id	sid	tid	token	lemma	pos
ajst-29874	1	1	academic	academic	ADJ
ajst-29874	1	2	journal	journal	NOUN
ajst-29874	1	3	of	of	ADP
ajst-29874	1	4	science	science	NOUN
ajst-29874	1	5	and	and	CCONJ
ajst-29874	1	6	technology	technology	NOUN
ajst-29874	1	7	issn	issn	NOUN
ajst-29874	1	8	:	:	PUNCT
ajst-29874	1	9	2771	2771	NUM
ajst-29874	1	10	-	-	SYM
ajst-29874	1	11	3032	3032	NUM
ajst-29874	1	12	|	|	NOUN
ajst-29874	1	13	vol	vol	NOUN
ajst-29874	1	14	.	.	PUNCT
ajst-29874	2	1	14	14	NUM
ajst-29874	2	2	,	,	PUNCT
ajst-29874	2	3	no	no	INTJ
ajst-29874	2	4	.	.	NOUN
ajst-29874	2	5	2	2	NUM
ajst-29874	2	6	,	,	PUNCT
ajst-29874	2	7	2025	2025	NUM
ajst-29874	2	8	174	174	NUM
ajst-29874	2	9	deep	deep	ADJ
ajst-29874	2	10	learning‐based	learning‐based	ADJ
ajst-29874	2	11	magnetic	magnetic	ADJ
ajst-29874	2	12	core	core	NOUN
ajst-29874	2	13	loss	loss	NOUN
ajst-29874	2	14	prediction	prediction	NOUN
ajst-29874	2	15	using	use	VERB
ajst-29874	2	16	a	a	DET
ajst-29874	2	17	fully	fully	ADV
ajst-29874	2	18	connected	connect	VERB
ajst-29874	2	19	neural	neural	ADJ
ajst-29874	2	20	network	network	NOUN
ajst-29874	2	21	wenqi	wenqi	PROPN
ajst-29874	2	22	wang1	wang1	PROPN
ajst-29874	2	23	,	,	PUNCT
ajst-29874	2	24	donghao	donghao	PROPN
ajst-29874	2	25	yang1	yang1	PROPN
ajst-29874	2	26	,	,	PUNCT
ajst-29874	2	27	binkai	binkai	PROPN
ajst-29874	2	28	sun2	sun2	PROPN
ajst-29874	2	29	,	,	PUNCT
ajst-29874	2	30	zhenzhen	zhenzhen	PROPN
ajst-29874	2	31	huang1	huang1	PROPN
ajst-29874	2	32	,	,	PUNCT
ajst-29874	2	33	xin	xin	PROPN
ajst-29874	2	34	huang1	huang1	PROPN
ajst-29874	2	35	,	,	PUNCT
ajst-29874	2	36	jiayi	jiayi	PROPN
ajst-29874	2	37	li2	li2	PROPN
ajst-29874	2	38	,	,	PUNCT
ajst-29874	3	1	xiuqin	xiuqin	PROPN
ajst-29874	3	2	huang3	huang3	PROPN
ajst-29874	4	1	and	and	CCONJ
ajst-29874	4	2	xinjian	xinjian	ADJ
ajst-29874	4	3	wang1	wang1	PROPN
ajst-29874	4	4	,	,	PUNCT
ajst-29874	4	5	*	*	PUNCT
ajst-29874	5	1	1school	1school	NUM
ajst-29874	5	2	of	of	ADP
ajst-29874	5	3	physics	physics	NOUN
ajst-29874	5	4	and	and	CCONJ
ajst-29874	5	5	electronic	electronic	ADJ
ajst-29874	5	6	engineering	engineering	NOUN
ajst-29874	5	7	,	,	PUNCT
ajst-29874	5	8	harbin	harbin	PROPN
ajst-29874	5	9	normal	normal	ADJ
ajst-29874	5	10	university	university	PROPN
ajst-29874	5	11	,	,	PUNCT
ajst-29874	5	12	harbin	harbin	PROPN
ajst-29874	5	13	,	,	PUNCT
ajst-29874	5	14	china	china	PROPN
ajst-29874	5	15	2school	2school	NUM
ajst-29874	5	16	of	of	ADP
ajst-29874	5	17	chemistry	chemistry	NOUN
ajst-29874	5	18	and	and	CCONJ
ajst-29874	5	19	chemical	chemical	PROPN
ajst-29874	5	20	engineering	engineering	NOUN
ajst-29874	5	21	,	,	PUNCT
ajst-29874	5	22	harbin	harbin	PROPN
ajst-29874	5	23	normal	normal	ADJ
ajst-29874	5	24	university	university	PROPN
ajst-29874	5	25	,	,	PUNCT
ajst-29874	5	26	harbin	harbin	PROPN
ajst-29874	5	27	,	,	PUNCT
ajst-29874	5	28	china	china	PROPN
ajst-29874	5	29	3school	3school	NUM
ajst-29874	5	30	of	of	ADP
ajst-29874	5	31	business	business	NOUN
ajst-29874	5	32	,	,	PUNCT
ajst-29874	5	33	minnan	minnan	PROPN
ajst-29874	5	34	normal	normal	ADJ
ajst-29874	5	35	university	university	NOUN
ajst-29874	5	36	,	,	PUNCT
ajst-29874	5	37	zhangzhou	zhangzhou	PROPN
ajst-29874	5	38	,	,	PUNCT
ajst-29874	5	39	china	china	PROPN
ajst-29874	5	40	*	*	PUNCT
ajst-29874	5	41	corresponding	correspond	VERB
ajst-29874	5	42	author	author	NOUN
ajst-29874	5	43	:	:	PUNCT
ajst-29874	5	44	abstract	abstract	ADJ
ajst-29874	5	45	:	:	PUNCT
ajst-29874	5	46	this	this	DET
ajst-29874	5	47	paper	paper	NOUN
ajst-29874	5	48	proposes	propose	VERB
ajst-29874	5	49	a	a	DET
ajst-29874	5	50	core	core	NOUN
ajst-29874	5	51	loss	loss	NOUN
ajst-29874	5	52	prediction	prediction	NOUN
ajst-29874	5	53	model	model	NOUN
ajst-29874	5	54	based	base	VERB
ajst-29874	5	55	on	on	ADP
ajst-29874	5	56	a	a	DET
ajst-29874	5	57	fully	fully	ADV
ajst-29874	5	58	connected	connected	ADJ
ajst-29874	5	59	neural	neural	ADJ
ajst-29874	5	60	network	network	NOUN
ajst-29874	5	61	(	(	PUNCT
ajst-29874	5	62	fcnn	fcnn	PROPN
ajst-29874	5	63	)	)	PUNCT
ajst-29874	5	64	.	.	PUNCT
ajst-29874	6	1	after	after	ADP
ajst-29874	6	2	preprocessing	preprocesse	VERB
ajst-29874	6	3	the	the	DET
ajst-29874	6	4	data	datum	NOUN
ajst-29874	6	5	,	,	PUNCT
ajst-29874	6	6	seven	seven	NUM
ajst-29874	6	7	key	key	ADJ
ajst-29874	6	8	features	feature	NOUN
ajst-29874	6	9	are	be	AUX
ajst-29874	6	10	extracted	extract	VERB
ajst-29874	6	11	,	,	PUNCT
ajst-29874	6	12	and	and	CCONJ
ajst-29874	6	13	feature	feature	NOUN
ajst-29874	6	14	importance	importance	NOUN
ajst-29874	6	15	is	be	AUX
ajst-29874	6	16	sorted	sort	VERB
ajst-29874	6	17	.	.	PUNCT
ajst-29874	7	1	the	the	DET
ajst-29874	7	2	excitation	excitation	NOUN
ajst-29874	7	3	waveform	waveform	VERB
ajst-29874	7	4	features	feature	NOUN
ajst-29874	7	5	are	be	AUX
ajst-29874	7	6	converted	convert	VERB
ajst-29874	7	7	into	into	ADP
ajst-29874	7	8	five	five	NUM
ajst-29874	7	9	variables	variable	NOUN
ajst-29874	7	10	and	and	CCONJ
ajst-29874	7	11	combined	combine	VERB
ajst-29874	7	12	with	with	ADP
ajst-29874	7	13	four	four	NUM
ajst-29874	7	14	additional	additional	ADJ
ajst-29874	7	15	features	feature	NOUN
ajst-29874	7	16	to	to	PART
ajst-29874	7	17	form	form	VERB
ajst-29874	7	18	the	the	DET
ajst-29874	7	19	final	final	ADJ
ajst-29874	7	20	input	input	NOUN
ajst-29874	7	21	feature	feature	NOUN
ajst-29874	7	22	set	set	NOUN
ajst-29874	7	23	.	.	PUNCT
ajst-29874	8	1	based	base	VERB
ajst-29874	8	2	on	on	ADP
ajst-29874	8	3	this	this	PRON
ajst-29874	8	4	,	,	PUNCT
ajst-29874	8	5	the	the	DET
ajst-29874	8	6	fcnn	fcnn	PROPN
ajst-29874	8	7	prediction	prediction	NOUN
ajst-29874	8	8	model	model	NOUN
ajst-29874	8	9	is	be	AUX
ajst-29874	8	10	constructed	construct	VERB
ajst-29874	8	11	,	,	PUNCT
ajst-29874	8	12	the	the	DET
ajst-29874	8	13	early	early	ADJ
ajst-29874	8	14	stop	stop	NOUN
ajst-29874	8	15	method	method	NOUN
ajst-29874	8	16	is	be	AUX
ajst-29874	8	17	used	use	VERB
ajst-29874	8	18	in	in	ADP
ajst-29874	8	19	the	the	DET
ajst-29874	8	20	training	training	NOUN
ajst-29874	8	21	process	process	NOUN
ajst-29874	8	22	to	to	PART
ajst-29874	8	23	prevent	prevent	VERB
ajst-29874	8	24	overfitting	overfitting	NOUN
ajst-29874	8	25	,	,	PUNCT
ajst-29874	8	26	and	and	CCONJ
ajst-29874	8	27	the	the	DET
ajst-29874	8	28	generalization	generalization	NOUN
ajst-29874	8	29	ability	ability	NOUN
ajst-29874	8	30	is	be	AUX
ajst-29874	8	31	evaluated	evaluate	VERB
ajst-29874	8	32	using	use	VERB
ajst-29874	8	33	cross	cross	NOUN
ajst-29874	8	34	-	-	ADJ
ajst-29874	8	35	validation[1	validation[1	PROPN
ajst-29874	8	36	]	]	PUNCT
ajst-29874	8	37	.	.	PUNCT
ajst-29874	9	1	the	the	DET
ajst-29874	9	2	r2	r2	PROPN
ajst-29874	9	3	of	of	ADP
ajst-29874	9	4	the	the	DET
ajst-29874	9	5	model	model	NOUN
ajst-29874	9	6	training	training	NOUN
ajst-29874	9	7	set	set	NOUN
ajst-29874	9	8	and	and	CCONJ
ajst-29874	9	9	the	the	DET
ajst-29874	9	10	test	test	NOUN
ajst-29874	9	11	set	set	NOUN
ajst-29874	9	12	are	be	AUX
ajst-29874	9	13	0.9889	0.9889	NUM
ajst-29874	9	14	and	and	CCONJ
ajst-29874	9	15	0.9843	0.9843	NUM
ajst-29874	9	16	,	,	PUNCT
ajst-29874	9	17	respectively	respectively	ADV
ajst-29874	9	18	,	,	PUNCT
ajst-29874	9	19	and	and	CCONJ
ajst-29874	9	20	the	the	DET
ajst-29874	9	21	prediction	prediction	NOUN
ajst-29874	9	22	error	error	NOUN
ajst-29874	9	23	of	of	ADP
ajst-29874	9	24	94.55	94.55	NUM
ajst-29874	9	25	%	%	NOUN
ajst-29874	9	26	of	of	ADP
ajst-29874	9	27	the	the	DET
ajst-29874	9	28	test	test	NOUN
ajst-29874	9	29	samples	sample	NOUN
ajst-29874	9	30	is	be	AUX
ajst-29874	9	31	less	less	ADJ
ajst-29874	9	32	than	than	ADP
ajst-29874	9	33	±10	±10	NOUN
ajst-29874	9	34	%	%	NOUN
ajst-29874	9	35	.	.	PUNCT
ajst-29874	10	1	fcnn	fcnn	PROPN
ajst-29874	10	2	shows	show	VERB
ajst-29874	10	3	higher	high	ADJ
ajst-29874	10	4	accuracy	accuracy	NOUN
ajst-29874	10	5	and	and	CCONJ
ajst-29874	10	6	stability	stability	NOUN
ajst-29874	10	7	in	in	ADP
ajst-29874	10	8	core	core	NOUN
ajst-29874	10	9	loss	loss	NOUN
ajst-29874	10	10	prediction	prediction	NOUN
ajst-29874	10	11	tasks	task	NOUN
ajst-29874	10	12	than	than	ADP
ajst-29874	10	13	traditional	traditional	ADJ
ajst-29874	10	14	machine	machine	NOUN
ajst-29874	10	15	learning	learning	NOUN
ajst-29874	10	16	methods	method	NOUN
ajst-29874	10	17	.	.	PUNCT
ajst-29874	11	1	keywords	keyword	NOUN
ajst-29874	11	2	:	:	PUNCT
ajst-29874	11	3	core	core	NOUN
ajst-29874	11	4	loss	loss	NOUN
ajst-29874	11	5	prediction	prediction	NOUN
ajst-29874	11	6	,	,	PUNCT
ajst-29874	11	7	fully	fully	ADV
ajst-29874	11	8	connected	connected	ADJ
ajst-29874	11	9	neural	neural	ADJ
ajst-29874	11	10	network	network	NOUN
ajst-29874	11	11	,	,	PUNCT
ajst-29874	11	12	random	random	ADJ
ajst-29874	11	13	forest	forest	NOUN
ajst-29874	11	14	,	,	PUNCT
ajst-29874	11	15	model	model	NOUN
ajst-29874	11	16	evaluation	evaluation	NOUN
ajst-29874	11	17	.	.	PUNCT
ajst-29874	12	1	1	1	X
ajst-29874	12	2	.	.	X
ajst-29874	12	3	introduction	introduction	NOUN
ajst-29874	12	4	the	the	DET
ajst-29874	12	5	core	core	NOUN
ajst-29874	12	6	loss	loss	NOUN
ajst-29874	12	7	is	be	AUX
ajst-29874	12	8	a	a	DET
ajst-29874	12	9	key	key	ADJ
ajst-29874	12	10	parameter	parameter	NOUN
ajst-29874	12	11	of	of	ADP
ajst-29874	12	12	power	power	NOUN
ajst-29874	12	13	electronic	electronic	ADJ
ajst-29874	12	14	equipment	equipment	NOUN
ajst-29874	12	15	,	,	PUNCT
ajst-29874	12	16	and	and	CCONJ
ajst-29874	12	17	its	its	PRON
ajst-29874	12	18	accurate	accurate	ADJ
ajst-29874	12	19	prediction	prediction	NOUN
ajst-29874	12	20	is	be	AUX
ajst-29874	12	21	very	very	ADV
ajst-29874	12	22	important	important	ADJ
ajst-29874	12	23	for	for	ADP
ajst-29874	12	24	equipment	equipment	NOUN
ajst-29874	12	25	design	design	NOUN
ajst-29874	12	26	,	,	PUNCT
ajst-29874	12	27	optimization	optimization	NOUN
ajst-29874	12	28	,	,	PUNCT
ajst-29874	12	29	and	and	CCONJ
ajst-29874	12	30	operation	operation	NOUN
ajst-29874	12	31	efficiency	efficiency	NOUN
ajst-29874	12	32	improvement[2	improvement[2	PROPN
ajst-29874	12	33	]	]	PUNCT
ajst-29874	12	34	.	.	PUNCT
ajst-29874	13	1	the	the	DET
ajst-29874	13	2	traditional	traditional	ADJ
ajst-29874	13	3	core	core	NOUN
ajst-29874	13	4	loss	loss	NOUN
ajst-29874	13	5	prediction	prediction	NOUN
ajst-29874	13	6	methods	method	NOUN
ajst-29874	13	7	are	be	AUX
ajst-29874	13	8	mainly	mainly	ADV
ajst-29874	13	9	based	base	VERB
ajst-29874	13	10	on	on	ADP
ajst-29874	13	11	empirical	empirical	ADJ
ajst-29874	13	12	formulas	formula	NOUN
ajst-29874	13	13	and	and	CCONJ
ajst-29874	13	14	experimental	experimental	ADJ
ajst-29874	13	15	data	datum	NOUN
ajst-29874	13	16	,	,	PUNCT
ajst-29874	13	17	but	but	CCONJ
ajst-29874	13	18	because	because	SCONJ
ajst-29874	13	19	the	the	DET
ajst-29874	13	20	core	core	NOUN
ajst-29874	13	21	loss	loss	NOUN
ajst-29874	13	22	is	be	AUX
ajst-29874	13	23	affected	affect	VERB
ajst-29874	13	24	by	by	ADP
ajst-29874	13	25	many	many	ADJ
ajst-29874	13	26	factors	factor	NOUN
ajst-29874	13	27	,	,	PUNCT
ajst-29874	13	28	these	these	DET
ajst-29874	13	29	methods	method	NOUN
ajst-29874	13	30	make	make	VERB
ajst-29874	13	31	it	it	PRON
ajst-29874	13	32	difficult	difficult	ADJ
ajst-29874	13	33	to	to	PART
ajst-29874	13	34	capture	capture	VERB
ajst-29874	13	35	its	its	PRON
ajst-29874	13	36	complex	complex	ADJ
ajst-29874	13	37	nonlinear	nonlinear	ADJ
ajst-29874	13	38	relationship	relationship	NOUN
ajst-29874	13	39	,	,	PUNCT
ajst-29874	13	40	resulting	result	VERB
ajst-29874	13	41	in	in	ADP
ajst-29874	13	42	limited	limited	ADJ
ajst-29874	13	43	prediction	prediction	NOUN
ajst-29874	13	44	accuracy	accuracy	NOUN
ajst-29874	13	45	.	.	PUNCT
ajst-29874	14	1	in	in	ADP
ajst-29874	14	2	recent	recent	ADJ
ajst-29874	14	3	years	year	NOUN
ajst-29874	14	4	,	,	PUNCT
ajst-29874	14	5	with	with	ADP
ajst-29874	14	6	the	the	DET
ajst-29874	14	7	development	development	NOUN
ajst-29874	14	8	of	of	ADP
ajst-29874	14	9	machine	machine	NOUN
ajst-29874	14	10	learning	learn	VERB
ajst-29874	14	11	technology	technology	NOUN
ajst-29874	14	12	,	,	PUNCT
ajst-29874	14	13	the	the	DET
ajst-29874	14	14	data	data	NOUN
ajst-29874	14	15	-	-	PUNCT
ajst-29874	14	16	driven	drive	VERB
ajst-29874	14	17	magnetic	magnetic	ADJ
ajst-29874	14	18	core	core	NOUN
ajst-29874	14	19	loss	loss	NOUN
ajst-29874	14	20	prediction	prediction	NOUN
ajst-29874	14	21	method	method	NOUN
ajst-29874	14	22	has	have	AUX
ajst-29874	14	23	been	be	AUX
ajst-29874	14	24	widely	widely	ADV
ajst-29874	14	25	used	use	VERB
ajst-29874	14	26	.	.	PUNCT
ajst-29874	15	1	deep	deep	ADJ
ajst-29874	15	2	learning	learning	NOUN
ajst-29874	15	3	methods	method	NOUN
ajst-29874	15	4	,	,	PUNCT
ajst-29874	15	5	especially	especially	ADV
ajst-29874	15	6	fully	fully	ADV
ajst-29874	15	7	connected	connect	VERB
ajst-29874	15	8	neural	neural	ADJ
ajst-29874	15	9	networks	network	NOUN
ajst-29874	15	10	(	(	PUNCT
ajst-29874	15	11	fcnns	fcnn	NOUN
ajst-29874	15	12	)	)	PUNCT
ajst-29874	15	13	,	,	PUNCT
ajst-29874	15	14	have	have	AUX
ajst-29874	15	15	demonstrated	demonstrate	VERB
ajst-29874	15	16	remarkable	remarkable	ADJ
ajst-29874	15	17	capabilities	capability	NOUN
ajst-29874	15	18	in	in	ADP
ajst-29874	15	19	modeling	model	VERB
ajst-29874	15	20	complex	complex	ADJ
ajst-29874	15	21	nonlinear	nonlinear	ADJ
ajst-29874	15	22	systems	system	NOUN
ajst-29874	15	23	.	.	PUNCT
ajst-29874	16	1	in	in	ADP
ajst-29874	16	2	this	this	DET
ajst-29874	16	3	paper	paper	NOUN
ajst-29874	16	4	,	,	PUNCT
ajst-29874	16	5	a	a	DET
ajst-29874	16	6	magnetic	magnetic	ADJ
ajst-29874	16	7	core	core	NOUN
ajst-29874	16	8	loss	loss	NOUN
ajst-29874	16	9	prediction	prediction	NOUN
ajst-29874	16	10	model	model	NOUN
ajst-29874	16	11	based	base	VERB
ajst-29874	16	12	on	on	ADP
ajst-29874	16	13	fcnn	fcnn	PROPN
ajst-29874	16	14	is	be	AUX
ajst-29874	16	15	proposed	propose	VERB
ajst-29874	16	16	,	,	PUNCT
ajst-29874	16	17	and	and	CCONJ
ajst-29874	16	18	the	the	DET
ajst-29874	16	19	feature	feature	NOUN
ajst-29874	16	20	selection	selection	NOUN
ajst-29874	16	21	is	be	AUX
ajst-29874	16	22	combined	combine	VERB
ajst-29874	16	23	with	with	ADP
ajst-29874	16	24	random	random	ADJ
ajst-29874	16	25	forest	forest	NOUN
ajst-29874	16	26	to	to	PART
ajst-29874	16	27	improve	improve	VERB
ajst-29874	16	28	the	the	DET
ajst-29874	16	29	prediction	prediction	NOUN
ajst-29874	16	30	accuracy	accuracy	NOUN
ajst-29874	16	31	and	and	CCONJ
ajst-29874	16	32	generalization	generalization	NOUN
ajst-29874	16	33	ability	ability	NOUN
ajst-29874	16	34	.	.	PUNCT
ajst-29874	17	1	by	by	ADP
ajst-29874	17	2	comparing	compare	VERB
ajst-29874	17	3	with	with	ADP
ajst-29874	17	4	traditional	traditional	ADJ
ajst-29874	17	5	machine	machine	NOUN
ajst-29874	17	6	learning	learning	NOUN
ajst-29874	17	7	models	model	NOUN
ajst-29874	17	8	(	(	PUNCT
ajst-29874	17	9	random	random	ADJ
ajst-29874	17	10	forest	forest	NOUN
ajst-29874	17	11	,	,	PUNCT
ajst-29874	17	12	xgboost	xgboost	ADV
ajst-29874	17	13	)	)	PUNCT
ajst-29874	17	14	,	,	PUNCT
ajst-29874	17	15	we	we	PRON
ajst-29874	17	16	verify	verify	VERB
ajst-29874	17	17	the	the	DET
ajst-29874	17	18	effectiveness	effectiveness	NOUN
ajst-29874	17	19	of	of	ADP
ajst-29874	17	20	this	this	DET
ajst-29874	17	21	method	method	NOUN
ajst-29874	17	22	and	and	CCONJ
ajst-29874	17	23	provide	provide	VERB
ajst-29874	17	24	an	an	DET
ajst-29874	17	25	efficient	efficient	ADJ
ajst-29874	17	26	and	and	CCONJ
ajst-29874	17	27	scalable	scalable	ADJ
ajst-29874	17	28	solution	solution	NOUN
ajst-29874	17	29	for	for	ADP
ajst-29874	17	30	core	core	NOUN
ajst-29874	17	31	loss	loss	NOUN
ajst-29874	17	32	prediction	prediction	NOUN
ajst-29874	17	33	.	.	PUNCT
ajst-29874	18	1	2	2	X
ajst-29874	18	2	.	.	X
ajst-29874	18	3	data	datum	NOUN
ajst-29874	18	4	cleaning	clean	VERB
ajst-29874	18	5	after	after	ADP
ajst-29874	18	6	data	data	NOUN
ajst-29874	18	7	collection	collection	NOUN
ajst-29874	18	8	,	,	PUNCT
ajst-29874	18	9	data	datum	NOUN
ajst-29874	18	10	cleaning	cleaning	NOUN
ajst-29874	18	11	is	be	AUX
ajst-29874	18	12	usually	usually	ADV
ajst-29874	18	13	required	require	VERB
ajst-29874	18	14	,	,	PUNCT
ajst-29874	18	15	including	include	VERB
ajst-29874	18	16	dealing	deal	VERB
ajst-29874	18	17	with	with	ADP
ajst-29874	18	18	missing	miss	VERB
ajst-29874	18	19	values	value	NOUN
ajst-29874	18	20	,	,	PUNCT
ajst-29874	18	21	duplicate	duplicate	ADJ
ajst-29874	18	22	values	value	NOUN
ajst-29874	18	23	,	,	PUNCT
ajst-29874	18	24	and	and	CCONJ
ajst-29874	18	25	outliers	outlier	NOUN
ajst-29874	18	26	.	.	PUNCT
ajst-29874	19	1	for	for	ADP
ajst-29874	19	2	data	datum	NOUN
ajst-29874	19	3	that	that	PRON
ajst-29874	19	4	does	do	AUX
ajst-29874	19	5	not	not	PART
ajst-29874	19	6	meet	meet	VERB
ajst-29874	19	7	the	the	DET
ajst-29874	19	8	quality	quality	NOUN
ajst-29874	19	9	requirements	requirement	NOUN
ajst-29874	19	10	,	,	PUNCT
ajst-29874	19	11	delete	delete	ADJ
ajst-29874	19	12	or	or	CCONJ
ajst-29874	19	13	fill	fill	VERB
ajst-29874	19	14	the	the	DET
ajst-29874	19	15	policy	policy	NOUN
ajst-29874	19	16	to	to	PART
ajst-29874	19	17	improve	improve	VERB
ajst-29874	19	18	the	the	DET
ajst-29874	19	19	accuracy	accuracy	NOUN
ajst-29874	19	20	and	and	CCONJ
ajst-29874	19	21	reliability	reliability	NOUN
ajst-29874	19	22	of	of	ADP
ajst-29874	19	23	the	the	DET
ajst-29874	19	24	data	datum	NOUN
ajst-29874	19	25	.	.	PUNCT
ajst-29874	20	1	through	through	ADP
ajst-29874	20	2	data	data	NOUN
ajst-29874	20	3	analysis	analysis	NOUN
ajst-29874	20	4	,	,	PUNCT
ajst-29874	20	5	it	it	PRON
ajst-29874	20	6	is	be	AUX
ajst-29874	20	7	found	find	VERB
ajst-29874	20	8	that	that	SCONJ
ajst-29874	20	9	there	there	PRON
ajst-29874	20	10	are	be	VERB
ajst-29874	20	11	discrete	discrete	ADJ
ajst-29874	20	12	and	and	CCONJ
ajst-29874	20	13	continuous	continuous	ADJ
ajst-29874	20	14	data	datum	NOUN
ajst-29874	20	15	in	in	ADP
ajst-29874	20	16	the	the	DET
ajst-29874	20	17	data	datum	NOUN
ajst-29874	20	18	set	set	VERB
ajst-29874	20	19	,	,	PUNCT
ajst-29874	20	20	and	and	CCONJ
ajst-29874	20	21	the	the	DET
ajst-29874	20	22	data	datum	NOUN
ajst-29874	20	23	amplitude	amplitude	NOUN
ajst-29874	20	24	changes	change	NOUN
ajst-29874	20	25	greatly	greatly	ADV
ajst-29874	20	26	.	.	PUNCT
ajst-29874	21	1	based	base	VERB
ajst-29874	21	2	on	on	ADP
ajst-29874	21	3	this	this	DET
ajst-29874	21	4	characteristic	characteristic	NOUN
ajst-29874	21	5	,	,	PUNCT
ajst-29874	21	6	we	we	PRON
ajst-29874	21	7	first	first	ADV
ajst-29874	21	8	cleaned	clean	VERB
ajst-29874	21	9	the	the	DET
ajst-29874	21	10	data	datum	NOUN
ajst-29874	21	11	.	.	PUNCT
ajst-29874	22	1	first	first	ADV
ajst-29874	22	2	of	of	ADP
ajst-29874	22	3	all	all	PRON
ajst-29874	22	4	,	,	PUNCT
ajst-29874	22	5	the	the	DET
ajst-29874	22	6	data	data	NOUN
ajst-29874	22	7	is	be	AUX
ajst-29874	22	8	distributed	distribute	VERB
ajst-29874	22	9	and	and	CCONJ
ajst-29874	22	10	processed	process	VERB
ajst-29874	22	11	.	.	PUNCT
ajst-29874	23	1	here	here	ADV
ajst-29874	23	2	,	,	PUNCT
ajst-29874	23	3	the	the	DET
ajst-29874	23	4	distribution	distribution	NOUN
ajst-29874	23	5	diagram	diagram	NOUN
ajst-29874	23	6	of	of	ADP
ajst-29874	23	7	the	the	DET
ajst-29874	23	8	magnetic	magnetic	ADJ
ajst-29874	23	9	core	core	NOUN
ajst-29874	23	10	loss	loss	NOUN
ajst-29874	23	11	data	datum	NOUN
ajst-29874	23	12	is	be	AUX
ajst-29874	23	13	used	use	VERB
ajst-29874	23	14	for	for	ADP
ajst-29874	23	15	preliminary	preliminary	ADJ
ajst-29874	23	16	observation	observation	NOUN
ajst-29874	23	17	:	:	PUNCT
ajst-29874	23	18	figure	figure	NOUN
ajst-29874	23	19	1	1	NUM
ajst-29874	23	20	.	.	PUNCT
ajst-29874	23	21	core	core	NOUN
ajst-29874	23	22	loss	loss	NOUN
ajst-29874	23	23	distribution	distribution	NOUN
ajst-29874	23	24	diagram	diagram	NOUN
ajst-29874	23	25	175	175	NUM
ajst-29874	23	26	the	the	DET
ajst-29874	23	27	data	data	NOUN
ajst-29874	23	28	presents	present	VERB
ajst-29874	23	29	a	a	DET
ajst-29874	23	30	skewed	skewed	ADJ
ajst-29874	23	31	distribution	distribution	NOUN
ajst-29874	23	32	and	and	CCONJ
ajst-29874	23	33	contains	contain	VERB
ajst-29874	23	34	more	more	ADV
ajst-29874	23	35	extreme	extreme	ADJ
ajst-29874	23	36	values	value	NOUN
ajst-29874	23	37	.	.	PUNCT
ajst-29874	24	1	in	in	ADP
ajst-29874	24	2	order	order	NOUN
ajst-29874	24	3	to	to	PART
ajst-29874	24	4	further	far	ADV
ajst-29874	24	5	identify	identify	VERB
ajst-29874	24	6	outliers	outlier	NOUN
ajst-29874	24	7	and	and	CCONJ
ajst-29874	24	8	abnormal	abnormal	ADJ
ajst-29874	24	9	data	datum	NOUN
ajst-29874	24	10	,	,	PUNCT
ajst-29874	24	11	this	this	DET
ajst-29874	24	12	paper	paper	NOUN
ajst-29874	24	13	uses	use	VERB
ajst-29874	24	14	the	the	DET
ajst-29874	24	15	boxplot	boxplot	NOUN
ajst-29874	24	16	method	method	NOUN
ajst-29874	24	17	to	to	PART
ajst-29874	24	18	divide	divide	VERB
ajst-29874	24	19	the	the	DET
ajst-29874	24	20	data	datum	NOUN
ajst-29874	24	21	set	set	VERB
ajst-29874	24	22	into	into	ADP
ajst-29874	24	23	four	four	NUM
ajst-29874	24	24	quartiles	quartile	NOUN
ajst-29874	24	25	to	to	PART
ajst-29874	24	26	evaluate	evaluate	VERB
ajst-29874	24	27	the	the	DET
ajst-29874	24	28	variability	variability	NOUN
ajst-29874	24	29	.	.	PUNCT
ajst-29874	25	1	as	as	ADP
ajst-29874	25	2	a	a	DET
ajst-29874	25	3	commonly	commonly	ADV
ajst-29874	25	4	used	use	VERB
ajst-29874	25	5	statistical	statistical	ADJ
ajst-29874	25	6	graph	graph	NOUN
ajst-29874	25	7	,	,	PUNCT
ajst-29874	25	8	boxplot	boxplot	NOUN
ajst-29874	25	9	is	be	AUX
ajst-29874	25	10	used	use	VERB
ajst-29874	25	11	to	to	PART
ajst-29874	25	12	describe	describe	VERB
ajst-29874	25	13	the	the	DET
ajst-29874	25	14	distribution	distribution	NOUN
ajst-29874	25	15	characteristics	characteristic	NOUN
ajst-29874	25	16	of	of	ADP
ajst-29874	25	17	data	datum	NOUN
ajst-29874	25	18	,	,	PUNCT
ajst-29874	25	19	especially	especially	ADV
ajst-29874	25	20	the	the	DET
ajst-29874	25	21	central	central	ADJ
ajst-29874	25	22	position	position	NOUN
ajst-29874	25	23	,	,	PUNCT
ajst-29874	25	24	the	the	DET
ajst-29874	25	25	degree	degree	NOUN
ajst-29874	25	26	of	of	ADP
ajst-29874	25	27	dispersion	dispersion	NOUN
ajst-29874	25	28	,	,	PUNCT
ajst-29874	25	29	and	and	CCONJ
ajst-29874	25	30	the	the	DET
ajst-29874	25	31	outlier	outlier	NOUN
ajst-29874	25	32	of	of	ADP
ajst-29874	25	33	the	the	DET
ajst-29874	25	34	data[3	data[3	NUM
ajst-29874	25	35	]	]	PUNCT
ajst-29874	25	36	.	.	PUNCT
ajst-29874	26	1	first	first	ADV
ajst-29874	26	2	,	,	PUNCT
ajst-29874	26	3	sort	sort	VERB
ajst-29874	26	4	the	the	DET
ajst-29874	26	5	entire	entire	ADJ
ajst-29874	26	6	data	datum	NOUN
ajst-29874	26	7	in	in	ADP
ajst-29874	26	8	ascending	ascend	VERB
ajst-29874	26	9	order	order	NOUN
ajst-29874	26	10	,	,	PUNCT
ajst-29874	26	11	then	then	ADV
ajst-29874	26	12	divide	divide	VERB
ajst-29874	26	13	it	it	PRON
ajst-29874	26	14	into	into	ADP
ajst-29874	26	15	four	four	NUM
ajst-29874	26	16	equal	equal	ADJ
ajst-29874	26	17	quartiles	quartile	NOUN
ajst-29874	26	18	,	,	PUNCT
ajst-29874	26	19	called	call	VERB
ajst-29874	26	20	q1	q1	PROPN
ajst-29874	26	21	,	,	PUNCT
ajst-29874	26	22	q2	q2	NOUN
ajst-29874	26	23	,	,	PUNCT
ajst-29874	26	24	q3	q3	PROPN
ajst-29874	26	25	,	,	PUNCT
ajst-29874	26	26	and	and	CCONJ
ajst-29874	26	27	q4	q4	PROPN
ajst-29874	26	28	,	,	PUNCT
ajst-29874	26	29	where	where	SCONJ
ajst-29874	26	30	q1	q1	PROPN
ajst-29874	26	31	and	and	CCONJ
ajst-29874	26	32	q3	q3	NOUN
ajst-29874	26	33	are	be	AUX
ajst-29874	26	34	the	the	DET
ajst-29874	26	35	first	first	ADJ
ajst-29874	26	36	and	and	CCONJ
ajst-29874	26	37	third	third	ADJ
ajst-29874	26	38	quartiles	quartile	NOUN
ajst-29874	26	39	of	of	ADP
ajst-29874	26	40	the	the	DET
ajst-29874	26	41	dataset	dataset	NOUN
ajst-29874	26	42	,	,	PUNCT
ajst-29874	26	43	and	and	CCONJ
ajst-29874	26	44	iqr	iqr	PROPN
ajst-29874	26	45	is	be	AUX
ajst-29874	26	46	the	the	DET
ajst-29874	26	47	interquartile	interquartile	NOUN
ajst-29874	26	48	.	.	PUNCT
ajst-29874	27	1	at	at	ADP
ajst-29874	27	2	the	the	DET
ajst-29874	27	3	same	same	ADJ
ajst-29874	27	4	time	time	NOUN
ajst-29874	27	5	,	,	PUNCT
ajst-29874	27	6	it	it	PRON
ajst-29874	27	7	is	be	AUX
ajst-29874	27	8	specified	specify	VERB
ajst-29874	27	9	that	that	SCONJ
ajst-29874	27	10	points	point	NOUN
ajst-29874	27	11	other	other	ADJ
ajst-29874	27	12	than	than	ADP
ajst-29874	27	13	q1	q1	PROPN
ajst-29874	27	14	and	and	CCONJ
ajst-29874	27	15	q3	q3	NOUN
ajst-29874	27	16	are	be	AUX
ajst-29874	27	17	distribution	distribution	NOUN
ajst-29874	27	18	outliers	outlier	NOUN
ajst-29874	27	19	,	,	PUNCT
ajst-29874	27	20	and	and	CCONJ
ajst-29874	27	21	the	the	DET
ajst-29874	27	22	results	result	NOUN
ajst-29874	27	23	of	of	ADP
ajst-29874	27	24	the	the	DET
ajst-29874	27	25	boxplot	boxplot	NOUN
ajst-29874	27	26	are	be	AUX
ajst-29874	27	27	shown	show	VERB
ajst-29874	27	28	in	in	ADP
ajst-29874	27	29	figure	figure	NOUN
ajst-29874	27	30	2	2	NUM
ajst-29874	27	31	:	:	PUNCT
ajst-29874	27	32	figure	figure	NOUN
ajst-29874	27	33	2	2	NUM
ajst-29874	27	34	.	.	PUNCT
ajst-29874	27	35	core	core	PROPN
ajst-29874	27	36	loss	loss	PROPN
ajst-29874	27	37	box	box	PROPN
ajst-29874	27	38	diagram	diagram	NOUN
ajst-29874	27	39	it	it	PRON
ajst-29874	27	40	can	can	AUX
ajst-29874	27	41	be	be	AUX
ajst-29874	27	42	found	find	VERB
ajst-29874	27	43	that	that	SCONJ
ajst-29874	27	44	there	there	PRON
ajst-29874	27	45	are	be	VERB
ajst-29874	27	46	a	a	DET
ajst-29874	27	47	lot	lot	NOUN
ajst-29874	27	48	of	of	ADP
ajst-29874	27	49	outliers	outlier	NOUN
ajst-29874	27	50	in	in	ADP
ajst-29874	27	51	the	the	DET
ajst-29874	27	52	data	datum	NOUN
ajst-29874	27	53	,	,	PUNCT
ajst-29874	27	54	and	and	CCONJ
ajst-29874	27	55	combined	combine	VERB
ajst-29874	27	56	with	with	ADP
ajst-29874	27	57	the	the	DET
ajst-29874	27	58	skewed	skewed	ADJ
ajst-29874	27	59	distribution	distribution	NOUN
ajst-29874	27	60	,	,	PUNCT
ajst-29874	27	61	these	these	DET
ajst-29874	27	62	outliers	outlier	NOUN
ajst-29874	27	63	have	have	VERB
ajst-29874	27	64	certain	certain	ADJ
ajst-29874	27	65	experimental	experimental	ADJ
ajst-29874	27	66	significance	significance	NOUN
ajst-29874	27	67	.	.	PUNCT
ajst-29874	28	1	to	to	PART
ajst-29874	28	2	prevent	prevent	VERB
ajst-29874	28	3	the	the	DET
ajst-29874	28	4	data	data	NOUN
ajst-29874	28	5	features	feature	NOUN
ajst-29874	28	6	of	of	ADP
ajst-29874	28	7	magnetic	magnetic	ADJ
ajst-29874	28	8	core	core	NOUN
ajst-29874	28	9	loss	loss	NOUN
ajst-29874	28	10	from	from	ADP
ajst-29874	28	11	being	be	AUX
ajst-29874	28	12	damaged	damage	VERB
ajst-29874	28	13	,	,	PUNCT
ajst-29874	28	14	this	this	DET
ajst-29874	28	15	paper	paper	NOUN
ajst-29874	28	16	chooses	choose	VERB
ajst-29874	28	17	the	the	DET
ajst-29874	28	18	value	value	NOUN
ajst-29874	28	19	with	with	ADP
ajst-29874	28	20	the	the	DET
ajst-29874	28	21	largest	large	ADJ
ajst-29874	28	22	outlier	outlier	NOUN
ajst-29874	28	23	degree	degree	NOUN
ajst-29874	28	24	at	at	ADP
ajst-29874	28	25	the	the	DET
ajst-29874	28	26	top	top	NOUN
ajst-29874	28	27	as	as	ADP
ajst-29874	28	28	the	the	DET
ajst-29874	28	29	outlier	outlier	NOUN
ajst-29874	28	30	value	value	NOUN
ajst-29874	28	31	,	,	PUNCT
ajst-29874	28	32	while	while	SCONJ
ajst-29874	28	33	the	the	DET
ajst-29874	28	34	data	datum	NOUN
ajst-29874	28	35	near	near	ADP
ajst-29874	28	36	the	the	DET
ajst-29874	28	37	box	box	NOUN
ajst-29874	28	38	type	type	NOUN
ajst-29874	28	39	is	be	AUX
ajst-29874	28	40	retained	retain	VERB
ajst-29874	28	41	.	.	PUNCT
ajst-29874	29	1	3	3	X
ajst-29874	29	2	.	.	NOUN
ajst-29874	29	3	feature	feature	NOUN
ajst-29874	29	4	importance	importance	NOUN
ajst-29874	29	5	sorting	sorting	NOUN
ajst-29874	29	6	and	and	CCONJ
ajst-29874	29	7	screening	screening	NOUN
ajst-29874	29	8	based	base	VERB
ajst-29874	29	9	on	on	ADP
ajst-29874	29	10	random	random	ADJ
ajst-29874	29	11	forest	forest	NOUN
ajst-29874	29	12	3.1	3.1	NUM
ajst-29874	29	13	.	.	PUNCT
ajst-29874	30	1	feature	feature	NOUN
ajst-29874	30	2	selection	selection	NOUN
ajst-29874	30	3	and	and	CCONJ
ajst-29874	30	4	data	datum	NOUN
ajst-29874	30	5	processing	processing	NOUN
ajst-29874	30	6	before	before	ADP
ajst-29874	30	7	establishing	establish	VERB
ajst-29874	30	8	the	the	DET
ajst-29874	30	9	core	core	NOUN
ajst-29874	30	10	loss	loss	NOUN
ajst-29874	30	11	prediction	prediction	NOUN
ajst-29874	30	12	model	model	NOUN
ajst-29874	30	13	,	,	PUNCT
ajst-29874	30	14	the	the	DET
ajst-29874	30	15	feature	feature	NOUN
ajst-29874	30	16	selection	selection	NOUN
ajst-29874	30	17	is	be	AUX
ajst-29874	30	18	needed	need	VERB
ajst-29874	30	19	first	first	ADV
ajst-29874	30	20	.	.	PUNCT
ajst-29874	31	1	to	to	PART
ajst-29874	31	2	improve	improve	VERB
ajst-29874	31	3	the	the	DET
ajst-29874	31	4	prediction	prediction	NOUN
ajst-29874	31	5	ability	ability	NOUN
ajst-29874	31	6	of	of	ADP
ajst-29874	31	7	the	the	DET
ajst-29874	31	8	model	model	NOUN
ajst-29874	31	9	,	,	PUNCT
ajst-29874	31	10	we	we	PRON
ajst-29874	31	11	extracted	extract	VERB
ajst-29874	31	12	seven	seven	NUM
ajst-29874	31	13	waveform	waveform	NOUN
ajst-29874	31	14	features	feature	NOUN
ajst-29874	31	15	,	,	PUNCT
ajst-29874	31	16	including	include	VERB
ajst-29874	31	17	standard	standard	ADJ
ajst-29874	31	18	deviation	deviation	NOUN
ajst-29874	31	19	,	,	PUNCT
ajst-29874	31	20	skewness	skewness	NOUN
ajst-29874	31	21	,	,	PUNCT
ajst-29874	31	22	amplitude	amplitude	NOUN
ajst-29874	31	23	,	,	PUNCT
ajst-29874	31	24	kurtosis	kurtosis	NOUN
ajst-29874	31	25	,	,	PUNCT
ajst-29874	31	26	turning	turn	VERB
ajst-29874	31	27	point	point	NOUN
ajst-29874	31	28	,	,	PUNCT
ajst-29874	31	29	mean	mean	NOUN
ajst-29874	31	30	value	value	NOUN
ajst-29874	31	31	,	,	PUNCT
ajst-29874	31	32	and	and	CCONJ
ajst-29874	31	33	spectral	spectral	ADJ
ajst-29874	31	34	center[4	center[4	NOUN
ajst-29874	31	35	]	]	X
ajst-29874	31	36	.	.	PUNCT
ajst-29874	32	1	due	due	ADP
ajst-29874	32	2	to	to	ADP
ajst-29874	32	3	the	the	DET
ajst-29874	32	4	large	large	ADJ
ajst-29874	32	5	number	number	NOUN
ajst-29874	32	6	of	of	ADP
ajst-29874	32	7	data	data	NOUN
ajst-29874	32	8	features	feature	NOUN
ajst-29874	32	9	,	,	PUNCT
ajst-29874	32	10	to	to	PART
ajst-29874	32	11	reduce	reduce	VERB
ajst-29874	32	12	redundancy	redundancy	NOUN
ajst-29874	32	13	,	,	PUNCT
ajst-29874	32	14	we	we	PRON
ajst-29874	32	15	use	use	VERB
ajst-29874	32	16	random	random	ADJ
ajst-29874	32	17	forest	forest	NOUN
ajst-29874	32	18	to	to	PART
ajst-29874	32	19	sort	sort	VERB
ajst-29874	32	20	and	and	CCONJ
ajst-29874	32	21	screen	screen	VERB
ajst-29874	32	22	the	the	DET
ajst-29874	32	23	importance	importance	NOUN
ajst-29874	32	24	of	of	ADP
ajst-29874	32	25	features	feature	NOUN
ajst-29874	32	26	.	.	PUNCT
ajst-29874	33	1	random	random	ADJ
ajst-29874	33	2	forest	forest	NOUN
ajst-29874	33	3	improves	improve	VERB
ajst-29874	33	4	model	model	NOUN
ajst-29874	33	5	accuracy	accuracy	NOUN
ajst-29874	33	6	and	and	CCONJ
ajst-29874	33	7	robustness	robustness	NOUN
ajst-29874	33	8	by	by	ADP
ajst-29874	33	9	building	build	VERB
ajst-29874	33	10	multiple	multiple	ADJ
ajst-29874	33	11	decision	decision	NOUN
ajst-29874	33	12	trees	tree	NOUN
ajst-29874	33	13	trained	train	VERB
ajst-29874	33	14	on	on	ADP
ajst-29874	33	15	different	different	ADJ
ajst-29874	33	16	subsamples	subsample	NOUN
ajst-29874	33	17	and	and	CCONJ
ajst-29874	33	18	randomly	randomly	ADV
ajst-29874	33	19	selected	select	VERB
ajst-29874	33	20	features	feature	NOUN
ajst-29874	33	21	.	.	PUNCT
ajst-29874	34	1	it	it	PRON
ajst-29874	34	2	calculates	calculate	VERB
ajst-29874	34	3	feature	feature	NOUN
ajst-29874	34	4	importance	importance	NOUN
ajst-29874	34	5	scores	score	NOUN
ajst-29874	34	6	for	for	ADP
ajst-29874	34	7	dimensionality	dimensionality	NOUN
ajst-29874	34	8	reduction	reduction	NOUN
ajst-29874	34	9	and	and	CCONJ
ajst-29874	34	10	feature	feature	NOUN
ajst-29874	34	11	selection	selection	NOUN
ajst-29874	34	12	.	.	PUNCT
ajst-29874	35	1	among	among	ADP
ajst-29874	35	2	the	the	DET
ajst-29874	35	3	seven	seven	NUM
ajst-29874	35	4	features	feature	NOUN
ajst-29874	35	5	—	—	PUNCT
ajst-29874	35	6	mean	mean	ADJ
ajst-29874	35	7	,	,	PUNCT
ajst-29874	35	8	standard	standard	ADJ
ajst-29874	35	9	deviation	deviation	NOUN
ajst-29874	35	10	,	,	PUNCT
ajst-29874	35	11	skewness	skewness	NOUN
ajst-29874	35	12	,	,	PUNCT
ajst-29874	35	13	amplitude	amplitude	NOUN
ajst-29874	35	14	,	,	PUNCT
ajst-29874	35	15	spectral	spectral	ADJ
ajst-29874	35	16	center	center	NOUN
ajst-29874	35	17	,	,	PUNCT
ajst-29874	35	18	kurtosis	kurtosis	NOUN
ajst-29874	35	19	,	,	PUNCT
ajst-29874	35	20	and	and	CCONJ
ajst-29874	35	21	number	number	NOUN
ajst-29874	35	22	of	of	ADP
ajst-29874	35	23	turning	turn	VERB
ajst-29874	35	24	points	point	NOUN
ajst-29874	35	25	—	—	PUNCT
ajst-29874	35	26	the	the	DET
ajst-29874	35	27	top	top	ADJ
ajst-29874	35	28	three	three	NUM
ajst-29874	35	29	most	most	ADV
ajst-29874	35	30	important	important	ADJ
ajst-29874	35	31	features	feature	NOUN
ajst-29874	35	32	can	can	AUX
ajst-29874	35	33	be	be	AUX
ajst-29874	35	34	selected	select	VERB
ajst-29874	35	35	using	use	VERB
ajst-29874	35	36	random	random	ADJ
ajst-29874	35	37	forest	forest	NOUN
ajst-29874	35	38	.	.	PUNCT
ajst-29874	36	1	importance	importance	NOUN
ajst-29874	36	2	rating	rating	NOUN
ajst-29874	36	3	and	and	CCONJ
ajst-29874	36	4	ranking	ranking	NOUN
ajst-29874	36	5	are	be	AUX
ajst-29874	36	6	shown	show	VERB
ajst-29874	36	7	in	in	ADP
ajst-29874	36	8	figure	figure	NOUN
ajst-29874	36	9	3	3	NUM
ajst-29874	36	10	below	below	ADV
ajst-29874	36	11	:	:	PUNCT
ajst-29874	36	12	figure	figure	VERB
ajst-29874	36	13	3	3	NUM
ajst-29874	36	14	.	.	NOUN
ajst-29874	36	15	feature	feature	NOUN
ajst-29874	36	16	importance	importance	NOUN
ajst-29874	36	17	ranking	rank	VERB
ajst-29874	36	18	176	176	NUM
ajst-29874	36	19	as	as	SCONJ
ajst-29874	36	20	seen	see	VERB
ajst-29874	36	21	from	from	ADP
ajst-29874	36	22	the	the	DET
ajst-29874	36	23	figure	figure	NOUN
ajst-29874	36	24	,	,	PUNCT
ajst-29874	36	25	the	the	DET
ajst-29874	36	26	feature	feature	NOUN
ajst-29874	36	27	importance	importance	NOUN
ajst-29874	36	28	ranking	ranking	NOUN
ajst-29874	36	29	is	be	AUX
ajst-29874	36	30	:	:	PUNCT
ajst-29874	36	31	number	number	NOUN
ajst-29874	36	32	of	of	ADP
ajst-29874	36	33	turning	turn	VERB
ajst-29874	36	34	points	point	NOUN
ajst-29874	36	35	>	>	X
ajst-29874	36	36	kurtosis	kurtosis	NOUN
ajst-29874	36	37	>	>	X
ajst-29874	36	38	skewness	skewness	PROPN
ajst-29874	36	39	>	>	X
ajst-29874	36	40	standard	standard	ADJ
ajst-29874	36	41	deviation	deviation	NOUN
ajst-29874	36	42	>	>	X
ajst-29874	36	43	amplitude	amplitude	NOUN
ajst-29874	36	44	>	>	PUNCT
ajst-29874	36	45	mean	mean	NOUN
ajst-29874	36	46	=	=	SYM
ajst-29874	36	47	spectrum	spectrum	NOUN
ajst-29874	36	48	center	center	NOUN
ajst-29874	36	49	.	.	PUNCT
ajst-29874	37	1	to	to	PART
ajst-29874	37	2	simplify	simplify	VERB
ajst-29874	37	3	the	the	DET
ajst-29874	37	4	model	model	NOUN
ajst-29874	37	5	,	,	PUNCT
ajst-29874	37	6	improve	improve	VERB
ajst-29874	37	7	interpretability	interpretability	NOUN
ajst-29874	37	8	,	,	PUNCT
ajst-29874	37	9	and	and	CCONJ
ajst-29874	37	10	prevent	prevent	VERB
ajst-29874	37	11	overfitting	overfitting	NOUN
ajst-29874	37	12	,	,	PUNCT
ajst-29874	37	13	we	we	PRON
ajst-29874	37	14	eliminated	eliminate	VERB
ajst-29874	37	15	the	the	DET
ajst-29874	37	16	mean	mean	ADJ
ajst-29874	37	17	value	value	NOUN
ajst-29874	37	18	and	and	CCONJ
ajst-29874	37	19	spectrum	spectrum	NOUN
ajst-29874	37	20	center	center	NOUN
ajst-29874	37	21	and	and	CCONJ
ajst-29874	37	22	replaced	replace	VERB
ajst-29874	37	23	waveform	waveform	NOUN
ajst-29874	37	24	features	feature	NOUN
ajst-29874	37	25	with	with	ADP
ajst-29874	37	26	residual	residual	ADJ
ajst-29874	37	27	features	feature	NOUN
ajst-29874	37	28	.	.	PUNCT
ajst-29874	38	1	for	for	ADP
ajst-29874	38	2	a	a	DET
ajst-29874	38	3	universal	universal	ADJ
ajst-29874	38	4	prediction	prediction	NOUN
ajst-29874	38	5	model	model	NOUN
ajst-29874	38	6	across	across	ADP
ajst-29874	38	7	different	different	ADJ
ajst-29874	38	8	material	material	NOUN
ajst-29874	38	9	types	type	NOUN
ajst-29874	38	10	and	and	CCONJ
ajst-29874	38	11	temperatures	temperature	NOUN
ajst-29874	38	12	,	,	PUNCT
ajst-29874	38	13	we	we	PRON
ajst-29874	38	14	included	include	VERB
ajst-29874	38	15	material	material	NOUN
ajst-29874	38	16	type	type	NOUN
ajst-29874	38	17	,	,	PUNCT
ajst-29874	38	18	excitation	excitation	NOUN
ajst-29874	38	19	wave	wave	NOUN
ajst-29874	38	20	frequency	frequency	NOUN
ajst-29874	38	21	,	,	PUNCT
ajst-29874	38	22	peak	peak	NOUN
ajst-29874	38	23	value	value	NOUN
ajst-29874	38	24	,	,	PUNCT
ajst-29874	38	25	and	and	CCONJ
ajst-29874	38	26	temperature	temperature	NOUN
ajst-29874	38	27	,	,	PUNCT
ajst-29874	38	28	finally	finally	ADV
ajst-29874	38	29	selecting	select	VERB
ajst-29874	38	30	nine	nine	NUM
ajst-29874	38	31	features	feature	NOUN
ajst-29874	38	32	as	as	ADP
ajst-29874	38	33	inputs	input	NOUN
ajst-29874	38	34	.	.	PUNCT
ajst-29874	39	1	for	for	ADP
ajst-29874	39	2	material	material	NOUN
ajst-29874	39	3	characteristics	characteristic	NOUN
ajst-29874	39	4	,	,	PUNCT
ajst-29874	39	5	we	we	PRON
ajst-29874	39	6	used	use	VERB
ajst-29874	39	7	onehot	onehot	NOUN
ajst-29874	39	8	encoding	encoding	NOUN
ajst-29874	39	9	to	to	PART
ajst-29874	39	10	convert	convert	VERB
ajst-29874	39	11	categorical	categorical	ADJ
ajst-29874	39	12	variables	variable	NOUN
ajst-29874	39	13	into	into	ADP
ajst-29874	39	14	numerical	numerical	ADJ
ajst-29874	39	15	representations	representation	NOUN
ajst-29874	39	16	,	,	PUNCT
ajst-29874	39	17	avoiding	avoid	VERB
ajst-29874	39	18	issues	issue	NOUN
ajst-29874	39	19	with	with	ADP
ajst-29874	39	20	integer	integer	NOUN
ajst-29874	39	21	encoding	encoding	NOUN
ajst-29874	39	22	.	.	PUNCT
ajst-29874	40	1	4	4	X
ajst-29874	40	2	.	.	X
ajst-29874	40	3	establishment	establishment	NOUN
ajst-29874	40	4	of	of	ADP
ajst-29874	40	5	core	core	NOUN
ajst-29874	40	6	loss	loss	NOUN
ajst-29874	40	7	prediction	prediction	NOUN
ajst-29874	40	8	model	model	NOUN
ajst-29874	40	9	based	base	VERB
ajst-29874	40	10	on	on	ADP
ajst-29874	40	11	fcnn	fcnn	NOUN
ajst-29874	40	12	in	in	ADP
ajst-29874	40	13	order	order	NOUN
ajst-29874	40	14	to	to	PART
ajst-29874	40	15	construct	construct	VERB
ajst-29874	40	16	a	a	DET
ajst-29874	40	17	fully	fully	ADV
ajst-29874	40	18	connected	connected	ADJ
ajst-29874	40	19	neural	neural	ADJ
ajst-29874	40	20	network	network	NOUN
ajst-29874	40	21	(	(	PUNCT
ajst-29874	40	22	fcnn	fcnn	PROPN
ajst-29874	40	23	)	)	PUNCT
ajst-29874	40	24	core	core	NOUN
ajst-29874	40	25	loss	loss	NOUN
ajst-29874	40	26	prediction	prediction	NOUN
ajst-29874	40	27	model	model	NOUN
ajst-29874	40	28	,	,	PUNCT
ajst-29874	40	29	we	we	PRON
ajst-29874	40	30	use	use	VERB
ajst-29874	40	31	data	datum	NOUN
ajst-29874	40	32	with	with	ADP
ajst-29874	40	33	nine	nine	NUM
ajst-29874	40	34	features	feature	NOUN
ajst-29874	40	35	to	to	ADP
ajst-29874	40	36	model	model	NOUN
ajst-29874	40	37	.	.	PUNCT
ajst-29874	41	1	fcnn	fcnn	PROPN
ajst-29874	41	2	is	be	AUX
ajst-29874	41	3	a	a	DET
ajst-29874	41	4	classical	classical	ADJ
ajst-29874	41	5	neural	neural	ADJ
ajst-29874	41	6	network	network	NOUN
ajst-29874	41	7	architecture	architecture	NOUN
ajst-29874	41	8	,	,	PUNCT
ajst-29874	41	9	which	which	PRON
ajst-29874	41	10	consists	consist	VERB
ajst-29874	41	11	of	of	ADP
ajst-29874	41	12	an	an	DET
ajst-29874	41	13	input	input	NOUN
ajst-29874	41	14	layer	layer	NOUN
ajst-29874	41	15	(	(	PUNCT
ajst-29874	41	16	nine	nine	NUM
ajst-29874	41	17	neurons	neuron	NOUN
ajst-29874	41	18	corresponding	correspond	VERB
ajst-29874	41	19	to	to	ADP
ajst-29874	41	20	independent	independent	ADJ
ajst-29874	41	21	variables	variable	NOUN
ajst-29874	41	22	such	such	ADJ
ajst-29874	41	23	as	as	ADP
ajst-29874	41	24	frequency	frequency	NOUN
ajst-29874	41	25	,	,	PUNCT
ajst-29874	41	26	peak	peak	NOUN
ajst-29874	41	27	value	value	NOUN
ajst-29874	41	28	,	,	PUNCT
ajst-29874	41	29	and	and	CCONJ
ajst-29874	41	30	temperature	temperature	NOUN
ajst-29874	41	31	)	)	PUNCT
ajst-29874	41	32	,	,	PUNCT
ajst-29874	41	33	one	one	NUM
ajst-29874	41	34	or	or	CCONJ
ajst-29874	41	35	more	more	ADV
ajst-29874	41	36	hidden	hidden	ADJ
ajst-29874	41	37	layers	layer	NOUN
ajst-29874	41	38	,	,	PUNCT
ajst-29874	41	39	and	and	CCONJ
ajst-29874	41	40	an	an	DET
ajst-29874	41	41	output	output	NOUN
ajst-29874	41	42	layer	layer	NOUN
ajst-29874	41	43	(	(	PUNCT
ajst-29874	41	44	one	one	NUM
ajst-29874	41	45	neuron	neuron	NOUN
ajst-29874	41	46	corresponding	correspond	VERB
ajst-29874	41	47	to	to	ADP
ajst-29874	41	48	the	the	DET
ajst-29874	41	49	predicted	predict	VERB
ajst-29874	41	50	value	value	NOUN
ajst-29874	41	51	of	of	ADP
ajst-29874	41	52	core	core	ADJ
ajst-29874	41	53	loss	loss	NOUN
ajst-29874	41	54	)	)	PUNCT
ajst-29874	41	55	.	.	PUNCT
ajst-29874	42	1	its	its	PRON
ajst-29874	42	2	basic	basic	ADJ
ajst-29874	42	3	principles	principle	NOUN
ajst-29874	42	4	include	include	VERB
ajst-29874	42	5	linear	linear	ADJ
ajst-29874	42	6	transformation	transformation	NOUN
ajst-29874	42	7	,	,	PUNCT
ajst-29874	42	8	nonlinear	nonlinear	ADJ
ajst-29874	42	9	activation	activation	NOUN
ajst-29874	42	10	function	function	NOUN
ajst-29874	42	11	,	,	PUNCT
ajst-29874	42	12	and	and	CCONJ
ajst-29874	42	13	backpropagation	backpropagation	NOUN
ajst-29874	42	14	training	training	NOUN
ajst-29874	42	15	algorithm	algorithm	NOUN
ajst-29874	42	16	.	.	PUNCT
ajst-29874	43	1	given	give	VERB
ajst-29874	43	2	the	the	DET
ajst-29874	43	3	large	large	ADJ
ajst-29874	43	4	amount	amount	NOUN
ajst-29874	43	5	of	of	ADP
ajst-29874	43	6	data	datum	NOUN
ajst-29874	43	7	,	,	PUNCT
ajst-29874	43	8	numerous	numerous	ADJ
ajst-29874	43	9	influencing	influence	VERB
ajst-29874	43	10	variables	variable	NOUN
ajst-29874	43	11	,	,	PUNCT
ajst-29874	43	12	and	and	CCONJ
ajst-29874	43	13	complex	complex	ADJ
ajst-29874	43	14	nonlinear	nonlinear	ADJ
ajst-29874	43	15	relationships	relationship	NOUN
ajst-29874	43	16	in	in	ADP
ajst-29874	43	17	this	this	DET
ajst-29874	43	18	study	study	NOUN
ajst-29874	43	19	,	,	PUNCT
ajst-29874	43	20	we	we	PRON
ajst-29874	43	21	adopted	adopt	VERB
ajst-29874	43	22	the	the	DET
ajst-29874	43	23	fcnn	fcnn	ADJ
ajst-29874	43	24	structure	structure	NOUN
ajst-29874	43	25	containing	contain	VERB
ajst-29874	43	26	three	three	NUM
ajst-29874	43	27	hidden	hidden	ADJ
ajst-29874	43	28	layers	layer	NOUN
ajst-29874	43	29	,	,	PUNCT
ajst-29874	43	30	with	with	ADP
ajst-29874	43	31	the	the	DET
ajst-29874	43	32	number	number	NOUN
ajst-29874	43	33	of	of	ADP
ajst-29874	43	34	neurons	neuron	NOUN
ajst-29874	43	35	being	be	AUX
ajst-29874	43	36	64	64	NUM
ajst-29874	43	37	,	,	PUNCT
ajst-29874	43	38	32	32	NUM
ajst-29874	43	39	,	,	PUNCT
ajst-29874	43	40	and	and	CCONJ
ajst-29874	43	41	64	64	NUM
ajst-29874	44	1	[	[	SYM
ajst-29874	44	2	5	5	NUM
ajst-29874	44	3	]	]	PUNCT
ajst-29874	44	4	,	,	PUNCT
ajst-29874	44	5	respectively	respectively	ADV
ajst-29874	44	6	,	,	PUNCT
ajst-29874	44	7	as	as	SCONJ
ajst-29874	44	8	shown	show	VERB
ajst-29874	44	9	in	in	ADP
ajst-29874	44	10	figure	figure	NOUN
ajst-29874	44	11	4	4	NUM
ajst-29874	44	12	.	.	PUNCT
ajst-29874	45	1	this	this	DET
ajst-29874	45	2	configuration	configuration	NOUN
ajst-29874	45	3	avoids	avoid	VERB
ajst-29874	45	4	overfitting	overfitte	VERB
ajst-29874	45	5	while	while	SCONJ
ajst-29874	45	6	maintaining	maintain	VERB
ajst-29874	45	7	the	the	DET
ajst-29874	45	8	complexity	complexity	NOUN
ajst-29874	45	9	of	of	ADP
ajst-29874	45	10	the	the	DET
ajst-29874	45	11	model	model	NOUN
ajst-29874	45	12	,	,	PUNCT
ajst-29874	45	13	helping	help	VERB
ajst-29874	45	14	to	to	PART
ajst-29874	45	15	capture	capture	VERB
ajst-29874	45	16	important	important	ADJ
ajst-29874	45	17	features	feature	NOUN
ajst-29874	45	18	in	in	ADP
ajst-29874	45	19	the	the	DET
ajst-29874	45	20	data	datum	NOUN
ajst-29874	45	21	.	.	PUNCT
ajst-29874	46	1	figure	figure	VERB
ajst-29874	46	2	4	4	NUM
ajst-29874	46	3	.	.	PUNCT
ajst-29874	47	1	network	network	NOUN
ajst-29874	47	2	structure	structure	NOUN
ajst-29874	47	3	the	the	DET
ajst-29874	47	4	specific	specific	ADJ
ajst-29874	47	5	training	training	NOUN
ajst-29874	47	6	process	process	NOUN
ajst-29874	47	7	is	be	AUX
ajst-29874	47	8	:	:	PUNCT
ajst-29874	47	9	1	1	X
ajst-29874	47	10	.	.	X
ajst-29874	47	11	enter	enter	VERB
ajst-29874	47	12	layer	layer	NOUN
ajst-29874	47	13	to	to	PART
ajst-29874	47	14	hide	hide	VERB
ajst-29874	47	15	layer	layer	NOUN
ajst-29874	47	16	1	1	NUM
ajst-29874	47	17	1.1	1.1	NUM
ajst-29874	47	18	linear	linear	NOUN
ajst-29874	47	19	transformation	transformation	NOUN
ajst-29874	47	20	1	1	NUM
ajst-29874	47	21	1	1	NUM
ajst-29874	47	22	(	(	PUNCT
ajst-29874	47	23	1)bz	1)bz	PROPN
ajst-29874	47	24	w	w	PROPN
ajst-29874	47	25	x	x	PROPN
ajst-29874	47	26			NUM
ajst-29874	47	27			ADV
ajst-29874	47	28	（	（	PROPN
ajst-29874	47	29	）	）	PUNCT
ajst-29874	48	1	（	（	PUNCT
ajst-29874	48	2	）	）	PUNCT
ajst-29874	49	1	among	among	ADP
ajst-29874	49	2	them	they	PRON
ajst-29874	49	3	,	,	PUNCT
ajst-29874	49	4	(	(	PUNCT
ajst-29874	49	5	1)w	1)w	NUM
ajst-29874	49	6	for	for	ADP
ajst-29874	49	7	the	the	DET
ajst-29874	49	8	weight	weight	NOUN
ajst-29874	49	9	matrix	matrix	NOUN
ajst-29874	49	10	,	,	PUNCT
ajst-29874	49	11	each	each	DET
ajst-29874	49	12	weight	weight	NOUN
ajst-29874	49	13	representation	representation	NOUN
ajst-29874	49	14	input	input	NOUN
ajst-29874	49	15	feature	feature	NOUN
ajst-29874	49	16	jx	jx	PROPN
ajst-29874	49	17	to	to	ADP
ajst-29874	49	18	what	what	DET
ajst-29874	49	19	extent	extent	NOUN
ajst-29874	49	20	it	it	PRON
ajst-29874	49	21	affects	affect	VERB
ajst-29874	49	22	the	the	DET
ajst-29874	49	23	ihidden	ihidden	PROPN
ajst-29874	49	24	neuron	neuron	PROPN
ajst-29874	49	25	,	,	PUNCT
ajst-29874	49	26	the	the	DET
ajst-29874	49	27	shape	shape	NOUN
ajst-29874	49	28	is	be	AUX
ajst-29874	49	29	(	(	PUNCT
ajst-29874	49	30	n1,9	n1,9	PROPN
ajst-29874	49	31	)	)	PUNCT
ajst-29874	49	32	,	,	PUNCT
ajst-29874	49	33	which	which	PRON
ajst-29874	49	34	is	be	AUX
ajst-29874	49	35	the	the	DET
ajst-29874	49	36	input	input	NOUN
ajst-29874	49	37	vector	vector	NOUN
ajst-29874	49	38	of	of	ADP
ajst-29874	49	39	9	9	NUM
ajst-29874	49	40	influence	influence	NOUN
ajst-29874	49	41	features	feature	VERB
ajst-29874	49	42	such	such	ADJ
ajst-29874	49	43	as	as	ADP
ajst-29874	49	44	frequency	frequency	NOUN
ajst-29874	49	45	,	,	PUNCT
ajst-29874	49	46	peak	peak	NOUN
ajst-29874	49	47	value	value	NOUN
ajst-29874	49	48	and	and	CCONJ
ajst-29874	49	49	temperature	temperature	NOUN
ajst-29874	49	50	,	,	PUNCT
ajst-29874	49	51	and	and	CCONJ
ajst-29874	49	52	the	the	DET
ajst-29874	49	53	shape	shape	NOUN
ajst-29874	49	54	is	be	AUX
ajst-29874	49	55	(	(	PUNCT
ajst-29874	49	56	9,1	9,1	NUM
ajst-29874	49	57	)	)	PUNCT
ajst-29874	49	58	,	,	PUNCT
ajst-29874	49	59	(	(	PUNCT
ajst-29874	49	60	1)b	1)b	X
ajst-29874	49	61	is	be	AUX
ajst-29874	49	62	a	a	DET
ajst-29874	49	63	bias	bias	NOUN
ajst-29874	49	64	vector	vector	NOUN
ajst-29874	49	65	of	of	ADP
ajst-29874	49	66	shape	shape	NOUN
ajst-29874	49	67	(	(	PUNCT
ajst-29874	49	68	n1,1	n1,1	NOUN
ajst-29874	49	69	)	)	PUNCT
ajst-29874	49	70	.	.	PUNCT
ajst-29874	50	1	1.2	1.2	NUM
ajst-29874	50	2	nonlinear	nonlinear	ADJ
ajst-29874	50	3	activation	activation	NOUN
ajst-29874	50	4	:	:	PUNCT
ajst-29874	50	5	for	for	ADP
ajst-29874	50	6	this	this	DET
ajst-29874	50	7	task	task	NOUN
ajst-29874	50	8	,	,	PUNCT
ajst-29874	50	9	we	we	PRON
ajst-29874	50	10	chose	choose	VERB
ajst-29874	50	11	the	the	DET
ajst-29874	50	12	relu	relu	NOUN
ajst-29874	50	13	activation	activation	NOUN
ajst-29874	50	14	function	function	VERB
ajst-29874	50	15	to	to	PART
ajst-29874	50	16	introduce	introduce	VERB
ajst-29874	50	17	nonlinearities	nonlinearitie	NOUN
ajst-29874	50	18	:	:	PUNCT
ajst-29874	50	19	1	1	NUM
ajst-29874	50	20	(	(	PUNCT
ajst-29874	50	21	1)max(0	1)max(0	NUM
ajst-29874	50	22	,	,	PUNCT
ajst-29874	50	23	)	)	PUNCT
ajst-29874	50	24	a	a	DET
ajst-29874	50	25	z	z	PROPN
ajst-29874	50	26	（	（	X
ajst-29874	50	27	）	）	PUNCT
ajst-29874	50	28	make	make	VERB
ajst-29874	50	29	sure	sure	ADJ
ajst-29874	50	30	only	only	ADV
ajst-29874	50	31	when	when	SCONJ
ajst-29874	50	32	linear	linear	ADJ
ajst-29874	50	33	combinations	combination	NOUN
ajst-29874	50	34	are	be	AUX
ajst-29874	50	35	available	available	ADJ
ajst-29874	50	36	when	when	SCONJ
ajst-29874	50	37	it	it	PRON
ajst-29874	50	38	is	be	AUX
ajst-29874	50	39	positive	positive	ADJ
ajst-29874	50	40	neurons	neuron	NOUN
ajst-29874	50	41	will	will	AUX
ajst-29874	50	42	be	be	AUX
ajst-29874	50	43	activated	activate	VERB
ajst-29874	50	44	.	.	PUNCT
ajst-29874	51	1	this	this	DET
ajst-29874	51	2	task	task	NOUN
ajst-29874	51	3	reflects	reflect	VERB
ajst-29874	51	4	the	the	DET
ajst-29874	51	5	influence	influence	NOUN
ajst-29874	51	6	of	of	ADP
ajst-29874	51	7	certain	certain	ADJ
ajst-29874	51	8	feature	feature	NOUN
ajst-29874	51	9	combinations	combination	NOUN
ajst-29874	51	10	on	on	ADP
ajst-29874	51	11	loss	loss	NOUN
ajst-29874	51	12	under	under	ADP
ajst-29874	51	13	specific	specific	ADJ
ajst-29874	51	14	conditions	condition	NOUN
ajst-29874	51	15	.	.	PUNCT
ajst-29874	52	1	2	2	X
ajst-29874	52	2	.	.	X
ajst-29874	52	3	hide	hide	VERB
ajst-29874	52	4	layers	layer	NOUN
ajst-29874	52	5	1	1	NUM
ajst-29874	52	6	to	to	PART
ajst-29874	52	7	2	2	NUM
ajst-29874	52	8	3	3	NUM
ajst-29874	52	9	.	.	PUNCT
ajst-29874	52	10	hide	hide	VERB
ajst-29874	52	11	layers	layer	NOUN
ajst-29874	52	12	2	2	NUM
ajst-29874	52	13	to	to	PART
ajst-29874	52	14	hide	hide	VERB
ajst-29874	52	15	layers	layer	NOUN
ajst-29874	52	16	3	3	NUM
ajst-29874	52	17	the	the	DET
ajst-29874	52	18	principle	principle	NOUN
ajst-29874	52	19	is	be	AUX
ajst-29874	52	20	the	the	DET
ajst-29874	52	21	same	same	ADJ
ajst-29874	52	22	as	as	ADP
ajst-29874	52	23	above	above	ADV
ajst-29874	52	24	;	;	PUNCT
ajst-29874	52	25	both	both	PRON
ajst-29874	52	26	are	be	AUX
ajst-29874	52	27	linear	linear	ADJ
ajst-29874	52	28	transformation	transformation	NOUN
ajst-29874	52	29	and	and	CCONJ
ajst-29874	52	30	nonlinear	nonlinear	ADJ
ajst-29874	52	31	activation	activation	NOUN
ajst-29874	52	32	processes	process	NOUN
ajst-29874	52	33	.	.	PUNCT
ajst-29874	53	1	the	the	DET
ajst-29874	53	2	second	second	ADJ
ajst-29874	53	3	layer	layer	NOUN
ajst-29874	53	4	of	of	ADP
ajst-29874	53	5	neurons	neuron	NOUN
ajst-29874	53	6	combines	combine	VERB
ajst-29874	53	7	the	the	DET
ajst-29874	53	8	output	output	NOUN
ajst-29874	53	9	of	of	ADP
ajst-29874	53	10	the	the	DET
ajst-29874	53	11	first	first	ADJ
ajst-29874	53	12	layer	layer	NOUN
ajst-29874	53	13	,	,	PUNCT
ajst-29874	53	14	and	and	CCONJ
ajst-29874	53	15	the	the	DET
ajst-29874	53	16	third	third	ADJ
ajst-29874	53	17	layer	layer	NOUN
ajst-29874	53	18	of	of	ADP
ajst-29874	53	19	neurons	neuron	NOUN
ajst-29874	53	20	combines	combine	VERB
ajst-29874	53	21	the	the	DET
ajst-29874	53	22	output	output	NOUN
ajst-29874	53	23	of	of	ADP
ajst-29874	53	24	the	the	DET
ajst-29874	53	25	second	second	ADJ
ajst-29874	53	26	layer	layer	NOUN
ajst-29874	53	27	to	to	PART
ajst-29874	53	28	extract	extract	VERB
ajst-29874	53	29	higher	high	ADJ
ajst-29874	53	30	-	-	PUNCT
ajst-29874	53	31	level	level	NOUN
ajst-29874	53	32	features	feature	NOUN
ajst-29874	53	33	,	,	PUNCT
ajst-29874	53	34	which	which	PRON
ajst-29874	53	35	is	be	AUX
ajst-29874	53	36	the	the	DET
ajst-29874	53	37	key	key	NOUN
ajst-29874	53	38	to	to	ADP
ajst-29874	53	39	capturing	capture	VERB
ajst-29874	53	40	complex	complex	ADJ
ajst-29874	53	41	nonlinear	nonlinear	ADJ
ajst-29874	53	42	relationships	relationship	NOUN
ajst-29874	53	43	.	.	PUNCT
ajst-29874	54	1	4	4	X
ajst-29874	54	2	.	.	X
ajst-29874	54	3	hide	hide	VERB
ajst-29874	54	4	layer	layer	NOUN
ajst-29874	54	5	3	3	NUM
ajst-29874	54	6	to	to	ADP
ajst-29874	54	7	the	the	DET
ajst-29874	54	8	output	output	NOUN
ajst-29874	54	9	layer	layer	NOUN
ajst-29874	54	10	4.1	4.1	NUM
ajst-29874	54	11	linear	linear	NOUN
ajst-29874	54	12	transformation	transformation	NOUN
ajst-29874	54	13	(	(	PUNCT
ajst-29874	54	14	3	3	NUM
ajst-29874	54	15	)	)	PUNCT
ajst-29874	54	16	(	(	PUNCT
ajst-29874	54	17	3	3	X
ajst-29874	54	18	)	)	PUNCT
ajst-29874	54	19	(	(	PUNCT
ajst-29874	54	20	2	2	NUM
ajst-29874	54	21	)	)	PUNCT
ajst-29874	54	22	(	(	PUNCT
ajst-29874	54	23	3)z	3)z	NUM
ajst-29874	54	24	w	w	ADP
ajst-29874	54	25	a	a	DET
ajst-29874	54	26	b	b	NOUN
ajst-29874	54	27			PROPN
ajst-29874	54	28			VERB
ajst-29874	54	29	among	among	ADP
ajst-29874	54	30	them	they	PRON
ajst-29874	54	31	(	(	PUNCT
ajst-29874	54	32	3)w	3)w	NOUN
ajst-29874	54	33	is	be	AUX
ajst-29874	54	34	a	a	DET
ajst-29874	54	35	weight	weight	NOUN
ajst-29874	54	36	matrix	matrix	NOUN
ajst-29874	54	37	of	of	ADP
ajst-29874	54	38	shape	shape	NOUN
ajst-29874	54	39	(	(	PUNCT
ajst-29874	54	40	1,n2	1,n2	NUM
ajst-29874	54	41	)	)	PUNCT
ajst-29874	54	42	that	that	PRON
ajst-29874	54	43	maps	map	VERB
ajst-29874	54	44	the	the	DET
ajst-29874	54	45	output	output	NOUN
ajst-29874	54	46	of	of	ADP
ajst-29874	54	47	hidden	hide	VERB
ajst-29874	54	48	layer	layer	NOUN
ajst-29874	54	49	3	3	NUM
ajst-29874	54	50	to	to	ADP
ajst-29874	54	51	the	the	DET
ajst-29874	54	52	output	output	NOUN
ajst-29874	54	53	layer	layer	NOUN
ajst-29874	54	54	.	.	PUNCT
ajst-29874	55	1	4.2	4.2	NUM
ajst-29874	55	2	output	output	NOUN
ajst-29874	55	3	(	(	PUNCT
ajst-29874	55	4	3)ŷ	3)ŷ	NUM
ajst-29874	55	5	z	z	NOUN
ajst-29874	55	6	the	the	DET
ajst-29874	55	7	output	output	NOUN
ajst-29874	55	8	is	be	AUX
ajst-29874	55	9	a	a	DET
ajst-29874	55	10	predicted	predict	VERB
ajst-29874	55	11	value	value	NOUN
ajst-29874	55	12	of	of	ADP
ajst-29874	55	13	the	the	DET
ajst-29874	55	14	core	core	NOUN
ajst-29874	55	15	loss	loss	NOUN
ajst-29874	55	16	directly	directly	ADV
ajst-29874	55	17	related	relate	VERB
ajst-29874	55	18	to	to	ADP
ajst-29874	55	19	the	the	DET
ajst-29874	55	20	previous	previous	ADJ
ajst-29874	55	21	feature	feature	NOUN
ajst-29874	55	22	combination	combination	NOUN
ajst-29874	55	23	.	.	PUNCT
ajst-29874	56	1	5	5	X
ajst-29874	56	2	.	.	X
ajst-29874	56	3	calculate	calculate	VERB
ajst-29874	56	4	the	the	DET
ajst-29874	56	5	loss	loss	NOUN
ajst-29874	56	6	function	function	NOUN
ajst-29874	56	7	and	and	CCONJ
ajst-29874	56	8	backpropagation	backpropagation	NOUN
ajst-29874	56	9	5.1	5.1	NUM
ajst-29874	56	10	calculate	calculate	NOUN
ajst-29874	56	11	losses	loss	NOUN
ajst-29874	56	12	using	use	VERB
ajst-29874	56	13	mean	mean	ADJ
ajst-29874	56	14	square	square	ADJ
ajst-29874	56	15	error	error	NOUN
ajst-29874	56	16	(	(	PUNCT
ajst-29874	56	17	mse	mse	NOUN
ajst-29874	56	18	)	)	PUNCT
ajst-29874	56	19	:	:	PUNCT
ajst-29874	57	1	2	2	NUM
ajst-29874	57	2	1	1	NUM
ajst-29874	57	3	1	1	NUM
ajst-29874	57	4	n	n	NUM
ajst-29874	57	5	i	i	PRON
ajst-29874	58	1	i	i	PRON
ajst-29874	58	2	l	l	VERB
ajst-29874	58	3	y	y	PROPN
ajst-29874	58	4	n	n	CCONJ
ajst-29874	58	5			PROPN
ajst-29874	59	1			PROPN
ajst-29874	59	2			NOUN
ajst-29874	59	3	the	the	DET
ajst-29874	59	4	loss	loss	NOUN
ajst-29874	59	5	function	function	NOUN
ajst-29874	59	6	quantifies	quantify	VERB
ajst-29874	59	7	the	the	DET
ajst-29874	59	8	gap	gap	NOUN
ajst-29874	59	9	between	between	ADP
ajst-29874	59	10	the	the	DET
ajst-29874	59	11	loss	loss	NOUN
ajst-29874	59	12	predicted	predict	VERB
ajst-29874	59	13	by	by	ADP
ajst-29874	59	14	the	the	DET
ajst-29874	59	15	model	model	NOUN
ajst-29874	59	16	and	and	CCONJ
ajst-29874	59	17	the	the	DET
ajst-29874	59	18	actual	actual	ADJ
ajst-29874	59	19	loss	loss	NOUN
ajst-29874	59	20	to	to	PART
ajst-29874	59	21	improve	improve	VERB
ajst-29874	59	22	the	the	DET
ajst-29874	59	23	prediction	prediction	NOUN
ajst-29874	59	24	accuracy	accuracy	NOUN
ajst-29874	59	25	by	by	ADP
ajst-29874	59	26	minimizing	minimize	VERB
ajst-29874	59	27	it	it	PRON
ajst-29874	59	28	.	.	PUNCT
ajst-29874	60	1	5.2	5.2	NUM
ajst-29874	60	2	back	back	ADJ
ajst-29874	60	3	propagation	propagation	NOUN
ajst-29874	60	4	backpropagation	backpropagation	NOUN
ajst-29874	60	5	calculates	calculate	VERB
ajst-29874	60	6	the	the	DET
ajst-29874	60	7	gradient	gradient	NOUN
ajst-29874	60	8	of	of	ADP
ajst-29874	60	9	each	each	DET
ajst-29874	60	10	layer	layer	NOUN
ajst-29874	60	11	by	by	ADP
ajst-29874	60	12	the	the	DET
ajst-29874	60	13	chain	chain	NOUN
ajst-29874	60	14	rule	rule	NOUN
ajst-29874	60	15	:	:	PUNCT
ajst-29874	60	16	(	(	PUNCT
ajst-29874	60	17	3	3	X
ajst-29874	60	18	)	)	PUNCT
ajst-29874	60	19	(	(	PUNCT
ajst-29874	60	20	3	3	X
ajst-29874	60	21	)	)	PUNCT
ajst-29874	60	22	l	l	NOUN
ajst-29874	61	1	l	l	NOUN
ajst-29874	62	1	y	y	PROPN
ajst-29874	62	2	w	w	PROPN
ajst-29874	62	3	y	y	PROPN
ajst-29874	62	4	w	w	PROPN
ajst-29874	62	5			ADJ
ajst-29874	62	6			ADJ
ajst-29874	62	7			ADJ
ajst-29874	62	8			ADJ
ajst-29874	62	9			ADJ
ajst-29874	62	10			ADJ
ajst-29874	62	11			ADJ
ajst-29874	62	12			ADJ
ajst-29874	62	13	177	177	NUM
ajst-29874	62	14	update	update	NOUN
ajst-29874	62	15	weights	weight	NOUN
ajst-29874	62	16	and	and	CCONJ
ajst-29874	62	17	bias	bias	NOUN
ajst-29874	62	18	:	:	PUNCT
ajst-29874	62	19	(	(	PUNCT
ajst-29874	62	20	)	)	PUNCT
ajst-29874	62	21	(	(	PUNCT
ajst-29874	62	22	)	)	PUNCT
ajst-29874	62	23	(	(	PUNCT
ajst-29874	62	24	)	)	PUNCT
ajst-29874	62	25	l	l	NOUN
ajst-29874	62	26	l	l	NOUN
ajst-29874	62	27	l	l	X
ajst-29874	62	28	l	l	X
ajst-29874	62	29	w	w	PROPN
ajst-29874	62	30	w	w	PROPN
ajst-29874	62	31	w	w	PROPN
ajst-29874	62	32			PROPN
ajst-29874	62	33			X
ajst-29874	62	34			PROPN
ajst-29874	62	35			PROPN
ajst-29874	62	36	this	this	DET
ajst-29874	62	37	process	process	NOUN
ajst-29874	62	38	adjusts	adjust	VERB
ajst-29874	62	39	the	the	DET
ajst-29874	62	40	weights	weight	NOUN
ajst-29874	62	41	according	accord	VERB
ajst-29874	62	42	to	to	ADP
ajst-29874	62	43	the	the	DET
ajst-29874	62	44	sensitivity	sensitivity	NOUN
ajst-29874	62	45	of	of	ADP
ajst-29874	62	46	the	the	DET
ajst-29874	62	47	losses	loss	NOUN
ajst-29874	62	48	to	to	ADP
ajst-29874	62	49	each	each	DET
ajst-29874	62	50	weight	weight	NOUN
ajst-29874	62	51	,	,	PUNCT
ajst-29874	62	52	ensuring	ensure	VERB
ajst-29874	62	53	that	that	SCONJ
ajst-29874	62	54	the	the	DET
ajst-29874	62	55	model	model	NOUN
ajst-29874	62	56	can	can	AUX
ajst-29874	62	57	predict	predict	VERB
ajst-29874	62	58	more	more	ADV
ajst-29874	62	59	accurately	accurately	ADV
ajst-29874	62	60	.	.	PUNCT
ajst-29874	63	1	after	after	ADP
ajst-29874	63	2	repeated	repeat	VERB
ajst-29874	63	3	backpropagation	backpropagation	NOUN
ajst-29874	63	4	and	and	CCONJ
ajst-29874	63	5	weight	weight	NOUN
ajst-29874	63	6	updating	updating	NOUN
ajst-29874	63	7	,	,	PUNCT
ajst-29874	63	8	the	the	DET
ajst-29874	63	9	model	model	NOUN
ajst-29874	63	10	can	can	AUX
ajst-29874	63	11	learn	learn	VERB
ajst-29874	63	12	the	the	DET
ajst-29874	63	13	complex	complex	ADJ
ajst-29874	63	14	nonlinear	nonlinear	ADJ
ajst-29874	63	15	relationship	relationship	NOUN
ajst-29874	63	16	between	between	ADP
ajst-29874	63	17	the	the	DET
ajst-29874	63	18	independent	independent	ADJ
ajst-29874	63	19	variable	variable	NOUN
ajst-29874	63	20	and	and	CCONJ
ajst-29874	63	21	the	the	DET
ajst-29874	63	22	dependent	dependent	ADJ
ajst-29874	63	23	variable	variable	NOUN
ajst-29874	63	24	.	.	PUNCT
ajst-29874	64	1	5	5	X
ajst-29874	64	2	.	.	X
ajst-29874	64	3	acknowledgements	acknowledgement	NOUN
ajst-29874	64	4	because	because	SCONJ
ajst-29874	64	5	the	the	DET
ajst-29874	64	6	numerical	numerical	ADJ
ajst-29874	64	7	scale	scale	NOUN
ajst-29874	64	8	of	of	ADP
ajst-29874	64	9	magnetic	magnetic	ADJ
ajst-29874	64	10	core	core	NOUN
ajst-29874	64	11	loss	loss	NOUN
ajst-29874	64	12	varies	vary	VERB
ajst-29874	64	13	greatly	greatly	ADV
ajst-29874	64	14	,	,	PUNCT
ajst-29874	64	15	direct	direct	ADJ
ajst-29874	64	16	training	training	NOUN
ajst-29874	64	17	may	may	AUX
ajst-29874	64	18	cause	cause	VERB
ajst-29874	64	19	the	the	DET
ajst-29874	64	20	model	model	NOUN
ajst-29874	64	21	to	to	PART
ajst-29874	64	22	be	be	AUX
ajst-29874	64	23	difficult	difficult	ADJ
ajst-29874	64	24	to	to	PART
ajst-29874	64	25	converge	converge	VERB
ajst-29874	64	26	.	.	PUNCT
ajst-29874	65	1	therefore	therefore	ADV
ajst-29874	65	2	,	,	PUNCT
ajst-29874	65	3	we	we	PRON
ajst-29874	65	4	input	input	VERB
ajst-29874	65	5	the	the	DET
ajst-29874	65	6	model	model	NOUN
ajst-29874	65	7	training	training	NOUN
ajst-29874	65	8	after	after	ADP
ajst-29874	65	9	normalizing	normalize	VERB
ajst-29874	65	10	the	the	DET
ajst-29874	65	11	independent	independent	ADJ
ajst-29874	65	12	variables	variable	NOUN
ajst-29874	65	13	and	and	CCONJ
ajst-29874	65	14	dependent	dependent	ADJ
ajst-29874	65	15	variables	variable	NOUN
ajst-29874	65	16	and	and	CCONJ
ajst-29874	65	17	reverse	reverse	NOUN
ajst-29874	65	18	normalize	normalize	VERB
ajst-29874	65	19	the	the	DET
ajst-29874	65	20	model	model	NOUN
ajst-29874	65	21	output	output	NOUN
ajst-29874	65	22	results	result	NOUN
ajst-29874	65	23	in	in	ADP
ajst-29874	65	24	the	the	DET
ajst-29874	65	25	prediction	prediction	NOUN
ajst-29874	65	26	stage	stage	NOUN
ajst-29874	65	27	to	to	PART
ajst-29874	65	28	restore	restore	VERB
ajst-29874	65	29	the	the	DET
ajst-29874	65	30	original	original	ADJ
ajst-29874	65	31	value	value	NOUN
ajst-29874	65	32	range	range	NOUN
ajst-29874	65	33	for	for	ADP
ajst-29874	65	34	comparison	comparison	NOUN
ajst-29874	65	35	with	with	ADP
ajst-29874	65	36	real	real	ADJ
ajst-29874	65	37	data	datum	NOUN
ajst-29874	65	38	or	or	CCONJ
ajst-29874	65	39	practical	practical	ADJ
ajst-29874	65	40	application	application	NOUN
ajst-29874	65	41	.	.	PUNCT
ajst-29874	66	1	because	because	SCONJ
ajst-29874	66	2	the	the	DET
ajst-29874	66	3	numerical	numerical	ADJ
ajst-29874	66	4	scale	scale	NOUN
ajst-29874	66	5	of	of	ADP
ajst-29874	66	6	magnetic	magnetic	ADJ
ajst-29874	66	7	core	core	NOUN
ajst-29874	66	8	loss	loss	NOUN
ajst-29874	66	9	varies	vary	VERB
ajst-29874	66	10	greatly	greatly	ADV
ajst-29874	66	11	,	,	PUNCT
ajst-29874	66	12	direct	direct	ADJ
ajst-29874	66	13	training	training	NOUN
ajst-29874	66	14	may	may	AUX
ajst-29874	66	15	cause	cause	VERB
ajst-29874	66	16	the	the	DET
ajst-29874	66	17	model	model	NOUN
ajst-29874	66	18	to	to	PART
ajst-29874	66	19	be	be	AUX
ajst-29874	66	20	difficult	difficult	ADJ
ajst-29874	66	21	to	to	PART
ajst-29874	66	22	converge	converge	VERB
ajst-29874	66	23	.	.	PUNCT
ajst-29874	67	1	therefore	therefore	ADV
ajst-29874	67	2	,	,	PUNCT
ajst-29874	67	3	we	we	PRON
ajst-29874	67	4	input	input	VERB
ajst-29874	67	5	the	the	DET
ajst-29874	67	6	model	model	NOUN
ajst-29874	67	7	training	training	NOUN
ajst-29874	67	8	after	after	ADP
ajst-29874	67	9	normalizing	normalize	VERB
ajst-29874	67	10	the	the	DET
ajst-29874	67	11	independent	independent	ADJ
ajst-29874	67	12	variables	variable	NOUN
ajst-29874	67	13	and	and	CCONJ
ajst-29874	67	14	dependent	dependent	ADJ
ajst-29874	67	15	variables	variable	NOUN
ajst-29874	67	16	and	and	CCONJ
ajst-29874	67	17	reverse	reverse	NOUN
ajst-29874	67	18	normalize	normalize	VERB
ajst-29874	67	19	the	the	DET
ajst-29874	67	20	model	model	NOUN
ajst-29874	67	21	output	output	NOUN
ajst-29874	67	22	results	result	NOUN
ajst-29874	67	23	in	in	ADP
ajst-29874	67	24	the	the	DET
ajst-29874	67	25	prediction	prediction	NOUN
ajst-29874	67	26	stage	stage	NOUN
ajst-29874	67	27	to	to	PART
ajst-29874	67	28	restore	restore	VERB
ajst-29874	67	29	the	the	DET
ajst-29874	67	30	original	original	ADJ
ajst-29874	67	31	value	value	NOUN
ajst-29874	67	32	range	range	NOUN
ajst-29874	67	33	for	for	ADP
ajst-29874	67	34	comparison	comparison	NOUN
ajst-29874	67	35	with	with	ADP
ajst-29874	67	36	real	real	ADJ
ajst-29874	67	37	data	datum	NOUN
ajst-29874	67	38	or	or	CCONJ
ajst-29874	67	39	practical	practical	ADJ
ajst-29874	67	40	application[6	application[6	PROPN
ajst-29874	67	41	]	]	PUNCT
ajst-29874	67	42	.	.	PUNCT
ajst-29874	68	1	after	after	ADP
ajst-29874	68	2	300	300	NUM
ajst-29874	68	3	rounds	round	NOUN
ajst-29874	68	4	of	of	ADP
ajst-29874	68	5	training	training	NOUN
ajst-29874	68	6	,	,	PUNCT
ajst-29874	68	7	the	the	DET
ajst-29874	68	8	loss	loss	NOUN
ajst-29874	68	9	of	of	ADP
ajst-29874	68	10	the	the	DET
ajst-29874	68	11	model	model	NOUN
ajst-29874	68	12	on	on	ADP
ajst-29874	68	13	the	the	DET
ajst-29874	68	14	normalized	normalize	VERB
ajst-29874	68	15	data	data	NOUN
ajst-29874	68	16	tends	tend	VERB
ajst-29874	68	17	to	to	PART
ajst-29874	68	18	stabilize	stabilize	VERB
ajst-29874	68	19	and	and	CCONJ
ajst-29874	68	20	converges	converge	VERB
ajst-29874	68	21	to	to	ADP
ajst-29874	68	22	0.0113	0.0113	NUM
ajst-29874	68	23	,	,	PUNCT
ajst-29874	68	24	where	where	SCONJ
ajst-29874	68	25	the	the	DET
ajst-29874	68	26	training	training	NOUN
ajst-29874	68	27	sets	set	NOUN
ajst-29874	68	28	and	and	CCONJ
ajst-29874	68	29	the	the	DET
ajst-29874	68	30	test	test	NOUN
ajst-29874	68	31	set	set	NOUN
ajst-29874	68	32	are	be	AUX
ajst-29874	68	33	.	.	PUNCT
ajst-29874	69	1	the	the	DET
ajst-29874	69	2	results	result	NOUN
ajst-29874	69	3	show	show	VERB
ajst-29874	69	4	that	that	SCONJ
ajst-29874	69	5	the	the	DET
ajst-29874	69	6	model	model	NOUN
ajst-29874	69	7	has	have	VERB
ajst-29874	69	8	high	high	ADJ
ajst-29874	69	9	prediction	prediction	NOUN
ajst-29874	69	10	accuracy	accuracy	NOUN
ajst-29874	69	11	,	,	PUNCT
ajst-29874	69	12	and	and	CCONJ
ajst-29874	69	13	the	the	DET
ajst-29874	69	14	results	result	NOUN
ajst-29874	69	15	of	of	ADP
ajst-29874	69	16	the	the	DET
ajst-29874	69	17	test	test	NOUN
ajst-29874	69	18	set	set	VERB
ajst-29874	69	19	further	far	ADV
ajst-29874	69	20	verify	verify	VERB
ajst-29874	69	21	its	its	PRON
ajst-29874	69	22	good	good	ADJ
ajst-29874	69	23	generalization	generalization	NOUN
ajst-29874	69	24	ability[7	ability[7	PROPN
ajst-29874	69	25	]	]	PUNCT
ajst-29874	69	26	.	.	PUNCT
ajst-29874	70	1	the	the	DET
ajst-29874	70	2	experimental	experimental	ADJ
ajst-29874	70	3	results	result	NOUN
ajst-29874	70	4	show	show	VERB
ajst-29874	70	5	that	that	SCONJ
ajst-29874	70	6	the	the	DET
ajst-29874	70	7	training	training	NOUN
ajst-29874	70	8	loss	loss	NOUN
ajst-29874	70	9	gradually	gradually	ADV
ajst-29874	70	10	decreases	decrease	VERB
ajst-29874	70	11	with	with	ADP
ajst-29874	70	12	the	the	DET
ajst-29874	70	13	increase	increase	NOUN
ajst-29874	70	14	in	in	ADP
ajst-29874	70	15	the	the	DET
ajst-29874	70	16	number	number	NOUN
ajst-29874	70	17	of	of	ADP
ajst-29874	70	18	iterations	iteration	NOUN
ajst-29874	70	19	and	and	CCONJ
ajst-29874	70	20	finally	finally	ADV
ajst-29874	70	21	stabilizes	stabilize	VERB
ajst-29874	70	22	at	at	ADP
ajst-29874	70	23	around	around	ADP
ajst-29874	70	24	0.0113	0.0113	NUM
ajst-29874	70	25	(	(	PUNCT
ajst-29874	70	26	figure	figure	NOUN
ajst-29874	70	27	5	5	NUM
ajst-29874	70	28	.	.	PUNCT
ajst-29874	71	1	(	(	PUNCT
ajst-29874	71	2	a	a	NOUN
ajst-29874	71	3	)	)	PUNCT
ajst-29874	71	4	)	)	PUNCT
ajst-29874	71	5	.	.	PUNCT
ajst-29874	72	1	in	in	ADP
ajst-29874	72	2	the	the	DET
ajst-29874	72	3	test	test	NOUN
ajst-29874	72	4	set	set	NOUN
ajst-29874	72	5	,	,	PUNCT
ajst-29874	72	6	we	we	PRON
ajst-29874	72	7	drew	draw	VERB
ajst-29874	72	8	a	a	DET
ajst-29874	72	9	comparison	comparison	NOUN
ajst-29874	72	10	graph	graph	NOUN
ajst-29874	72	11	between	between	ADP
ajst-29874	72	12	the	the	DET
ajst-29874	72	13	real	real	ADJ
ajst-29874	72	14	value	value	NOUN
ajst-29874	72	15	and	and	CCONJ
ajst-29874	72	16	the	the	DET
ajst-29874	72	17	predicted	predict	VERB
ajst-29874	72	18	value	value	NOUN
ajst-29874	72	19	and	and	CCONJ
ajst-29874	72	20	marked	mark	VERB
ajst-29874	72	21	the	the	DET
ajst-29874	72	22	error	error	NOUN
ajst-29874	72	23	interval	interval	NOUN
ajst-29874	72	24	of	of	ADP
ajst-29874	72	25	±10	±10	NOUN
ajst-29874	72	26	%	%	NOUN
ajst-29874	72	27	(	(	PUNCT
ajst-29874	72	28	figure	figure	NOUN
ajst-29874	72	29	5	5	NUM
ajst-29874	72	30	.	.	PUNCT
ajst-29874	73	1	(	(	PUNCT
ajst-29874	73	2	b	b	NOUN
ajst-29874	73	3	)	)	PUNCT
ajst-29874	73	4	)	)	PUNCT
ajst-29874	73	5	.	.	PUNCT
ajst-29874	74	1	the	the	DET
ajst-29874	74	2	results	result	NOUN
ajst-29874	74	3	show	show	VERB
ajst-29874	74	4	that	that	SCONJ
ajst-29874	74	5	94.55	94.55	NUM
ajst-29874	74	6	%	%	NOUN
ajst-29874	74	7	of	of	ADP
ajst-29874	74	8	the	the	DET
ajst-29874	74	9	predicted	predict	VERB
ajst-29874	74	10	values	value	NOUN
ajst-29874	74	11	fall	fall	VERB
ajst-29874	74	12	within	within	ADP
ajst-29874	74	13	the	the	DET
ajst-29874	74	14	error	error	NOUN
ajst-29874	74	15	range	range	NOUN
ajst-29874	74	16	,	,	PUNCT
ajst-29874	74	17	which	which	PRON
ajst-29874	74	18	indicates	indicate	VERB
ajst-29874	74	19	that	that	SCONJ
ajst-29874	74	20	fcnn	fcnn	PROPN
ajst-29874	74	21	can	can	AUX
ajst-29874	74	22	accurately	accurately	ADV
ajst-29874	74	23	predict	predict	VERB
ajst-29874	74	24	the	the	DET
ajst-29874	74	25	core	core	NOUN
ajst-29874	74	26	loss	loss	NOUN
ajst-29874	74	27	and	and	CCONJ
ajst-29874	74	28	has	have	VERB
ajst-29874	74	29	good	good	ADJ
ajst-29874	74	30	generalization	generalization	NOUN
ajst-29874	74	31	ability	ability	NOUN
ajst-29874	74	32	.	.	PUNCT
ajst-29874	75	1	a.	a.	NOUN
ajst-29874	75	2	training	training	NOUN
ajst-29874	75	3	loss	loss	NOUN
ajst-29874	75	4	changes	change	NOUN
ajst-29874	75	5	with	with	ADP
ajst-29874	75	6	the	the	DET
ajst-29874	75	7	number	number	NOUN
ajst-29874	75	8	of	of	ADP
ajst-29874	75	9	iterations	iteration	NOUN
ajst-29874	75	10	b.	b.	PROPN
ajst-29874	75	11	graph	graph	NOUN
ajst-29874	75	12	comparing	compare	VERB
ajst-29874	75	13	the	the	DET
ajst-29874	75	14	true	true	ADJ
ajst-29874	75	15	value	value	NOUN
ajst-29874	75	16	with	with	ADP
ajst-29874	75	17	the	the	DET
ajst-29874	75	18	predicted	predict	VERB
ajst-29874	75	19	value	value	NOUN
ajst-29874	75	20	figure	figure	NOUN
ajst-29874	75	21	5	5	NUM
ajst-29874	75	22	.	.	PUNCT
ajst-29874	75	23	iterative	iterative	NOUN
ajst-29874	75	24	loss	loss	NOUN
ajst-29874	75	25	versus	versus	ADP
ajst-29874	75	26	prediction	prediction	NOUN
ajst-29874	75	27	graph	graph	NOUN
ajst-29874	75	28	due	due	ADP
ajst-29874	75	29	to	to	ADP
ajst-29874	75	30	the	the	DET
ajst-29874	75	31	large	large	ADJ
ajst-29874	75	32	range	range	NOUN
ajst-29874	75	33	and	and	CCONJ
ajst-29874	75	34	fluctuation	fluctuation	NOUN
ajst-29874	75	35	of	of	ADP
ajst-29874	75	36	core	core	NOUN
ajst-29874	75	37	loss	loss	NOUN
ajst-29874	75	38	,	,	PUNCT
ajst-29874	75	39	mse	mse	PROPN
ajst-29874	75	40	,	,	PUNCT
ajst-29874	75	41	rmse	rmse	NOUN
ajst-29874	75	42	,	,	PUNCT
ajst-29874	75	43	and	and	CCONJ
ajst-29874	75	44	mape	mape	NOUN
ajst-29874	75	45	as	as	ADP
ajst-29874	75	46	evaluation	evaluation	NOUN
ajst-29874	75	47	indexes	index	NOUN
ajst-29874	75	48	may	may	AUX
ajst-29874	75	49	not	not	PART
ajst-29874	75	50	be	be	AUX
ajst-29874	75	51	enough	enough	ADJ
ajst-29874	75	52	to	to	PART
ajst-29874	75	53	fully	fully	ADV
ajst-29874	75	54	reflect	reflect	VERB
ajst-29874	75	55	the	the	DET
ajst-29874	75	56	model	model	NOUN
ajst-29874	75	57	performance	performance	NOUN
ajst-29874	75	58	.	.	PUNCT
ajst-29874	76	1	therefore	therefore	ADV
ajst-29874	76	2	,	,	PUNCT
ajst-29874	76	3	we	we	PRON
ajst-29874	76	4	further	far	ADV
ajst-29874	76	5	calculated	calculate	VERB
ajst-29874	76	6	the	the	DET
ajst-29874	76	7	symmetric	symmetric	ADJ
ajst-29874	76	8	mean	mean	ADJ
ajst-29874	76	9	absolute	absolute	ADJ
ajst-29874	76	10	percentage	percentage	NOUN
ajst-29874	76	11	error	error	NOUN
ajst-29874	76	12	(	(	PUNCT
ajst-29874	76	13	smape	smape	NOUN
ajst-29874	76	14	)	)	PUNCT
ajst-29874	76	15	,	,	PUNCT
ajst-29874	76	16	which	which	PRON
ajst-29874	76	17	measures	measure	VERB
ajst-29874	76	18	the	the	DET
ajst-29874	76	19	relative	relative	ADJ
ajst-29874	76	20	error	error	NOUN
ajst-29874	76	21	between	between	ADP
ajst-29874	76	22	the	the	DET
ajst-29874	76	23	predicted	predict	VERB
ajst-29874	76	24	value	value	NOUN
ajst-29874	76	25	and	and	CCONJ
ajst-29874	76	26	the	the	DET
ajst-29874	76	27	true	true	ADJ
ajst-29874	76	28	value	value	NOUN
ajst-29874	76	29	.	.	PUNCT
ajst-29874	77	1	smape	smape	NOUN
ajst-29874	77	2	is	be	AUX
ajst-29874	77	3	more	more	ADV
ajst-29874	77	4	robust	robust	ADJ
ajst-29874	77	5	when	when	SCONJ
ajst-29874	77	6	processing	processing	NOUN
ajst-29874	77	7	data	datum	NOUN
ajst-29874	77	8	with	with	ADP
ajst-29874	77	9	large	large	ADJ
ajst-29874	77	10	fluctuations	fluctuation	NOUN
ajst-29874	77	11	or	or	CCONJ
ajst-29874	77	12	extreme	extreme	ADJ
ajst-29874	77	13	values	value	NOUN
ajst-29874	77	14	,	,	PUNCT
ajst-29874	77	15	and	and	CCONJ
ajst-29874	77	16	it	it	PRON
ajst-29874	77	17	has	have	VERB
ajst-29874	77	18	symmetry	symmetry	NOUN
ajst-29874	77	19	,	,	PUNCT
ajst-29874	77	20	which	which	PRON
ajst-29874	77	21	can	can	AUX
ajst-29874	77	22	balance	balance	VERB
ajst-29874	77	23	the	the	DET
ajst-29874	77	24	impact	impact	NOUN
ajst-29874	77	25	of	of	ADP
ajst-29874	77	26	predicted	predict	VERB
ajst-29874	77	27	value	value	NOUN
ajst-29874	77	28	and	and	CCONJ
ajst-29874	77	29	actual	actual	ADJ
ajst-29874	77	30	value	value	NOUN
ajst-29874	77	31	on	on	ADP
ajst-29874	77	32	the	the	DET
ajst-29874	77	33	error	error	NOUN
ajst-29874	77	34	to	to	PART
ajst-29874	77	35	avoid	avoid	VERB
ajst-29874	77	36	the	the	DET
ajst-29874	77	37	traditional	traditional	ADJ
ajst-29874	77	38	error	error	NOUN
ajst-29874	77	39	index	index	NOUN
ajst-29874	77	40	being	be	AUX
ajst-29874	77	41	too	too	ADV
ajst-29874	77	42	sensitive	sensitive	ADJ
ajst-29874	77	43	to	to	ADP
ajst-29874	77	44	extreme	extreme	ADJ
ajst-29874	77	45	value	value	NOUN
ajst-29874	77	46	data	datum	NOUN
ajst-29874	77	47	.	.	PUNCT
ajst-29874	78	1	the	the	DET
ajst-29874	78	2	calculation	calculation	NOUN
ajst-29874	78	3	formula	formula	NOUN
ajst-29874	78	4	is	be	AUX
ajst-29874	78	5	:	:	PUNCT
ajst-29874	78	6	where	where	SCONJ
ajst-29874	78	7	is	be	AUX
ajst-29874	78	8	the	the	DET
ajst-29874	78	9	actual	actual	ADJ
ajst-29874	78	10	value	value	NOUN
ajst-29874	78	11	,	,	PUNCT
ajst-29874	78	12	is	be	AUX
ajst-29874	78	13	the	the	DET
ajst-29874	78	14	predicted	predict	VERB
ajst-29874	78	15	value	value	NOUN
ajst-29874	78	16	,	,	PUNCT
ajst-29874	78	17	and	and	CCONJ
ajst-29874	78	18	is	be	AUX
ajst-29874	78	19	the	the	DET
ajst-29874	78	20	total	total	ADJ
ajst-29874	78	21	number	number	NOUN
ajst-29874	78	22	of	of	ADP
ajst-29874	78	23	samples	sample	NOUN
ajst-29874	78	24	.	.	PUNCT
ajst-29874	79	1	in	in	ADP
ajst-29874	79	2	this	this	DET
ajst-29874	79	3	model	model	NOUN
ajst-29874	79	4	training	training	NOUN
ajst-29874	79	5	,	,	PUNCT
ajst-29874	79	6	test	test	NOUN
ajst-29874	79	7	set	set	NOUN
ajst-29874	79	8	.	.	PUNCT
ajst-29874	80	1	the	the	DET
ajst-29874	80	2	results	result	NOUN
ajst-29874	80	3	show	show	VERB
ajst-29874	80	4	that	that	SCONJ
ajst-29874	80	5	both	both	CCONJ
ajst-29874	80	6	the	the	DET
ajst-29874	80	7	training	training	NOUN
ajst-29874	80	8	set	set	NOUN
ajst-29874	80	9	and	and	CCONJ
ajst-29874	80	10	the	the	DET
ajst-29874	80	11	test	test	NOUN
ajst-29874	80	12	set	set	NOUN
ajst-29874	80	13	have	have	VERB
ajst-29874	80	14	excellent	excellent	ADJ
ajst-29874	80	15	prediction	prediction	NOUN
ajst-29874	80	16	accuracy	accuracy	NOUN
ajst-29874	80	17	and	and	CCONJ
ajst-29874	80	18	generalization	generalization	NOUN
ajst-29874	80	19	ability	ability	NOUN
ajst-29874	80	20	.	.	PUNCT
ajst-29874	81	1	to	to	PART
ajst-29874	81	2	analyze	analyze	VERB
ajst-29874	81	3	the	the	DET
ajst-29874	81	4	model	model	NOUN
ajst-29874	81	5	's	's	PART
ajst-29874	81	6	losses	loss	NOUN
ajst-29874	81	7	,	,	PUNCT
ajst-29874	81	8	we	we	PRON
ajst-29874	81	9	visualized	visualize	VERB
ajst-29874	81	10	the	the	DET
ajst-29874	81	11	residuals	residual	NOUN
ajst-29874	81	12	using	use	VERB
ajst-29874	81	13	scatter	scatter	NOUN
ajst-29874	81	14	plots	plot	NOUN
ajst-29874	81	15	,	,	PUNCT
ajst-29874	81	16	histograms	histogram	NOUN
ajst-29874	81	17	,	,	PUNCT
ajst-29874	81	18	and	and	CCONJ
ajst-29874	81	19	boxplots	boxplot	NOUN
ajst-29874	81	20	(	(	PUNCT
ajst-29874	81	21	figure	figure	VERB
ajst-29874	81	22	6	6	NUM
ajst-29874	81	23	)	)	PUNCT
ajst-29874	81	24	.	.	PUNCT
ajst-29874	82	1	the	the	DET
ajst-29874	82	2	scatter	scatter	NOUN
ajst-29874	82	3	plot	plot	NOUN
ajst-29874	82	4	shows	show	VERB
ajst-29874	82	5	that	that	SCONJ
ajst-29874	82	6	residuals	residual	NOUN
ajst-29874	82	7	are	be	AUX
ajst-29874	82	8	mostly	mostly	ADV
ajst-29874	82	9	concentrated	concentrate	VERB
ajst-29874	82	10	around	around	ADP
ajst-29874	82	11	low	low	ADJ
ajst-29874	82	12	predicted	predict	VERB
ajst-29874	82	13	values	value	NOUN
ajst-29874	82	14	,	,	PUNCT
ajst-29874	82	15	with	with	ADP
ajst-29874	82	16	no	no	DET
ajst-29874	82	17	obvious	obvious	ADJ
ajst-29874	82	18	systematic	systematic	ADJ
ajst-29874	82	19	bias	bias	NOUN
ajst-29874	82	20	or	or	CCONJ
ajst-29874	82	21	nonlinear	nonlinear	ADJ
ajst-29874	82	22	trends	trend	NOUN
ajst-29874	82	23	,	,	PUNCT
ajst-29874	82	24	indicating	indicate	VERB
ajst-29874	82	25	stable	stable	ADJ
ajst-29874	82	26	model	model	NOUN
ajst-29874	82	27	fitting	fit	VERB
ajst-29874	82	28	without	without	ADP
ajst-29874	82	29	significant	significant	ADJ
ajst-29874	82	30	underfitting	underfitting	NOUN
ajst-29874	82	31	or	or	CCONJ
ajst-29874	82	32	overfitting	overfitting	NOUN
ajst-29874	82	33	.	.	PUNCT
ajst-29874	83	1	the	the	DET
ajst-29874	83	2	histogram	histogram	NOUN
ajst-29874	83	3	reveals	reveal	VERB
ajst-29874	83	4	a	a	DET
ajst-29874	83	5	roughly	roughly	ADV
ajst-29874	83	6	normal	normal	ADJ
ajst-29874	83	7	distribution	distribution	NOUN
ajst-29874	83	8	of	of	ADP
ajst-29874	83	9	residuals	residual	NOUN
ajst-29874	83	10	centered	center	VERB
ajst-29874	83	11	around	around	ADP
ajst-29874	83	12	zero	zero	NUM
ajst-29874	83	13	,	,	PUNCT
ajst-29874	83	14	suggesting	suggest	VERB
ajst-29874	83	15	small	small	ADJ
ajst-29874	83	16	overall	overall	ADJ
ajst-29874	83	17	prediction	prediction	NOUN
ajst-29874	83	18	errors	error	NOUN
ajst-29874	83	19	and	and	CCONJ
ajst-29874	83	20	good	good	ADJ
ajst-29874	83	21	model	model	NOUN
ajst-29874	83	22	performance	performance	NOUN
ajst-29874	83	23	.	.	PUNCT
ajst-29874	84	1	the	the	DET
ajst-29874	84	2	boxplot	boxplot	NOUN
ajst-29874	84	3	further	far	ADV
ajst-29874	84	4	confirms	confirm	VERB
ajst-29874	84	5	this	this	PRON
ajst-29874	84	6	,	,	PUNCT
ajst-29874	84	7	with	with	ADP
ajst-29874	84	8	a	a	DET
ajst-29874	84	9	median	median	ADJ
ajst-29874	84	10	close	close	ADJ
ajst-29874	84	11	to	to	ADP
ajst-29874	84	12	zero	zero	NUM
ajst-29874	84	13	and	and	CCONJ
ajst-29874	84	14	a	a	DET
ajst-29874	84	15	small	small	ADJ
ajst-29874	84	16	interquartile	interquartile	NOUN
ajst-29874	84	17	range	range	NOUN
ajst-29874	84	18	,	,	PUNCT
ajst-29874	84	19	indicating	indicate	VERB
ajst-29874	84	20	accurate	accurate	ADJ
ajst-29874	84	21	predictions	prediction	NOUN
ajst-29874	84	22	for	for	ADP
ajst-29874	84	23	most	most	ADJ
ajst-29874	84	24	samples	sample	NOUN
ajst-29874	84	25	and	and	CCONJ
ajst-29874	84	26	no	no	DET
ajst-29874	84	27	significant	significant	ADJ
ajst-29874	84	28	bias	bias	NOUN
ajst-29874	84	29	.	.	PUNCT
ajst-29874	85	1	in	in	ADP
ajst-29874	85	2	summary	summary	NOUN
ajst-29874	85	3	,	,	PUNCT
ajst-29874	85	4	our	our	PRON
ajst-29874	85	5	model	model	NOUN
ajst-29874	85	6	has	have	VERB
ajst-29874	85	7	a	a	DET
ajst-29874	85	8	good	good	ADJ
ajst-29874	85	9	residual	residual	ADJ
ajst-29874	85	10	distribution	distribution	NOUN
ajst-29874	85	11	,	,	PUNCT
ajst-29874	85	12	small	small	ADJ
ajst-29874	85	13	prediction	prediction	NOUN
ajst-29874	85	14	errors	error	NOUN
ajst-29874	85	15	,	,	PUNCT
ajst-29874	85	16	and	and	CCONJ
ajst-29874	85	17	meets	meet	VERB
ajst-29874	85	18	the	the	DET
ajst-29874	85	19	normality	normality	NOUN
ajst-29874	85	20	assumption	assumption	NOUN
ajst-29874	85	21	,	,	PUNCT
ajst-29874	85	22	demonstrating	demonstrate	VERB
ajst-29874	85	23	strong	strong	ADJ
ajst-29874	85	24	fitting	fitting	ADJ
ajst-29874	85	25	ability	ability	NOUN
ajst-29874	85	26	.	.	PUNCT
ajst-29874	86	1	a	a	DET
ajst-29874	86	2	few	few	ADJ
ajst-29874	86	3	outliers	outlier	NOUN
ajst-29874	86	4	do	do	AUX
ajst-29874	86	5	not	not	PART
ajst-29874	86	6	significantly	significantly	ADV
ajst-29874	86	7	impact	impact	VERB
ajst-29874	86	8	the	the	DET
ajst-29874	86	9	overall	overall	ADJ
ajst-29874	86	10	model	model	NOUN
ajst-29874	86	11	performance	performance	NOUN
ajst-29874	86	12	.	.	PUNCT
ajst-29874	87	1	2	2	NUM
ajst-29874	88	1	0.9889r	0.9889r	NUM
ajst-29874	88	2			NUM
ajst-29874	88	3	2	2	NUM
ajst-29874	88	4	0.9843r	0.9843r	NOUN
ajst-29874	88	5			NUM
ajst-29874	88	6	1	1	NUM
ajst-29874	88	7	ˆ|	ˆ|	PROPN
ajst-29874	88	8	|100	|100	PROPN
ajst-29874	88	9	%	%	NOUN
ajst-29874	88	10	smape	smape	NOUN
ajst-29874	88	11	ˆ|	ˆ|	PROPN
ajst-29874	89	1	|	|	ADV
ajst-29874	89	2	|	|	ADV
ajst-29874	89	3	|	|	ADV
ajst-29874	89	4	2	2	NUM
ajst-29874	89	5	n	n	NOUN
ajst-29874	90	1	i	i	PRON
ajst-29874	90	2	i	i	PRON
ajst-29874	91	1	i	i	PRON
ajst-29874	91	2	ii	ii	VERB
ajst-29874	92	1	y	y	VERB
ajst-29874	92	2	y	y	PROPN
ajst-29874	92	3	y	y	PROPN
ajst-29874	92	4	yn	yn	PROPN
ajst-29874	92	5			PROPN
ajst-29874	92	6			PROPN
ajst-29874	92	7			PROPN
ajst-29874	92	8			NOUN
ajst-29874	92	9	i	i	PRON
ajst-29874	92	10	ˆiy	ˆiy	VERB
ajst-29874	92	11	n	n	DET
ajst-29874	92	12	178	178	NUM
ajst-29874	92	13	figure	figure	NOUN
ajst-29874	92	14	6	6	NUM
ajst-29874	92	15	.	.	PUNCT
ajst-29874	92	16	residual	residual	ADJ
ajst-29874	92	17	diagram	diagram	NOUN
ajst-29874	92	18	to	to	PART
ajst-29874	92	19	further	far	ADV
ajst-29874	92	20	test	test	VERB
ajst-29874	92	21	the	the	DET
ajst-29874	92	22	excellence	excellence	NOUN
ajst-29874	92	23	of	of	ADP
ajst-29874	92	24	the	the	DET
ajst-29874	92	25	core	core	NOUN
ajst-29874	92	26	loss	loss	NOUN
ajst-29874	92	27	prediction	prediction	NOUN
ajst-29874	92	28	model	model	NOUN
ajst-29874	92	29	we	we	PRON
ajst-29874	92	30	built	build	VERB
ajst-29874	92	31	,	,	PUNCT
ajst-29874	92	32	we	we	PRON
ajst-29874	92	33	selected	select	VERB
ajst-29874	92	34	the	the	DET
ajst-29874	92	35	traditional	traditional	ADJ
ajst-29874	92	36	machine	machine	NOUN
ajst-29874	92	37	learning	learn	VERB
ajst-29874	92	38	algorithm	algorithm	NOUN
ajst-29874	92	39	to	to	PART
ajst-29874	92	40	build	build	VERB
ajst-29874	92	41	the	the	DET
ajst-29874	92	42	model	model	NOUN
ajst-29874	92	43	as	as	ADP
ajst-29874	92	44	a	a	DET
ajst-29874	92	45	comparison	comparison	NOUN
ajst-29874	92	46	.	.	PUNCT
ajst-29874	93	1	fig	fig	NOUN
ajst-29874	93	2	.	.	PUNCT
ajst-29874	94	1	7	7	NUM
ajst-29874	94	2	shows	show	VERB
ajst-29874	94	3	the	the	DET
ajst-29874	94	4	visual	visual	ADJ
ajst-29874	94	5	results	result	NOUN
ajst-29874	94	6	of	of	ADP
ajst-29874	94	7	the	the	DET
ajst-29874	94	8	real	real	ADJ
ajst-29874	94	9	value	value	NOUN
ajst-29874	94	10	and	and	CCONJ
ajst-29874	94	11	predicted	predict	VERB
ajst-29874	94	12	value	value	NOUN
ajst-29874	94	13	of	of	ADP
ajst-29874	94	14	random	random	ADJ
ajst-29874	94	15	forest	forest	NOUN
ajst-29874	94	16	and	and	CCONJ
ajst-29874	94	17	xgboost	xgboost	PRON
ajst-29874	94	18	,	,	PUNCT
ajst-29874	94	19	79.26	79.26	NUM
ajst-29874	94	20	%	%	NOUN
ajst-29874	94	21	and	and	CCONJ
ajst-29874	94	22	80.87	80.87	NUM
ajst-29874	94	23	%	%	NOUN
ajst-29874	94	24	of	of	ADP
ajst-29874	94	25	the	the	DET
ajst-29874	94	26	data	datum	NOUN
ajst-29874	94	27	fall	fall	VERB
ajst-29874	94	28	within	within	ADP
ajst-29874	94	29	the	the	DET
ajst-29874	94	30	±10	±10	NOUN
ajst-29874	94	31	%	%	NOUN
ajst-29874	94	32	error	error	NOUN
ajst-29874	94	33	line	line	NOUN
ajst-29874	94	34	,	,	PUNCT
ajst-29874	94	35	respectively	respectively	ADV
ajst-29874	94	36	.	.	PUNCT
ajst-29874	95	1	the	the	DET
ajst-29874	95	2	prediction	prediction	NOUN
ajst-29874	95	3	effect	effect	NOUN
ajst-29874	95	4	and	and	CCONJ
ajst-29874	95	5	generalization	generalization	NOUN
ajst-29874	95	6	ability	ability	NOUN
ajst-29874	95	7	are	be	AUX
ajst-29874	95	8	worse	bad	ADJ
ajst-29874	95	9	than	than	ADP
ajst-29874	95	10	the	the	DET
ajst-29874	95	11	fcnn	fcnn	ADJ
ajst-29874	95	12	loss	loss	NOUN
ajst-29874	95	13	prediction	prediction	NOUN
ajst-29874	95	14	model	model	NOUN
ajst-29874	95	15	.	.	PUNCT
ajst-29874	96	1	a.	a.	PROPN
ajst-29874	96	2	random	random	PROPN
ajst-29874	96	3	forest	forest	NOUN
ajst-29874	96	4	prediction	prediction	NOUN
ajst-29874	96	5	map	map	NOUN
ajst-29874	96	6	b.	b.	PROPN
ajst-29874	96	7	xgboost	xgboost	PROPN
ajst-29874	96	8	prediction	prediction	NOUN
ajst-29874	96	9	graph	graph	NOUN
ajst-29874	96	10	figure	figure	NOUN
ajst-29874	96	11	7	7	NUM
ajst-29874	96	12	.	.	PUNCT
ajst-29874	96	13	neural	neural	ADJ
ajst-29874	96	14	network	network	NOUN
ajst-29874	96	15	prediction	prediction	NOUN
ajst-29874	96	16	graph	graph	NOUN
ajst-29874	96	17	the	the	DET
ajst-29874	96	18	detailed	detailed	ADJ
ajst-29874	96	19	evaluation	evaluation	NOUN
ajst-29874	96	20	index	index	NOUN
ajst-29874	96	21	comparison	comparison	NOUN
ajst-29874	96	22	is	be	AUX
ajst-29874	96	23	shown	show	VERB
ajst-29874	96	24	in	in	ADP
ajst-29874	96	25	figure	figure	NOUN
ajst-29874	96	26	8	8	NUM
ajst-29874	96	27	.	.	PUNCT
ajst-29874	97	1	through	through	ADP
ajst-29874	97	2	the	the	DET
ajst-29874	97	3	comparison	comparison	NOUN
ajst-29874	97	4	of	of	ADP
ajst-29874	97	5	multiple	multiple	ADJ
ajst-29874	97	6	machine	machine	NOUN
ajst-29874	97	7	learning	learning	NOUN
ajst-29874	97	8	models	model	NOUN
ajst-29874	97	9	,	,	PUNCT
ajst-29874	97	10	it	it	PRON
ajst-29874	97	11	can	can	AUX
ajst-29874	97	12	be	be	AUX
ajst-29874	97	13	found	find	VERB
ajst-29874	97	14	that	that	SCONJ
ajst-29874	97	15	the	the	DET
ajst-29874	97	16	general	general	ADJ
ajst-29874	97	17	core	core	NOUN
ajst-29874	97	18	loss	loss	NOUN
ajst-29874	97	19	prediction	prediction	NOUN
ajst-29874	97	20	model	model	NOUN
ajst-29874	97	21	built	build	VERB
ajst-29874	97	22	by	by	ADP
ajst-29874	97	23	us	we	PRON
ajst-29874	97	24	based	base	VERB
ajst-29874	97	25	on	on	ADP
ajst-29874	97	26	fcnn	fcnn	PROPN
ajst-29874	97	27	is	be	AUX
ajst-29874	97	28	higher	high	ADJ
ajst-29874	97	29	than	than	ADP
ajst-29874	97	30	the	the	DET
ajst-29874	97	31	traditional	traditional	ADJ
ajst-29874	97	32	machine	machine	NOUN
ajst-29874	97	33	learning	learning	NOUN
ajst-29874	97	34	model	model	NOUN
ajst-29874	97	35	in	in	ADP
ajst-29874	97	36	terms	term	NOUN
ajst-29874	97	37	of	of	ADP
ajst-29874	97	38	prediction	prediction	NOUN
ajst-29874	97	39	accuracy	accuracy	NOUN
ajst-29874	97	40	and	and	CCONJ
ajst-29874	97	41	generalization	generalization	NOUN
ajst-29874	97	42	ability	ability	NOUN
ajst-29874	97	43	.	.	PUNCT
ajst-29874	98	1	figure	figure	NOUN
ajst-29874	98	2	8	8	NUM
ajst-29874	98	3	.	.	PUNCT
ajst-29874	98	4	comparison	comparison	NOUN
ajst-29874	98	5	diagram	diagram	NOUN
ajst-29874	98	6	of	of	ADP
ajst-29874	98	7	model	model	NOUN
ajst-29874	98	8	indicators	indicator	NOUN
ajst-29874	98	9	6	6	NUM
ajst-29874	98	10	.	.	PUNCT
ajst-29874	99	1	conclusion	conclusion	NOUN
ajst-29874	99	2	in	in	ADP
ajst-29874	99	3	this	this	DET
ajst-29874	99	4	study	study	NOUN
ajst-29874	99	5	,	,	PUNCT
ajst-29874	99	6	a	a	DET
ajst-29874	99	7	magnetic	magnetic	ADJ
ajst-29874	99	8	core	core	NOUN
ajst-29874	99	9	loss	loss	NOUN
ajst-29874	99	10	prediction	prediction	NOUN
ajst-29874	99	11	model	model	NOUN
ajst-29874	99	12	based	base	VERB
ajst-29874	99	13	on	on	ADP
ajst-29874	99	14	a	a	DET
ajst-29874	99	15	fully	fully	ADV
ajst-29874	99	16	connected	connected	ADJ
ajst-29874	99	17	neural	neural	ADJ
ajst-29874	99	18	network	network	NOUN
ajst-29874	99	19	(	(	PUNCT
ajst-29874	99	20	fcnn	fcnn	PROPN
ajst-29874	99	21	)	)	PUNCT
ajst-29874	99	22	was	be	AUX
ajst-29874	99	23	constructed	construct	VERB
ajst-29874	99	24	,	,	PUNCT
ajst-29874	99	25	and	and	CCONJ
ajst-29874	99	26	the	the	DET
ajst-29874	99	27	prediction	prediction	NOUN
ajst-29874	99	28	accuracy	accuracy	NOUN
ajst-29874	99	29	and	and	CCONJ
ajst-29874	99	30	generalization	generalization	NOUN
ajst-29874	99	31	ability	ability	NOUN
ajst-29874	99	32	of	of	ADP
ajst-29874	99	33	the	the	DET
ajst-29874	99	34	model	model	NOUN
ajst-29874	99	35	were	be	AUX
ajst-29874	99	36	improved	improve	VERB
ajst-29874	99	37	through	through	ADP
ajst-29874	99	38	data	datum	NOUN
ajst-29874	99	39	preprocessing	preprocessing	NOUN
ajst-29874	99	40	,	,	PUNCT
ajst-29874	99	41	feature	feature	NOUN
ajst-29874	99	42	extraction	extraction	NOUN
ajst-29874	99	43	,	,	PUNCT
ajst-29874	99	44	and	and	CCONJ
ajst-29874	99	45	optimization	optimization	NOUN
ajst-29874	99	46	training	training	NOUN
ajst-29874	99	47	process	process	NOUN
ajst-29874	99	48	.	.	PUNCT
ajst-29874	100	1	the	the	DET
ajst-29874	100	2	experimental	experimental	ADJ
ajst-29874	100	3	results	result	NOUN
ajst-29874	100	4	show	show	VERB
ajst-29874	100	5	that	that	SCONJ
ajst-29874	100	6	fcnn	fcnn	PROPN
ajst-29874	100	7	has	have	VERB
ajst-29874	100	8	better	well	ADJ
ajst-29874	100	9	prediction	prediction	NOUN
ajst-29874	100	10	performance	performance	NOUN
ajst-29874	100	11	than	than	ADP
ajst-29874	100	12	the	the	DET
ajst-29874	100	13	traditional	traditional	ADJ
ajst-29874	100	14	random	random	ADJ
ajst-29874	100	15	forest	forest	NOUN
ajst-29874	100	16	and	and	CCONJ
ajst-29874	100	17	xgboost	xgboost	NOUN
ajst-29874	100	18	models	model	NOUN
ajst-29874	100	19	.	.	PUNCT
ajst-29874	101	1	in	in	ADP
ajst-29874	101	2	the	the	DET
ajst-29874	101	3	test	test	NOUN
ajst-29874	101	4	set	set	NOUN
ajst-29874	101	5	,	,	PUNCT
ajst-29874	101	6	94.55	94.55	NUM
ajst-29874	101	7	%	%	NOUN
ajst-29874	101	8	of	of	ADP
ajst-29874	101	9	the	the	DET
ajst-29874	101	10	sample	sample	NOUN
ajst-29874	101	11	prediction	prediction	NOUN
ajst-29874	101	12	error	error	NOUN
ajst-29874	101	13	is	be	AUX
ajst-29874	101	14	less	less	ADJ
ajst-29874	101	15	than	than	ADP
ajst-29874	101	16	±10	±10	VERB
ajst-29874	101	17	%	%	NOUN
ajst-29874	101	18	,	,	PUNCT
ajst-29874	101	19	indicating	indicate	VERB
ajst-29874	101	20	that	that	SCONJ
ajst-29874	101	21	the	the	DET
ajst-29874	101	22	model	model	NOUN
ajst-29874	101	23	can	can	AUX
ajst-29874	101	24	effectively	effectively	ADV
ajst-29874	101	25	capture	capture	VERB
ajst-29874	101	26	the	the	DET
ajst-29874	101	27	complex	complex	ADJ
ajst-29874	101	28	nonlinear	nonlinear	ADJ
ajst-29874	101	29	characteristics	characteristic	NOUN
ajst-29874	101	30	of	of	ADP
ajst-29874	101	31	the	the	DET
ajst-29874	101	32	core	core	NOUN
ajst-29874	101	33	loss	loss	NOUN
ajst-29874	101	34	.	.	PUNCT
ajst-29874	102	1	the	the	DET
ajst-29874	102	2	results	result	NOUN
ajst-29874	102	3	show	show	VERB
ajst-29874	102	4	that	that	SCONJ
ajst-29874	102	5	this	this	DET
ajst-29874	102	6	method	method	NOUN
ajst-29874	102	7	can	can	AUX
ajst-29874	102	8	provide	provide	VERB
ajst-29874	102	9	reliable	reliable	ADJ
ajst-29874	102	10	support	support	NOUN
ajst-29874	102	11	for	for	ADP
ajst-29874	102	12	the	the	DET
ajst-29874	102	13	design	design	NOUN
ajst-29874	102	14	and	and	CCONJ
ajst-29874	102	15	optimization	optimization	NOUN
ajst-29874	102	16	of	of	ADP
ajst-29874	102	17	power	power	NOUN
ajst-29874	102	18	electronic	electronic	ADJ
ajst-29874	102	19	equipment	equipment	NOUN
ajst-29874	102	20	and	and	CCONJ
ajst-29874	102	21	has	have	VERB
ajst-29874	102	22	a	a	DET
ajst-29874	102	23	wide	wide	ADJ
ajst-29874	102	24	range	range	NOUN
ajst-29874	102	25	of	of	ADP
ajst-29874	102	26	engineering	engineering	NOUN
ajst-29874	102	27	applications	application	NOUN
ajst-29874	102	28	.	.	PUNCT
ajst-29874	103	1	references	reference	NOUN
ajst-29874	103	2	[	[	X
ajst-29874	103	3	1	1	NUM
ajst-29874	103	4	]	]	PUNCT
ajst-29874	103	5	h.	h.	PROPN
ajst-29874	103	6	sun	sun	PROPN
ajst-29874	103	7	,	,	PUNCT
ajst-29874	103	8	l.	l.	PROPN
ajst-29874	103	9	yu	yu	PROPN
ajst-29874	103	10	and	and	CCONJ
ajst-29874	103	11	j.	j.	PROPN
ajst-29874	103	12	katto	katto	PROPN
ajst-29874	103	13	,	,	PUNCT
ajst-29874	103	14	"	"	PUNCT
ajst-29874	103	15	fully	fully	ADV
ajst-29874	103	16	neural	neural	ADJ
ajst-29874	103	17	network	network	NOUN
ajst-29874	103	18	mode	mode	PROPN
ajst-29874	103	19	based	base	VERB
ajst-29874	103	20	intra	intra	ADJ
ajst-29874	103	21	prediction	prediction	NOUN
ajst-29874	103	22	of	of	ADP
ajst-29874	103	23	variable	variable	ADJ
ajst-29874	103	24	block	block	NOUN
ajst-29874	103	25	size	size	NOUN
ajst-29874	103	26	,	,	PUNCT
ajst-29874	103	27	"	"	PUNCT
ajst-29874	103	28	2020	2020	NUM
ajst-29874	103	29	ieee	ieee	PROPN
ajst-29874	103	30	international	international	ADJ
ajst-29874	103	31	conference	conference	NOUN
ajst-29874	103	32	on	on	ADP
ajst-29874	103	33	visual	visual	ADJ
ajst-29874	103	34	communications	communication	NOUN
ajst-29874	103	35	and	and	CCONJ
ajst-29874	103	36	image	image	NOUN
ajst-29874	103	37	processing	processing	NOUN
ajst-29874	103	38	(	(	PUNCT
ajst-29874	103	39	vcip	vcip	NOUN
ajst-29874	103	40	)	)	PUNCT
ajst-29874	103	41	,	,	PUNCT
ajst-29874	103	42	macau	macau	PROPN
ajst-29874	103	43	,	,	PUNCT
ajst-29874	103	44	china	china	PROPN
ajst-29874	103	45	,	,	PUNCT
ajst-29874	103	46	2020	2020	NUM
ajst-29874	103	47	,	,	PUNCT
ajst-29874	103	48	pp	pp	ADV
ajst-29874	103	49	.	.	PUNCT
ajst-29874	104	1	21	21	NUM
ajst-29874	104	2	-	-	SYM
ajst-29874	104	3	24	24	NUM
ajst-29874	104	4	179	179	NUM
ajst-29874	105	1	[	[	X
ajst-29874	105	2	2	2	NUM
ajst-29874	105	3	]	]	PUNCT
ajst-29874	105	4	z.	z.	PROPN
ajst-29874	105	5	li	li	PROPN
ajst-29874	105	6	,	,	PUNCT
ajst-29874	105	7	w.	w.	PROPN
ajst-29874	105	8	han	han	PROPN
ajst-29874	105	9	,	,	PUNCT
ajst-29874	105	10	z.	z.	PROPN
ajst-29874	105	11	xin	xin	PROPN
ajst-29874	105	12	,	,	PUNCT
ajst-29874	105	13	q.	q.	PROPN
ajst-29874	105	14	liu	liu	PROPN
ajst-29874	105	15	,	,	PUNCT
ajst-29874	105	16	j.	j.	PROPN
ajst-29874	105	17	chen	chen	PROPN
ajst-29874	105	18	and	and	CCONJ
ajst-29874	105	19	p.	p.	PROPN
ajst-29874	105	20	c.	c.	PROPN
ajst-29874	105	21	loh	loh	PROPN
ajst-29874	105	22	,	,	PUNCT
ajst-29874	105	23	"	"	PUNCT
ajst-29874	105	24	a	a	DET
ajst-29874	105	25	review	review	NOUN
ajst-29874	105	26	of	of	ADP
ajst-29874	105	27	magnetic	magnetic	ADJ
ajst-29874	105	28	core	core	NOUN
ajst-29874	105	29	materials	material	NOUN
ajst-29874	105	30	,	,	PUNCT
ajst-29874	105	31	core	core	NOUN
ajst-29874	105	32	loss	loss	NOUN
ajst-29874	105	33	modeling	modeling	NOUN
ajst-29874	105	34	and	and	CCONJ
ajst-29874	105	35	measurements	measurement	NOUN
ajst-29874	105	36	in	in	ADP
ajst-29874	105	37	high	high	ADJ
ajst-29874	105	38	-	-	PUNCT
ajst-29874	105	39	power	power	NOUN
ajst-29874	105	40	high	high	ADJ
ajst-29874	105	41	-	-	PUNCT
ajst-29874	105	42	frequency	frequency	NOUN
ajst-29874	105	43	transformers	transformer	NOUN
ajst-29874	105	44	,	,	PUNCT
ajst-29874	105	45	"	"	PUNCT
ajst-29874	105	46	in	in	ADP
ajst-29874	105	47	cpss	cpss	NOUN
ajst-29874	105	48	transactions	transaction	NOUN
ajst-29874	105	49	on	on	ADP
ajst-29874	105	50	power	power	NOUN
ajst-29874	105	51	electronics	electronic	NOUN
ajst-29874	105	52	and	and	CCONJ
ajst-29874	105	53	applications	application	NOUN
ajst-29874	105	54	,	,	PUNCT
ajst-29874	105	55	vol	vol	NOUN
ajst-29874	105	56	.	.	PROPN
ajst-29874	105	57	7	7	NUM
ajst-29874	105	58	,	,	PUNCT
ajst-29874	105	59	no	no	INTJ
ajst-29874	105	60	.	.	NOUN
ajst-29874	105	61	4	4	NUM
ajst-29874	105	62	,	,	PUNCT
ajst-29874	105	63	pp	pp	ADJ
ajst-29874	105	64	.	.	PUNCT
ajst-29874	106	1	359	359	NUM
ajst-29874	106	2	-	-	SYM
ajst-29874	106	3	373	373	NUM
ajst-29874	106	4	,	,	PUNCT
ajst-29874	106	5	december	december	PROPN
ajst-29874	106	6	2022	2022	NUM
ajst-29874	106	7	[	[	X
ajst-29874	106	8	3	3	X
ajst-29874	106	9	]	]	X
ajst-29874	106	10	l.	l.	PROPN
ajst-29874	106	11	a.	a.	PROPN
ajst-29874	106	12	demidova	demidova	PROPN
ajst-29874	106	13	and	and	CCONJ
ajst-29874	106	14	a.	a.	NOUN
ajst-29874	106	15	a.	a.	NOUN
ajst-29874	106	16	misailidi	misailidi	PROPN
ajst-29874	106	17	,	,	PUNCT
ajst-29874	106	18	"	"	PUNCT
ajst-29874	106	19	algorithm	algorithm	NOUN
ajst-29874	106	20	for	for	ADP
ajst-29874	106	21	detecting	detect	VERB
ajst-29874	106	22	anomalous	anomalous	ADJ
ajst-29874	106	23	student	student	NOUN
ajst-29874	106	24	activities	activity	NOUN
ajst-29874	106	25	in	in	ADP
ajst-29874	106	26	the	the	DET
ajst-29874	106	27	online	online	ADJ
ajst-29874	106	28	learning	learning	NOUN
ajst-29874	106	29	process	process	NOUN
ajst-29874	106	30	based	base	VERB
ajst-29874	106	31	on	on	ADP
ajst-29874	106	32	box	box	NOUN
ajst-29874	106	33	plots	plot	NOUN
ajst-29874	106	34	,	,	PUNCT
ajst-29874	106	35	"	"	PUNCT
ajst-29874	106	36	2023	2023	NUM
ajst-29874	106	37	5th	5th	ADJ
ajst-29874	106	38	international	international	ADJ
ajst-29874	106	39	conference	conference	NOUN
ajst-29874	106	40	on	on	ADP
ajst-29874	106	41	control	control	NOUN
ajst-29874	106	42	systems	system	NOUN
ajst-29874	106	43	,	,	PUNCT
ajst-29874	106	44	mathematical	mathematical	ADJ
ajst-29874	106	45	modeling	modeling	NOUN
ajst-29874	106	46	,	,	PUNCT
ajst-29874	106	47	automation	automation	NOUN
ajst-29874	106	48	and	and	CCONJ
ajst-29874	106	49	energy	energy	NOUN
ajst-29874	106	50	efficiency	efficiency	NOUN
ajst-29874	106	51	(	(	PUNCT
ajst-29874	106	52	summa	summa	PROPN
ajst-29874	106	53	)	)	PUNCT
ajst-29874	106	54	,	,	PUNCT
ajst-29874	106	55	lipetsk	lipetsk	ADV
ajst-29874	106	56	,	,	PUNCT
ajst-29874	106	57	russian	russian	PROPN
ajst-29874	106	58	federation	federation	PROPN
ajst-29874	106	59	,	,	PUNCT
ajst-29874	106	60	2023	2023	NUM
ajst-29874	106	61	,	,	PUNCT
ajst-29874	106	62	pp	pp	ADV
ajst-29874	106	63	.	.	PUNCT
ajst-29874	107	1	309	309	NUM
ajst-29874	107	2	-	-	SYM
ajst-29874	107	3	314	314	NUM
ajst-29874	107	4	,	,	PUNCT
ajst-29874	107	5	[	[	X
ajst-29874	107	6	4	4	X
ajst-29874	107	7	]	]	PUNCT
ajst-29874	107	8	s.	s.	PROPN
ajst-29874	107	9	khalid	khalid	PROPN
ajst-29874	107	10	,	,	PUNCT
ajst-29874	107	11	t.	t.	PROPN
ajst-29874	107	12	khalil	khalil	PROPN
ajst-29874	107	13	,	,	PUNCT
ajst-29874	107	14	and	and	CCONJ
ajst-29874	107	15	s.	s.	PROPN
ajst-29874	107	16	nasreen	nasreen	PROPN
ajst-29874	107	17	,	,	PUNCT
ajst-29874	107	18	"	"	PUNCT
ajst-29874	107	19	a	a	DET
ajst-29874	107	20	survey	survey	NOUN
ajst-29874	107	21	of	of	ADP
ajst-29874	107	22	feature	feature	NOUN
ajst-29874	107	23	selection	selection	NOUN
ajst-29874	107	24	and	and	CCONJ
ajst-29874	107	25	feature	feature	NOUN
ajst-29874	107	26	extraction	extraction	NOUN
ajst-29874	107	27	techniques	technique	NOUN
ajst-29874	107	28	in	in	ADP
ajst-29874	107	29	machine	machine	NOUN
ajst-29874	107	30	learning	learning	NOUN
ajst-29874	107	31	,	,	PUNCT
ajst-29874	107	32	"	"	PUNCT
ajst-29874	107	33	2014	2014	NUM
ajst-29874	107	34	science	science	NOUN
ajst-29874	107	35	and	and	CCONJ
ajst-29874	107	36	information	information	NOUN
ajst-29874	107	37	conference	conference	NOUN
ajst-29874	107	38	,	,	PUNCT
ajst-29874	107	39	london	london	PROPN
ajst-29874	107	40	,	,	PUNCT
ajst-29874	107	41	uk	uk	PROPN
ajst-29874	107	42	,	,	PUNCT
ajst-29874	107	43	2014	2014	NUM
ajst-29874	107	44	,	,	PUNCT
ajst-29874	107	45	pp	pp	ADP
ajst-29874	107	46	.	.	PUNCT
ajst-29874	108	1	372	372	NUM
ajst-29874	108	2	-	-	SYM
ajst-29874	108	3	378	378	NUM
ajst-29874	108	4	[	[	SYM
ajst-29874	108	5	5	5	NUM
ajst-29874	108	6	]	]	PUNCT
ajst-29874	108	7	k.	k.	PROPN
ajst-29874	108	8	z.	z.	PROPN
ajst-29874	108	9	mao	mao	PROPN
ajst-29874	108	10	and	and	CCONJ
ajst-29874	108	11	guang	guang	PROPN
ajst-29874	108	12	-	-	PUNCT
ajst-29874	108	13	bin	bin	PROPN
ajst-29874	108	14	huang	huang	PROPN
ajst-29874	108	15	,	,	PUNCT
ajst-29874	108	16	"	"	PUNCT
ajst-29874	108	17	neuron	neuron	NOUN
ajst-29874	108	18	selection	selection	NOUN
ajst-29874	108	19	for	for	ADP
ajst-29874	108	20	rbf	rbf	PROPN
ajst-29874	108	21	neural	neural	ADJ
ajst-29874	108	22	network	network	NOUN
ajst-29874	108	23	classifier	classifier	NOUN
ajst-29874	108	24	based	base	VERB
ajst-29874	108	25	on	on	ADP
ajst-29874	108	26	data	data	NOUN
ajst-29874	108	27	structure	structure	NOUN
ajst-29874	108	28	preserving	preserve	VERB
ajst-29874	108	29	criterion	criterion	NOUN
ajst-29874	108	30	,	,	PUNCT
ajst-29874	108	31	"	"	PUNCT
ajst-29874	108	32	in	in	ADP
ajst-29874	108	33	ieee	ieee	NOUN
ajst-29874	108	34	transactions	transaction	NOUN
ajst-29874	108	35	on	on	ADP
ajst-29874	108	36	neural	neural	ADJ
ajst-29874	108	37	networks	network	NOUN
ajst-29874	108	38	,	,	PUNCT
ajst-29874	108	39	vol	vol	NOUN
ajst-29874	108	40	.	.	PROPN
ajst-29874	109	1	16	16	NUM
ajst-29874	109	2	,	,	PUNCT
ajst-29874	109	3	no	no	INTJ
ajst-29874	109	4	.	.	NOUN
ajst-29874	109	5	6	6	NUM
ajst-29874	109	6	,	,	PUNCT
ajst-29874	109	7	pp	pp	ADJ
ajst-29874	109	8	.	.	PUNCT
ajst-29874	110	1	1531	1531	NUM
ajst-29874	110	2	-	-	SYM
ajst-29874	110	3	1540	1540	NUM
ajst-29874	110	4	,	,	PUNCT
ajst-29874	110	5	nov	nov	PROPN
ajst-29874	110	6	.	.	PROPN
ajst-29874	110	7	2005	2005	NUM
ajst-29874	110	8	[	[	X
ajst-29874	110	9	6	6	NUM
ajst-29874	110	10	]	]	PUNCT
ajst-29874	110	11	m.	m.	NOUN
ajst-29874	110	12	kolarik	kolarik	PROPN
ajst-29874	110	13	,	,	PUNCT
ajst-29874	110	14	r.	r.	PROPN
ajst-29874	110	15	burget	burget	PROPN
ajst-29874	110	16	,	,	PUNCT
ajst-29874	110	17	and	and	CCONJ
ajst-29874	110	18	k.	k.	PROPN
ajst-29874	110	19	riha	riha	PROPN
ajst-29874	110	20	,	,	PUNCT
ajst-29874	110	21	"	"	PUNCT
ajst-29874	110	22	comparing	compare	VERB
ajst-29874	110	23	normalization	normalization	NOUN
ajst-29874	110	24	methods	method	NOUN
ajst-29874	110	25	for	for	ADP
ajst-29874	110	26	limited	limited	ADJ
ajst-29874	110	27	batch	batch	NOUN
ajst-29874	110	28	size	size	NOUN
ajst-29874	110	29	segmentation	segmentation	NOUN
ajst-29874	110	30	neural	neural	ADJ
ajst-29874	110	31	networks	network	NOUN
ajst-29874	110	32	,	,	PUNCT
ajst-29874	110	33	"	"	PUNCT
ajst-29874	110	34	2020	2020	NUM
ajst-29874	110	35	43rd	43rd	ADJ
ajst-29874	110	36	international	international	ADJ
ajst-29874	110	37	conference	conference	NOUN
ajst-29874	110	38	on	on	ADP
ajst-29874	110	39	telecommunications	telecommunication	NOUN
ajst-29874	110	40	and	and	CCONJ
ajst-29874	110	41	signal	signal	NOUN
ajst-29874	110	42	processing	processing	NOUN
ajst-29874	110	43	(	(	PUNCT
ajst-29874	110	44	tsp	tsp	NOUN
ajst-29874	110	45	)	)	PUNCT
ajst-29874	110	46	,	,	PUNCT
ajst-29874	110	47	milan	milan	PROPN
ajst-29874	110	48	,	,	PUNCT
ajst-29874	110	49	italy	italy	PROPN
ajst-29874	110	50	,	,	PUNCT
ajst-29874	110	51	2020	2020	NUM
ajst-29874	110	52	,	,	PUNCT
ajst-29874	110	53	pp	pp	ADV
ajst-29874	110	54	.	.	PUNCT
ajst-29874	111	1	677	677	NUM
ajst-29874	111	2	-	-	SYM
ajst-29874	111	3	680	680	NUM
ajst-29874	111	4	[	[	X
ajst-29874	111	5	7	7	NUM
ajst-29874	111	6	]	]	PUNCT
ajst-29874	111	7	z.	z.	PROPN
ajst-29874	111	8	zhang	zhang	PROPN
ajst-29874	111	9	and	and	CCONJ
ajst-29874	111	10	x.	x.	PROPN
ajst-29874	111	11	zhang	zhang	PROPN
ajst-29874	111	12	,	,	PUNCT
ajst-29874	111	13	"	"	PUNCT
ajst-29874	111	14	a	a	DET
ajst-29874	111	15	review	review	NOUN
ajst-29874	111	16	of	of	ADP
ajst-29874	111	17	research	research	NOUN
ajst-29874	111	18	on	on	ADP
ajst-29874	111	19	generalization	generalization	NOUN
ajst-29874	111	20	error	error	NOUN
ajst-29874	111	21	analysis	analysis	NOUN
ajst-29874	111	22	of	of	ADP
ajst-29874	111	23	deep	deep	ADJ
ajst-29874	111	24	learning	learning	NOUN
ajst-29874	111	25	models	model	NOUN
ajst-29874	111	26	,	,	PUNCT
ajst-29874	111	27	"	"	PUNCT
ajst-29874	111	28	2023	2023	NUM
ajst-29874	111	29	international	international	ADJ
ajst-29874	111	30	conference	conference	NOUN
ajst-29874	111	31	on	on	ADP
ajst-29874	111	32	image	image	NOUN
ajst-29874	111	33	processing	processing	NOUN
ajst-29874	111	34	,	,	PUNCT
ajst-29874	111	35	computer	computer	NOUN
ajst-29874	111	36	vision	vision	NOUN
ajst-29874	111	37	and	and	CCONJ
ajst-29874	111	38	machine	machine	NOUN
ajst-29874	111	39	learning	learning	NOUN
ajst-29874	111	40	(	(	PUNCT
ajst-29874	111	41	icicml	icicml	NOUN
ajst-29874	111	42	)	)	PUNCT
ajst-29874	111	43	,	,	PUNCT
ajst-29874	111	44	chengdu	chengdu	PROPN
ajst-29874	111	45	,	,	PUNCT
ajst-29874	111	46	china	china	PROPN
ajst-29874	111	47	,	,	PUNCT
ajst-29874	111	48	2023	2023	NUM
ajst-29874	111	49	,	,	PUNCT
ajst-29874	111	50	pp	pp	ADJ
ajst-29874	111	51	.	.	PUNCT
ajst-29874	112	1	906	906	NUM
ajst-29874	112	2	-	-	SYM
ajst-29874	112	3	912	912	NUM
ajst-29874	112	4	.	.	PUNCT