	192	8	2	2	NUM
fcis-16180	192	9	.	.	PUNCT
fcis-16180	192	10	comparison	comparison	NOUN
fcis-16180	192	11	of	of	ADP
fcis-16180	192	12	classification	classification	NOUN
fcis-16180	192	13	accuracy	accuracy	NOUN
fcis-16180	192	14	on	on	ADP
fcis-16180	192	15	the	the	DET
fcis-16180	192	16	cifar-10	cifar-10	PROPN
fcis-16180	192	17	and	and	CCONJ
fcis-16180	192	18	mnist	mnist	NOUN
fcis-16180	192	19	datasets	dataset	NOUN
fcis-16180	192	20	between	between	ADP
fcis-16180	192	21	ppbrfl	ppbrfl	NOUN
fcis-16180	192	22	and	and	CCONJ
fcis-16180	192	23	fl	fl	X
fcis-16180	193	1	[	[	X
fcis-16180	193	2	5	5	NUM
fcis-16180	193	3	]	]	PUNCT
fcis-16180	193	4	(	(	PUNCT
fcis-16180	193	5	a	a	NOUN
fcis-16180	193	6	)	)	PUNCT
fcis-16180	193	7	on	on	ADP
fcis-16180	193	8	the	the	DET
fcis-16180	193	9	cifar-10	cifar-10	PROPN
fcis-16180	193	10	dataset	dataset	NOUN
fcis-16180	193	11	.	.	PUNCT
fcis-16180	194	1	(	(	PUNCT
fcis-16180	194	2	b	b	X
fcis-16180	194	3	)	)	PUNCT
fcis-16180	194	4	on	on	ADP
fcis-16180	194	5	the	the	DET
fcis-16180	194	6	mnist	mnist	NOUN
fcis-16180	194	7	dataset	dataset	VERB
fcis-16180	194	8	cifar-10	cifar-10	PROPN
fcis-16180	194	9	dataset	dataset	NOUN
fcis-16180	195	1	[	[	X
fcis-16180	195	2	40	40	NUM
fcis-16180	195	3	]	]	X
fcis-16180	195	4	:	:	PUNCT
fcis-16180	195	5	the	the	DET
fcis-16180	195	6	cifar-10	cifar-10	PROPN
fcis-16180	195	7	dataset	dataset	NOUN
fcis-16180	195	8	is	be	AUX
fcis-16180	195	9	an	an	DET
fcis-16180	195	10	image	image	NOUN
fcis-16180	195	11	dataset	dataset	NOUN
fcis-16180	195	12	consisting	consist	VERB
fcis-16180	195	13	of	of	ADP
fcis-16180	195	14	various	various	ADJ
fcis-16180	195	15	items	item	NOUN
fcis-16180	195	16	.	.	PUNCT
fcis-16180	196	1	it	it	PRON
fcis-16180	196	2	comprises	comprise	VERB
fcis-16180	196	3	a	a	DET
fcis-16180	196	4	total	total	NOUN
fcis-16180	196	5	of	of	ADP
fcis-16180	196	6	60,000	60,000	NUM
fcis-16180	196	7	samples	sample	NOUN
fcis-16180	196	8	,	,	PUNCT
fcis-16180	196	9	with	with	ADP
fcis-16180	196	10	10	10	NUM
fcis-16180	196	11	categories	category	NOUN
fcis-16180	196	12	,	,	PUNCT
fcis-16180	196	13	each	each	PRON
fcis-16180	196	14	having	have	VERB
fcis-16180	196	15	6,000	6,000	NUM
fcis-16180	196	16	samples	sample	NOUN
fcis-16180	196	17	(	(	PUNCT
fcis-16180	196	18	5000	5000	NUM
fcis-16180	196	19	for	for	ADP
fcis-16180	196	20	training	training	NOUN
fcis-16180	196	21	and	and	CCONJ
fcis-16180	196	22	1000	1000	NUM
fcis-16180	196	23	for	for	ADP
fcis-16180	196	24	testing	testing	NOUN
fcis-16180	196	25	)	)	PUNCT
fcis-16180	196	26	.	.	PUNCT
fcis-16180	197	1	each	each	DET
fcis-16180	197	2	sample	sample	NOUN
fcis-16180	197	3	in	in	ADP
fcis-16180	197	4	the	the	DET
fcis-16180	197	5	dataset	dataset	NOUN
fcis-16180	197	6	is	be	AUX
fcis-16180	197	7	a	a	DET
fcis-16180	197	8	32x32	32x32	NUM
fcis-16180	197	9	pixel	pixel	PROPN
fcis-16180	197	10	rgb	rgb	PROPN
fcis-16180	197	11	image	image	NOUN
fcis-16180	197	12	with	with	ADP
fcis-16180	197	13	three	three	NUM
fcis-16180	197	14	channels	channel	NOUN
fcis-16180	197	15	.	.	PUNCT
fcis-16180	198	1	in	in	ADP
fcis-16180	198	2	this	this	DET
fcis-16180	198	3	experiment	experiment	NOUN
fcis-16180	198	4	,	,	PUNCT
fcis-16180	198	5	resnet18	resnet18	NOUN
fcis-16180	198	6	[	[	X
fcis-16180	198	7	41	41	NUM
fcis-16180	198	8	]	]	X
fcis-16180	198	9	neural	neural	ADJ
fcis-16180	198	10	network	network	NOUN
fcis-16180	198	11	is	be	AUX
fcis-16180	198	12	trained	train	VERB
fcis-16180	198	13	using	use	VERB
fcis-16180	198	14	the	the	DET
fcis-16180	198	15	above	above	ADJ
fcis-16180	198	16	two	two	NUM
fcis-16180	198	17	datasets	dataset	NOUN
fcis-16180	198	18	.	.	PUNCT
fcis-16180	199	1	resnet18	resnet18	NOUN
fcis-16180	200	1	[	[	X
fcis-16180	200	2	41	41	NUM
fcis-16180	200	3	]	]	PUNCT
fcis-16180	200	4	is	be	AUX
fcis-16180	200	5	a	a	DET
fcis-16180	200	6	residual	residual	ADJ
fcis-16180	200	7	neural	neural	ADJ
fcis-16180	200	8	network	network	NOUN
fcis-16180	200	9	which	which	PRON
fcis-16180	200	10	is	be	AUX
fcis-16180	200	11	composed	compose	VERB
fcis-16180	200	12	of	of	ADP
fcis-16180	200	13	a	a	DET
fcis-16180	200	14	series	series	NOUN
fcis-16180	200	15	of	of	ADP
fcis-16180	200	16	residual	residual	ADJ
fcis-16180	200	17	blocks	block	NOUN
fcis-16180	200	18	.	.	PUNCT
fcis-16180	201	1	it	it	PRON
fcis-16180	201	2	comprises	comprise	VERB
fcis-16180	201	3	17	17	NUM
fcis-16180	201	4	internal	internal	ADJ
fcis-16180	201	5	convolutional	convolutional	ADJ
fcis-16180	201	6	layers	layer	NOUN
fcis-16180	201	7	and	and	CCONJ
fcis-16180	201	8	1	1	NUM
fcis-16180	201	9	fully	fully	ADV
fcis-16180	201	10	connected	connected	ADJ
fcis-16180	201	11	layer	layer	NOUN
fcis-16180	201	12	.	.	PUNCT
fcis-16180	202	1	resnet18	resnet18	NOUN
fcis-16180	202	2	neural	neural	ADJ
fcis-16180	202	3	network	network	NOUN
fcis-16180	202	4	is	be	AUX
fcis-16180	202	5	widely	widely	ADV
fcis-16180	202	6	used	use	VERB
fcis-16180	202	7	for	for	ADP
fcis-16180	202	8	image	image	NOUN
fcis-16180	202	9	classification	classification	NOUN
fcis-16180	202	10	tasks	task	NOUN
fcis-16180	202	11	.	.	PUNCT
fcis-16180	203	1	in	in	ADP
fcis-16180	203	2	the	the	DET
fcis-16180	203	3	following	following	NOUN
fcis-16180	203	4	,	,	PUNCT
fcis-16180	203	5	we	we	PRON
fcis-16180	203	6	choose	choose	VERB
fcis-16180	203	7	the	the	DET
fcis-16180	203	8	standard	standard	ADJ
fcis-16180	203	9	federated	federated	ADJ
fcis-16180	203	10	learning	learning	NOUN
fcis-16180	203	11	scheme	scheme	NOUN
fcis-16180	204	1	[	[	X
fcis-16180	204	2	5	5	NUM
fcis-16180	204	3	]	]	PUNCT
fcis-16180	204	4	,	,	PUNCT
fcis-16180	204	5	which	which	PRON
fcis-16180	204	6	does	do	AUX
fcis-16180	204	7	not	not	PART
fcis-16180	204	8	support	support	VERB
fcis-16180	204	9	byzantine	byzantine	ADJ
fcis-16180	204	10	robustness	robustness	NOUN
fcis-16180	204	11	and	and	CCONJ
fcis-16180	204	12	privacy	privacy	NOUN
fcis-16180	204	13	protection	protection	NOUN
fcis-16180	204	14	,	,	PUNCT
fcis-16180	204	15	as	as	ADP
fcis-16180	204	16	a	a	DET
fcis-16180	204	17	benchmark	benchmark	NOUN
fcis-16180	204	18	for	for	ADP
fcis-16180	204	19	comparison	comparison	NOUN
fcis-16180	204	20	with	with	ADP
fcis-16180	204	21	the	the	DET
fcis-16180	204	22	proposed	proposed	ADJ
fcis-16180	204	23	ppbrfl	ppbrfl	NOUN
fcis-16180	204	24	.	.	PUNCT
fcis-16180	205	1	in	in	ADP
fcis-16180	205	2	order	order	NOUN
fcis-16180	205	3	to	to	PART
fcis-16180	205	4	evaluate	evaluate	VERB
fcis-16180	205	5	the	the	DET
fcis-16180	205	6	impact	impact	NOUN
fcis-16180	205	7	of	of	ADP
fcis-16180	205	8	the	the	DET
fcis-16180	205	9	proposed	propose	VERB
fcis-16180	205	10	ppbrfl	ppbrfl	NOUN
fcis-16180	205	11	on	on	ADP
fcis-16180	205	12	the	the	DET
fcis-16180	205	13	model	model	NOUN
fcis-16180	205	14	classification	classification	NOUN
fcis-16180	205	15	accuracy	accuracy	NOUN
fcis-16180	205	16	,	,	PUNCT
fcis-16180	205	17	we	we	PRON
fcis-16180	205	18	conduct	conduct	VERB
fcis-16180	205	19	a	a	DET
fcis-16180	205	20	comparative	comparative	ADJ
fcis-16180	205	21	analysis	analysis	NOUN
fcis-16180	205	22	between	between	ADP
fcis-16180	205	23	ppbrfl	ppbrfl	NOUN
fcis-16180	205	24	and	and	CCONJ
fcis-16180	205	25	the	the	DET
fcis-16180	205	26	scheme	scheme	NOUN
fcis-16180	205	27	[	[	X
fcis-16180	205	28	5	5	NUM
fcis-16180	205	29	]	]	PUNCT
fcis-16180	205	30	.	.	PUNCT
fcis-16180	206	1	this	this	DET
fcis-16180	206	2	analysis	analysis	NOUN
fcis-16180	206	3	is	be	AUX
fcis-16180	206	4	executed	execute	VERB
fcis-16180	206	5	under	under	ADP
fcis-16180	206	6	conditions	condition	NOUN
fcis-16180	206	7	where	where	SCONJ
fcis-16180	206	8	byzantine	byzantine	ADJ
fcis-16180	206	9	users	user	NOUN
fcis-16180	206	10	are	be	AUX
fcis-16180	206	11	absent	absent	ADJ
fcis-16180	206	12	,	,	PUNCT
fcis-16180	206	13	allowing	allow	VERB
fcis-16180	206	14	us	we	PRON
fcis-16180	206	15	to	to	PART
fcis-16180	206	16	exclude	exclude	VERB
fcis-16180	206	17	the	the	DET
fcis-16180	206	18	impact	impact	NOUN
fcis-16180	206	19	of	of	ADP
fcis-16180	206	20	byzantine	byzantine	ADJ
fcis-16180	206	21	users	user	NOUN
fcis-16180	206	22	on	on	ADP
fcis-16180	206	23	classification	classification	NOUN
fcis-16180	206	24	accuracy	accuracy	NOUN
fcis-16180	206	25	.	.	PUNCT
fcis-16180	207	1	(	(	PUNCT
fcis-16180	207	2	a	a	X
fcis-16180	207	3	)	)	PUNCT
fcis-16180	207	4	(	(	PUNCT
fcis-16180	207	5	b	b	X
fcis-16180	207	6	)	)	PUNCT
fcis-16180	207	7	(	(	PUNCT
fcis-16180	207	8	c	c	X
fcis-16180	207	9	)	)	PUNCT
fcis-16180	207	10	(	(	PUNCT
fcis-16180	207	11	d	d	X
fcis-16180	207	12	)	)	PUNCT
fcis-16180	207	13	(	(	PUNCT
fcis-16180	207	14	e	e	NOUN
fcis-16180	207	15	)	)	PUNCT
fcis-16180	207	16	(	(	PUNCT
fcis-16180	207	17	f	f	X
fcis-16180	207	18	)	)	PUNCT
fcis-16180	207	19	figure	figure	NOUN
fcis-16180	207	20	3	3	NUM
fcis-16180	207	21	.	.	NOUN
fcis-16180	207	22	comparison	comparison	NOUN
fcis-16180	207	23	of	of	ADP
fcis-16180	207	24	classification	classification	NOUN
fcis-16180	207	25	accuracy	accuracy	NOUN
fcis-16180	207	26	on	on	ADP
fcis-16180	207	27	the	the	DET
fcis-16180	207	28	cifar-10	cifar-10	PROPN
fcis-16180	207	29	dataset	dataset	NOUN
fcis-16180	207	30	between	between	ADP
fcis-16180	207	31	ppbrfl	ppbrfl	NOUN
fcis-16180	207	32	and	and	CCONJ
fcis-16180	207	33	fl	fl	X
fcis-16180	208	1	[	[	X
fcis-16180	208	2	5	5	X
fcis-16180	208	3	]	]	PUNCT
fcis-16180	208	4	under	under	ADP
fcis-16180	208	5	byzantine	byzantine	ADJ
fcis-16180	208	6	attack	attack	NOUN
fcis-16180	208	7	.	.	PUNCT
fcis-16180	209	1	(	(	PUNCT
fcis-16180	209	2	a	a	X
fcis-16180	209	3	)	)	PUNCT
fcis-16180	209	4	fl	fl	NOUN
fcis-16180	210	1	[	[	X
fcis-16180	210	2	5	5	NUM
fcis-16180	210	3	]	]	PUNCT
fcis-16180	210	4	with	with	ADP
fcis-16180	210	5	0	0	NUM
fcis-16180	210	6	%	%	NOUN
fcis-16180	210	7	byzantine	byzantine	ADJ
fcis-16180	210	8	users	user	NOUN
fcis-16180	210	9	and	and	CCONJ
fcis-16180	210	10	fl	fl	PRON
fcis-16180	211	1	[	[	X
fcis-16180	211	2	5	5	NUM
fcis-16180	211	3	]	]	PUNCT
fcis-16180	211	4	with	with	ADP
fcis-16180	211	5	20	20	NUM
fcis-16180	211	6	%	%	NOUN
fcis-16180	211	7	byzantine	byzantine	ADJ
fcis-16180	211	8	users	user	NOUN
fcis-16180	211	9	.	.	PUNCT
fcis-16180	212	1	(	(	PUNCT
fcis-16180	212	2	b	b	X
fcis-16180	212	3	)	)	PUNCT
fcis-16180	212	4	fl	fl	NOUN
fcis-16180	213	1	[	[	X
fcis-16180	213	2	5	5	NUM
fcis-16180	213	3	]	]	PUNCT
fcis-16180	213	4	with	with	ADP
fcis-16180	213	5	0	0	NUM
fcis-16180	213	6	%	%	NOUN
fcis-16180	213	7	byzantine	byzantine	ADJ
fcis-16180	213	8	users	user	NOUN
fcis-16180	213	9	and	and	CCONJ
fcis-16180	213	10	fl	fl	PRON
fcis-16180	214	1	[	[	X
fcis-16180	214	2	5	5	NUM
fcis-16180	214	3	]	]	PUNCT
fcis-16180	214	4	with	with	ADP
fcis-16180	214	5	40	40	NUM
fcis-16180	214	6	%	%	NOUN
fcis-16180	214	7	byzantine	byzantine	ADJ
fcis-16180	214	8	users	user	NOUN
fcis-16180	214	9	.	.	PUNCT
fcis-16180	215	1	(	(	PUNCT
fcis-16180	215	2	c	c	X
fcis-16180	215	3	)	)	PUNCT
fcis-16180	215	4	ppbrfl	ppbrfl	NOUN
fcis-16180	215	5	with	with	ADP
fcis-16180	215	6	0	0	NUM
fcis-16180	215	7	%	%	NOUN
fcis-16180	215	8	byzantine	byzantine	ADJ
fcis-16180	215	9	users	user	NOUN
fcis-16180	215	10	and	and	CCONJ
fcis-16180	215	11	ppbrfl	ppbrfl	VERB
fcis-16180	215	12	with	with	ADP
fcis-16180	215	13	20	20	NUM
fcis-16180	215	14	%	%	NOUN
fcis-16180	215	15	byzantine	byzantine	ADJ
fcis-16180	215	16	users	user	NOUN
fcis-16180	215	17	.	.	PUNCT
fcis-16180	216	1	(	(	PUNCT
fcis-16180	216	2	d	d	X
fcis-16180	216	3	)	)	PUNCT
fcis-16180	216	4	ppbrfl	ppbrfl	NOUN
fcis-16180	216	5	with	with	ADP
fcis-16180	216	6	0	0	NUM
fcis-16180	216	7	%	%	NOUN
fcis-16180	216	8	byzantine	byzantine	ADJ
fcis-16180	216	9	users	user	NOUN
fcis-16180	216	10	and	and	CCONJ
fcis-16180	216	11	ppbrfl	ppbrfl	VERB
fcis-16180	216	12	with	with	ADP
fcis-16180	216	13	40	40	NUM
fcis-16180	216	14	%	%	NOUN
fcis-16180	216	15	byzantine	byzantine	ADJ
fcis-16180	216	16	users	user	NOUN
fcis-16180	216	17	.	.	PUNCT
fcis-16180	217	1	(	(	PUNCT
fcis-16180	217	2	e	e	X
fcis-16180	217	3	)	)	PUNCT
fcis-16180	217	4	ppbrfl	ppbrfl	NOUN
fcis-16180	217	5	with	with	ADP
fcis-16180	217	6	20	20	NUM
fcis-16180	217	7	%	%	NOUN
fcis-16180	217	8	byzantine	byzantine	ADJ
fcis-16180	217	9	users	user	NOUN
fcis-16180	217	10	and	and	CCONJ
fcis-16180	217	11	fl	fl	PRON
fcis-16180	218	1	[	[	X
fcis-16180	218	2	5	5	NUM
fcis-16180	218	3	]	]	PUNCT
fcis-16180	218	4	with	with	ADP
fcis-16180	218	5	20	20	NUM
fcis-16180	218	6	%	%	NOUN
fcis-16180	218	7	byzantine	byzantine	ADJ
fcis-16180	218	8	users	user	NOUN
fcis-16180	218	9	.	.	PUNCT
fcis-16180	219	1	(	(	PUNCT
fcis-16180	219	2	f	f	X
fcis-16180	219	3	)	)	PUNCT
fcis-16180	219	4	ppbrfl	ppbrfl	NOUN
fcis-16180	219	5	with	with	ADP
fcis-16180	219	6	40	40	NUM
fcis-16180	219	7	%	%	NOUN
fcis-16180	219	8	byzantine	byzantine	ADJ
fcis-16180	219	9	users	user	NOUN
fcis-16180	219	10	and	and	CCONJ
fcis-16180	219	11	fl	fl	PRON
fcis-16180	220	1	[	[	X
fcis-16180	220	2	5	5	NUM
fcis-16180	220	3	]	]	PUNCT
fcis-16180	220	4	with	with	ADP
fcis-16180	220	5	40	40	NUM
fcis-16180	220	6	%	%	NOUN
fcis-16180	220	7	byzantine	byzantine	ADJ
fcis-16180	220	8	users	user	NOUN
fcis-16180	220	9	fig	fig	NOUN
fcis-16180	220	10	.	.	PUNCT
fcis-16180	221	1	2(a	2(a	NUM
fcis-16180	221	2	)	)	PUNCT
fcis-16180	221	3	and	and	CCONJ
fcis-16180	221	4	fig	fig	NOUN
fcis-16180	221	5	.	.	PUNCT
fcis-16180	222	1	2(b	2(b	NUM
fcis-16180	222	2	)	)	PUNCT
fcis-16180	222	3	illustrate	illustrate	VERB
fcis-16180	222	4	the	the	DET
fcis-16180	222	5	model	model	NOUN
fcis-16180	222	6	classification	classification	NOUN
fcis-16180	222	7	accuracy	accuracy	NOUN
fcis-16180	222	8	results	result	NOUN
fcis-16180	222	9	of	of	ADP
fcis-16180	222	10	ppbrfl	ppbrfl	NOUN
fcis-16180	222	11	and	and	CCONJ
fcis-16180	222	12	the	the	DET
fcis-16180	222	13	scheme	scheme	NOUN
fcis-16180	223	1	[	[	X
fcis-16180	223	2	5	5	NUM
fcis-16180	223	3	]	]	PUNCT
fcis-16180	223	4	on	on	ADP
fcis-16180	223	5	the	the	DET
fcis-16180	223	6	cifar-10	cifar-10	PROPN
fcis-16180	223	7	and	and	CCONJ
fcis-16180	223	8	mnist	mnist	NOUN
fcis-16180	223	9	datasets	dataset	NOUN
fcis-16180	223	10	,	,	PUNCT
fcis-16180	223	11	respectively	respectively	ADV
fcis-16180	223	12	.	.	PUNCT
fcis-16180	224	1	the	the	DET
fcis-16180	224	2	comparison	comparison	NOUN
fcis-16180	224	3	reveals	reveal	VERB
fcis-16180	224	4	that	that	SCONJ
fcis-16180	224	5	the	the	DET
fcis-16180	224	6	classification	classification	NOUN
fcis-16180	224	7	accuracy	accuracy	NOUN
fcis-16180	224	8	achieved	achieve	VERB
fcis-16180	224	9	by	by	ADP
fcis-16180	224	10	ppbrfl	ppbrfl	NOUN
fcis-16180	224	11	is	be	AUX
fcis-16180	224	12	nearly	nearly	ADV
fcis-16180	224	13	identical	identical	ADJ
fcis-16180	224	14	to	to	ADP
fcis-16180	224	15	that	that	PRON
fcis-16180	224	16	of	of	ADP
fcis-16180	224	17	the	the	DET
fcis-16180	224	18	scheme	scheme	NOUN
fcis-16180	225	1	[	[	X
fcis-16180	225	2	5	5	NUM
fcis-16180	225	3	]	]	PUNCT
fcis-16180	225	4	for	for	ADP
fcis-16180	225	5	both	both	DET
fcis-16180	225	6	datasets	dataset	NOUN
fcis-16180	225	7	.	.	PUNCT
fcis-16180	226	1	we	we	PRON
fcis-16180	226	2	can	can	AUX
fcis-16180	226	3	conclude	conclude	VERB
fcis-16180	226	4	that	that	SCONJ
fcis-16180	226	5	ppbrfl	ppbrfl	NOUN
fcis-16180	226	6	achieves	achieve	VERB
fcis-16180	226	7	privacy	privacy	NOUN
fcis-16180	226	8	preservation	preservation	NOUN
fcis-16180	226	9	without	without	ADP
fcis-16180	226	10	compromising	compromise	VERB
fcis-16180	226	11	the	the	DET
fcis-16180	226	12	model	model	NOUN
fcis-16180	226	13	classification	classification	NOUN
fcis-16180	226	14	accuracy	accuracy	NOUN
fcis-16180	226	15	since	since	SCONJ
fcis-16180	226	16	ppbrfl	ppbrfl	NOUN
fcis-16180	226	17	employs	employ	VERB
fcis-16180	226	18	symmetric	symmetric	ADJ
fcis-16180	226	19	homomorphic	homomorphic	ADJ
fcis-16180	226	20	encryption	encryption	NOUN
fcis-16180	226	21	to	to	PART
fcis-16180	226	22	achieve	achieve	VERB
fcis-16180	226	23	privacy	privacy	NOUN
fcis-16180	226	24	protection	protection	NOUN
fcis-16180	226	25	without	without	ADP
fcis-16180	226	26	introducing	introduce	VERB
fcis-16180	226	27	any	any	DET
fcis-16180	226	28	noise	noise	NOUN
fcis-16180	226	29	,	,	PUNCT
fcis-16180	226	30	thus	thus	ADV
fcis-16180	226	31	maintaining	maintain	VERB
fcis-16180	226	32	the	the	DET
fcis-16180	226	33	classification	classification	NOUN
fcis-16180	226	34	accuracy	accuracy	NOUN
fcis-16180	226	35	of	of	ADP
fcis-16180	226	36	the	the	DET
fcis-16180	226	37	model	model	NOUN
fcis-16180	226	38	.	.	PUNCT
fcis-16180	227	1	to	to	PART
fcis-16180	227	2	comprehensively	comprehensively	ADV
fcis-16180	227	3	assess	assess	VERB
fcis-16180	227	4	the	the	DET
fcis-16180	227	5	influence	influence	NOUN
fcis-16180	227	6	of	of	ADP
fcis-16180	227	7	byzantine	byzantine	ADJ
fcis-16180	227	8	robustness	robustness	NOUN
fcis-16180	227	9	on	on	ADP
fcis-16180	227	10	the	the	DET
fcis-16180	227	11	proposed	proposed	ADJ
fcis-16180	227	12	ppbrfl	ppbrfl	NOUN
fcis-16180	227	13	,	,	PUNCT
fcis-16180	227	14	we	we	PRON
fcis-16180	227	15	undertake	undertake	VERB
fcis-16180	227	16	a	a	DET
fcis-16180	227	17	comparative	comparative	ADJ
fcis-16180	227	18	analysis	analysis	NOUN
fcis-16180	227	19	between	between	ADP
fcis-16180	227	20	ppbrfl	ppbrfl	NOUN
fcis-16180	227	21	and	and	CCONJ
fcis-16180	227	22	the	the	DET
fcis-16180	227	23	scheme	scheme	NOUN
fcis-16180	227	24	[	[	X
fcis-16180	227	25	5	5	NUM
fcis-16180	227	26	]	]	PUNCT
fcis-16180	227	27	.	.	PUNCT
fcis-16180	228	1	we	we	PRON
fcis-16180	228	2	evaluate	evaluate	VERB
fcis-16180	228	3	the	the	DET
fcis-16180	228	4	classification	classification	NOUN
fcis-16180	228	5	accuracy	accuracy	NOUN
fcis-16180	228	6	of	of	ADP
fcis-16180	228	7	the	the	DET
fcis-16180	228	8	same	same	ADJ
fcis-16180	228	9	model	model	NOUN
fcis-16180	228	10	under	under	ADP
fcis-16180	228	11	different	different	ADJ
fcis-16180	228	12	datasets	dataset	NOUN
fcis-16180	228	13	,	,	PUNCT
fcis-16180	228	14	and	and	CCONJ
fcis-16180	228	15	choose	choose	VERB
fcis-16180	228	16	different	different	ADJ
fcis-16180	228	17	numbers	number	NOUN
fcis-16180	228	18	of	of	ADP
fcis-16180	228	19	byzantine	byzantine	ADJ
fcis-16180	228	20	users	user	NOUN
fcis-16180	228	21	with	with	ADP
fcis-16180	228	22	0	0	NUM
fcis-16180	228	23	%	%	NOUN
fcis-16180	228	24	,	,	PUNCT
fcis-16180	228	25	20	20	NUM
fcis-16180	228	26	%	%	NOUN
fcis-16180	228	27	22	22	NUM
fcis-16180	228	28	and	and	CCONJ
fcis-16180	228	29	40	40	NUM
fcis-16180	228	30	%	%	NOUN
fcis-16180	228	31	,	,	PUNCT
fcis-16180	228	32	respectively	respectively	ADV
fcis-16180	228	33	.	.	PUNCT
fcis-16180	228	34	fig.3(a	fig.3(a	ADJ
fcis-16180	228	35	)	)	PUNCT
fcis-16180	228	36	and	and	CCONJ
fcis-16180	228	37	fig.3(b	fig.3(b	X
fcis-16180	228	38	)	)	PUNCT
fcis-16180	228	39	illustrate	illustrate	VERB
fcis-16180	228	40	a	a	DET
fcis-16180	228	41	comparison	comparison	NOUN
fcis-16180	228	42	of	of	ADP
fcis-16180	228	43	the	the	DET
fcis-16180	228	44	model	model	NOUN
fcis-16180	228	45	classification	classification	NOUN
fcis-16180	228	46	accuracy	accuracy	NOUN
fcis-16180	228	47	achieved	achieve	VERB
fcis-16180	228	48	by	by	ADP
fcis-16180	228	49	the	the	DET
fcis-16180	228	50	scheme	scheme	NOUN
fcis-16180	229	1	[	[	X
fcis-16180	229	2	5	5	NUM
fcis-16180	229	3	]	]	PUNCT
fcis-16180	229	4	under	under	ADP
fcis-16180	229	5	two	two	NUM
fcis-16180	229	6	distinct	distinct	ADJ
fcis-16180	229	7	scenarios	scenario	NOUN
fcis-16180	229	8	:	:	PUNCT
fcis-16180	229	9	one	one	NUM
fcis-16180	229	10	in	in	ADP
fcis-16180	229	11	which	which	PRON
fcis-16180	229	12	the	the	DET
fcis-16180	229	13	scheme	scheme	NOUN
fcis-16180	229	14	[	[	X
fcis-16180	229	15	5	5	NUM
fcis-16180	229	16	]	]	PUNCT
fcis-16180	229	17	does	do	AUX
fcis-16180	229	18	not	not	PART
fcis-16180	229	19	suffer	suffer	VERB
fcis-16180	229	20	byzantine	byzantine	ADJ
fcis-16180	229	21	attack	attack	NOUN
fcis-16180	229	22	,	,	PUNCT
fcis-16180	229	23	and	and	CCONJ
fcis-16180	229	24	the	the	DET
fcis-16180	229	25	other	other	ADJ
fcis-16180	229	26	in	in	ADP
fcis-16180	229	27	which	which	PRON
fcis-16180	229	28	it	it	PRON
fcis-16180	229	29	is	be	AUX
fcis-16180	229	30	subjected	subject	VERB
fcis-16180	229	31	to	to	ADP
fcis-16180	229	32	byzantine	byzantine	ADJ
fcis-16180	229	33	attack	attack	NOUN
fcis-16180	229	34	(	(	PUNCT
fcis-16180	229	35	20	20	NUM
fcis-16180	229	36	%	%	NOUN
fcis-16180	229	37	byzantine	byzantine	ADJ
fcis-16180	229	38	users	user	NOUN
fcis-16180	229	39	and	and	CCONJ
fcis-16180	229	40	40	40	NUM
fcis-16180	229	41	%	%	NOUN
fcis-16180	229	42	byzantine	byzantine	ADJ
fcis-16180	229	43	users	user	NOUN
fcis-16180	229	44	)	)	PUNCT
fcis-16180	229	45	.	.	PUNCT
fcis-16180	230	1	the	the	DET
fcis-16180	230	2	results	result	NOUN
fcis-16180	230	3	show	show	VERB
fcis-16180	230	4	that	that	SCONJ
fcis-16180	230	5	the	the	DET
fcis-16180	230	6	model	model	NOUN
fcis-16180	230	7	classification	classification	NOUN
fcis-16180	230	8	accuracy	accuracy	NOUN
fcis-16180	230	9	of	of	ADP
fcis-16180	230	10	the	the	DET
fcis-16180	230	11	scheme	scheme	NOUN
fcis-16180	230	12	[	[	X
fcis-16180	230	13	5	5	NUM
fcis-16180	230	14	]	]	PUNCT
fcis-16180	230	15	decreases	decrease	NOUN
fcis-16180	230	16	when	when	SCONJ
fcis-16180	230	17	attacked	attack	VERB
fcis-16180	230	18	by	by	ADP
fcis-16180	230	19	byzantine	byzantine	ADJ
fcis-16180	230	20	users	user	NOUN
fcis-16180	230	21	,	,	PUNCT
fcis-16180	230	22	and	and	CCONJ
fcis-16180	230	23	decreases	decrease	VERB
fcis-16180	230	24	more	more	ADV
fcis-16180	230	25	significantly	significantly	ADV
fcis-16180	230	26	as	as	ADP
fcis-16180	230	27	the	the	DET
fcis-16180	230	28	proportion	proportion	NOUN
fcis-16180	230	29	of	of	ADP
fcis-16180	230	30	byzantine	byzantine	ADJ
fcis-16180	230	31	users	user	NOUN
fcis-16180	230	32	increases	increase	NOUN
fcis-16180	230	33	.	.	PUNCT
fcis-16180	231	1	this	this	DET
fcis-16180	231	2	observed	observed	ADJ
fcis-16180	231	3	trend	trend	NOUN
fcis-16180	231	4	can	can	AUX
fcis-16180	231	5	be	be	AUX
fcis-16180	231	6	attributed	attribute	VERB
fcis-16180	231	7	to	to	ADP
fcis-16180	231	8	the	the	DET
fcis-16180	231	9	inherent	inherent	ADJ
fcis-16180	231	10	absence	absence	NOUN
fcis-16180	231	11	of	of	ADP
fcis-16180	231	12	byzantine	byzantine	ADJ
fcis-16180	231	13	robustness	robustness	NOUN
fcis-16180	231	14	within	within	ADP
fcis-16180	231	15	the	the	DET
fcis-16180	231	16	scheme	scheme	NOUN
fcis-16180	231	17	[	[	X
fcis-16180	231	18	5	5	NUM
fcis-16180	231	19	]	]	PUNCT
fcis-16180	231	20	,	,	PUNCT
fcis-16180	231	21	rendering	render	VERB
fcis-16180	231	22	it	it	PRON
fcis-16180	231	23	considerably	considerably	ADV
fcis-16180	231	24	vulnerable	vulnerable	ADJ
fcis-16180	231	25	to	to	ADP
fcis-16180	231	26	byzantine	byzantine	ADJ
fcis-16180	231	27	user	user	NOUN
fcis-16180	231	28	-	-	PUNCT
fcis-16180	231	29	induced	induce	VERB
fcis-16180	231	30	disturbances	disturbance	NOUN
fcis-16180	231	31	.	.	PUNCT
fcis-16180	232	1	(	(	PUNCT
fcis-16180	232	2	a	a	X
fcis-16180	232	3	)	)	PUNCT
fcis-16180	232	4	(	(	PUNCT
fcis-16180	232	5	b	b	X
fcis-16180	232	6	)	)	PUNCT
fcis-16180	232	7	(	(	PUNCT
fcis-16180	232	8	c	c	X
fcis-16180	232	9	)	)	PUNCT
fcis-16180	232	10	(	(	PUNCT
fcis-16180	232	11	d	d	X
fcis-16180	232	12	)	)	PUNCT
fcis-16180	232	13	(	(	PUNCT
fcis-16180	232	14	e	e	NOUN
fcis-16180	232	15	)	)	PUNCT
fcis-16180	232	16	(	(	PUNCT
fcis-16180	232	17	f	f	X
fcis-16180	232	18	)	)	PUNCT
fcis-16180	232	19	figure	figure	NOUN
fcis-16180	232	20	4	4	NUM
fcis-16180	232	21	.	.	NOUN
fcis-16180	232	22	comparison	comparison	NOUN
fcis-16180	232	23	of	of	ADP
fcis-16180	232	24	classification	classification	NOUN
fcis-16180	232	25	accuracy	accuracy	NOUN
fcis-16180	232	26	on	on	ADP
fcis-16180	232	27	the	the	DET
fcis-16180	232	28	mnist	mnist	NOUN
fcis-16180	232	29	dataset	dataset	VERB
fcis-16180	232	30	between	between	ADP
fcis-16180	232	31	ppbrfl	ppbrfl	NOUN
fcis-16180	232	32	and	and	CCONJ
fcis-16180	232	33	fl	fl	X
fcis-16180	233	1	[	[	X
fcis-16180	233	2	5	5	X
fcis-16180	233	3	]	]	PUNCT
fcis-16180	233	4	under	under	ADP
fcis-16180	233	5	byzantine	byzantine	ADJ
fcis-16180	233	6	attack	attack	NOUN
fcis-16180	233	7	.	.	PUNCT
fcis-16180	234	1	(	(	PUNCT
fcis-16180	234	2	a	a	X
fcis-16180	234	3	)	)	PUNCT
fcis-16180	234	4	fl	fl	NOUN
fcis-16180	235	1	[	[	X
fcis-16180	235	2	5	5	NUM
fcis-16180	235	3	]	]	PUNCT
fcis-16180	235	4	with	with	ADP
fcis-16180	235	5	0	0	NUM
fcis-16180	235	6	%	%	NOUN
fcis-16180	235	7	byzantine	byzantine	ADJ
fcis-16180	235	8	users	user	NOUN
fcis-16180	235	9	and	and	CCONJ
fcis-16180	235	10	fl	fl	PRON
fcis-16180	236	1	[	[	X
fcis-16180	236	2	5	5	NUM
fcis-16180	236	3	]	]	PUNCT
fcis-16180	236	4	with	with	ADP
fcis-16180	236	5	20	20	NUM
fcis-16180	236	6	%	%	NOUN
fcis-16180	236	7	byzantine	byzantine	ADJ
fcis-16180	236	8	users	user	NOUN
fcis-16180	236	9	.	.	PUNCT
fcis-16180	237	1	(	(	PUNCT
fcis-16180	237	2	b	b	X
fcis-16180	237	3	)	)	PUNCT
fcis-16180	237	4	fl	fl	NOUN
fcis-16180	238	1	[	[	X
fcis-16180	238	2	5	5	NUM
fcis-16180	238	3	]	]	PUNCT
fcis-16180	238	4	with	with	ADP
fcis-16180	238	5	0	0	NUM
fcis-16180	238	6	%	%	NOUN
fcis-16180	238	7	byzantine	byzantine	ADJ
fcis-16180	238	8	users	user	NOUN
fcis-16180	238	9	and	and	CCONJ
fcis-16180	238	10	fl	fl	PRON
fcis-16180	239	1	[	[	X
fcis-16180	239	2	5	5	NUM
fcis-16180	239	3	]	]	PUNCT
fcis-16180	239	4	with	with	ADP
fcis-16180	239	5	40	40	NUM
fcis-16180	239	6	%	%	NOUN
fcis-16180	239	7	byzantine	byzantine	ADJ
fcis-16180	239	8	users	user	NOUN
fcis-16180	239	9	.	.	PUNCT
fcis-16180	240	1	(	(	PUNCT
fcis-16180	240	2	c	c	X
fcis-16180	240	3	)	)	PUNCT
fcis-16180	240	4	ppbrfl	ppbrfl	NOUN
fcis-16180	240	5	with	with	ADP
fcis-16180	240	6	0	0	NUM
fcis-16180	240	7	%	%	NOUN
fcis-16180	240	8	byzantine	byzantine	ADJ
fcis-16180	240	9	users	user	NOUN
fcis-16180	240	10	and	and	CCONJ
fcis-16180	240	11	ppbrfl	ppbrfl	VERB
fcis-16180	240	12	with	with	ADP
fcis-16180	240	13	20	20	NUM
fcis-16180	240	14	%	%	NOUN
fcis-16180	240	15	byzantine	byzantine	ADJ
fcis-16180	240	16	users	user	NOUN
fcis-16180	240	17	.	.	PUNCT
fcis-16180	241	1	(	(	PUNCT
fcis-16180	241	2	d	d	X
fcis-16180	241	3	)	)	PUNCT
fcis-16180	241	4	ppbrfl	ppbrfl	NOUN
fcis-16180	241	5	with	with	ADP
fcis-16180	241	6	0	0	NUM
fcis-16180	241	7	%	%	NOUN
fcis-16180	241	8	byzantine	byzantine	ADJ
fcis-16180	241	9	users	user	NOUN
fcis-16180	241	10	and	and	CCONJ
fcis-16180	241	11	ppbrfl	ppbrfl	VERB
fcis-16180	241	12	with	with	ADP
fcis-16180	241	13	40	40	NUM
fcis-16180	241	14	%	%	NOUN
fcis-16180	241	15	byzantine	byzantine	ADJ
fcis-16180	241	16	users	user	NOUN
fcis-16180	241	17	.	.	PUNCT
fcis-16180	242	1	(	(	PUNCT
fcis-16180	242	2	e	e	X
fcis-16180	242	3	)	)	PUNCT
fcis-16180	242	4	ppbrfl	ppbrfl	NOUN
fcis-16180	242	5	with	with	ADP
fcis-16180	242	6	20	20	NUM
fcis-16180	242	7	%	%	NOUN
fcis-16180	242	8	byzantine	byzantine	ADJ
fcis-16180	242	9	users	user	NOUN
fcis-16180	242	10	and	and	CCONJ
fcis-16180	242	11	fl	fl	PRON
fcis-16180	243	1	[	[	X
fcis-16180	243	2	5	5	NUM
fcis-16180	243	3	]	]	PUNCT
fcis-16180	243	4	with	with	ADP
fcis-16180	243	5	20	20	NUM
fcis-16180	243	6	%	%	NOUN
fcis-16180	243	7	byzantine	byzantine	ADJ
fcis-16180	243	8	users	user	NOUN
fcis-16180	243	9	.	.	PUNCT
fcis-16180	244	1	(	(	PUNCT
fcis-16180	244	2	f	f	X
fcis-16180	244	3	)	)	PUNCT
fcis-16180	244	4	ppbrfl	ppbrfl	NOUN
fcis-16180	244	5	with	with	ADP
fcis-16180	244	6	40	40	NUM
fcis-16180	244	7	%	%	NOUN
fcis-16180	244	8	byzantine	byzantine	ADJ
fcis-16180	244	9	users	user	NOUN
fcis-16180	244	10	and	and	CCONJ
fcis-16180	244	11	fl	fl	PRON
fcis-16180	245	1	[	[	X
fcis-16180	245	2	5	5	NUM
fcis-16180	245	3	]	]	PUNCT
fcis-16180	245	4	with	with	ADP
fcis-16180	245	5	40	40	NUM
fcis-16180	245	6	%	%	NOUN
fcis-16180	245	7	byzantine	byzantine	ADJ
fcis-16180	245	8	users	user	NOUN
fcis-16180	245	9	fig.3(c	fig.3(c	VERB
fcis-16180	245	10	)	)	PUNCT
fcis-16180	245	11	and	and	CCONJ
fcis-16180	245	12	fig.3(d	fig.3(d	NOUN
fcis-16180	245	13	)	)	PUNCT
fcis-16180	245	14	show	show	VERB
fcis-16180	245	15	a	a	DET
fcis-16180	245	16	comparison	comparison	NOUN
fcis-16180	245	17	of	of	ADP
fcis-16180	245	18	the	the	DET
fcis-16180	245	19	model	model	NOUN
fcis-16180	245	20	classification	classification	NOUN
fcis-16180	245	21	accuracy	accuracy	NOUN
fcis-16180	245	22	achieved	achieve	VERB
fcis-16180	245	23	by	by	ADP
fcis-16180	245	24	ppbrfl	ppbrfl	NOUN
fcis-16180	245	25	under	under	ADP
fcis-16180	245	26	two	two	NUM
fcis-16180	245	27	different	different	ADJ
fcis-16180	245	28	scenarios	scenario	NOUN
fcis-16180	245	29	:	:	PUNCT
fcis-16180	245	30	one	one	NUM
fcis-16180	245	31	in	in	ADP
fcis-16180	245	32	which	which	PRON
fcis-16180	245	33	ppbrfl	ppbrfl	NOUN
fcis-16180	245	34	is	be	AUX
fcis-16180	245	35	free	free	ADJ
fcis-16180	245	36	of	of	ADP
fcis-16180	245	37	byzantine	byzantine	ADJ
fcis-16180	245	38	attack	attack	NOUN
fcis-16180	245	39	,	,	PUNCT
fcis-16180	245	40	and	and	CCONJ
fcis-16180	245	41	the	the	DET
fcis-16180	245	42	other	other	ADJ
fcis-16180	245	43	in	in	ADP
fcis-16180	245	44	which	which	PRON
fcis-16180	245	45	it	it	PRON
fcis-16180	245	46	is	be	AUX
fcis-16180	245	47	subjected	subject	VERB
fcis-16180	245	48	to	to	ADP
fcis-16180	245	49	byzantine	byzantine	ADJ
fcis-16180	245	50	attack	attack	NOUN
fcis-16180	245	51	(	(	PUNCT
fcis-16180	245	52	20	20	NUM
fcis-16180	245	53	%	%	NOUN
fcis-16180	245	54	byzantine	byzantine	ADJ
fcis-16180	245	55	users	user	NOUN
fcis-16180	245	56	and	and	CCONJ
fcis-16180	245	57	40	40	NUM
fcis-16180	245	58	%	%	NOUN
fcis-16180	245	59	byzantine	byzantine	ADJ
fcis-16180	245	60	users	user	NOUN
fcis-16180	245	61	)	)	PUNCT
fcis-16180	245	62	.	.	PUNCT
fcis-16180	246	1	we	we	PRON
fcis-16180	246	2	can	can	AUX
fcis-16180	246	3	observe	observe	VERB
fcis-16180	246	4	that	that	SCONJ
fcis-16180	246	5	the	the	DET
fcis-16180	246	6	classification	classification	NOUN
fcis-16180	246	7	accuracy	accuracy	NOUN
fcis-16180	246	8	of	of	ADP
fcis-16180	246	9	the	the	DET
fcis-16180	246	10	ppbrfl	ppbrfl	NOUN
fcis-16180	246	11	attacked	attack	VERB
fcis-16180	246	12	by	by	ADP
fcis-16180	246	13	byzantine	byzantine	ADJ
fcis-16180	246	14	users	user	NOUN
fcis-16180	246	15	is	be	AUX
fcis-16180	246	16	almost	almost	ADV
fcis-16180	246	17	as	as	ADV
fcis-16180	246	18	accurate	accurate	ADJ
fcis-16180	246	19	as	as	ADP
fcis-16180	246	20	the	the	DET
fcis-16180	246	21	classification	classification	NOUN
fcis-16180	246	22	accuracy	accuracy	NOUN
fcis-16180	246	23	of	of	ADP
fcis-16180	246	24	the	the	DET
fcis-16180	246	25	ppbrfl	ppbrfl	NOUN
fcis-16180	246	26	not	not	PART
fcis-16180	246	27	attacked	attack	VERB
fcis-16180	246	28	by	by	ADP
fcis-16180	246	29	byzantine	byzantine	ADJ
fcis-16180	246	30	users	user	NOUN
fcis-16180	246	31	.	.	PUNCT
fcis-16180	247	1	as	as	SCONJ
fcis-16180	247	2	illustrated	illustrate	VERB
fcis-16180	247	3	in	in	ADP
fcis-16180	247	4	fig.3(e	fig.3(e	PROPN
fcis-16180	247	5	)	)	PUNCT
fcis-16180	247	6	and	and	CCONJ
fcis-16180	247	7	fig.3(f	fig.3(f	NUM
fcis-16180	247	8	)	)	PUNCT
fcis-16180	247	9	,	,	PUNCT
fcis-16180	247	10	the	the	DET
fcis-16180	247	11	model	model	NOUN
fcis-16180	247	12	classification	classification	NOUN
fcis-16180	247	13	accuracy	accuracy	NOUN
fcis-16180	247	14	of	of	ADP
fcis-16180	247	15	ppbrfl	ppbrfl	NOUN
fcis-16180	247	16	demonstrates	demonstrate	VERB
fcis-16180	247	17	greater	great	ADJ
fcis-16180	247	18	accuracy	accuracy	NOUN
fcis-16180	247	19	in	in	ADP
fcis-16180	247	20	comparison	comparison	NOUN
fcis-16180	247	21	to	to	ADP
fcis-16180	247	22	the	the	DET
fcis-16180	247	23	scheme	scheme	NOUN
fcis-16180	247	24	[	[	X
fcis-16180	247	25	5	5	NUM
fcis-16180	247	26	]	]	PUNCT
fcis-16180	247	27	when	when	SCONJ
fcis-16180	247	28	attacked	attack	VERB
fcis-16180	247	29	by	by	ADP
fcis-16180	247	30	the	the	DET
fcis-16180	247	31	same	same	ADJ
fcis-16180	247	32	number	number	NOUN
fcis-16180	247	33	of	of	ADP
fcis-16180	247	34	byzantine	byzantine	ADJ
fcis-16180	247	35	users	user	NOUN
fcis-16180	247	36	.	.	PUNCT
fcis-16180	248	1	the	the	DET
fcis-16180	248	2	above	above	ADJ
fcis-16180	248	3	analyses	analysis	NOUN
fcis-16180	248	4	indicate	indicate	VERB
fcis-16180	248	5	that	that	SCONJ
fcis-16180	248	6	ppbrfl	ppbrfl	NOUN
fcis-16180	248	7	exhibits	exhibit	VERB
fcis-16180	248	8	remarkable	remarkable	ADJ
fcis-16180	248	9	byzantine	byzantine	ADJ
fcis-16180	248	10	robustness	robustness	NOUN
fcis-16180	248	11	and	and	CCONJ
fcis-16180	248	12	minimally	minimally	ADV
fcis-16180	248	13	impacts	impact	VERB
fcis-16180	248	14	the	the	DET
fcis-16180	248	15	model	model	NOUN
fcis-16180	248	16	classification	classification	NOUN
fcis-16180	248	17	accuracy	accuracy	NOUN
fcis-16180	248	18	when	when	SCONJ
fcis-16180	248	19	attacked	attack	VERB
fcis-16180	248	20	by	by	ADP
fcis-16180	248	21	byzantine	byzantine	ADJ
fcis-16180	248	22	users	user	NOUN
fcis-16180	248	23	.	.	PUNCT
fcis-16180	249	1	fig.4	fig.4	PROPN
fcis-16180	249	2	shows	show	VERB
fcis-16180	249	3	that	that	SCONJ
fcis-16180	249	4	ppbrfl	ppbrfl	NOUN
fcis-16180	249	5	maintains	maintain	VERB
fcis-16180	249	6	exceptional	exceptional	ADJ
fcis-16180	249	7	byzantine	byzantine	ADJ
fcis-16180	249	8	robustness	robustness	NOUN
fcis-16180	249	9	under	under	ADP
fcis-16180	249	10	mnist	mnist	NOUN
fcis-16180	249	11	dataset	dataset	NOUN
fcis-16180	249	12	.	.	PUNCT
fcis-16180	250	1	4.2	4.2	NUM
fcis-16180	250	2	.	.	PUNCT
fcis-16180	250	3	computation	computation	NOUN
fcis-16180	250	4	efficiency	efficiency	NOUN
fcis-16180	250	5	on	on	ADP
fcis-16180	250	6	the	the	DET
fcis-16180	250	7	user	user	NOUN
fcis-16180	250	8	side	side	NOUN
fcis-16180	250	9	we	we	PRON
fcis-16180	250	10	pick	pick	VERB
fcis-16180	250	11	the	the	DET
fcis-16180	250	12	privacy	privacy	NOUN
fcis-16180	250	13	-	-	PUNCT
fcis-16180	250	14	preserving	preserve	VERB
fcis-16180	250	15	federated	federated	ADJ
fcis-16180	250	16	learning	learn	VERB
fcis-16180	250	17	scheme	scheme	NOUN
fcis-16180	251	1	[	[	X
fcis-16180	251	2	7	7	NUM
fcis-16180	251	3	]	]	PUNCT
fcis-16180	251	4	as	as	ADP
fcis-16180	251	5	the	the	DET
fcis-16180	251	6	benchmark	benchmark	NOUN
fcis-16180	251	7	,	,	PUNCT
fcis-16180	251	8	as	as	SCONJ
fcis-16180	251	9	it	it	PRON
fcis-16180	251	10	also	also	ADV
fcis-16180	251	11	uses	use	VERB
fcis-16180	251	12	homomorphic	homomorphic	ADJ
fcis-16180	251	13	encryption	encryption	NOUN
fcis-16180	251	14	to	to	PART
fcis-16180	251	15	protect	protect	VERB
fcis-16180	251	16	the	the	DET
fcis-16180	251	17	user	user	NOUN
fcis-16180	251	18	’s	’s	PART
fcis-16180	251	19	data	data	PROPN
fcis-16180	251	20	privacy	privacy	NOUN
fcis-16180	251	21	during	during	ADP
fcis-16180	251	22	federated	federated	ADJ
fcis-16180	251	23	learning	learning	NOUN
fcis-16180	251	24	process	process	NOUN
fcis-16180	251	25	.	.	PUNCT
fcis-16180	252	1	in	in	ADP
fcis-16180	252	2	this	this	DET
fcis-16180	252	3	experiment	experiment	NOUN
fcis-16180	252	4	,	,	PUNCT
fcis-16180	252	5	we	we	PRON
fcis-16180	252	6	evaluate	evaluate	VERB
fcis-16180	252	7	the	the	DET
fcis-16180	252	8	computation	computation	NOUN
fcis-16180	252	9	efficiency	efficiency	NOUN
fcis-16180	252	10	on	on	ADP
fcis-16180	252	11	the	the	DET
fcis-16180	252	12	user	user	NOUN
fcis-16180	252	13	side	side	NOUN
fcis-16180	252	14	by	by	ADP
fcis-16180	252	15	progressively	progressively	ADV
fcis-16180	252	16	increasing	increase	VERB
fcis-16180	252	17	the	the	DET
fcis-16180	252	18	number	number	NOUN
fcis-16180	252	19	of	of	ADP
fcis-16180	252	20	users	user	NOUN
fcis-16180	252	21	for	for	ADP
fcis-16180	252	22	models	model	NOUN
fcis-16180	252	23	with	with	ADP
fcis-16180	252	24	5k	5k	NUM
fcis-16180	252	25	entry	entry	NOUN
fcis-16180	252	26	parameters	parameter	NOUN
fcis-16180	252	27	per	per	ADP
fcis-16180	252	28	user	user	NOUN
fcis-16180	252	29	and	and	CCONJ
fcis-16180	252	30	10k	10k	NOUN
fcis-16180	252	31	entry	entry	NOUN
fcis-16180	252	32	parameters	parameter	NOUN
fcis-16180	252	33	per	per	ADP
fcis-16180	252	34	user	user	NOUN
fcis-16180	252	35	.	.	PUNCT
fcis-16180	253	1	from	from	ADP
fcis-16180	253	2	section	section	NOUN
fcis-16180	253	3	3	3	NUM
fcis-16180	253	4	,	,	PUNCT
fcis-16180	253	5	we	we	PRON
fcis-16180	253	6	can	can	AUX
fcis-16180	253	7	know	know	VERB
fcis-16180	253	8	that	that	SCONJ
fcis-16180	253	9	the	the	DET
fcis-16180	253	10	computation	computation	NOUN
fcis-16180	253	11	overhead	overhead	ADV
fcis-16180	253	12	on	on	ADP
fcis-16180	253	13	the	the	DET
fcis-16180	253	14	user	user	NOUN
fcis-16180	253	15	side	side	NOUN
fcis-16180	253	16	primarily	primarily	ADV
fcis-16180	253	17	arises	arise	VERB
fcis-16180	253	18	from	from	ADP
fcis-16180	253	19	encrypting	encrypt	VERB
fcis-16180	253	20	the	the	DET
fcis-16180	253	21	model	model	NOUN
fcis-16180	253	22	.	.	PUNCT
fcis-16180	254	1	as	as	SCONJ
fcis-16180	254	2	depicted	depict	VERB
fcis-16180	254	3	in	in	ADP
fcis-16180	254	4	fig.5(a	fig.5(a	ADJ
fcis-16180	254	5	)	)	PUNCT
fcis-16180	254	6	and	and	CCONJ
fcis-16180	254	7	fig.5(b	fig.5(b	NOUN
fcis-16180	254	8	)	)	PUNCT
fcis-16180	254	9	show	show	VERB
fcis-16180	254	10	that	that	SCONJ
fcis-16180	254	11	as	as	ADP
fcis-16180	254	12	the	the	DET
fcis-16180	254	13	number	number	NOUN
fcis-16180	254	14	of	of	ADP
fcis-16180	254	15	users	user	NOUN
fcis-16180	254	16	increases	increase	NOUN
fcis-16180	254	17	,	,	PUNCT
fcis-16180	254	18	the	the	DET
fcis-16180	254	19	computational	computational	ADJ
fcis-16180	254	20	overheads	overhead	NOUN
fcis-16180	254	21	on	on	ADP
fcis-16180	254	22	the	the	DET
fcis-16180	254	23	user	user	NOUN
fcis-16180	254	24	side	side	NOUN
fcis-16180	254	25	of	of	ADP
fcis-16180	254	26	ppbrfl	ppbrfl	NOUN
fcis-16180	254	27	and	and	CCONJ
fcis-16180	254	28	the	the	DET
fcis-16180	254	29	scheme	scheme	NOUN
fcis-16180	254	30	[	[	X
fcis-16180	254	31	7	7	X
fcis-16180	254	32	]	]	PUNCT
fcis-16180	254	33	also	also	ADV
fcis-16180	254	34	grow	grow	VERB
fcis-16180	254	35	accordingly	accordingly	ADV
fcis-16180	254	36	.	.	PUNCT
fcis-16180	255	1	furthermore	furthermore	ADV
fcis-16180	255	2	,	,	PUNCT
fcis-16180	255	3	the	the	DET
fcis-16180	255	4	proposed	propose	VERB
fcis-16180	255	5	ppbrfl	ppbrfl	NOUN
fcis-16180	255	6	demonstrates	demonstrate	VERB
fcis-16180	255	7	significantly	significantly	ADV
fcis-16180	255	8	higher	high	ADJ
fcis-16180	255	9	efficiency	efficiency	NOUN
fcis-16180	255	10	on	on	ADP
fcis-16180	255	11	the	the	DET
fcis-16180	255	12	user	user	NOUN
fcis-16180	255	13	side	side	NOUN
fcis-16180	255	14	compared	compare	VERB
fcis-16180	255	15	to	to	ADP
fcis-16180	255	16	the	the	DET
fcis-16180	255	17	scheme	scheme	NOUN
fcis-16180	255	18	[	[	X
fcis-16180	255	19	7	7	NUM
fcis-16180	255	20	]	]	PUNCT
fcis-16180	255	21	.	.	PUNCT
fcis-16180	256	1	(	(	PUNCT
fcis-16180	256	2	a	a	X
fcis-16180	256	3	)	)	PUNCT
fcis-16180	256	4	(	(	PUNCT
fcis-16180	256	5	b	b	X
fcis-16180	256	6	)	)	PUNCT
fcis-16180	256	7	figure	figure	NOUN
fcis-16180	256	8	5	5	NUM
fcis-16180	256	9	.	.	PUNCT
fcis-16180	257	1	comparison	comparison	NOUN
fcis-16180	257	2	of	of	ADP
fcis-16180	257	3	model	model	NOUN
fcis-16180	257	4	encryption	encryption	NOUN
fcis-16180	257	5	computation	computation	NOUN
fcis-16180	257	6	overhead	overhead	ADV
fcis-16180	257	7	on	on	ADP
fcis-16180	257	8	the	the	DET
fcis-16180	257	9	user	user	NOUN
fcis-16180	257	10	side	side	NOUN
fcis-16180	257	11	between	between	ADP
fcis-16180	257	12	ppbrfl	ppbrfl	NOUN
fcis-16180	257	13	and	and	CCONJ
fcis-16180	257	14	the	the	DET
fcis-16180	257	15	scheme	scheme	NOUN
fcis-16180	258	1	[	[	X
fcis-16180	258	2	7	7	NUM
fcis-16180	258	3	]	]	PUNCT
fcis-16180	258	4	.	.	PUNCT
fcis-16180	259	1	(	(	PUNCT
fcis-16180	259	2	a	a	X
fcis-16180	259	3	)	)	PUNCT
fcis-16180	259	4	5k	5k	NUM
fcis-16180	259	5	-	-	PUNCT
fcis-16180	259	6	entry	entry	NOUN
fcis-16180	259	7	parameters	parameter	NOUN
fcis-16180	259	8	per	per	ADP
fcis-16180	259	9	user	user	NOUN
fcis-16180	259	10	(	(	PUNCT
fcis-16180	259	11	b	b	NOUN
fcis-16180	259	12	)	)	PUNCT
fcis-16180	259	13	10k	10k	NOUN
fcis-16180	259	14	-	-	PUNCT
fcis-16180	259	15	entry	entry	NOUN
fcis-16180	259	16	parameters	parameter	NOUN
fcis-16180	259	17	per	per	ADP
fcis-16180	259	18	user	user	NOUN
fcis-16180	259	19	5	5	NUM
fcis-16180	259	20	.	.	PUNCT
fcis-16180	259	21	conclusion	conclusion	NOUN
fcis-16180	259	22	in	in	ADP
fcis-16180	259	23	this	this	DET
fcis-16180	259	24	paper	paper	NOUN
fcis-16180	259	25	,	,	PUNCT
fcis-16180	259	26	we	we	PRON
fcis-16180	259	27	propose	propose	VERB
fcis-16180	259	28	a	a	DET
fcis-16180	259	29	new	new	ADJ
fcis-16180	259	30	privacy	privacy	NOUN
fcis-16180	259	31	-	-	PUNCT
fcis-16180	259	32	preserving	preserve	VERB
fcis-16180	259	33	byzantine	byzantine	ADJ
fcis-16180	259	34	robust	robust	ADJ
fcis-16180	259	35	federated	federated	ADJ
fcis-16180	259	36	learning	learning	NOUN
fcis-16180	259	37	scheme	scheme	NOUN
fcis-16180	259	38	(	(	PUNCT
fcis-16180	259	39	ppbrfl	ppbrfl	NOUN
fcis-16180	259	40	)	)	PUNCT
fcis-16180	259	41	.	.	PUNCT
fcis-16180	260	1	this	this	DET
fcis-16180	260	2	scheme	scheme	NOUN
fcis-16180	260	3	is	be	AUX
fcis-16180	260	4	able	able	ADJ
fcis-16180	260	5	to	to	PART
fcis-16180	260	6	support	support	VERB
fcis-16180	260	7	both	both	DET
fcis-16180	260	8	privacy	privacy	NOUN
fcis-16180	260	9	preserving	preserve	VERB
fcis-16180	260	10	and	and	CCONJ
fcis-16180	260	11	byzantine	byzantine	ADJ
fcis-16180	260	12	robustness	robustness	NOUN
fcis-16180	260	13	.	.	PUNCT
fcis-16180	261	1	specifically	specifically	ADV
fcis-16180	261	2	,	,	PUNCT
fcis-16180	261	3	we	we	PRON
fcis-16180	261	4	utilize	utilize	VERB
fcis-16180	261	5	symmetric	symmetric	ADJ
fcis-16180	261	6	homomorphic	homomorphic	ADJ
fcis-16180	261	7	encryption	encryption	NOUN
fcis-16180	261	8	technology	technology	NOUN
fcis-16180	261	9	to	to	PART
fcis-16180	261	10	encrypt	encrypt	VERB
fcis-16180	261	11	the	the	DET
fcis-16180	261	12	trained	train	VERB
fcis-16180	261	13	model	model	NOUN
fcis-16180	261	14	.	.	PUNCT
fcis-16180	262	1	the	the	DET
fcis-16180	262	2	cloud	cloud	PROPN
fcis-16180	262	3	server	server	NOUN
fcis-16180	262	4	can	can	AUX
fcis-16180	262	5	aggregate	aggregate	VERB
fcis-16180	262	6	the	the	DET
fcis-16180	262	7	encrypted	encrypt	VERB
fcis-16180	262	8	trained	train	VERB
fcis-16180	262	9	models	model	NOUN
fcis-16180	262	10	but	but	CCONJ
fcis-16180	262	11	can	can	AUX
fcis-16180	262	12	not	not	PART
fcis-16180	262	13	obtain	obtain	VERB
fcis-16180	262	14	users	user	NOUN
fcis-16180	262	15	’	’	PART
fcis-16180	262	16	data	datum	NOUN
fcis-16180	262	17	privacy	privacy	NOUN
fcis-16180	262	18	from	from	ADP
fcis-16180	262	19	the	the	DET
fcis-16180	262	20	encrypted	encrypt	VERB
fcis-16180	262	21	trained	train	VERB
fcis-16180	262	22	models	model	NOUN
fcis-16180	262	23	.	.	PUNCT
fcis-16180	263	1	moreover	moreover	ADV
fcis-16180	263	2	,	,	PUNCT
fcis-16180	263	3	we	we	PRON
fcis-16180	263	4	use	use	VERB
fcis-16180	263	5	the	the	DET
fcis-16180	263	6	cosine	cosine	NOUN
fcis-16180	263	7	similarity	similarity	NOUN
fcis-16180	263	8	value	value	NOUN
fcis-16180	263	9	as	as	ADP
fcis-16180	263	10	a	a	DET
fcis-16180	263	11	benchmark	benchmark	NOUN
fcis-16180	263	12	and	and	CCONJ
fcis-16180	263	13	utilize	utilize	VERB
fcis-16180	263	14	a	a	DET
fcis-16180	263	15	grouping	group	VERB
fcis-16180	263	16	mechanism	mechanism	NOUN
fcis-16180	263	17	to	to	PART
fcis-16180	263	18	select	select	VERB
fcis-16180	263	19	users	user	NOUN
fcis-16180	263	20	for	for	ADP
fcis-16180	263	21	aggregation	aggregation	NOUN
fcis-16180	263	22	.	.	PUNCT
fcis-16180	264	1	using	use	VERB
fcis-16180	264	2	the	the	DET
fcis-16180	264	3	above	above	ADJ
fcis-16180	264	4	method	method	NOUN
fcis-16180	264	5	,	,	PUNCT
fcis-16180	264	6	our	our	PRON
fcis-16180	264	7	scheme	scheme	NOUN
fcis-16180	264	8	can	can	AUX
fcis-16180	264	9	effectively	effectively	ADV
fcis-16180	264	10	resist	resist	VERB
fcis-16180	264	11	the	the	DET
fcis-16180	264	12	attacks	attack	NOUN
fcis-16180	264	13	from	from	ADP
fcis-16180	264	14	byzantine	byzantine	ADJ
fcis-16180	264	15	users	user	NOUN
fcis-16180	264	16	.	.	PUNCT
fcis-16180	265	1	experiments	experiment	NOUN
fcis-16180	265	2	illustrate	illustrate	VERB
fcis-16180	265	3	that	that	SCONJ
fcis-16180	265	4	ppbrfl	ppbrfl	NOUN
fcis-16180	265	5	is	be	AUX
fcis-16180	265	6	able	able	ADJ
fcis-16180	265	7	to	to	PART
fcis-16180	265	8	achieve	achieve	VERB
fcis-16180	265	9	almost	almost	ADV
fcis-16180	265	10	the	the	DET
fcis-16180	265	11	same	same	ADJ
fcis-16180	265	12	model	model	NOUN
fcis-16180	265	13	classification	classification	NOUN
fcis-16180	265	14	accuracy	accuracy	NOUN
fcis-16180	265	15	as	as	ADP
fcis-16180	265	16	traditional	traditional	ADJ
fcis-16180	265	17	federated	federated	ADJ
fcis-16180	265	18	learning	learning	NOUN
fcis-16180	265	19	scheme	scheme	NOUN
fcis-16180	265	20	while	while	SCONJ
fcis-16180	265	21	guaranteeing	guarantee	VERB
fcis-16180	265	22	user	user	NOUN
fcis-16180	265	23	privacy	privacy	NOUN
fcis-16180	265	24	and	and	CCONJ
fcis-16180	265	25	byzantine	byzantine	ADJ
fcis-16180	265	26	robustness	robustness	NOUN
fcis-16180	265	27	.	.	PUNCT
fcis-16180	266	1	the	the	DET
fcis-16180	266	2	performance	performance	NOUN
fcis-16180	266	3	on	on	ADP
fcis-16180	266	4	the	the	DET
fcis-16180	266	5	user	user	NOUN
fcis-16180	266	6	side	side	NOUN
fcis-16180	266	7	exhibits	exhibit	VERB
fcis-16180	266	8	favorable	favorable	ADJ
fcis-16180	266	9	efficiency	efficiency	NOUN
fcis-16180	266	10	.	.	PUNCT
fcis-16180	267	1	references	reference	NOUN
fcis-16180	267	2	[	[	X
fcis-16180	267	3	1	1	X
fcis-16180	267	4	]	]	PUNCT
fcis-16180	267	5	j.	j.	PROPN
fcis-16180	267	6	roman	roman	PROPN
fcis-16180	267	7	and	and	CCONJ
fcis-16180	267	8	a.	a.	PROPN
fcis-16180	267	9	jameel	jameel	PROPN
fcis-16180	267	10	,	,	PUNCT
fcis-16180	267	11	“	"	PUNCT
fcis-16180	267	12	backpropagation	backpropagation	NOUN
fcis-16180	267	13	and	and	CCONJ
fcis-16180	267	14	recurrent	recurrent	ADJ
fcis-16180	267	15	neural	neural	ADJ
fcis-16180	267	16	networks	network	NOUN
fcis-16180	267	17	in	in	ADP
fcis-16180	267	18	financial	financial	ADJ
fcis-16180	267	19	analysis	analysis	NOUN
fcis-16180	267	20	of	of	ADP
fcis-16180	267	21	multiple	multiple	ADJ
fcis-16180	267	22	stock	stock	NOUN
fcis-16180	267	23	market	market	NOUN
fcis-16180	267	24	returns	return	NOUN
fcis-16180	267	25	,	,	PUNCT
fcis-16180	267	26	”	"	PUNCT
fcis-16180	267	27	in	in	ADP
fcis-16180	267	28	proceedings	proceeding	NOUN
fcis-16180	267	29	of	of	ADP
fcis-16180	267	30	hicss-29	hicss-29	PROPN
fcis-16180	267	31	:	:	PUNCT
fcis-16180	267	32	29th	29th	ADJ
fcis-16180	267	33	hawaii	hawaii	PROPN
fcis-16180	267	34	international	international	ADJ
fcis-16180	267	35	conference	conference	NOUN
fcis-16180	267	36	on	on	ADP
fcis-16180	267	37	system	system	NOUN
fcis-16180	267	38	sciences	science	NOUN
fcis-16180	267	39	,	,	PUNCT
fcis-16180	267	40	vol	vol	NOUN
fcis-16180	267	41	.	.	PROPN
fcis-16180	268	1	2	2	NUM
fcis-16180	268	2	.	.	X
fcis-16180	268	3	ieee	ieee	NOUN
fcis-16180	268	4	,	,	PUNCT
fcis-16180	268	5	1996	1996	NUM
fcis-16180	268	6	,	,	PUNCT
fcis-16180	268	7	pp	pp	ADV
fcis-16180	268	8	.	.	PUNCT
fcis-16180	269	1	454–460	454–460	NUM
fcis-16180	269	2	.	.	PUNCT
fcis-16180	270	1	[	[	X
fcis-16180	270	2	2	2	NUM
fcis-16180	270	3	]	]	PUNCT
fcis-16180	270	4	m.	m.	NOUN
fcis-16180	270	5	bakator	bakator	NOUN
fcis-16180	270	6	and	and	CCONJ
fcis-16180	270	7	d.	d.	PROPN
fcis-16180	270	8	radosav	radosav	PROPN
fcis-16180	270	9	,	,	PUNCT
fcis-16180	270	10	“	"	PUNCT
fcis-16180	270	11	deep	deep	ADJ
fcis-16180	270	12	learning	learning	NOUN
fcis-16180	270	13	and	and	CCONJ
fcis-16180	270	14	medical	medical	ADJ
fcis-16180	270	15	diagnosis	diagnosis	NOUN
fcis-16180	270	16	:	:	PUNCT
fcis-16180	270	17	a	a	DET
fcis-16180	270	18	review	review	NOUN
fcis-16180	270	19	of	of	ADP
fcis-16180	270	20	literature	literature	NOUN
fcis-16180	270	21	,	,	PUNCT
fcis-16180	270	22	”	"	PUNCT
fcis-16180	270	23	multimodal	multimodal	NOUN
fcis-16180	270	24	technologies	technology	NOUN
fcis-16180	270	25	and	and	CCONJ
fcis-16180	270	26	interaction	interaction	NOUN
fcis-16180	270	27	,	,	PUNCT
fcis-16180	270	28	vol	vol	NOUN
fcis-16180	270	29	.	.	PROPN
fcis-16180	270	30	2	2	NUM
fcis-16180	270	31	,	,	PUNCT
fcis-16180	270	32	no	no	INTJ
fcis-16180	270	33	.	.	NOUN
fcis-16180	270	34	3	3	NUM
fcis-16180	270	35	,	,	PUNCT
fcis-16180	270	36	p.	p.	NOUN
fcis-16180	270	37	47	47	NUM
fcis-16180	270	38	,	,	PUNCT
fcis-16180	270	39	2018	2018	NUM
fcis-16180	270	40	.	.	PUNCT
fcis-16180	271	1	23	23	NUM
fcis-16180	272	1	[	[	X
fcis-16180	272	2	3	3	NUM
fcis-16180	272	3	]	]	PUNCT
fcis-16180	272	4	a.	a.	PROPN
fcis-16180	272	5	b.	b.	PROPN
fcis-16180	272	6	nassif	nassif	PROPN
fcis-16180	272	7	,	,	PUNCT
fcis-16180	272	8	i.	i.	PROPN
fcis-16180	272	9	shahin	shahin	PROPN
fcis-16180	272	10	,	,	PUNCT
fcis-16180	272	11	i.	i.	PROPN
fcis-16180	272	12	attili	attili	PROPN
fcis-16180	272	13	,	,	PUNCT
fcis-16180	272	14	m.	m.	NOUN
fcis-16180	272	15	azzeh	azzeh	PROPN
fcis-16180	272	16	,	,	PUNCT
fcis-16180	272	17	and	and	CCONJ
fcis-16180	272	18	k.	k.	PROPN
fcis-16180	272	19	shaalan	shaalan	PROPN
fcis-16180	272	20	,	,	PUNCT
fcis-16180	272	21	“	"	PUNCT
fcis-16180	272	22	speech	speech	NOUN
fcis-16180	272	23	recognition	recognition	NOUN
fcis-16180	272	24	using	use	VERB
fcis-16180	272	25	deep	deep	ADJ
fcis-16180	272	26	neural	neural	ADJ
fcis-16180	272	27	networks	network	NOUN
fcis-16180	272	28	:	:	PUNCT
fcis-16180	272	29	a	a	DET
fcis-16180	272	30	systematic	systematic	ADJ
fcis-16180	272	31	review	review	NOUN
fcis-16180	272	32	,	,	PUNCT
fcis-16180	272	33	”	"	PUNCT
fcis-16180	272	34	ieee	ieee	NOUN
fcis-16180	272	35	access	access	NOUN
fcis-16180	272	36	,	,	PUNCT
fcis-16180	272	37	vol	vol	NOUN
fcis-16180	272	38	.	.	PROPN
fcis-16180	272	39	7	7	NUM
fcis-16180	272	40	,	,	PUNCT
fcis-16180	272	41	pp	pp	ADJ
fcis-16180	272	42	.	.	PUNCT
fcis-16180	273	1	19	19	NUM
fcis-16180	273	2	143–19	143–19	PROPN
fcis-16180	273	3	165	165	NUM
fcis-16180	273	4	,	,	PUNCT
fcis-16180	273	5	2019	2019	NUM
fcis-16180	273	6	.	.	PUNCT
fcis-16180	274	1	[	[	X
fcis-16180	274	2	4	4	X
fcis-16180	274	3	]	]	X
fcis-16180	274	4	g.	g.	PROPN
fcis-16180	274	5	xu	xu	PROPN
fcis-16180	274	6	,	,	PUNCT
fcis-16180	274	7	h.	h.	PROPN
fcis-16180	274	8	li	li	PROPN
fcis-16180	274	9	,	,	PUNCT
fcis-16180	274	10	s.	s.	PROPN
fcis-16180	274	11	liu	liu	PROPN
fcis-16180	274	12	,	,	PUNCT
fcis-16180	274	13	k.	k.	PROPN
fcis-16180	274	14	yang	yang	PROPN
fcis-16180	274	15	,	,	PUNCT
fcis-16180	274	16	and	and	CCONJ
fcis-16180	274	17	x.	x.	PROPN
fcis-16180	274	18	lin	lin	PROPN
fcis-16180	274	19	,	,	PUNCT
fcis-16180	274	20	“	"	PUNCT
fcis-16180	274	21	verifynet	verifynet	NOUN
fcis-16180	274	22	:	:	PUNCT
fcis-16180	274	23	secure	secure	ADJ
fcis-16180	274	24	and	and	CCONJ
fcis-16180	274	25	verifiable	verifiable	ADJ
fcis-16180	274	26	federated	federated	ADJ
fcis-16180	274	27	learning	learning	NOUN
fcis-16180	274	28	,	,	PUNCT
fcis-16180	274	29	”	"	PUNCT
fcis-16180	274	30	ieee	ieee	NOUN
fcis-16180	274	31	transactions	transaction	NOUN
fcis-16180	274	32	on	on	ADP
fcis-16180	274	33	information	information	NOUN
fcis-16180	274	34	forensics	forensic	NOUN
fcis-16180	274	35	and	and	CCONJ
fcis-16180	274	36	security	security	NOUN
fcis-16180	274	37	,	,	PUNCT
fcis-16180	274	38	vol	vol	NOUN
fcis-16180	274	39	.	.	PROPN
fcis-16180	274	40	15	15	NUM
fcis-16180	274	41	,	,	PUNCT
fcis-16180	274	42	pp	pp	ADJ
fcis-16180	274	43	.	.	PUNCT
fcis-16180	275	1	911–926	911–926	NUM
fcis-16180	275	2	,	,	PUNCT
fcis-16180	275	3	2019	2019	NUM
fcis-16180	275	4	.	.	PUNCT
fcis-16180	276	1	[	[	X
fcis-16180	276	2	5	5	X
fcis-16180	276	3	]	]	PUNCT
fcis-16180	276	4	j.	j.	PROPN
fcis-16180	276	5	konecnˇ	konecnˇ	PROPN
fcis-16180	276	6	y	y	PROPN
fcis-16180	276	7	,	,	PUNCT
fcis-16180	276	8	h.	h.	PROPN
fcis-16180	276	9	b.	b.	PROPN
fcis-16180	276	10	mcmahan	mcmahan	PROPN
fcis-16180	276	11	,	,	PUNCT
fcis-16180	276	12	f.	f.	PROPN
fcis-16180	276	13	x.	x.	PROPN
fcis-16180	276	14	yu	yu	PROPN
fcis-16180	276	15	,	,	PUNCT
fcis-16180	276	16	p.	p.	PROPN
fcis-16180	276	17	richt	richt	PROPN
fcis-16180	276	18	`	`	PUNCT
fcis-16180	276	19	arik	arik	PROPN
fcis-16180	276	20	,	,	PUNCT
fcis-16180	276	21	a.	a.	PROPN
fcis-16180	276	22	t.	t.	PROPN
fcis-16180	276	23	´	´	PROPN
fcis-16180	276	24	suresh	suresh	PROPN
fcis-16180	276	25	,	,	PUNCT
fcis-16180	276	26	and	and	CCONJ
fcis-16180	276	27	d.	d.	PROPN
fcis-16180	276	28	bacon	bacon	NOUN
fcis-16180	276	29	,	,	PUNCT
fcis-16180	276	30	“	"	PUNCT
fcis-16180	276	31	federated	federated	ADJ
fcis-16180	276	32	learning	learning	NOUN
fcis-16180	276	33	:	:	PUNCT
fcis-16180	276	34	strategies	strategy	NOUN
fcis-16180	276	35	for	for	ADP
fcis-16180	276	36	improving	improve	VERB
fcis-16180	276	37	communication	communication	NOUN
fcis-16180	276	38	efficiency	efficiency	NOUN
fcis-16180	276	39	,	,	PUNCT
fcis-16180	276	40	”	"	PUNCT
fcis-16180	276	41	arxiv	arxiv	PROPN
fcis-16180	276	42	preprint	preprint	VERB
fcis-16180	276	43	arxiv:1610.05492	arxiv:1610.05492	PROPN
fcis-16180	276	44	,	,	PUNCT
fcis-16180	276	45	2016	2016	NUM
fcis-16180	276	46	.	.	PUNCT
fcis-16180	277	1	[	[	X
fcis-16180	277	2	6	6	NUM
fcis-16180	277	3	]	]	PUNCT
fcis-16180	277	4	k.	k.	PROPN
fcis-16180	277	5	bonawitz	bonawitz	PROPN
fcis-16180	277	6	,	,	PUNCT
fcis-16180	277	7	h.	h.	PROPN
fcis-16180	277	8	eichner	eichner	PROPN
fcis-16180	277	9	,	,	PUNCT
fcis-16180	277	10	w.	w.	PROPN
fcis-16180	277	11	grieskamp	grieskamp	PROPN
fcis-16180	277	12	,	,	PUNCT
fcis-16180	277	13	d.	d.	PROPN
fcis-16180	277	14	huba	huba	PROPN
fcis-16180	277	15	,	,	PUNCT
fcis-16180	277	16	a.	a.	NOUN
fcis-16180	277	17	ingerman	ingerman	NOUN
fcis-16180	277	18	,	,	PUNCT
fcis-16180	277	19	v.	v.	ADP
fcis-16180	277	20	ivanov	ivanov	PROPN
fcis-16180	277	21	,	,	PUNCT
fcis-16180	277	22	c.	c.	PROPN
fcis-16180	277	23	kiddon	kiddon	PROPN
fcis-16180	277	24	,	,	PUNCT
fcis-16180	277	25	j.	j.	PROPN
fcis-16180	277	26	konecnˇ	konecnˇ	PROPN
fcis-16180	277	27	y	y	PROPN
fcis-16180	277	28	,	,	PUNCT
fcis-16180	277	29	s.	s.	PROPN
fcis-16180	277	30	mazzocchi	mazzocchi	PROPN
fcis-16180	277	31	,	,	PUNCT
fcis-16180	277	32	`	`	PUNCT
fcis-16180	277	33	b.	b.	PROPN
fcis-16180	277	34	mcmahan	mcmahan	PROPN
fcis-16180	277	35	et	et	PROPN
fcis-16180	277	36	al	al	PROPN
fcis-16180	277	37	.	.	PROPN
fcis-16180	277	38	,	,	PUNCT
fcis-16180	277	39	“	"	PUNCT
fcis-16180	277	40	towards	towards	ADP
fcis-16180	277	41	federated	federated	ADJ
fcis-16180	277	42	learning	learning	NOUN
fcis-16180	277	43	at	at	ADP
fcis-16180	277	44	scale	scale	NOUN
fcis-16180	277	45	:	:	PUNCT
fcis-16180	277	46	9	9	NUM
fcis-16180	277	47	system	system	NOUN
fcis-16180	277	48	design	design	NOUN
fcis-16180	277	49	,	,	PUNCT
fcis-16180	277	50	”	"	PUNCT
fcis-16180	277	51	proceedings	proceeding	NOUN
fcis-16180	277	52	of	of	ADP
fcis-16180	277	53	machine	machine	NOUN
fcis-16180	277	54	learning	learning	NOUN
fcis-16180	277	55	and	and	CCONJ
fcis-16180	277	56	systems	system	NOUN
fcis-16180	277	57	,	,	PUNCT
fcis-16180	277	58	vol	vol	NOUN
fcis-16180	277	59	.	.	PROPN
fcis-16180	277	60	1	1	NUM
fcis-16180	277	61	,	,	PUNCT
fcis-16180	277	62	pp	pp	ADJ
fcis-16180	277	63	.	.	PUNCT
fcis-16180	278	1	374–388	374–388	NUM
fcis-16180	278	2	,	,	PUNCT
fcis-16180	278	3	2019	2019	NUM
fcis-16180	278	4	.	.	PUNCT
fcis-16180	279	1	[	[	X
fcis-16180	279	2	7	7	X
fcis-16180	279	3	]	]	X
fcis-16180	279	4	l.	l.	PROPN
fcis-16180	279	5	t.	t.	PROPN
fcis-16180	279	6	phong	phong	PROPN
fcis-16180	279	7	,	,	PUNCT
fcis-16180	279	8	y.	y.	PROPN
fcis-16180	279	9	aono	aono	PROPN
fcis-16180	279	10	,	,	PUNCT
fcis-16180	279	11	t.	t.	PROPN
fcis-16180	279	12	hayashi	hayashi	PROPN
fcis-16180	279	13	,	,	PUNCT
fcis-16180	279	14	l.	l.	PROPN
fcis-16180	279	15	wang	wang	PROPN
fcis-16180	279	16	,	,	PUNCT
fcis-16180	279	17	and	and	CCONJ
fcis-16180	279	18	s.	s.	PROPN
fcis-16180	279	19	moriai	moriai	PROPN
fcis-16180	279	20	,	,	PUNCT
fcis-16180	279	21	“	"	PUNCT
fcis-16180	279	22	privacy	privacy	NOUN
fcis-16180	279	23	-	-	PUNCT
fcis-16180	279	24	preserving	preserve	VERB
fcis-16180	279	25	deep	deep	ADJ
fcis-16180	279	26	learning	learning	NOUN
fcis-16180	279	27	via	via	ADP
fcis-16180	279	28	additively	additively	ADV
fcis-16180	279	29	homomorphic	homomorphic	ADJ
fcis-16180	279	30	encryption	encryption	NOUN
fcis-16180	279	31	,	,	PUNCT
fcis-16180	279	32	”	"	PUNCT
fcis-16180	279	33	ieee	ieee	NOUN
fcis-16180	279	34	transactions	transaction	NOUN
fcis-16180	279	35	on	on	ADP
fcis-16180	279	36	information	information	NOUN
fcis-16180	279	37	forensics	forensic	NOUN
fcis-16180	279	38	and	and	CCONJ
fcis-16180	279	39	security	security	NOUN
fcis-16180	279	40	,	,	PUNCT
fcis-16180	279	41	vol	vol	NOUN
fcis-16180	279	42	.	.	PROPN
fcis-16180	279	43	13	13	NUM
fcis-16180	279	44	,	,	PUNCT
fcis-16180	279	45	no	no	INTJ
fcis-16180	279	46	.	.	NOUN
fcis-16180	279	47	5	5	NUM
fcis-16180	279	48	,	,	PUNCT
fcis-16180	279	49	pp	pp	ADJ
fcis-16180	279	50	.	.	PUNCT
fcis-16180	279	51	1333	1333	NUM
fcis-16180	279	52	–	–	PUNCT
fcis-16180	279	53	1345	1345	NUM
fcis-16180	279	54	,	,	PUNCT
fcis-16180	279	55	2018	2018	NUM
fcis-16180	279	56	.	.	PUNCT
fcis-16180	280	1	[	[	X
fcis-16180	280	2	8	8	NUM
fcis-16180	280	3	]	]	X
fcis-16180	280	4	r.	r.	PROPN
fcis-16180	280	5	shokri	shokri	PROPN
fcis-16180	280	6	and	and	CCONJ
fcis-16180	280	7	v.	v.	ADP
fcis-16180	280	8	shmatikov	shmatikov	PROPN
fcis-16180	280	9	,	,	PUNCT
fcis-16180	280	10	“	"	PUNCT
fcis-16180	280	11	privacy	privacy	NOUN
fcis-16180	280	12	-	-	PUNCT
fcis-16180	280	13	preserving	preserve	VERB
fcis-16180	280	14	deep	deep	ADJ
fcis-16180	280	15	learning	learning	NOUN
fcis-16180	280	16	,	,	PUNCT
fcis-16180	280	17	”	"	PUNCT
fcis-16180	280	18	in	in	ADP
fcis-16180	280	19	proceedings	proceeding	NOUN
fcis-16180	280	20	of	of	ADP
fcis-16180	280	21	the	the	DET
fcis-16180	280	22	22nd	22nd	PROPN
fcis-16180	280	23	acm	acm	PROPN
fcis-16180	280	24	sigsac	sigsac	NOUN
fcis-16180	280	25	conference	conference	PROPN
fcis-16180	280	26	on	on	ADP
fcis-16180	280	27	computer	computer	NOUN
fcis-16180	280	28	and	and	CCONJ
fcis-16180	280	29	communications	communication	NOUN
fcis-16180	280	30	security	security	NOUN
fcis-16180	280	31	,	,	PUNCT
fcis-16180	280	32	2015	2015	NUM
fcis-16180	280	33	,	,	PUNCT
fcis-16180	280	34	pp	pp	ADJ
fcis-16180	280	35	.	.	PUNCT
fcis-16180	280	36	1310–1321	1310–1321	NUM
fcis-16180	280	37	.	.	PUNCT
fcis-16180	281	1	[	[	X
fcis-16180	281	2	9	9	NUM
fcis-16180	281	3	]	]	X
fcis-16180	281	4	f.	f.	PROPN
fcis-16180	281	5	mo	mo	PROPN
fcis-16180	281	6	,	,	PUNCT
fcis-16180	281	7	h.	h.	PROPN
fcis-16180	281	8	haddadi	haddadi	PROPN
fcis-16180	281	9	,	,	PUNCT
fcis-16180	281	10	k.	k.	PROPN
fcis-16180	281	11	katevas	katevas	PROPN
fcis-16180	281	12	,	,	PUNCT
fcis-16180	281	13	e.	e.	PROPN
fcis-16180	281	14	marin	marin	PROPN
fcis-16180	281	15	,	,	PUNCT
fcis-16180	281	16	d.	d.	PROPN
fcis-16180	281	17	perino	perino	PROPN
fcis-16180	281	18	,	,	PUNCT
fcis-16180	281	19	and	and	CCONJ
fcis-16180	281	20	n.	n.	PROPN
fcis-16180	281	21	kourtellis	kourtellis	PROPN
fcis-16180	281	22	,	,	PUNCT
fcis-16180	281	23	“	"	PUNCT
fcis-16180	281	24	ppfl	ppfl	NOUN
fcis-16180	281	25	:	:	PUNCT
fcis-16180	281	26	privacy	privacy	NOUN
fcis-16180	281	27	-	-	PUNCT
fcis-16180	281	28	preserving	preserve	VERB
fcis-16180	281	29	federated	federated	ADJ
fcis-16180	281	30	learning	learning	NOUN
fcis-16180	281	31	with	with	ADP
fcis-16180	281	32	trusted	trust	VERB
fcis-16180	281	33	execution	execution	NOUN
fcis-16180	281	34	environments	environment	NOUN
fcis-16180	281	35	,	,	PUNCT
fcis-16180	281	36	”	"	PUNCT
fcis-16180	281	37	in	in	ADP
fcis-16180	281	38	proceedings	proceeding	NOUN
fcis-16180	281	39	of	of	ADP
fcis-16180	281	40	the	the	DET
fcis-16180	281	41	19th	19th	ADJ
fcis-16180	281	42	annual	annual	ADJ
fcis-16180	281	43	international	international	ADJ
fcis-16180	281	44	conference	conference	NOUN
fcis-16180	281	45	on	on	ADP
fcis-16180	281	46	mobile	mobile	ADJ
fcis-16180	281	47	systems	system	NOUN
fcis-16180	281	48	,	,	PUNCT
fcis-16180	281	49	applications	application	NOUN
fcis-16180	281	50	,	,	PUNCT
fcis-16180	281	51	and	and	CCONJ
fcis-16180	281	52	services	service	NOUN
fcis-16180	281	53	,	,	PUNCT
fcis-16180	281	54	2021	2021	NUM
fcis-16180	281	55	,	,	PUNCT
fcis-16180	281	56	pp	pp	ADV
fcis-16180	281	57	.	.	PUNCT
fcis-16180	282	1	94–108	94–108	NUM
fcis-16180	282	2	.	.	PUNCT
fcis-16180	283	1	[	[	X
fcis-16180	283	2	10	10	NUM
fcis-16180	283	3	]	]	X
fcis-16180	283	4	h.	h.	PROPN
fcis-16180	283	5	fang	fang	PROPN
fcis-16180	283	6	and	and	CCONJ
fcis-16180	283	7	q.	q.	PROPN
fcis-16180	283	8	qian	qian	PROPN
fcis-16180	283	9	,	,	PUNCT
fcis-16180	283	10	“	"	PUNCT
fcis-16180	283	11	privacy	privacy	NOUN
fcis-16180	283	12	preserving	preserve	VERB
fcis-16180	283	13	machine	machine	NOUN
fcis-16180	283	14	learning	learn	VERB
fcis-16180	283	15	with	with	ADP
fcis-16180	283	16	homomorphic	homomorphic	ADJ
fcis-16180	283	17	encryption	encryption	NOUN
fcis-16180	283	18	and	and	CCONJ
fcis-16180	283	19	federated	federated	ADJ
fcis-16180	283	20	learning	learning	NOUN
fcis-16180	283	21	,	,	PUNCT
fcis-16180	283	22	”	"	PUNCT
fcis-16180	283	23	future	future	ADJ
fcis-16180	283	24	internet	internet	NOUN
fcis-16180	283	25	,	,	PUNCT
fcis-16180	283	26	vol	vol	NOUN
fcis-16180	283	27	.	.	PROPN
fcis-16180	283	28	13	13	NUM
fcis-16180	283	29	,	,	PUNCT
fcis-16180	283	30	no	no	INTJ
fcis-16180	283	31	.	.	NOUN
fcis-16180	283	32	4	4	NUM
fcis-16180	283	33	,	,	PUNCT
fcis-16180	283	34	p.	p.	NOUN
fcis-16180	283	35	94	94	NUM
fcis-16180	283	36	,	,	PUNCT
fcis-16180	283	37	2021	2021	NUM
fcis-16180	283	38	.	.	PUNCT
fcis-16180	284	1	[	[	X
fcis-16180	284	2	11	11	NUM
fcis-16180	284	3	]	]	PUNCT
fcis-16180	284	4	j.	j.	PROPN
fcis-16180	284	5	ma	ma	PROPN
fcis-16180	284	6	,	,	PUNCT
fcis-16180	284	7	s.-a	s.-a	PROPN
fcis-16180	284	8	.	.	PUNCT
fcis-16180	285	1	naas	naas	PROPN
fcis-16180	285	2	,	,	PUNCT
fcis-16180	285	3	s.	s.	PROPN
fcis-16180	285	4	sigg	sigg	PROPN
fcis-16180	285	5	,	,	PUNCT
fcis-16180	285	6	and	and	CCONJ
fcis-16180	285	7	x.	x.	NOUN
fcis-16180	285	8	lyu	lyu	NOUN
fcis-16180	285	9	,	,	PUNCT
fcis-16180	285	10	“	"	PUNCT
fcis-16180	285	11	privacy	privacy	NOUN
fcis-16180	285	12	-	-	PUNCT
fcis-16180	285	13	preserving	preserve	VERB
fcis-16180	285	14	federated	federated	ADJ
fcis-16180	285	15	learning	learning	NOUN
fcis-16180	285	16	based	base	VERB
fcis-16180	285	17	on	on	ADP
fcis-16180	285	18	multi	multi	ADJ
fcis-16180	285	19	-	-	ADJ
fcis-16180	285	20	key	key	ADJ
fcis-16180	285	21	homomorphic	homomorphic	ADJ
fcis-16180	285	22	encryption	encryption	NOUN
fcis-16180	285	23	,	,	PUNCT
fcis-16180	285	24	”	"	PUNCT
fcis-16180	285	25	international	international	ADJ
fcis-16180	285	26	journal	journal	NOUN
fcis-16180	285	27	of	of	ADP
fcis-16180	285	28	intelligent	intelligent	ADJ
fcis-16180	285	29	systems	system	NOUN
fcis-16180	285	30	,	,	PUNCT
fcis-16180	285	31	vol	vol	NOUN
fcis-16180	285	32	.	.	PROPN
fcis-16180	285	33	37	37	NUM
fcis-16180	285	34	,	,	PUNCT
fcis-16180	285	35	no	no	INTJ
fcis-16180	285	36	.	.	NOUN
fcis-16180	285	37	9	9	NUM
fcis-16180	285	38	,	,	PUNCT
fcis-16180	285	39	pp	pp	ADJ
fcis-16180	285	40	.	.	PUNCT
fcis-16180	285	41	5880–5901	5880–5901	NUM
fcis-16180	285	42	,	,	PUNCT
fcis-16180	285	43	2022	2022	NUM
fcis-16180	285	44	.	.	PUNCT
fcis-16180	286	1	[	[	X
fcis-16180	286	2	12	12	NUM
fcis-16180	286	3	]	]	PUNCT
fcis-16180	286	4	x.	x.	PROPN
fcis-16180	286	5	zhang	zhang	PROPN
fcis-16180	286	6	,	,	PUNCT
fcis-16180	286	7	a.	a.	PROPN
fcis-16180	286	8	fu	fu	PROPN
fcis-16180	286	9	,	,	PUNCT
fcis-16180	286	10	h.	h.	PROPN
fcis-16180	286	11	wang	wang	PROPN
fcis-16180	286	12	,	,	PUNCT
fcis-16180	286	13	c.	c.	PROPN
fcis-16180	286	14	zhou	zhou	PROPN
fcis-16180	286	15	,	,	PUNCT
fcis-16180	286	16	and	and	CCONJ
fcis-16180	286	17	z.	z.	PROPN
fcis-16180	286	18	chen	chen	PROPN
fcis-16180	286	19	,	,	PUNCT
fcis-16180	286	20	“	"	PUNCT
fcis-16180	286	21	a	a	DET
fcis-16180	286	22	privacypreserving	privacypreserving	NOUN
fcis-16180	286	23	and	and	CCONJ
fcis-16180	286	24	verifiable	verifiable	ADJ
fcis-16180	286	25	federated	federated	ADJ
fcis-16180	286	26	learning	learning	NOUN
fcis-16180	286	27	scheme	scheme	NOUN
fcis-16180	286	28	,	,	PUNCT
fcis-16180	286	29	”	"	PUNCT
fcis-16180	286	30	in	in	ADP
fcis-16180	286	31	icc	icc	PROPN
fcis-16180	286	32	2020	2020	NUM
fcis-16180	286	33	-	-	SYM
fcis-16180	286	34	2020	2020	NUM
fcis-16180	286	35	ieee	ieee	NOUN
fcis-16180	286	36	international	international	ADJ
fcis-16180	286	37	conference	conference	NOUN
fcis-16180	286	38	on	on	ADP
fcis-16180	286	39	communications	communication	NOUN
fcis-16180	286	40	(	(	PUNCT
fcis-16180	286	41	icc	icc	PROPN
fcis-16180	286	42	)	)	PUNCT
fcis-16180	286	43	.	.	PUNCT
fcis-16180	287	1	ieee	ieee	PROPN
fcis-16180	287	2	,	,	PUNCT
fcis-16180	287	3	2020	2020	NUM
fcis-16180	287	4	,	,	PUNCT
fcis-16180	287	5	pp	pp	ADV
fcis-16180	287	6	.	.	PUNCT
fcis-16180	288	1	1–6	1–6	X
fcis-16180	288	2	.	.	PUNCT
fcis-16180	289	1	[	[	X
fcis-16180	289	2	13	13	NUM
fcis-16180	289	3	]	]	X
fcis-16180	289	4	w.	w.	PROPN
fcis-16180	289	5	wang	wang	PROPN
fcis-16180	289	6	,	,	PUNCT
fcis-16180	289	7	x.	x.	PROPN
fcis-16180	289	8	li	li	PROPN
fcis-16180	289	9	,	,	PUNCT
fcis-16180	289	10	x.	x.	PROPN
fcis-16180	289	11	qiu	qiu	PROPN
fcis-16180	289	12	,	,	PUNCT
fcis-16180	289	13	x.	x.	PROPN
fcis-16180	289	14	zhang	zhang	PROPN
fcis-16180	289	15	,	,	PUNCT
fcis-16180	289	16	j.	j.	PROPN
fcis-16180	289	17	zhao	zhao	PROPN
fcis-16180	289	18	,	,	PUNCT
fcis-16180	289	19	and	and	CCONJ
fcis-16180	289	20	v.	v.	ADP
fcis-16180	289	21	brusic	brusic	ADJ
fcis-16180	289	22	,	,	PUNCT
fcis-16180	289	23	“	"	PUNCT
fcis-16180	289	24	a	a	DET
fcis-16180	289	25	privacy	privacy	NOUN
fcis-16180	289	26	preserving	preserve	VERB
fcis-16180	289	27	framework	framework	NOUN
fcis-16180	289	28	for	for	ADP
fcis-16180	289	29	federated	federated	ADJ
fcis-16180	289	30	learning	learning	NOUN
fcis-16180	289	31	in	in	ADP
fcis-16180	289	32	smart	smart	ADJ
fcis-16180	289	33	healthcare	healthcare	NOUN
fcis-16180	289	34	systems	system	NOUN
fcis-16180	289	35	,	,	PUNCT
fcis-16180	289	36	”	"	PUNCT
fcis-16180	289	37	information	information	NOUN
fcis-16180	289	38	processing	processing	NOUN
fcis-16180	289	39	&	&	CCONJ
fcis-16180	289	40	management	management	NOUN
fcis-16180	289	41	,	,	PUNCT
fcis-16180	289	42	vol	vol	NOUN
fcis-16180	289	43	.	.	PROPN
fcis-16180	289	44	60	60	NUM
fcis-16180	289	45	,	,	PUNCT
fcis-16180	289	46	no	no	INTJ
fcis-16180	289	47	.	.	NOUN
fcis-16180	289	48	1	1	NUM
fcis-16180	289	49	,	,	PUNCT
fcis-16180	289	50	p.	p.	NOUN
fcis-16180	289	51	103167	103167	NUM
fcis-16180	289	52	,	,	PUNCT
fcis-16180	289	53	2023	2023	NUM
fcis-16180	289	54	.	.	PUNCT
fcis-16180	290	1	[	[	X
fcis-16180	290	2	14	14	NUM
fcis-16180	290	3	]	]	X
fcis-16180	290	4	s.	s.	PROPN
fcis-16180	290	5	truex	truex	PROPN
fcis-16180	290	6	,	,	PUNCT
fcis-16180	290	7	n.	n.	PROPN
fcis-16180	290	8	baracaldo	baracaldo	PROPN
fcis-16180	290	9	,	,	PUNCT
fcis-16180	290	10	a.	a.	PROPN
fcis-16180	290	11	anwar	anwar	PROPN
fcis-16180	290	12	,	,	PUNCT
fcis-16180	290	13	t.	t.	NOUN
fcis-16180	290	14	steinke	steinke	PROPN
fcis-16180	290	15	,	,	PUNCT
fcis-16180	290	16	h.	h.	PROPN
fcis-16180	290	17	ludwig	ludwig	PROPN
fcis-16180	290	18	,	,	PUNCT
fcis-16180	290	19	r.	r.	PROPN
fcis-16180	290	20	zhang	zhang	PROPN
fcis-16180	290	21	,	,	PUNCT
fcis-16180	290	22	and	and	CCONJ
fcis-16180	290	23	y.	y.	PROPN
fcis-16180	290	24	zhou	zhou	PROPN
fcis-16180	290	25	,	,	PUNCT
fcis-16180	290	26	“	"	PUNCT
fcis-16180	290	27	a	a	DET
fcis-16180	290	28	hybrid	hybrid	ADJ
fcis-16180	290	29	approach	approach	NOUN
fcis-16180	290	30	to	to	ADP
fcis-16180	290	31	privacy	privacy	NOUN
fcis-16180	290	32	-	-	PUNCT
fcis-16180	290	33	preserving	preserve	VERB
fcis-16180	290	34	federated	federated	ADJ
fcis-16180	290	35	learning	learning	NOUN
fcis-16180	290	36	,	,	PUNCT
fcis-16180	290	37	”	"	PUNCT
fcis-16180	290	38	in	in	ADP
fcis-16180	290	39	proceedings	proceeding	NOUN
fcis-16180	290	40	of	of	ADP
fcis-16180	290	41	the	the	DET
fcis-16180	290	42	12th	12th	ADJ
fcis-16180	290	43	acm	acm	NOUN
fcis-16180	290	44	workshop	workshop	NOUN
fcis-16180	290	45	on	on	ADP
fcis-16180	290	46	artificial	artificial	ADJ
fcis-16180	290	47	intelligence	intelligence	NOUN
fcis-16180	290	48	and	and	CCONJ
fcis-16180	290	49	security	security	NOUN
fcis-16180	290	50	,	,	PUNCT
fcis-16180	290	51	2019	2019	NUM
fcis-16180	290	52	,	,	PUNCT
fcis-16180	290	53	pp	pp	ADP
fcis-16180	290	54	.	.	PUNCT
fcis-16180	291	1	1–11	1–11	NOUN
fcis-16180	291	2	.	.	PUNCT
fcis-16180	292	1	[	[	X
fcis-16180	292	2	15	15	NUM
fcis-16180	292	3	]	]	X
fcis-16180	292	4	j.	j.	PROPN
fcis-16180	292	5	xu	xu	PROPN
fcis-16180	292	6	,	,	PUNCT
fcis-16180	292	7	s.-l	s.-l	PROPN
fcis-16180	292	8	.	.	PUNCT
fcis-16180	293	1	huang	huang	PROPN
fcis-16180	293	2	,	,	PUNCT
fcis-16180	293	3	l.	l.	PROPN
fcis-16180	293	4	song	song	PROPN
fcis-16180	293	5	,	,	PUNCT
fcis-16180	293	6	and	and	CCONJ
fcis-16180	293	7	t.	t.	PROPN
fcis-16180	293	8	lan	lan	PROPN
fcis-16180	293	9	,	,	PUNCT
fcis-16180	293	10	“	"	PUNCT
fcis-16180	293	11	signguard	signguard	NOUN
fcis-16180	293	12	:	:	PUNCT
fcis-16180	293	13	byzantine	byzantine	ADJ
fcis-16180	293	14	-	-	PUNCT
fcis-16180	293	15	robust	robust	ADJ
fcis-16180	293	16	federated	federated	ADJ
fcis-16180	293	17	learning	learning	NOUN
fcis-16180	293	18	through	through	ADP
fcis-16180	293	19	collaborative	collaborative	ADJ
fcis-16180	293	20	malicious	malicious	ADJ
fcis-16180	293	21	gradient	gradient	NOUN
fcis-16180	293	22	filtering	filtering	NOUN
fcis-16180	293	23	,	,	PUNCT
fcis-16180	293	24	”	"	PUNCT
fcis-16180	293	25	arxiv	arxiv	PROPN
fcis-16180	293	26	preprint	preprint	NOUN
fcis-16180	293	27	arxiv:2109.05872	arxiv:2109.05872	NOUN
fcis-16180	293	28	,	,	PUNCT
fcis-16180	293	29	2021	2021	NUM
fcis-16180	293	30	.	.	PUNCT
fcis-16180	294	1	[	[	X
fcis-16180	294	2	16	16	NUM
fcis-16180	294	3	]	]	X
fcis-16180	294	4	p.	p.	NOUN
fcis-16180	294	5	blanchard	blanchard	PROPN
fcis-16180	294	6	,	,	PUNCT
fcis-16180	294	7	e.	e.	PROPN
fcis-16180	294	8	m.	m.	PROPN
fcis-16180	294	9	el	el	PROPN
fcis-16180	294	10	mhamdi	mhamdi	PROPN
fcis-16180	294	11	,	,	PUNCT
fcis-16180	294	12	r.	r.	PROPN
fcis-16180	294	13	guerraoui	guerraoui	PROPN
fcis-16180	294	14	,	,	PUNCT
fcis-16180	294	15	and	and	CCONJ
fcis-16180	294	16	j.	j.	PROPN
fcis-16180	294	17	stainer	stainer	PROPN
fcis-16180	294	18	,	,	PUNCT
fcis-16180	294	19	“	"	PUNCT
fcis-16180	294	20	machine	machine	NOUN
fcis-16180	294	21	learning	learn	VERB
fcis-16180	294	22	with	with	ADP
fcis-16180	294	23	adversaries	adversary	NOUN
fcis-16180	294	24	:	:	PUNCT
fcis-16180	294	25	byzantine	byzantine	ADJ
fcis-16180	294	26	tolerant	tolerant	ADJ
fcis-16180	294	27	gradient	gradient	ADJ
fcis-16180	294	28	descent	descent	NOUN
fcis-16180	294	29	,	,	PUNCT
fcis-16180	294	30	”	"	PUNCT
fcis-16180	294	31	advances	advance	NOUN
fcis-16180	294	32	in	in	ADP
fcis-16180	294	33	neural	neural	ADJ
fcis-16180	294	34	information	information	NOUN
fcis-16180	294	35	processing	processing	NOUN
fcis-16180	294	36	systems	system	NOUN
fcis-16180	294	37	,	,	PUNCT
fcis-16180	294	38	vol	vol	NOUN
fcis-16180	294	39	.	.	PROPN
fcis-16180	294	40	30	30	NUM
fcis-16180	294	41	,	,	PUNCT
fcis-16180	294	42	2017	2017	NUM
fcis-16180	294	43	.	.	PUNCT
fcis-16180	295	1	[	[	X
fcis-16180	295	2	17	17	NUM
fcis-16180	295	3	]	]	X
fcis-16180	295	4	d.	d.	PROPN
fcis-16180	295	5	yin	yin	PROPN
fcis-16180	295	6	,	,	PUNCT
fcis-16180	295	7	y.	y.	PROPN
fcis-16180	295	8	chen	chen	PROPN
fcis-16180	295	9	,	,	PUNCT
fcis-16180	295	10	r.	r.	PROPN
fcis-16180	295	11	kannan	kannan	PROPN
fcis-16180	295	12	,	,	PUNCT
fcis-16180	295	13	and	and	CCONJ
fcis-16180	295	14	p.	p.	PROPN
fcis-16180	295	15	bartlett	bartlett	PROPN
fcis-16180	295	16	,	,	PUNCT
fcis-16180	295	17	“	"	PUNCT
fcis-16180	295	18	byzantine	byzantine	ADJ
fcis-16180	295	19	-	-	ADJ
fcis-16180	295	20	robust	robust	ADJ
fcis-16180	295	21	distributed	distributed	ADJ
fcis-16180	295	22	learning	learning	NOUN
fcis-16180	295	23	:	:	PUNCT
fcis-16180	295	24	towards	towards	ADP
fcis-16180	295	25	optimal	optimal	ADJ
fcis-16180	295	26	statistical	statistical	ADJ
fcis-16180	295	27	rates	rate	NOUN
fcis-16180	295	28	,	,	PUNCT
fcis-16180	295	29	”	"	PUNCT
fcis-16180	295	30	in	in	ADP
fcis-16180	295	31	international	international	ADJ
fcis-16180	295	32	conference	conference	NOUN
fcis-16180	295	33	on	on	ADP
fcis-16180	295	34	machine	machine	NOUN
fcis-16180	295	35	learning	learning	NOUN
fcis-16180	295	36	.	.	PUNCT
fcis-16180	296	1	pmlr	pmlr	NOUN
fcis-16180	296	2	,	,	PUNCT
fcis-16180	296	3	2018	2018	NUM
fcis-16180	296	4	,	,	PUNCT
fcis-16180	296	5	pp	pp	ADJ
fcis-16180	296	6	.	.	PUNCT
fcis-16180	296	7	5650–5659	5650–5659	NUM
fcis-16180	296	8	.	.	PUNCT
fcis-16180	297	1	[	[	X
fcis-16180	297	2	18	18	NUM
fcis-16180	297	3	]	]	PUNCT
fcis-16180	297	4	x.	x.	PROPN
fcis-16180	297	5	cao	cao	PROPN
fcis-16180	297	6	,	,	PUNCT
fcis-16180	297	7	m.	m.	NOUN
fcis-16180	297	8	fang	fang	PROPN
fcis-16180	297	9	,	,	PUNCT
fcis-16180	297	10	j.	j.	PROPN
fcis-16180	297	11	liu	liu	PROPN
fcis-16180	297	12	,	,	PUNCT
fcis-16180	297	13	and	and	CCONJ
fcis-16180	297	14	n.	n.	PROPN
fcis-16180	297	15	z.	z.	PROPN
fcis-16180	297	16	gong	gong	PROPN
fcis-16180	297	17	,	,	PUNCT
fcis-16180	297	18	“	"	PUNCT
fcis-16180	297	19	fltrust	fltrust	NOUN
fcis-16180	297	20	:	:	PUNCT
fcis-16180	297	21	byzantinerobust	byzantinerobust	ADJ
fcis-16180	297	22	federated	federated	ADJ
fcis-16180	297	23	learning	learning	NOUN
fcis-16180	297	24	via	via	ADP
fcis-16180	297	25	trust	trust	NOUN
fcis-16180	297	26	bootstrapping	bootstrapping	PROPN
fcis-16180	297	27	,	,	PUNCT
fcis-16180	297	28	”	"	PUNCT
fcis-16180	297	29	arxiv	arxiv	PROPN
fcis-16180	297	30	preprint	preprint	NOUN
fcis-16180	297	31	arxiv:2012.13995	arxiv:2012.13995	NOUN
fcis-16180	297	32	,	,	PUNCT
fcis-16180	297	33	2020	2020	NUM
fcis-16180	297	34	.	.	PUNCT
fcis-16180	298	1	[	[	X
fcis-16180	298	2	19	19	NUM
fcis-16180	298	3	]	]	PUNCT
fcis-16180	298	4	a.	a.	NOUN
fcis-16180	298	5	fu	fu	PROPN
fcis-16180	298	6	,	,	PUNCT
fcis-16180	298	7	x.	x.	PROPN
fcis-16180	298	8	zhang	zhang	PROPN
fcis-16180	298	9	,	,	PUNCT
fcis-16180	298	10	n.	n.	PROPN
fcis-16180	298	11	xiong	xiong	PROPN
fcis-16180	298	12	,	,	PUNCT
fcis-16180	298	13	y.	y.	PROPN
fcis-16180	298	14	gao	gao	PROPN
fcis-16180	298	15	,	,	PUNCT
fcis-16180	298	16	h.	h.	PROPN
fcis-16180	298	17	wang	wang	PROPN
fcis-16180	298	18	,	,	PUNCT
fcis-16180	298	19	and	and	CCONJ
fcis-16180	298	20	j.	j.	PROPN
fcis-16180	298	21	zhang	zhang	PROPN
fcis-16180	298	22	,	,	PUNCT
fcis-16180	298	23	“	"	PUNCT
fcis-16180	298	24	vfl	vfl	PROPN
fcis-16180	298	25	:	:	PUNCT
fcis-16180	298	26	a	a	DET
fcis-16180	298	27	verifiable	verifiable	ADJ
fcis-16180	298	28	federated	federated	ADJ
fcis-16180	298	29	learning	learning	NOUN
fcis-16180	298	30	with	with	ADP
fcis-16180	298	31	privacy	privacy	NOUN
fcis-16180	298	32	-	-	PUNCT
fcis-16180	298	33	preserving	preserve	VERB
fcis-16180	298	34	for	for	ADP
fcis-16180	298	35	big	big	ADJ
fcis-16180	298	36	data	datum	NOUN
fcis-16180	298	37	in	in	ADP
fcis-16180	298	38	industrial	industrial	ADJ
fcis-16180	298	39	iot	iot	NOUN
fcis-16180	298	40	,	,	PUNCT
fcis-16180	298	41	”	"	PUNCT
fcis-16180	298	42	ieee	ieee	NOUN
fcis-16180	298	43	transactions	transaction	NOUN
fcis-16180	298	44	on	on	ADP
fcis-16180	298	45	industrial	industrial	ADJ
fcis-16180	298	46	informatics	informatic	NOUN
fcis-16180	298	47	,	,	PUNCT
fcis-16180	298	48	vol	vol	NOUN
fcis-16180	298	49	.	.	PROPN
fcis-16180	298	50	18	18	NUM
fcis-16180	298	51	,	,	PUNCT
fcis-16180	298	52	no	no	INTJ
fcis-16180	298	53	.	.	NOUN
fcis-16180	298	54	5	5	NUM
fcis-16180	298	55	,	,	PUNCT
fcis-16180	298	56	pp	pp	ADJ
fcis-16180	298	57	.	.	PUNCT
fcis-16180	299	1	3316–3326	3316–3326	NUM
fcis-16180	299	2	,	,	PUNCT
fcis-16180	299	3	2020	2020	NUM
fcis-16180	299	4	.	.	PUNCT
fcis-16180	300	1	[	[	X
fcis-16180	300	2	20	20	NUM
fcis-16180	300	3	]	]	PUNCT
fcis-16180	300	4	h.	h.	PROPN
fcis-16180	300	5	gao	gao	PROPN
fcis-16180	300	6	,	,	PUNCT
fcis-16180	300	7	n.	n.	PROPN
fcis-16180	300	8	he	he	PRON
fcis-16180	300	9	,	,	PUNCT
fcis-16180	300	10	and	and	CCONJ
fcis-16180	300	11	t.	t.	PROPN
fcis-16180	300	12	gao	gao	PROPN
fcis-16180	300	13	,	,	PUNCT
fcis-16180	300	14	“	"	PUNCT
fcis-16180	300	15	sverifl	sverifl	NOUN
fcis-16180	300	16	:	:	PUNCT
fcis-16180	300	17	successive	successive	ADJ
fcis-16180	300	18	verifiable	verifiable	ADJ
fcis-16180	300	19	federated	federated	ADJ
fcis-16180	300	20	learning	learning	NOUN
fcis-16180	300	21	with	with	ADP
fcis-16180	300	22	privacy	privacy	NOUN
fcis-16180	300	23	-	-	PUNCT
fcis-16180	300	24	preserving	preserving	NOUN
fcis-16180	300	25	,	,	PUNCT
fcis-16180	300	26	”	"	PUNCT
fcis-16180	300	27	information	information	NOUN
fcis-16180	300	28	sciences	science	NOUN
fcis-16180	300	29	,	,	PUNCT
fcis-16180	300	30	vol	vol	NOUN
fcis-16180	300	31	.	.	PROPN
fcis-16180	300	32	622	622	NUM
fcis-16180	300	33	,	,	PUNCT
fcis-16180	300	34	pp	pp	ADJ
fcis-16180	300	35	.	.	PUNCT
fcis-16180	301	1	98–114	98–114	NUM
fcis-16180	301	2	,	,	PUNCT
fcis-16180	301	3	2023	2023	NUM
fcis-16180	301	4	.	.	PUNCT
fcis-16180	302	1	[	[	X
fcis-16180	302	2	21	21	NUM
fcis-16180	302	3	]	]	X
fcis-16180	302	4	r.	r.	PROPN
fcis-16180	302	5	xu	xu	PROPN
fcis-16180	302	6	,	,	PUNCT
fcis-16180	302	7	n.	n.	PROPN
fcis-16180	302	8	baracaldo	baracaldo	PROPN
fcis-16180	302	9	,	,	PUNCT
fcis-16180	302	10	y.	y.	PROPN
fcis-16180	302	11	zhou	zhou	PROPN
fcis-16180	302	12	,	,	PUNCT
fcis-16180	302	13	a.	a.	PROPN
fcis-16180	302	14	anwar	anwar	PROPN
fcis-16180	302	15	,	,	PUNCT
fcis-16180	302	16	and	and	CCONJ
fcis-16180	302	17	h.	h.	PROPN
fcis-16180	302	18	ludwig	ludwig	PROPN
fcis-16180	302	19	,	,	PUNCT
fcis-16180	302	20	“	"	PUNCT
fcis-16180	302	21	hybridalpha	hybridalpha	NOUN
fcis-16180	302	22	:	:	PUNCT
fcis-16180	302	23	an	an	DET
fcis-16180	302	24	efficient	efficient	ADJ
fcis-16180	302	25	approach	approach	NOUN
fcis-16180	302	26	for	for	ADP
fcis-16180	302	27	privacypreserving	privacypreserve	VERB
fcis-16180	302	28	federated	federated	ADJ
fcis-16180	302	29	learning	learning	NOUN
fcis-16180	302	30	,	,	PUNCT
fcis-16180	302	31	”	"	PUNCT
fcis-16180	302	32	in	in	ADP
fcis-16180	302	33	proceedings	proceeding	NOUN
fcis-16180	302	34	of	of	ADP
fcis-16180	302	35	the	the	DET
fcis-16180	302	36	12th	12th	ADJ
fcis-16180	302	37	acm	acm	NOUN
fcis-16180	302	38	workshop	workshop	NOUN
fcis-16180	302	39	on	on	ADP
fcis-16180	302	40	artificial	artificial	ADJ
fcis-16180	302	41	intelligence	intelligence	NOUN
fcis-16180	302	42	and	and	CCONJ
fcis-16180	302	43	security	security	NOUN
fcis-16180	302	44	,	,	PUNCT
fcis-16180	302	45	2019	2019	NUM
fcis-16180	302	46	,	,	PUNCT
fcis-16180	302	47	pp	pp	ADJ
fcis-16180	302	48	.	.	PUNCT
fcis-16180	303	1	13–23	13–23	X
fcis-16180	303	2	.	.	PUNCT
fcis-16180	304	1	[	[	X
fcis-16180	304	2	22	22	NUM
fcis-16180	304	3	]	]	PUNCT
fcis-16180	304	4	r.	r.	PROPN
fcis-16180	304	5	hu	hu	PROPN
fcis-16180	304	6	,	,	PUNCT
fcis-16180	304	7	y.	y.	PROPN
fcis-16180	304	8	guo	guo	PROPN
fcis-16180	304	9	,	,	PUNCT
fcis-16180	304	10	h.	h.	PROPN
fcis-16180	304	11	li	li	PROPN
fcis-16180	304	12	,	,	PUNCT
fcis-16180	304	13	q.	q.	PROPN
fcis-16180	304	14	pei	pei	PROPN
fcis-16180	304	15	,	,	PUNCT
fcis-16180	304	16	and	and	CCONJ
fcis-16180	304	17	y.	y.	PROPN
fcis-16180	304	18	gong	gong	PROPN
fcis-16180	304	19	,	,	PUNCT
fcis-16180	304	20	“	"	PUNCT
fcis-16180	304	21	personalized	personalize	VERB
fcis-16180	304	22	federated	federated	ADJ
fcis-16180	304	23	learning	learning	NOUN
fcis-16180	304	24	with	with	ADP
fcis-16180	304	25	differential	differential	ADJ
fcis-16180	304	26	privacy	privacy	NOUN
fcis-16180	304	27	,	,	PUNCT
fcis-16180	304	28	”	"	PUNCT
fcis-16180	304	29	ieee	ieee	NOUN
fcis-16180	304	30	internet	internet	NOUN
fcis-16180	304	31	of	of	ADP
fcis-16180	304	32	things	thing	NOUN
fcis-16180	304	33	journal	journal	NOUN
fcis-16180	304	34	,	,	PUNCT
fcis-16180	304	35	vol	vol	NOUN
fcis-16180	304	36	.	.	PROPN
fcis-16180	305	1	7	7	NUM
fcis-16180	305	2	,	,	PUNCT
fcis-16180	305	3	no	no	INTJ
fcis-16180	305	4	.	.	NOUN
fcis-16180	305	5	10	10	NUM
fcis-16180	305	6	,	,	PUNCT
fcis-16180	305	7	pp	pp	ADJ
fcis-16180	305	8	.	.	PUNCT
fcis-16180	306	1	9530–9539	9530–9539	NUM
fcis-16180	306	2	,	,	PUNCT
fcis-16180	306	3	2020	2020	NUM
fcis-16180	306	4	.	.	PUNCT
fcis-16180	307	1	[	[	X
fcis-16180	307	2	23	23	NUM
fcis-16180	307	3	]	]	PUNCT
fcis-16180	307	4	a.	a.	NOUN
fcis-16180	307	5	triastcyn	triastcyn	NOUN
fcis-16180	307	6	and	and	CCONJ
fcis-16180	307	7	b.	b.	NOUN
fcis-16180	307	8	faltings	falting	NOUN
fcis-16180	307	9	,	,	PUNCT
fcis-16180	307	10	“	"	PUNCT
fcis-16180	307	11	federated	federate	VERB
fcis-16180	307	12	learning	learning	NOUN
fcis-16180	307	13	with	with	ADP
fcis-16180	307	14	bayesian	bayesian	NOUN
fcis-16180	307	15	differential	differential	NOUN
fcis-16180	307	16	privacy	privacy	NOUN
fcis-16180	307	17	,	,	PUNCT
fcis-16180	307	18	”	"	PUNCT
fcis-16180	307	19	in	in	ADP
fcis-16180	307	20	2019	2019	NUM
fcis-16180	307	21	ieee	ieee	NOUN
fcis-16180	307	22	international	international	ADJ
fcis-16180	307	23	conference	conference	NOUN
fcis-16180	307	24	on	on	ADP
fcis-16180	307	25	big	big	ADJ
fcis-16180	307	26	data	datum	NOUN
fcis-16180	307	27	(	(	PUNCT
fcis-16180	307	28	big	big	ADJ
fcis-16180	307	29	data	datum	NOUN
fcis-16180	307	30	)	)	PUNCT
fcis-16180	307	31	.	.	PUNCT
fcis-16180	308	1	ieee	ieee	PROPN
fcis-16180	308	2	,	,	PUNCT
fcis-16180	308	3	2019	2019	NUM
fcis-16180	308	4	,	,	PUNCT
fcis-16180	308	5	pp	pp	ADJ
fcis-16180	308	6	.	.	PUNCT
fcis-16180	308	7	2587–2596	2587–2596	NOUN
fcis-16180	308	8	.	.	PUNCT
fcis-16180	309	1	[	[	X
fcis-16180	309	2	24	24	NUM
fcis-16180	309	3	]	]	X
fcis-16180	309	4	h.	h.	PROPN
fcis-16180	309	5	ye	ye	PROPN
fcis-16180	309	6	,	,	PUNCT
fcis-16180	309	7	j.	j.	PROPN
fcis-16180	309	8	liu	liu	PROPN
fcis-16180	309	9	,	,	PUNCT
fcis-16180	309	10	h.	h.	PROPN
fcis-16180	309	11	zhen	zhen	PROPN
fcis-16180	309	12	,	,	PUNCT
fcis-16180	309	13	w.	w.	PROPN
fcis-16180	309	14	jiang	jiang	PROPN
fcis-16180	309	15	,	,	PUNCT
fcis-16180	309	16	b.	b.	PROPN
fcis-16180	309	17	wang	wang	PROPN
fcis-16180	309	18	,	,	PUNCT
fcis-16180	309	19	and	and	CCONJ
fcis-16180	309	20	w.	w.	PROPN
fcis-16180	309	21	wang	wang	PROPN
fcis-16180	309	22	,	,	PUNCT
fcis-16180	309	23	“	"	PUNCT
fcis-16180	309	24	vrefl	vrefl	NOUN
fcis-16180	309	25	:	:	PUNCT
fcis-16180	309	26	verifiable	verifiable	ADJ
fcis-16180	309	27	and	and	CCONJ
fcis-16180	309	28	reconnection	reconnection	NOUN
fcis-16180	309	29	-	-	PUNCT
fcis-16180	309	30	efficient	efficient	ADJ
fcis-16180	309	31	federated	federated	ADJ
fcis-16180	309	32	learning	learning	NOUN
fcis-16180	309	33	in	in	ADP
fcis-16180	309	34	iot	iot	ADJ
fcis-16180	309	35	scenarios	scenario	NOUN
fcis-16180	309	36	,	,	PUNCT
fcis-16180	309	37	”	"	PUNCT
fcis-16180	309	38	journal	journal	NOUN
fcis-16180	309	39	of	of	ADP
fcis-16180	309	40	network	network	NOUN
fcis-16180	309	41	and	and	CCONJ
fcis-16180	309	42	computer	computer	NOUN
fcis-16180	309	43	applications	application	NOUN
fcis-16180	309	44	,	,	PUNCT
fcis-16180	309	45	vol	vol	NOUN
fcis-16180	309	46	.	.	PROPN
fcis-16180	309	47	207	207	NUM
fcis-16180	309	48	,	,	PUNCT
fcis-16180	309	49	p.	p.	NOUN
fcis-16180	309	50	103486	103486	NUM
fcis-16180	309	51	,	,	PUNCT
fcis-16180	309	52	2022	2022	NUM
fcis-16180	309	53	.	.	PUNCT
fcis-16180	310	1	[	[	X
fcis-16180	310	2	25	25	NUM
fcis-16180	310	3	]	]	PUNCT
fcis-16180	310	4	z.	z.	PROPN
fcis-16180	310	5	ma	ma	PROPN
fcis-16180	310	6	,	,	PUNCT
fcis-16180	310	7	j.	j.	PROPN
fcis-16180	310	8	ma	ma	PROPN
fcis-16180	310	9	,	,	PUNCT
fcis-16180	310	10	y.	y.	PROPN
fcis-16180	310	11	miao	miao	PROPN
fcis-16180	310	12	,	,	PUNCT
fcis-16180	310	13	y.	y.	PROPN
fcis-16180	310	14	li	li	PROPN
fcis-16180	310	15	,	,	PUNCT
fcis-16180	310	16	and	and	CCONJ
fcis-16180	310	17	r.	r.	PROPN
fcis-16180	310	18	h.	h.	PROPN
fcis-16180	310	19	deng	deng	PROPN
fcis-16180	310	20	,	,	PUNCT
fcis-16180	310	21	“	"	PUNCT
fcis-16180	310	22	shieldfl	shieldfl	NOUN
fcis-16180	310	23	:	:	PUNCT
fcis-16180	310	24	mitigating	mitigate	VERB
fcis-16180	310	25	model	model	NOUN
fcis-16180	310	26	poisoning	poisoning	NOUN
fcis-16180	310	27	attacks	attack	NOUN
fcis-16180	310	28	in	in	ADP
fcis-16180	310	29	privacy	privacy	NOUN
fcis-16180	310	30	-	-	PUNCT
fcis-16180	310	31	preserving	preserve	VERB
fcis-16180	310	32	federated	federated	ADJ
fcis-16180	310	33	learning	learning	NOUN
fcis-16180	310	34	,	,	PUNCT
fcis-16180	310	35	”	"	PUNCT
fcis-16180	310	36	ieee	ieee	NOUN
fcis-16180	310	37	transactions	transaction	NOUN
fcis-16180	310	38	on	on	ADP
fcis-16180	310	39	information	information	NOUN
fcis-16180	310	40	forensics	forensic	NOUN
fcis-16180	310	41	and	and	CCONJ
fcis-16180	310	42	security	security	NOUN
fcis-16180	310	43	,	,	PUNCT
fcis-16180	310	44	vol	vol	NOUN
fcis-16180	310	45	.	.	PROPN
fcis-16180	310	46	17	17	NUM
fcis-16180	310	47	,	,	PUNCT
fcis-16180	310	48	pp	pp	ADJ
fcis-16180	310	49	.	.	PUNCT
fcis-16180	311	1	1639–1654	1639–1654	NUM
fcis-16180	311	2	,	,	PUNCT
fcis-16180	311	3	2022	2022	NUM
fcis-16180	311	4	.	.	PUNCT
fcis-16180	312	1	[	[	X
fcis-16180	312	2	26	26	NUM
fcis-16180	312	3	]	]	X
fcis-16180	312	4	c.	c.	PROPN
fcis-16180	312	5	fang	fang	PROPN
fcis-16180	312	6	,	,	PUNCT
fcis-16180	312	7	y.	y.	PROPN
fcis-16180	312	8	guo	guo	PROPN
fcis-16180	312	9	,	,	PUNCT
fcis-16180	312	10	n.	n.	PROPN
fcis-16180	312	11	wang	wang	PROPN
fcis-16180	312	12	,	,	PUNCT
fcis-16180	312	13	and	and	CCONJ
fcis-16180	312	14	a.	a.	PROPN
fcis-16180	312	15	ju	ju	PROPN
fcis-16180	312	16	,	,	PUNCT
fcis-16180	312	17	“	"	PUNCT
fcis-16180	312	18	highly	highly	ADV
fcis-16180	312	19	efficient	efficient	ADJ
fcis-16180	312	20	federated	federated	ADJ
fcis-16180	312	21	learning	learning	NOUN
fcis-16180	312	22	with	with	ADP
fcis-16180	312	23	strong	strong	ADJ
fcis-16180	312	24	privacy	privacy	NOUN
fcis-16180	312	25	preservation	preservation	NOUN
fcis-16180	312	26	in	in	ADP
fcis-16180	312	27	cloud	cloud	NOUN
fcis-16180	312	28	computing	computing	NOUN
fcis-16180	312	29	,	,	PUNCT
fcis-16180	312	30	”	"	PUNCT
fcis-16180	312	31	computers	computer	NOUN
fcis-16180	312	32	&	&	CCONJ
fcis-16180	312	33	security	security	NOUN
fcis-16180	312	34	,	,	PUNCT
fcis-16180	312	35	vol	vol	NOUN
fcis-16180	312	36	.	.	PROPN
fcis-16180	312	37	96	96	NUM
fcis-16180	312	38	,	,	PUNCT
fcis-16180	312	39	p.	p.	NOUN
fcis-16180	312	40	101889	101889	NUM
fcis-16180	312	41	,	,	PUNCT
fcis-16180	312	42	2020	2020	NUM
fcis-16180	312	43	.	.	PUNCT
fcis-16180	313	1	[	[	X
fcis-16180	313	2	27	27	NUM
fcis-16180	313	3	]	]	X
fcis-16180	313	4	b.	b.	PROPN
fcis-16180	313	5	jia	jia	PROPN
fcis-16180	313	6	,	,	PUNCT
fcis-16180	313	7	x.	x.	PROPN
fcis-16180	313	8	zhang	zhang	PROPN
fcis-16180	313	9	,	,	PUNCT
fcis-16180	313	10	j.	j.	PROPN
fcis-16180	313	11	liu	liu	PROPN
fcis-16180	313	12	,	,	PUNCT
fcis-16180	313	13	y.	y.	PROPN
fcis-16180	313	14	zhang	zhang	PROPN
fcis-16180	313	15	,	,	PUNCT
fcis-16180	313	16	k.	k.	PROPN
fcis-16180	313	17	huang	huang	PROPN
fcis-16180	313	18	,	,	PUNCT
fcis-16180	313	19	and	and	CCONJ
fcis-16180	313	20	y.	y.	PROPN
fcis-16180	313	21	liang	liang	PROPN
fcis-16180	313	22	,	,	PUNCT
fcis-16180	313	23	“	"	PUNCT
fcis-16180	313	24	blockchain	blockchain	PROPN
fcis-16180	313	25	-	-	PUNCT
fcis-16180	313	26	enabled	enable	VERB
fcis-16180	313	27	federated	federated	ADJ
fcis-16180	313	28	learning	learning	NOUN
fcis-16180	313	29	data	data	PROPN
fcis-16180	313	30	protection	protection	NOUN
fcis-16180	313	31	aggregation	aggregation	NOUN
fcis-16180	313	32	scheme	scheme	NOUN
fcis-16180	313	33	with	with	ADP
fcis-16180	313	34	differential	differential	NOUN
fcis-16180	313	35	privacy	privacy	NOUN
fcis-16180	313	36	and	and	CCONJ
fcis-16180	313	37	homomorphic	homomorphic	ADJ
fcis-16180	313	38	encryption	encryption	NOUN
fcis-16180	313	39	in	in	ADP
fcis-16180	313	40	iiot	iiot	NOUN
fcis-16180	313	41	,	,	PUNCT
fcis-16180	313	42	”	"	PUNCT
fcis-16180	313	43	ieee	ieee	NOUN
fcis-16180	313	44	transactions	transaction	NOUN
fcis-16180	313	45	on	on	ADP
fcis-16180	313	46	industrial	industrial	ADJ
fcis-16180	313	47	informatics	informatic	NOUN
fcis-16180	313	48	,	,	PUNCT
fcis-16180	313	49	vol	vol	NOUN
fcis-16180	313	50	.	.	PROPN
fcis-16180	313	51	18	18	NUM
fcis-16180	313	52	,	,	PUNCT
fcis-16180	313	53	no	no	INTJ
fcis-16180	313	54	.	.	NOUN
fcis-16180	313	55	6	6	NUM
fcis-16180	313	56	,	,	PUNCT
fcis-16180	313	57	pp	pp	ADJ
fcis-16180	313	58	.	.	PUNCT
fcis-16180	313	59	4049–4058	4049–4058	NUM
fcis-16180	313	60	,	,	PUNCT
fcis-16180	313	61	2021	2021	NUM
fcis-16180	313	62	.	.	PUNCT
fcis-16180	314	1	[	[	X
fcis-16180	314	2	28	28	NUM
fcis-16180	314	3	]	]	X
fcis-16180	314	4	h.	h.	PROPN
fcis-16180	314	5	zhou	zhou	PROPN
fcis-16180	314	6	,	,	PUNCT
fcis-16180	314	7	g.	g.	PROPN
fcis-16180	314	8	yang	yang	PROPN
fcis-16180	314	9	,	,	PUNCT
fcis-16180	314	10	h.	h.	PROPN
fcis-16180	314	11	dai	dai	PROPN
fcis-16180	314	12	,	,	PUNCT
fcis-16180	314	13	and	and	CCONJ
fcis-16180	314	14	g.	g.	PROPN
fcis-16180	314	15	liu	liu	PROPN
fcis-16180	314	16	,	,	PUNCT
fcis-16180	314	17	“	"	PUNCT
fcis-16180	314	18	pflf	pflf	ADJ
fcis-16180	314	19	:	:	PUNCT
fcis-16180	314	20	privacypreserving	privacypreserve	VERB
fcis-16180	314	21	federated	federated	ADJ
fcis-16180	314	22	learning	learning	NOUN
fcis-16180	314	23	framework	framework	NOUN
fcis-16180	314	24	for	for	ADP
fcis-16180	314	25	edge	edge	NOUN
fcis-16180	314	26	computing	computing	NOUN
fcis-16180	314	27	,	,	PUNCT
fcis-16180	314	28	”	"	PUNCT
fcis-16180	314	29	ieee	ieee	NOUN
fcis-16180	314	30	transactions	transaction	NOUN
fcis-16180	314	31	on	on	ADP
fcis-16180	314	32	information	information	NOUN
fcis-16180	314	33	forensics	forensic	NOUN
fcis-16180	314	34	and	and	CCONJ
fcis-16180	314	35	security	security	NOUN
fcis-16180	314	36	,	,	PUNCT
fcis-16180	314	37	vol	vol	NOUN
fcis-16180	314	38	.	.	PROPN
fcis-16180	314	39	17	17	NUM
fcis-16180	314	40	,	,	PUNCT
fcis-16180	314	41	pp	pp	ADJ
fcis-16180	314	42	.	.	PUNCT
fcis-16180	314	43	1905–1918	1905–1918	NUM
fcis-16180	314	44	,	,	PUNCT
fcis-16180	314	45	2022	2022	NUM
fcis-16180	314	46	.	.	PUNCT
fcis-16180	315	1	[	[	X
fcis-16180	315	2	29	29	NUM
fcis-16180	315	3	]	]	X
fcis-16180	315	4	v.	v.	CCONJ
fcis-16180	315	5	mugunthan	mugunthan	PROPN
fcis-16180	315	6	,	,	PUNCT
fcis-16180	315	7	a.	a.	NOUN
fcis-16180	315	8	polychroniadou	polychroniadou	NOUN
fcis-16180	315	9	,	,	PUNCT
fcis-16180	315	10	d.	d.	PROPN
fcis-16180	315	11	byrd	byrd	PROPN
fcis-16180	315	12	,	,	PUNCT
fcis-16180	315	13	and	and	CCONJ
fcis-16180	315	14	t.	t.	PROPN
fcis-16180	315	15	h.	h.	PROPN
fcis-16180	315	16	balch	balch	PROPN
fcis-16180	315	17	,	,	PUNCT
fcis-16180	315	18	“	"	PUNCT
fcis-16180	315	19	smpai	smpai	NOUN
fcis-16180	315	20	:	:	PUNCT
fcis-16180	315	21	secure	secure	VERB
fcis-16180	315	22	multi	multi	ADJ
fcis-16180	315	23	-	-	ADJ
fcis-16180	315	24	party	party	ADJ
fcis-16180	315	25	computation	computation	NOUN
fcis-16180	315	26	for	for	ADP
fcis-16180	315	27	federated	federated	ADJ
fcis-16180	315	28	learning	learning	NOUN
fcis-16180	315	29	,	,	PUNCT
fcis-16180	315	30	”	"	PUNCT
fcis-16180	315	31	in	in	ADP
fcis-16180	315	32	proceedings	proceeding	NOUN
fcis-16180	315	33	of	of	ADP
fcis-16180	315	34	the	the	DET
fcis-16180	315	35	neurips	neurip	NOUN
fcis-16180	315	36	2019	2019	NUM
fcis-16180	315	37	workshop	workshop	NOUN
fcis-16180	315	38	on	on	ADP
fcis-16180	315	39	robust	robust	ADJ
fcis-16180	315	40	ai	ai	NOUN
fcis-16180	315	41	in	in	ADP
fcis-16180	315	42	financial	financial	ADJ
fcis-16180	315	43	services	service	NOUN
fcis-16180	315	44	.	.	PUNCT
fcis-16180	316	1	mit	mit	PROPN
fcis-16180	316	2	press	press	PROPN
fcis-16180	316	3	cambridge	cambridge	PROPN
fcis-16180	316	4	,	,	PUNCT
fcis-16180	316	5	ma	ma	PROPN
fcis-16180	316	6	,	,	PUNCT
fcis-16180	316	7	usa	usa	PROPN
fcis-16180	316	8	,	,	PUNCT
fcis-16180	316	9	2019	2019	NUM
fcis-16180	316	10	,	,	PUNCT
fcis-16180	316	11	pp	pp	ADJ
fcis-16180	316	12	.	.	PUNCT
fcis-16180	317	1	1–9	1–9	NOUN
fcis-16180	317	2	.	.	PUNCT
fcis-16180	318	1	[	[	X
fcis-16180	318	2	30	30	NUM
fcis-16180	318	3	]	]	X
fcis-16180	318	4	c.	c.	PROPN
fcis-16180	318	5	zhou	zhou	PROPN
fcis-16180	318	6	,	,	PUNCT
fcis-16180	318	7	a.	a.	NOUN
fcis-16180	318	8	fu	fu	PROPN
fcis-16180	318	9	,	,	PUNCT
fcis-16180	318	10	s.	s.	PROPN
fcis-16180	318	11	yu	yu	PROPN
fcis-16180	318	12	,	,	PUNCT
fcis-16180	318	13	w.	w.	PROPN
fcis-16180	318	14	yang	yang	PROPN
fcis-16180	318	15	,	,	PUNCT
fcis-16180	318	16	h.	h.	PROPN
fcis-16180	318	17	wang	wang	PROPN
fcis-16180	318	18	,	,	PUNCT
fcis-16180	318	19	and	and	CCONJ
fcis-16180	318	20	y.	y.	PROPN
fcis-16180	318	21	zhang	zhang	PROPN
fcis-16180	318	22	,	,	PUNCT
fcis-16180	318	23	“	"	PUNCT
fcis-16180	318	24	privacy	privacy	NOUN
fcis-16180	318	25	-	-	PUNCT
fcis-16180	318	26	preserving	preserve	VERB
fcis-16180	318	27	federated	federated	ADJ
fcis-16180	318	28	learning	learning	NOUN
fcis-16180	318	29	in	in	ADP
fcis-16180	318	30	fog	fog	PROPN
fcis-16180	318	31	computing	computing	NOUN
fcis-16180	318	32	,	,	PUNCT
fcis-16180	318	33	”	"	PUNCT
fcis-16180	318	34	ieee	ieee	NOUN
fcis-16180	318	35	internet	internet	NOUN
fcis-16180	318	36	of	of	ADP
fcis-16180	318	37	things	thing	NOUN
fcis-16180	318	38	journal	journal	NOUN
fcis-16180	318	39	,	,	PUNCT
fcis-16180	318	40	vol	vol	NOUN
fcis-16180	318	41	.	.	PROPN
fcis-16180	318	42	7	7	NUM
fcis-16180	318	43	,	,	PUNCT
fcis-16180	318	44	no	no	INTJ
fcis-16180	318	45	.	.	NOUN
fcis-16180	318	46	11	11	NUM
fcis-16180	318	47	,	,	PUNCT
fcis-16180	318	48	pp	pp	ADJ
fcis-16180	318	49	.	.	PUNCT
fcis-16180	319	1	10	10	NUM
fcis-16180	319	2	782–10	782–10	NOUN
fcis-16180	319	3	793	793	NUM
fcis-16180	319	4	,	,	PUNCT
fcis-16180	319	5	2020	2020	NUM
fcis-16180	319	6	.	.	PUNCT
fcis-16180	320	1	[	[	X
fcis-16180	320	2	31	31	NUM
fcis-16180	320	3	]	]	PUNCT
fcis-16180	320	4	l.	l.	PROPN
fcis-16180	320	5	zhao	zhao	PROPN
fcis-16180	320	6	,	,	PUNCT
fcis-16180	320	7	j.	j.	PROPN
fcis-16180	320	8	jiang	jiang	PROPN
fcis-16180	320	9	,	,	PUNCT
fcis-16180	320	10	b.	b.	PROPN
fcis-16180	320	11	feng	feng	PROPN
fcis-16180	320	12	,	,	PUNCT
fcis-16180	320	13	q.	q.	PROPN
fcis-16180	320	14	wang	wang	PROPN
fcis-16180	320	15	,	,	PUNCT
fcis-16180	320	16	c.	c.	PROPN
fcis-16180	320	17	shen	shen	PROPN
fcis-16180	320	18	,	,	PUNCT
fcis-16180	320	19	and	and	CCONJ
fcis-16180	320	20	q.	q.	PROPN
fcis-16180	320	21	li	li	PROPN
fcis-16180	320	22	,	,	PUNCT
fcis-16180	320	23	“	"	PUNCT
fcis-16180	320	24	sear	sear	NOUN
fcis-16180	320	25	:	:	PUNCT
fcis-16180	320	26	secure	secure	ADJ
fcis-16180	320	27	and	and	CCONJ
fcis-16180	320	28	efficient	efficient	ADJ
fcis-16180	320	29	aggregation	aggregation	NOUN
fcis-16180	320	30	for	for	ADP
fcis-16180	320	31	byzantine	byzantine	ADJ
fcis-16180	320	32	-	-	PUNCT
fcis-16180	320	33	robust	robust	ADJ
fcis-16180	320	34	federated	federated	ADJ
fcis-16180	320	35	learning	learning	NOUN
fcis-16180	320	36	,	,	PUNCT
fcis-16180	320	37	”	"	PUNCT
fcis-16180	320	38	ieee	ieee	NOUN
fcis-16180	320	39	transactions	transaction	NOUN
fcis-16180	320	40	on	on	ADP
fcis-16180	320	41	dependable	dependable	ADJ
fcis-16180	320	42	and	and	CCONJ
fcis-16180	320	43	secure	secure	ADJ
fcis-16180	320	44	computing	computing	NOUN
fcis-16180	320	45	,	,	PUNCT
fcis-16180	320	46	vol	vol	NOUN
fcis-16180	320	47	.	.	PROPN
fcis-16180	320	48	19	19	NUM
fcis-16180	320	49	,	,	PUNCT
fcis-16180	320	50	no	no	INTJ
fcis-16180	320	51	.	.	NOUN
fcis-16180	320	52	5	5	NUM
fcis-16180	320	53	,	,	PUNCT
fcis-16180	320	54	pp	pp	ADJ
fcis-16180	320	55	.	.	PUNCT
fcis-16180	320	56	3329–3342	3329–3342	NUM
fcis-16180	320	57	,	,	PUNCT
fcis-16180	320	58	2021.n	2021.n	NUM
fcis-16180	320	59	.	.	PUNCT
fcis-16180	320	60	kawasaki	kawasaki	PROPN
fcis-16180	320	61	,	,	PUNCT
fcis-16180	320	62	“	"	PUNCT
fcis-16180	320	63	parametric	parametric	ADJ
fcis-16180	320	64	study	study	NOUN
fcis-16180	320	65	of	of	ADP
fcis-16180	320	66	thermal	thermal	ADJ
fcis-16180	320	67	and	and	CCONJ
fcis-16180	320	68	chemical	chemical	NOUN
fcis-16180	320	69	nonequilibrium	nonequilibrium	NOUN
fcis-16180	320	70	nozzle	nozzle	NOUN
fcis-16180	320	71	flow	flow	NOUN
fcis-16180	320	72	,	,	PUNCT
fcis-16180	320	73	”	"	PUNCT
fcis-16180	320	74	m.s	m.s	PROPN
fcis-16180	320	75	.	.	PROPN
fcis-16180	320	76	thesis	thesis	PROPN
fcis-16180	320	77	,	,	PUNCT
fcis-16180	320	78	dept	dept	NOUN
fcis-16180	320	79	.	.	PROPN
fcis-16180	320	80	electron	electron	PROPN
fcis-16180	320	81	.	.	PUNCT
fcis-16180	321	1	eng	eng	PROPN
fcis-16180	321	2	.	.	PROPN
fcis-16180	321	3	,	,	PUNCT
fcis-16180	321	4	osaka	osaka	PROPN
fcis-16180	321	5	univ	univ	PROPN
fcis-16180	321	6	.	.	PROPN
fcis-16180	321	7	,	,	PUNCT
fcis-16180	321	8	osaka	osaka	PROPN
fcis-16180	321	9	,	,	PUNCT
fcis-16180	321	10	japan	japan	PROPN
fcis-16180	321	11	,	,	PUNCT
fcis-16180	321	12	1993	1993	NUM
fcis-16180	321	13	.	.	PUNCT
fcis-16180	322	1	[	[	X
fcis-16180	322	2	32	32	NUM
fcis-16180	322	3	]	]	PUNCT
fcis-16180	322	4	z.	z.	PROPN
fcis-16180	322	5	zhang	zhang	PROPN
fcis-16180	322	6	,	,	PUNCT
fcis-16180	322	7	l.	l.	PROPN
fcis-16180	322	8	wu	wu	PROPN
fcis-16180	322	9	,	,	PUNCT
fcis-16180	322	10	c.	c.	PROPN
fcis-16180	322	11	ma	ma	PROPN
fcis-16180	322	12	,	,	PUNCT
fcis-16180	322	13	j.	j.	PROPN
fcis-16180	322	14	li	li	PROPN
fcis-16180	322	15	,	,	PUNCT
fcis-16180	322	16	j.	j.	PROPN
fcis-16180	322	17	wang	wang	PROPN
fcis-16180	322	18	,	,	PUNCT
fcis-16180	322	19	q.	q.	PROPN
fcis-16180	322	20	wang	wang	PROPN
fcis-16180	322	21	,	,	PUNCT
fcis-16180	322	22	and	and	CCONJ
fcis-16180	322	23	s.	s.	PROPN
fcis-16180	322	24	yu	yu	PROPN
fcis-16180	322	25	,	,	PUNCT
fcis-16180	322	26	“	"	PUNCT
fcis-16180	322	27	lsfl	lsfl	NOUN
fcis-16180	322	28	:	:	PUNCT
fcis-16180	322	29	a	a	DET
fcis-16180	322	30	lightweight	lightweight	ADJ
fcis-16180	322	31	and	and	CCONJ
fcis-16180	322	32	secure	secure	VERB
fcis-16180	322	33	federated	federated	ADJ
fcis-16180	322	34	learning	learning	NOUN
fcis-16180	322	35	scheme	scheme	NOUN
fcis-16180	322	36	for	for	ADP
fcis-16180	322	37	edge	edge	NOUN
fcis-16180	322	38	computing	computing	NOUN
fcis-16180	322	39	,	,	PUNCT
fcis-16180	322	40	”	"	PUNCT
fcis-16180	322	41	ieee	ieee	NOUN
fcis-16180	322	42	transactions	transaction	NOUN
fcis-16180	322	43	on	on	ADP
fcis-16180	322	44	information	information	NOUN
fcis-16180	322	45	forensics	forensic	NOUN
fcis-16180	322	46	and	and	CCONJ
fcis-16180	322	47	security	security	NOUN
fcis-16180	322	48	,	,	PUNCT
fcis-16180	322	49	vol	vol	NOUN
fcis-16180	322	50	.	.	PROPN
fcis-16180	322	51	18	18	NUM
fcis-16180	322	52	,	,	PUNCT
fcis-16180	322	53	pp	pp	ADJ
fcis-16180	322	54	.	.	PUNCT
fcis-16180	323	1	365	365	NUM
fcis-16180	323	2	–	–	PUNCT
fcis-16180	323	3	379	379	NUM
fcis-16180	323	4	,	,	PUNCT
fcis-16180	323	5	2022	2022	NUM
fcis-16180	323	6	.	.	PUNCT
fcis-16180	324	1	[	[	X
fcis-16180	324	2	33	33	NUM
fcis-16180	324	3	]	]	PUNCT
fcis-16180	324	4	b.	b.	PROPN
fcis-16180	324	5	zhao	zhao	PROPN
fcis-16180	324	6	,	,	PUNCT
fcis-16180	324	7	p.	p.	PROPN
fcis-16180	324	8	sun	sun	PROPN
fcis-16180	324	9	,	,	PUNCT
fcis-16180	324	10	l.	l.	PROPN
fcis-16180	324	11	fang	fang	PROPN
fcis-16180	324	12	,	,	PUNCT
fcis-16180	324	13	t.	t.	PROPN
fcis-16180	324	14	wang	wang	PROPN
fcis-16180	324	15	,	,	PUNCT
fcis-16180	324	16	and	and	CCONJ
fcis-16180	324	17	k.	k.	PROPN
fcis-16180	324	18	jiang	jiang	PROPN
fcis-16180	324	19	,	,	PUNCT
fcis-16180	324	20	“	"	PUNCT
fcis-16180	324	21	fedcom	fedcom	PROPN
fcis-16180	324	22	:	:	PUNCT
fcis-16180	324	23	a	a	DET
fcis-16180	324	24	byzantine	byzantine	ADJ
fcis-16180	324	25	-	-	PUNCT
fcis-16180	324	26	robust	robust	ADJ
fcis-16180	324	27	local	local	ADJ
fcis-16180	324	28	model	model	NOUN
fcis-16180	324	29	aggregation	aggregation	NOUN
fcis-16180	324	30	rule	rule	NOUN
fcis-16180	324	31	10	10	NUM
fcis-16180	324	32	using	use	VERB
fcis-16180	324	33	data	datum	NOUN
fcis-16180	324	34	commitment	commitment	NOUN
fcis-16180	324	35	for	for	ADP
fcis-16180	324	36	federated	federated	ADJ
fcis-16180	324	37	learning	learning	NOUN
fcis-16180	324	38	,	,	PUNCT
fcis-16180	324	39	”	"	PUNCT
fcis-16180	324	40	arxiv	arxiv	PROPN
fcis-16180	324	41	preprint	preprint	NOUN
fcis-16180	324	42	arxiv:2104.08020	arxiv:2104.08020	NOUN
fcis-16180	324	43	,	,	PUNCT
fcis-16180	324	44	2021	2021	NUM
fcis-16180	324	45	.	.	PUNCT
fcis-16180	325	1	[	[	X
fcis-16180	325	2	34	34	NUM
fcis-16180	325	3	]	]	X
fcis-16180	325	4	h.	h.	PROPN
fcis-16180	325	5	guo	guo	PROPN
fcis-16180	325	6	,	,	PUNCT
fcis-16180	325	7	h.	h.	PROPN
fcis-16180	325	8	wang	wang	PROPN
fcis-16180	325	9	,	,	PUNCT
fcis-16180	325	10	t.	t.	NOUN
fcis-16180	325	11	song	song	NOUN
fcis-16180	325	12	,	,	PUNCT
fcis-16180	325	13	y.	y.	PROPN
fcis-16180	325	14	hua	hua	PROPN
fcis-16180	325	15	,	,	PUNCT
fcis-16180	325	16	z.	z.	PROPN
fcis-16180	325	17	lv	lv	PROPN
fcis-16180	325	18	,	,	PUNCT
fcis-16180	325	19	x.	x.	PROPN
fcis-16180	325	20	jin	jin	PROPN
fcis-16180	325	21	,	,	PUNCT
fcis-16180	325	22	z.	z.	PROPN
fcis-16180	325	23	xue	xue	PROPN
fcis-16180	325	24	,	,	PUNCT
fcis-16180	325	25	r.	r.	PROPN
fcis-16180	325	26	ma	ma	PROPN
fcis-16180	325	27	,	,	PUNCT
fcis-16180	325	28	and	and	CCONJ
fcis-16180	325	29	h.	h.	PROPN
fcis-16180	325	30	guan	guan	PROPN
fcis-16180	325	31	,	,	PUNCT
fcis-16180	325	32	“	"	PUNCT
fcis-16180	325	33	siren	siren	ADJ
fcis-16180	325	34	:	:	PUNCT
fcis-16180	325	35	byzantine	byzantine	ADJ
fcis-16180	325	36	-	-	PUNCT
fcis-16180	325	37	robust	robust	ADJ
fcis-16180	325	38	federated	federated	ADJ
fcis-16180	325	39	learning	learning	NOUN
fcis-16180	325	40	via	via	ADP
fcis-16180	325	41	proactive	proactive	ADJ
fcis-16180	325	42	alarming	alarming	ADJ
fcis-16180	325	43	,	,	PUNCT
fcis-16180	325	44	”	"	PUNCT
fcis-16180	325	45	in	in	ADP
fcis-16180	325	46	proceedings	proceeding	NOUN
fcis-16180	325	47	of	of	ADP
fcis-16180	325	48	the	the	DET
fcis-16180	325	49	acm	acm	NOUN
fcis-16180	325	50	symposium	symposium	NOUN
fcis-16180	325	51	on	on	ADP
fcis-16180	325	52	cloud	cloud	NOUN
fcis-16180	325	53	computing	computing	NOUN
fcis-16180	325	54	,	,	PUNCT
fcis-16180	325	55	2021	2021	NUM
fcis-16180	325	56	,	,	PUNCT
fcis-16180	325	57	pp	pp	ADV
fcis-16180	325	58	.	.	PUNCT
fcis-16180	326	1	47–60	47–60	NOUN
fcis-16180	326	2	.	.	PUNCT
fcis-16180	327	1	24	24	NUM
fcis-16180	328	1	[	[	SYM
fcis-16180	328	2	35	35	NUM
fcis-16180	328	3	]	]	X
fcis-16180	328	4	l.	l.	PROPN
fcis-16180	328	5	munoz	munoz	PROPN
fcis-16180	328	6	-	-	PUNCT
fcis-16180	328	7	gonz	gonz	PROPN
fcis-16180	328	8	˜	˜	PROPN
fcis-16180	328	9	alez	alez	PROPN
fcis-16180	328	10	,	,	PUNCT
fcis-16180	328	11	k.	k.	PROPN
fcis-16180	328	12	t.	t.	PROPN
fcis-16180	328	13	co	co	PROPN
fcis-16180	328	14	,	,	PUNCT
fcis-16180	328	15	and	and	CCONJ
fcis-16180	328	16	e.	e.	PROPN
fcis-16180	328	17	c.	c.	PROPN
fcis-16180	328	18	lupu	lupu	PROPN
fcis-16180	328	19	,	,	PUNCT
fcis-16180	328	20	´	´	NOUN
fcis-16180	328	21	“	"	PUNCT
fcis-16180	328	22	byzantinerobust	byzantinerobust	VERB
fcis-16180	328	23	federated	federate	VERB
fcis-16180	328	24	machine	machine	NOUN
fcis-16180	328	25	learning	learn	VERB
fcis-16180	328	26	through	through	ADP
fcis-16180	328	27	adaptive	adaptive	ADJ
fcis-16180	328	28	model	model	NOUN
fcis-16180	328	29	averaging	averaging	NOUN
fcis-16180	328	30	,	,	PUNCT
fcis-16180	328	31	”	"	PUNCT
fcis-16180	328	32	arxiv	arxiv	PROPN
fcis-16180	328	33	preprint	preprint	PROPN
fcis-16180	328	34	arxiv:1909.05125	arxiv:1909.05125	NUM
fcis-16180	328	35	,	,	PUNCT
fcis-16180	328	36	2019.e	2019.e	PROPN
fcis-16180	328	37	.	.	PUNCT
fcis-16180	328	38	e.	e.	PROPN
fcis-16180	328	39	reber	reber	PROPN
fcis-16180	328	40	,	,	PUNCT
fcis-16180	328	41	r.	r.	PROPN
fcis-16180	328	42	l.	l.	PROPN
fcis-16180	328	43	michell	michell	PROPN
fcis-16180	328	44	,	,	PUNCT
fcis-16180	328	45	and	and	CCONJ
fcis-16180	328	46	c.	c.	PROPN
fcis-16180	328	47	j.	j.	PROPN
fcis-16180	328	48	carter	carter	PROPN
fcis-16180	328	49	,	,	PUNCT
fcis-16180	328	50	“	"	PUNCT
fcis-16180	328	51	oxygen	oxygen	NOUN
fcis-16180	328	52	absorption	absorption	NOUN
fcis-16180	328	53	in	in	ADP
fcis-16180	328	54	the	the	DET
fcis-16180	328	55	earth	earth	NOUN
fcis-16180	328	56	’s	’s	PART
fcis-16180	328	57	atmosphere	atmosphere	NOUN
fcis-16180	328	58	,	,	PUNCT
fcis-16180	328	59	”	"	PUNCT
fcis-16180	328	60	aerospace	aerospace	PROPN
fcis-16180	328	61	corp	corp	PROPN
fcis-16180	328	62	.	.	PROPN
fcis-16180	328	63	,	,	PUNCT
fcis-16180	328	64	los	los	PROPN
fcis-16180	328	65	angeles	angeles	PROPN
fcis-16180	328	66	,	,	PUNCT
fcis-16180	328	67	ca	can	AUX
fcis-16180	328	68	,	,	PUNCT
fcis-16180	328	69	tech	tech	NOUN
fcis-16180	328	70	.	.	PUNCT
fcis-16180	329	1	rep	rep	PROPN
fcis-16180	329	2	.	.	PROPN
fcis-16180	329	3	tr-0200	tr-0200	PROPN
fcis-16180	329	4	(	(	PUNCT
fcis-16180	329	5	420	420	NUM
fcis-16180	329	6	-	-	PUNCT
fcis-16180	329	7	46)-3	46)-3	NUM
fcis-16180	329	8	,	,	PUNCT
fcis-16180	329	9	nov	nov	PROPN
fcis-16180	329	10	.	.	PROPN
fcis-16180	329	11	1988	1988	NUM
fcis-16180	329	12	.	.	PUNCT
fcis-16180	330	1	[	[	X
fcis-16180	330	2	36	36	NUM
fcis-16180	330	3	]	]	X
fcis-16180	330	4	r.	r.	PROPN
fcis-16180	330	5	wang	wang	PROPN
fcis-16180	330	6	,	,	PUNCT
fcis-16180	330	7	x.	x.	PROPN
fcis-16180	330	8	wang	wang	PROPN
fcis-16180	330	9	,	,	PUNCT
fcis-16180	330	10	h.	h.	PROPN
fcis-16180	330	11	chen	chen	PROPN
fcis-16180	330	12	,	,	PUNCT
fcis-16180	330	13	s.	s.	PROPN
fcis-16180	330	14	picek	picek	PROPN
fcis-16180	330	15	,	,	PUNCT
fcis-16180	330	16	z.	z.	PROPN
fcis-16180	330	17	liu	liu	PROPN
fcis-16180	330	18	,	,	PUNCT
fcis-16180	330	19	and	and	CCONJ
fcis-16180	330	20	k.	k.	PROPN
fcis-16180	330	21	liang	liang	PROPN
fcis-16180	330	22	,	,	PUNCT
fcis-16180	330	23	“	"	PUNCT
fcis-16180	330	24	brief	brief	ADJ
fcis-16180	330	25	but	but	CCONJ
fcis-16180	330	26	powerful	powerful	ADJ
fcis-16180	330	27	:	:	PUNCT
fcis-16180	330	28	byzantine	byzantine	ADJ
fcis-16180	330	29	-	-	PUNCT
fcis-16180	330	30	robust	robust	ADJ
fcis-16180	330	31	and	and	CCONJ
fcis-16180	330	32	privacy	privacy	NOUN
fcis-16180	330	33	-	-	PUNCT
fcis-16180	330	34	preserving	preserve	VERB
fcis-16180	330	35	federated	federated	ADJ
fcis-16180	330	36	learning	learning	NOUN
fcis-16180	330	37	via	via	ADP
fcis-16180	330	38	model	model	NOUN
fcis-16180	330	39	segmentation	segmentation	NOUN
fcis-16180	330	40	and	and	CCONJ
fcis-16180	330	41	secure	secure	ADJ
fcis-16180	330	42	clustering	clustering	NOUN
fcis-16180	330	43	,	,	PUNCT
fcis-16180	330	44	”	"	PUNCT
fcis-16180	330	45	arxiv	arxiv	PROPN
fcis-16180	330	46	preprint	preprint	NOUN
fcis-16180	330	47	arxiv:2208.10161	arxiv:2208.10161	PROPN
fcis-16180	330	48	,	,	PUNCT
fcis-16180	330	49	2022	2022	NUM
fcis-16180	330	50	.	.	PUNCT
fcis-16180	331	1	[	[	X
fcis-16180	331	2	37	37	NUM
fcis-16180	331	3	]	]	PUNCT
fcis-16180	331	4	s.	s.	PROPN
fcis-16180	331	5	li	li	PROPN
fcis-16180	331	6	,	,	PUNCT
fcis-16180	331	7	e.	e.	PROPN
fcis-16180	331	8	ngai	ngai	PROPN
fcis-16180	331	9	,	,	PUNCT
fcis-16180	331	10	and	and	CCONJ
fcis-16180	331	11	t.	t.	PROPN
fcis-16180	331	12	voigt	voigt	PROPN
fcis-16180	331	13	,	,	PUNCT
fcis-16180	331	14	“	"	PUNCT
fcis-16180	331	15	byzantine	byzantine	ADJ
fcis-16180	331	16	-	-	ADJ
fcis-16180	331	17	robust	robust	ADJ
fcis-16180	331	18	aggregation	aggregation	NOUN
fcis-16180	331	19	in	in	ADP
fcis-16180	331	20	federated	federated	ADJ
fcis-16180	331	21	learning	learning	NOUN
fcis-16180	331	22	empowered	empower	VERB
fcis-16180	331	23	industrial	industrial	ADJ
fcis-16180	331	24	iot	iot	NOUN
fcis-16180	331	25	,	,	PUNCT
fcis-16180	331	26	”	"	PUNCT
fcis-16180	331	27	ieee	ieee	NOUN
fcis-16180	331	28	transactions	transaction	NOUN
fcis-16180	331	29	on	on	ADP
fcis-16180	331	30	industrial	industrial	ADJ
fcis-16180	331	31	informatics	informatic	NOUN
fcis-16180	331	32	,	,	PUNCT
fcis-16180	331	33	vol	vol	NOUN
fcis-16180	331	34	.	.	PROPN
fcis-16180	331	35	19	19	NUM
fcis-16180	331	36	,	,	PUNCT
fcis-16180	331	37	no	no	INTJ
fcis-16180	331	38	.	.	NOUN
fcis-16180	331	39	2	2	NUM
fcis-16180	331	40	,	,	PUNCT
fcis-16180	331	41	pp	pp	ADJ
fcis-16180	331	42	.	.	PUNCT
fcis-16180	332	1	1165	1165	NUM
fcis-16180	332	2	–	–	PUNCT
fcis-16180	332	3	1175	1175	NUM
fcis-16180	332	4	,	,	PUNCT
fcis-16180	332	5	2021	2021	NUM
fcis-16180	332	6	.	.	PUNCT
fcis-16180	333	1	[	[	X
fcis-16180	333	2	38	38	NUM
fcis-16180	333	3	]	]	PUNCT
fcis-16180	333	4	x.	x.	NOUN
fcis-16180	333	5	tang	tang	PROPN
fcis-16180	333	6	,	,	PUNCT
fcis-16180	333	7	m.	m.	NOUN
fcis-16180	333	8	shen	shen	PROPN
fcis-16180	333	9	,	,	PUNCT
fcis-16180	333	10	q.	q.	PROPN
fcis-16180	333	11	li	li	PROPN
fcis-16180	333	12	,	,	PUNCT
fcis-16180	333	13	l.	l.	PROPN
fcis-16180	333	14	zhu	zhu	PROPN
fcis-16180	333	15	,	,	PUNCT
fcis-16180	333	16	t.	t.	PROPN
fcis-16180	333	17	xue	xue	PROPN
fcis-16180	333	18	,	,	PUNCT
fcis-16180	333	19	and	and	CCONJ
fcis-16180	333	20	q.	q.	PROPN
fcis-16180	333	21	qu	qu	PROPN
fcis-16180	333	22	,	,	PUNCT
fcis-16180	333	23	“	"	PUNCT
fcis-16180	333	24	pile	pile	ADJ
fcis-16180	333	25	:	:	PUNCT
fcis-16180	333	26	robust	robust	ADJ
fcis-16180	333	27	privacy	privacy	NOUN
fcis-16180	333	28	-	-	PUNCT
fcis-16180	333	29	preserving	preserve	VERB
fcis-16180	333	30	federated	federated	ADJ
fcis-16180	333	31	learning	learning	NOUN
fcis-16180	333	32	via	via	ADP
fcis-16180	333	33	verifiable	verifiable	ADJ
fcis-16180	333	34	perturbations	perturbation	NOUN
fcis-16180	333	35	,	,	PUNCT
fcis-16180	333	36	”	"	PUNCT
fcis-16180	333	37	ieee	ieee	NOUN
fcis-16180	333	38	transactions	transaction	NOUN
fcis-16180	333	39	on	on	ADP
fcis-16180	333	40	dependable	dependable	ADJ
fcis-16180	333	41	and	and	CCONJ
fcis-16180	333	42	secure	secure	ADJ
fcis-16180	333	43	computing	computing	NOUN
fcis-16180	333	44	,	,	PUNCT
fcis-16180	333	45	2023	2023	NUM
fcis-16180	333	46	.	.	PUNCT
fcis-16180	334	1	[	[	X
fcis-16180	334	2	39	39	NUM
fcis-16180	334	3	]	]	PUNCT
fcis-16180	334	4	l.	l.	PROPN
fcis-16180	334	5	deng	deng	PROPN
fcis-16180	334	6	,	,	PUNCT
fcis-16180	334	7	“	"	PUNCT
fcis-16180	334	8	the	the	DET
fcis-16180	334	9	mnist	mnist	NOUN
fcis-16180	334	10	database	database	NOUN
fcis-16180	334	11	of	of	ADP
fcis-16180	334	12	handwritten	handwritten	ADJ
fcis-16180	334	13	digit	digit	NOUN
fcis-16180	334	14	images	image	NOUN
fcis-16180	334	15	for	for	ADP
fcis-16180	334	16	machine	machine	NOUN
fcis-16180	334	17	learning	learn	VERB
fcis-16180	334	18	research	research	NOUN
fcis-16180	335	1	[	[	X
fcis-16180	335	2	best	good	ADJ
fcis-16180	335	3	of	of	ADP
fcis-16180	335	4	the	the	DET
fcis-16180	335	5	web	web	NOUN
fcis-16180	335	6	]	]	PUNCT
fcis-16180	335	7	,	,	PUNCT
fcis-16180	335	8	”	"	PUNCT
fcis-16180	335	9	ieee	ieee	NOUN
fcis-16180	335	10	signal	signal	NOUN
fcis-16180	335	11	processing	processing	NOUN
fcis-16180	335	12	magazine	magazine	NOUN
fcis-16180	335	13	,	,	PUNCT
fcis-16180	335	14	vol	vol	NOUN
fcis-16180	335	15	.	.	PROPN
fcis-16180	335	16	29	29	NUM
fcis-16180	335	17	,	,	PUNCT
fcis-16180	335	18	no	no	INTJ
fcis-16180	335	19	.	.	NOUN
fcis-16180	335	20	6	6	NUM
fcis-16180	335	21	,	,	PUNCT
fcis-16180	335	22	pp	pp	ADJ
fcis-16180	335	23	.	.	PUNCT
fcis-16180	336	1	141–142	141–142	NUM
fcis-16180	336	2	,	,	PUNCT
fcis-16180	336	3	2012	2012	NUM
fcis-16180	336	4	.	.	PUNCT
fcis-16180	337	1	[	[	X
fcis-16180	337	2	40	40	NUM
fcis-16180	337	3	]	]	PUNCT
fcis-16180	337	4	a.	a.	NOUN
fcis-16180	337	5	krizhevsky	krizhevsky	PROPN
fcis-16180	337	6	,	,	PUNCT
fcis-16180	337	7	g.	g.	PROPN
fcis-16180	337	8	hinton	hinton	PROPN
fcis-16180	337	9	et	et	PROPN
fcis-16180	337	10	al	al	PROPN
fcis-16180	337	11	.	.	PROPN
fcis-16180	337	12	,	,	PUNCT
fcis-16180	337	13	“	"	PUNCT
fcis-16180	337	14	learning	learn	VERB
fcis-16180	337	15	multiple	multiple	ADJ
fcis-16180	337	16	layers	layer	NOUN
fcis-16180	337	17	of	of	ADP
fcis-16180	337	18	features	feature	NOUN
fcis-16180	337	19	from	from	ADP
fcis-16180	337	20	tiny	tiny	ADJ
fcis-16180	337	21	images	image	NOUN
fcis-16180	337	22	,	,	PUNCT
fcis-16180	337	23	”	"	PUNCT
fcis-16180	337	24	2009	2009	NUM
fcis-16180	337	25	.	.	PUNCT
fcis-16180	338	1	[	[	X
fcis-16180	338	2	41	41	NUM
fcis-16180	338	3	]	]	PUNCT
fcis-16180	338	4	k.	k.	NOUN
fcis-16180	339	1	he	he	PRON
fcis-16180	339	2	,	,	PUNCT
fcis-16180	339	3	x.	x.	PROPN
fcis-16180	339	4	zhang	zhang	PROPN
fcis-16180	339	5	,	,	PUNCT
fcis-16180	339	6	s.	s.	PROPN
fcis-16180	339	7	ren	ren	PROPN
fcis-16180	339	8	,	,	PUNCT
fcis-16180	339	9	and	and	CCONJ
fcis-16180	339	10	j.	j.	PROPN
fcis-16180	339	11	sun	sun	PROPN
fcis-16180	339	12	,	,	PUNCT
fcis-16180	339	13	“	"	PUNCT
fcis-16180	339	14	deep	deep	ADJ
fcis-16180	339	15	residual	residual	ADJ
fcis-16180	339	16	learning	learning	NOUN
fcis-16180	339	17	for	for	ADP
fcis-16180	339	18	image	image	NOUN
fcis-16180	339	19	recognition	recognition	NOUN
fcis-16180	339	20	,	,	PUNCT
fcis-16180	339	21	”	"	PUNCT
fcis-16180	339	22	in	in	ADP
fcis-16180	339	23	proceedings	proceeding	NOUN
fcis-16180	339	24	of	of	ADP
fcis-16180	339	25	the	the	DET
fcis-16180	339	26	ieee	ieee	NOUN
fcis-16180	339	27	conference	conference	NOUN
fcis-16180	339	28	on	on	ADP
fcis-16180	339	29	computer	computer	NOUN
fcis-16180	339	30	vision	vision	NOUN
fcis-16180	339	31	and	and	CCONJ
fcis-16180	339	32	pattern	pattern	NOUN
fcis-16180	339	33	recognition	recognition	NOUN
fcis-16180	339	34	,	,	PUNCT
fcis-16180	339	35	2016	2016	NUM
fcis-16180	339	36	,	,	PUNCT
fcis-16180	339	37	pp	pp	ADJ
fcis-16180	339	38	.	.	PUNCT
fcis-16180	340	1	770–778	770–778	NUM