id	sid	tid	token	lemma	pos
ejpam-3176	1	1	european	european	PROPN
ejpam-3176	1	2	journal	journal	PROPN
ejpam-3176	1	3	of	of	ADP
ejpam-3176	1	4	pure	pure	ADJ
ejpam-3176	1	5	and	and	CCONJ
ejpam-3176	1	6	applied	apply	VERB
ejpam-3176	1	7	mathematics	mathematic	NOUN
ejpam-3176	1	8	vol	vol	NOUN
ejpam-3176	1	9	.	.	PUNCT
ejpam-3176	2	1	11	11	NUM
ejpam-3176	2	2	,	,	PUNCT
ejpam-3176	2	3	no	no	INTJ
ejpam-3176	2	4	.	.	NOUN
ejpam-3176	2	5	1	1	NUM
ejpam-3176	2	6	,	,	PUNCT
ejpam-3176	2	7	2018	2018	NUM
ejpam-3176	2	8	,	,	PUNCT
ejpam-3176	2	9	284	284	NUM
ejpam-3176	2	10	-	-	SYM
ejpam-3176	2	11	298	298	NUM
ejpam-3176	2	12	issn	issn	PROPN
ejpam-3176	2	13	1307	1307	NUM
ejpam-3176	2	14	-	-	SYM
ejpam-3176	2	15	5543	5543	NUM
ejpam-3176	2	16	–	–	PUNCT
ejpam-3176	2	17	www.ejpam.com	www.ejpam.com	X
ejpam-3176	2	18	published	publish	VERB
ejpam-3176	2	19	by	by	ADP
ejpam-3176	2	20	new	new	PROPN
ejpam-3176	2	21	york	york	PROPN
ejpam-3176	2	22	business	business	PROPN
ejpam-3176	2	23	global	global	ADJ
ejpam-3176	2	24	performance	performance	NOUN
ejpam-3176	2	25	of	of	ADP
ejpam-3176	2	26	linear	linear	ADJ
ejpam-3176	2	27	discriminant	discriminant	ADJ
ejpam-3176	2	28	analysis	analysis	NOUN
ejpam-3176	2	29	using	use	VERB
ejpam-3176	2	30	different	different	ADJ
ejpam-3176	2	31	robust	robust	ADJ
ejpam-3176	2	32	methods	method	NOUN
ejpam-3176	2	33	mufda	mufda	PROPN
ejpam-3176	2	34	j.	j.	PROPN
ejpam-3176	2	35	alrawashdeh1,∗	alrawashdeh1,∗	PROPN
ejpam-3176	2	36	,	,	PUNCT
ejpam-3176	2	37	taha	taha	PROPN
ejpam-3176	2	38	radwan1,2	radwan1,2	PROPN
ejpam-3176	2	39	and	and	CCONJ
ejpam-3176	2	40	khalid	khalid	PROPN
ejpam-3176	2	41	abunawas1	abunawas1	PROPN
ejpam-3176	2	42	1	1	NUM
ejpam-3176	2	43	department	department	NOUN
ejpam-3176	2	44	of	of	ADP
ejpam-3176	2	45	mathematics	mathematic	NOUN
ejpam-3176	2	46	,	,	PUNCT
ejpam-3176	2	47	college	college	NOUN
ejpam-3176	2	48	of	of	ADP
ejpam-3176	2	49	sciences	science	NOUN
ejpam-3176	2	50	and	and	CCONJ
ejpam-3176	2	51	arts	art	NOUN
ejpam-3176	2	52	,	,	PUNCT
ejpam-3176	2	53	al	al	PROPN
ejpam-3176	2	54	-	-	PUNCT
ejpam-3176	2	55	rass	rass	PROPN
ejpam-3176	2	56	,	,	PUNCT
ejpam-3176	2	57	qassim	qassim	PROPN
ejpam-3176	2	58	university	university	NOUN
ejpam-3176	2	59	,	,	PUNCT
ejpam-3176	2	60	kingdom	kingdom	NOUN
ejpam-3176	2	61	of	of	ADP
ejpam-3176	2	62	saudi	saudi	PROPN
ejpam-3176	2	63	arabia	arabia	PROPN
ejpam-3176	2	64	.	.	PUNCT
ejpam-3176	3	1	2	2	NUM
ejpam-3176	3	2	department	department	NOUN
ejpam-3176	3	3	of	of	ADP
ejpam-3176	3	4	mathematics	mathematic	NOUN
ejpam-3176	3	5	and	and	CCONJ
ejpam-3176	3	6	statistics	statistic	NOUN
ejpam-3176	3	7	,	,	PUNCT
ejpam-3176	3	8	port	port	NOUN
ejpam-3176	3	9	said	say	VERB
ejpam-3176	3	10	university	university	NOUN
ejpam-3176	3	11	,	,	PUNCT
ejpam-3176	3	12	port	port	NOUN
ejpam-3176	3	13	said	say	VERB
ejpam-3176	3	14	,	,	PUNCT
ejpam-3176	3	15	egypt	egypt	PROPN
ejpam-3176	3	16	abstract	abstract	PROPN
ejpam-3176	3	17	.	.	PUNCT
ejpam-3176	4	1	this	this	DET
ejpam-3176	4	2	study	study	NOUN
ejpam-3176	4	3	aims	aim	VERB
ejpam-3176	4	4	to	to	PART
ejpam-3176	4	5	combine	combine	VERB
ejpam-3176	4	6	the	the	DET
ejpam-3176	4	7	new	new	ADJ
ejpam-3176	4	8	deterministic	deterministic	ADJ
ejpam-3176	4	9	minimum	minimum	NOUN
ejpam-3176	4	10	covariance	covariance	NOUN
ejpam-3176	4	11	determinant	determinant	ADJ
ejpam-3176	4	12	(	(	PUNCT
ejpam-3176	4	13	detmcd	detmcd	NOUN
ejpam-3176	4	14	)	)	PUNCT
ejpam-3176	4	15	algorithm	algorithm	NOUN
ejpam-3176	4	16	with	with	ADP
ejpam-3176	4	17	linear	linear	ADJ
ejpam-3176	4	18	discriminant	discriminant	ADJ
ejpam-3176	4	19	analysis	analysis	NOUN
ejpam-3176	4	20	(	(	PUNCT
ejpam-3176	4	21	lda	lda	PROPN
ejpam-3176	4	22	)	)	PUNCT
ejpam-3176	4	23	and	and	CCONJ
ejpam-3176	4	24	compare	compare	VERB
ejpam-3176	4	25	it	it	PRON
ejpam-3176	4	26	with	with	ADP
ejpam-3176	4	27	the	the	DET
ejpam-3176	4	28	fast	fast	ADJ
ejpam-3176	4	29	minimum	minimum	NOUN
ejpam-3176	4	30	covariance	covariance	NOUN
ejpam-3176	4	31	determinant	determinant	ADJ
ejpam-3176	4	32	(	(	PUNCT
ejpam-3176	4	33	fastmcd	fastmcd	NOUN
ejpam-3176	4	34	)	)	PUNCT
ejpam-3176	4	35	,	,	PUNCT
ejpam-3176	4	36	fast	fast	ADV
ejpam-3176	4	37	consistent	consistent	ADJ
ejpam-3176	4	38	high	high	ADJ
ejpam-3176	4	39	breakdown	breakdown	NOUN
ejpam-3176	4	40	(	(	PUNCT
ejpam-3176	4	41	fch	fch	PROPN
ejpam-3176	4	42	)	)	PUNCT
ejpam-3176	4	43	,	,	PUNCT
ejpam-3176	4	44	and	and	CCONJ
ejpam-3176	4	45	robust	robust	ADJ
ejpam-3176	4	46	fch	fch	NOUN
ejpam-3176	4	47	(	(	PUNCT
ejpam-3176	4	48	rfch	rfch	NOUN
ejpam-3176	4	49	)	)	PUNCT
ejpam-3176	4	50	algorithms	algorithm	NOUN
ejpam-3176	4	51	.	.	PUNCT
ejpam-3176	5	1	lda	lda	PROPN
ejpam-3176	5	2	classifies	classify	VERB
ejpam-3176	5	3	new	new	ADJ
ejpam-3176	5	4	observations	observation	NOUN
ejpam-3176	5	5	into	into	ADP
ejpam-3176	5	6	one	one	NUM
ejpam-3176	5	7	of	of	ADP
ejpam-3176	5	8	the	the	DET
ejpam-3176	5	9	unknown	unknown	ADJ
ejpam-3176	5	10	groups	group	NOUN
ejpam-3176	5	11	and	and	CCONJ
ejpam-3176	5	12	it	it	PRON
ejpam-3176	5	13	is	be	AUX
ejpam-3176	5	14	widely	widely	ADV
ejpam-3176	5	15	used	use	VERB
ejpam-3176	5	16	in	in	ADP
ejpam-3176	5	17	multivariate	multivariate	NOUN
ejpam-3176	5	18	statistical	statistical	ADJ
ejpam-3176	5	19	analysis	analysis	NOUN
ejpam-3176	5	20	.	.	PUNCT
ejpam-3176	6	1	the	the	DET
ejpam-3176	6	2	lda	lda	PROPN
ejpam-3176	6	3	mean	mean	VERB
ejpam-3176	6	4	and	and	CCONJ
ejpam-3176	6	5	covariance	covariance	NOUN
ejpam-3176	6	6	matrix	matrix	NOUN
ejpam-3176	6	7	parameters	parameter	NOUN
ejpam-3176	6	8	are	be	AUX
ejpam-3176	6	9	highly	highly	ADV
ejpam-3176	6	10	influenced	influence	VERB
ejpam-3176	6	11	by	by	ADP
ejpam-3176	6	12	outliers	outlier	NOUN
ejpam-3176	6	13	.	.	PUNCT
ejpam-3176	7	1	the	the	DET
ejpam-3176	7	2	detmcd	detmcd	PROPN
ejpam-3176	7	3	algorithm	algorithm	NOUN
ejpam-3176	7	4	is	be	AUX
ejpam-3176	7	5	highly	highly	ADV
ejpam-3176	7	6	robust	robust	ADJ
ejpam-3176	7	7	and	and	CCONJ
ejpam-3176	7	8	resistant	resistant	ADJ
ejpam-3176	7	9	to	to	ADP
ejpam-3176	7	10	outliers	outlier	NOUN
ejpam-3176	7	11	and	and	CCONJ
ejpam-3176	7	12	it	it	PRON
ejpam-3176	7	13	is	be	AUX
ejpam-3176	7	14	constructed	construct	VERB
ejpam-3176	7	15	to	to	PART
ejpam-3176	7	16	overcome	overcome	VERB
ejpam-3176	7	17	the	the	DET
ejpam-3176	7	18	outlier	outlier	ADJ
ejpam-3176	7	19	problem	problem	NOUN
ejpam-3176	7	20	.	.	PUNCT
ejpam-3176	8	1	moreover	moreover	ADV
ejpam-3176	8	2	,	,	PUNCT
ejpam-3176	8	3	the	the	DET
ejpam-3176	8	4	detmcd	detmcd	NOUN
ejpam-3176	8	5	algorithm	algorithm	NOUN
ejpam-3176	8	6	is	be	AUX
ejpam-3176	8	7	used	use	VERB
ejpam-3176	8	8	to	to	PART
ejpam-3176	8	9	estimate	estimate	VERB
ejpam-3176	8	10	location	location	NOUN
ejpam-3176	8	11	and	and	CCONJ
ejpam-3176	8	12	scatter	scatter	NOUN
ejpam-3176	8	13	matrices	matrix	NOUN
ejpam-3176	8	14	.	.	PUNCT
ejpam-3176	9	1	the	the	DET
ejpam-3176	9	2	detmcd	detmcd	NOUN
ejpam-3176	9	3	,	,	PUNCT
ejpam-3176	9	4	fastmcd	fastmcd	NOUN
ejpam-3176	9	5	,	,	PUNCT
ejpam-3176	9	6	fch	fch	PROPN
ejpam-3176	9	7	,	,	PUNCT
ejpam-3176	9	8	and	and	CCONJ
ejpam-3176	9	9	rfch	rfch	NOUN
ejpam-3176	9	10	algorithms	algorithm	NOUN
ejpam-3176	9	11	are	be	AUX
ejpam-3176	9	12	applied	apply	VERB
ejpam-3176	9	13	to	to	PART
ejpam-3176	9	14	estimate	estimate	VERB
ejpam-3176	9	15	misclassification	misclassification	NOUN
ejpam-3176	9	16	probability	probability	NOUN
ejpam-3176	9	17	using	use	VERB
ejpam-3176	9	18	robust	robust	ADJ
ejpam-3176	9	19	lda	lda	PROPN
ejpam-3176	9	20	.	.	PUNCT
ejpam-3176	10	1	all	all	DET
ejpam-3176	10	2	the	the	DET
ejpam-3176	10	3	algorithms	algorithm	NOUN
ejpam-3176	10	4	are	be	AUX
ejpam-3176	10	5	expected	expect	VERB
ejpam-3176	10	6	to	to	PART
ejpam-3176	10	7	improve	improve	VERB
ejpam-3176	10	8	the	the	DET
ejpam-3176	10	9	lda	lda	PROPN
ejpam-3176	10	10	model	model	NOUN
ejpam-3176	10	11	for	for	ADP
ejpam-3176	10	12	classification	classification	NOUN
ejpam-3176	10	13	purposes	purpose	NOUN
ejpam-3176	10	14	in	in	ADP
ejpam-3176	10	15	banks	bank	NOUN
ejpam-3176	10	16	,	,	PUNCT
ejpam-3176	10	17	such	such	ADJ
ejpam-3176	10	18	as	as	ADP
ejpam-3176	10	19	bankruptcy	bankruptcy	NOUN
ejpam-3176	10	20	and	and	CCONJ
ejpam-3176	10	21	failures	failure	NOUN
ejpam-3176	10	22	,	,	PUNCT
ejpam-3176	10	23	and	and	CCONJ
ejpam-3176	10	24	to	to	PART
ejpam-3176	10	25	distinguish	distinguish	VERB
ejpam-3176	10	26	between	between	ADP
ejpam-3176	10	27	islamic	islamic	ADJ
ejpam-3176	10	28	and	and	CCONJ
ejpam-3176	10	29	conventional	conventional	ADJ
ejpam-3176	10	30	banks	bank	NOUN
ejpam-3176	10	31	.	.	PUNCT
ejpam-3176	11	1	the	the	DET
ejpam-3176	11	2	performances	performance	NOUN
ejpam-3176	11	3	of	of	ADP
ejpam-3176	11	4	the	the	DET
ejpam-3176	11	5	estimators	estimator	NOUN
ejpam-3176	11	6	are	be	AUX
ejpam-3176	11	7	investigated	investigate	VERB
ejpam-3176	11	8	through	through	ADP
ejpam-3176	11	9	simulation	simulation	NOUN
ejpam-3176	11	10	and	and	CCONJ
ejpam-3176	11	11	actual	actual	ADJ
ejpam-3176	11	12	data	datum	NOUN
ejpam-3176	11	13	..	..	PUNCT
ejpam-3176	11	14	key	key	ADJ
ejpam-3176	11	15	words	word	NOUN
ejpam-3176	11	16	and	and	CCONJ
ejpam-3176	11	17	phrases	phrase	NOUN
ejpam-3176	11	18	:	:	PUNCT
ejpam-3176	11	19	detmcd	detmcd	ADJ
ejpam-3176	11	20	,	,	PUNCT
ejpam-3176	11	21	fastmcd	fastmcd	NOUN
ejpam-3176	11	22	,	,	PUNCT
ejpam-3176	11	23	financial	financial	ADJ
ejpam-3176	11	24	ratios	ratio	NOUN
ejpam-3176	11	25	,	,	PUNCT
ejpam-3176	11	26	linear	linear	ADJ
ejpam-3176	11	27	discriminant	discriminant	ADJ
ejpam-3176	11	28	analysis	analysis	NOUN
ejpam-3176	11	29	,	,	PUNCT
ejpam-3176	11	30	orthogonalized	orthogonalized	ADJ
ejpam-3176	11	31	gnanadesikankettenring	gnanadesikankettenring	NOUN
ejpam-3176	11	32	,	,	PUNCT
ejpam-3176	11	33	fch	fch	PROPN
ejpam-3176	11	34	,	,	PUNCT
ejpam-3176	11	35	rfch	rfch	NOUN
ejpam-3176	11	36	.	.	PUNCT
ejpam-3176	12	1	1	1	X
ejpam-3176	12	2	.	.	X
ejpam-3176	12	3	introduction	introduction	NOUN
ejpam-3176	12	4	linear	linear	PROPN
ejpam-3176	12	5	discriminant	discriminant	ADJ
ejpam-3176	12	6	analysis	analysis	NOUN
ejpam-3176	12	7	(	(	PUNCT
ejpam-3176	12	8	lda	lda	PROPN
ejpam-3176	12	9	)	)	PUNCT
ejpam-3176	12	10	is	be	AUX
ejpam-3176	12	11	widely	widely	ADV
ejpam-3176	12	12	used	use	VERB
ejpam-3176	12	13	in	in	ADP
ejpam-3176	12	14	multivariate	multivariate	NOUN
ejpam-3176	12	15	statistical	statistical	ADJ
ejpam-3176	12	16	techniques	technique	NOUN
ejpam-3176	12	17	for	for	ADP
ejpam-3176	12	18	data	datum	NOUN
ejpam-3176	12	19	analysis	analysis	NOUN
ejpam-3176	12	20	.	.	PUNCT
ejpam-3176	13	1	rules	rule	NOUN
ejpam-3176	13	2	that	that	PRON
ejpam-3176	13	3	describe	describe	VERB
ejpam-3176	13	4	separation	separation	NOUN
ejpam-3176	13	5	among	among	ADP
ejpam-3176	13	6	groups	group	NOUN
ejpam-3176	13	7	are	be	AUX
ejpam-3176	13	8	obtained	obtain	VERB
ejpam-3176	13	9	through	through	ADP
ejpam-3176	13	10	lda	lda	PROPN
ejpam-3176	13	11	.	.	PUNCT
ejpam-3176	14	1	variables	variable	NOUN
ejpam-3176	14	2	are	be	AUX
ejpam-3176	14	3	assumed	assume	VERB
ejpam-3176	14	4	to	to	PART
ejpam-3176	14	5	be	be	AUX
ejpam-3176	14	6	normally	normally	ADV
ejpam-3176	14	7	distributed	distribute	VERB
ejpam-3176	14	8	with	with	ADP
ejpam-3176	14	9	the	the	DET
ejpam-3176	14	10	equal	equal	ADJ
ejpam-3176	14	11	covariance	covariance	NOUN
ejpam-3176	14	12	matrix	matrix	NOUN
ejpam-3176	14	13	∑	∑	PUNCT
ejpam-3176	14	14	.	.	PUNCT
ejpam-3176	15	1	lda	lda	PROPN
ejpam-3176	15	2	is	be	AUX
ejpam-3176	15	3	highly	highly	ADV
ejpam-3176	15	4	sensitive	sensitive	ADJ
ejpam-3176	15	5	to	to	ADP
ejpam-3176	15	6	outlier	outlier	NOUN
ejpam-3176	15	7	observations	observation	NOUN
ejpam-3176	15	8	.	.	PUNCT
ejpam-3176	16	1	hence	hence	ADV
ejpam-3176	16	2	,	,	PUNCT
ejpam-3176	16	3	estimating	estimate	VERB
ejpam-3176	16	4	lda	lda	ADJ
ejpam-3176	16	5	parameters	parameter	NOUN
ejpam-3176	16	6	using	use	VERB
ejpam-3176	16	7	the	the	DET
ejpam-3176	16	8	classical	classical	ADJ
ejpam-3176	16	9	approach	approach	NOUN
ejpam-3176	16	10	will	will	AUX
ejpam-3176	16	11	affect	affect	VERB
ejpam-3176	16	12	the	the	DET
ejpam-3176	16	13	values	value	NOUN
ejpam-3176	16	14	of	of	ADP
ejpam-3176	16	15	parameters	parameter	NOUN
ejpam-3176	16	16	.	.	PUNCT
ejpam-3176	17	1	robust	robust	ADJ
ejpam-3176	17	2	estimators	estimator	NOUN
ejpam-3176	17	3	have	have	AUX
ejpam-3176	17	4	been	be	AUX
ejpam-3176	17	5	proposed	propose	VERB
ejpam-3176	17	6	to	to	PART
ejpam-3176	17	7	limit	limit	VERB
ejpam-3176	17	8	the	the	DET
ejpam-3176	17	9	effects	effect	NOUN
ejpam-3176	17	10	of	of	ADP
ejpam-3176	17	11	outlier	outlier	NOUN
ejpam-3176	17	12	observations	observation	NOUN
ejpam-3176	17	13	,	,	PUNCT
ejpam-3176	17	14	and	and	CCONJ
ejpam-3176	17	15	certain	certain	ADJ
ejpam-3176	17	16	methods	method	NOUN
ejpam-3176	17	17	have	have	AUX
ejpam-3176	17	18	been	be	AUX
ejpam-3176	17	19	presented	present	VERB
ejpam-3176	17	20	to	to	PART
ejpam-3176	17	21	overcome	overcome	VERB
ejpam-3176	17	22	the	the	DET
ejpam-3176	17	23	outlier	outlier	ADJ
ejpam-3176	17	24	problem	problem	NOUN
ejpam-3176	17	25	,	,	PUNCT
ejpam-3176	17	26	such	such	ADJ
ejpam-3176	17	27	as	as	ADP
ejpam-3176	17	28	the	the	DET
ejpam-3176	17	29	high	high	ADJ
ejpam-3176	17	30	breakdown	breakdown	ADJ
ejpam-3176	17	31	criterion	criterion	NOUN
ejpam-3176	17	32	developed	develop	VERB
ejpam-3176	17	33	by	by	ADP
ejpam-3176	17	34	hawkins	hawkin	NOUN
ejpam-3176	17	35	and	and	CCONJ
ejpam-3176	17	36	mclachlan	mclachlan	NOUN
ejpam-3176	18	1	[	[	X
ejpam-3176	18	2	11	11	NUM
ejpam-3176	18	3	]	]	PUNCT
ejpam-3176	18	4	.	.	PUNCT
ejpam-3176	19	1	croux	croux	VERB
ejpam-3176	19	2	et	et	PROPN
ejpam-3176	19	3	al	al	PROPN
ejpam-3176	19	4	.	.	PUNCT
ejpam-3176	20	1	[	[	X
ejpam-3176	20	2	5	5	NUM
ejpam-3176	20	3	]	]	PUNCT
ejpam-3176	20	4	investigated	investigate	VERB
ejpam-3176	20	5	the	the	DET
ejpam-3176	20	6	classification	classification	NOUN
ejpam-3176	20	7	efficiencies	efficiency	NOUN
ejpam-3176	20	8	∗corresponding	∗corresponde	VERB
ejpam-3176	20	9	author	author	NOUN
ejpam-3176	20	10	.	.	PUNCT
ejpam-3176	21	1	email	email	NOUN
ejpam-3176	21	2	addresses	address	NOUN
ejpam-3176	21	3	:	:	PUNCT
ejpam-3176	21	4	mufdajr@yahoo.com	mufdajr@yahoo.com	X
ejpam-3176	21	5	(	(	PUNCT
ejpam-3176	21	6	m.	m.	NOUN
ejpam-3176	21	7	j.	j.	PROPN
ejpam-3176	21	8	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	21	9	)	)	PUNCT
ejpam-3176	21	10	,	,	PUNCT
ejpam-3176	21	11	taha	taha	PROPN
ejpam-3176	21	12	ali	ali	PROPN
ejpam-3176	21	13	2003@hotmail.com	2003@hotmail.com	X
ejpam-3176	21	14	(	(	PUNCT
ejpam-3176	21	15	t.	t.	NOUN
ejpam-3176	21	16	radwan	radwan	PROPN
ejpam-3176	21	17	)	)	PUNCT
ejpam-3176	21	18	and	and	CCONJ
ejpam-3176	21	19	xsx6666@gmail.com	xsx6666@gmail.com	PUNCT
ejpam-3176	22	1	(	(	PUNCT
ejpam-3176	22	2	kh	kh	PROPN
ejpam-3176	22	3	.	.	PROPN
ejpam-3176	22	4	abunawas	abunawas	PROPN
ejpam-3176	22	5	)	)	PUNCT
ejpam-3176	22	6	http://www.ejpam.com	http://www.ejpam.com	X
ejpam-3176	23	1	284	284	NUM
ejpam-3176	24	1	c	c	X
ejpam-3176	24	2	©	©	PROPN
ejpam-3176	24	3	2018	2018	NUM
ejpam-3176	24	4	ejpam	ejpam	VERB
ejpam-3176	24	5	all	all	DET
ejpam-3176	24	6	rights	right	NOUN
ejpam-3176	24	7	reserved	reserve	VERB
ejpam-3176	24	8	.	.	PUNCT
ejpam-3176	25	1	mufda	mufda	PROPN
ejpam-3176	25	2	j.	j.	PROPN
ejpam-3176	25	3	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	25	4	,	,	PUNCT
ejpam-3176	25	5	taha	taha	PROPN
ejpam-3176	25	6	radwan	radwan	PROPN
ejpam-3176	25	7	and	and	CCONJ
ejpam-3176	25	8	khalid	khalid	PROPN
ejpam-3176	25	9	abunawas	abunawas	PROPN
ejpam-3176	25	10	/	/	SYM
ejpam-3176	25	11	eur	eur	PROPN
ejpam-3176	25	12	.	.	PUNCT
ejpam-3176	26	1	j.	j.	PROPN
ejpam-3176	26	2	pure	pure	PROPN
ejpam-3176	26	3	appl	appl	PROPN
ejpam-3176	26	4	.	.	PROPN
ejpam-3176	26	5	math	math	PROPN
ejpam-3176	26	6	,	,	PUNCT
ejpam-3176	26	7	11	11	NUM
ejpam-3176	26	8	(	(	PUNCT
ejpam-3176	26	9	1	1	NUM
ejpam-3176	26	10	)	)	PUNCT
ejpam-3176	26	11	(	(	PUNCT
ejpam-3176	26	12	2018	2018	NUM
ejpam-3176	26	13	)	)	PUNCT
ejpam-3176	26	14	,	,	PUNCT
ejpam-3176	26	15	284	284	NUM
ejpam-3176	26	16	-	-	SYM
ejpam-3176	26	17	298	298	NUM
ejpam-3176	26	18	285	285	NUM
ejpam-3176	26	19	of	of	ADP
ejpam-3176	26	20	robust	robust	ADJ
ejpam-3176	26	21	procedures	procedure	NOUN
ejpam-3176	26	22	with	with	ADP
ejpam-3176	26	23	respect	respect	NOUN
ejpam-3176	26	24	to	to	ADP
ejpam-3176	26	25	the	the	DET
ejpam-3176	26	26	classical	classical	ADJ
ejpam-3176	26	27	method	method	NOUN
ejpam-3176	26	28	.	.	PUNCT
ejpam-3176	27	1	rousseeuw	rousseeuw	PROPN
ejpam-3176	28	1	[	[	X
ejpam-3176	28	2	23	23	NUM
ejpam-3176	28	3	]	]	PUNCT
ejpam-3176	28	4	introduced	introduce	VERB
ejpam-3176	28	5	the	the	DET
ejpam-3176	28	6	minimum	minimum	ADJ
ejpam-3176	28	7	covariance	covariance	NOUN
ejpam-3176	28	8	determinant	determinant	ADJ
ejpam-3176	28	9	(	(	PUNCT
ejpam-3176	28	10	mcd	mcd	NOUN
ejpam-3176	28	11	)	)	PUNCT
ejpam-3176	28	12	estimator	estimator	NOUN
ejpam-3176	28	13	.	.	PUNCT
ejpam-3176	28	14	rousseeuw	rousseeuw	PROPN
ejpam-3176	28	15	and	and	CCONJ
ejpam-3176	28	16	driessen	driessen	NOUN
ejpam-3176	29	1	[	[	X
ejpam-3176	29	2	25	25	NUM
ejpam-3176	29	3	]	]	PUNCT
ejpam-3176	29	4	developed	develop	VERB
ejpam-3176	29	5	a	a	DET
ejpam-3176	29	6	new	new	ADJ
ejpam-3176	29	7	estimator	estimator	NOUN
ejpam-3176	29	8	called	call	VERB
ejpam-3176	29	9	fast	fast	ADJ
ejpam-3176	29	10	minimum	minimum	NOUN
ejpam-3176	29	11	covariance	covariance	NOUN
ejpam-3176	29	12	determinant	determinant	ADJ
ejpam-3176	29	13	(	(	PUNCT
ejpam-3176	29	14	fastmcd	fastmcd	NOUN
ejpam-3176	29	15	)	)	PUNCT
ejpam-3176	29	16	,	,	PUNCT
ejpam-3176	29	17	which	which	PRON
ejpam-3176	29	18	is	be	AUX
ejpam-3176	29	19	a	a	DET
ejpam-3176	29	20	highly	highly	ADV
ejpam-3176	29	21	robust	robust	ADJ
ejpam-3176	29	22	estimator	estimator	NOUN
ejpam-3176	29	23	for	for	ADP
ejpam-3176	29	24	observing	observe	VERB
ejpam-3176	29	25	outliers	outlier	NOUN
ejpam-3176	29	26	,	,	PUNCT
ejpam-3176	29	27	to	to	PART
ejpam-3176	29	28	fill	fill	VERB
ejpam-3176	29	29	the	the	DET
ejpam-3176	29	30	gap	gap	NOUN
ejpam-3176	29	31	in	in	ADP
ejpam-3176	29	32	contaminated	contaminated	ADJ
ejpam-3176	29	33	datasets	dataset	NOUN
ejpam-3176	29	34	.	.	PUNCT
ejpam-3176	30	1	todorov	todorov	PROPN
ejpam-3176	31	1	[	[	X
ejpam-3176	31	2	27	27	NUM
ejpam-3176	31	3	]	]	PUNCT
ejpam-3176	31	4	constructed	construct	VERB
ejpam-3176	31	5	the	the	DET
ejpam-3176	31	6	robust	robust	ADJ
ejpam-3176	31	7	wilks	wilk	NOUN
ejpam-3176	31	8	lambda	lambda	NOUN
ejpam-3176	31	9	,	,	PUNCT
ejpam-3176	31	10	which	which	PRON
ejpam-3176	31	11	is	be	AUX
ejpam-3176	31	12	uninfluenced	uninfluence	VERB
ejpam-3176	31	13	by	by	ADP
ejpam-3176	31	14	contaminated	contaminate	VERB
ejpam-3176	31	15	data	datum	NOUN
ejpam-3176	31	16	,	,	PUNCT
ejpam-3176	31	17	based	base	VERB
ejpam-3176	31	18	on	on	ADP
ejpam-3176	31	19	fastmcd	fastmcd	NOUN
ejpam-3176	31	20	to	to	PART
ejpam-3176	31	21	avoid	avoid	VERB
ejpam-3176	31	22	the	the	DET
ejpam-3176	31	23	outlier	outlier	ADJ
ejpam-3176	31	24	problem	problem	NOUN
ejpam-3176	31	25	.	.	PUNCT
ejpam-3176	32	1	fastmcd	fastmcd	NOUN
ejpam-3176	32	2	has	have	AUX
ejpam-3176	32	3	been	be	AUX
ejpam-3176	32	4	applied	apply	VERB
ejpam-3176	32	5	to	to	ADP
ejpam-3176	32	6	lda	lda	PROPN
ejpam-3176	32	7	(	(	PUNCT
ejpam-3176	32	8	he	he	PRON
ejpam-3176	32	9	and	and	CCONJ
ejpam-3176	32	10	fung	fung	PROPN
ejpam-3176	33	1	[	[	X
ejpam-3176	33	2	12	12	NUM
ejpam-3176	33	3	]	]	X
ejpam-3176	33	4	;	;	PUNCT
ejpam-3176	33	5	hubert	hubert	PROPN
ejpam-3176	33	6	and	and	CCONJ
ejpam-3176	33	7	van	van	PROPN
ejpam-3176	33	8	driessen	driessen	PROPN
ejpam-3176	33	9	[	[	X
ejpam-3176	33	10	14	14	NUM
ejpam-3176	33	11	]	]	PUNCT
ejpam-3176	33	12	)	)	PUNCT
ejpam-3176	33	13	and	and	CCONJ
ejpam-3176	33	14	to	to	ADP
ejpam-3176	33	15	different	different	ADJ
ejpam-3176	33	16	aspects	aspect	NOUN
ejpam-3176	33	17	of	of	ADP
ejpam-3176	33	18	science	science	NOUN
ejpam-3176	33	19	(	(	PUNCT
ejpam-3176	33	20	hubert	hubert	PROPN
ejpam-3176	33	21	et	et	PROPN
ejpam-3176	33	22	al	al	PROPN
ejpam-3176	33	23	.	.	PUNCT
ejpam-3176	34	1	[	[	X
ejpam-3176	34	2	16	16	NUM
ejpam-3176	34	3	]	]	SYM
ejpam-3176	34	4	)	)	PUNCT
ejpam-3176	34	5	.	.	PUNCT
ejpam-3176	35	1	fastmcd	fastmcd	NOUN
ejpam-3176	35	2	is	be	AUX
ejpam-3176	35	3	also	also	ADV
ejpam-3176	35	4	used	use	VERB
ejpam-3176	35	5	in	in	ADP
ejpam-3176	35	6	many	many	ADJ
ejpam-3176	35	7	multivariate	multivariate	NOUN
ejpam-3176	35	8	techniques	technique	NOUN
ejpam-3176	35	9	,	,	PUNCT
ejpam-3176	35	10	such	such	ADJ
ejpam-3176	35	11	as	as	ADP
ejpam-3176	35	12	the	the	DET
ejpam-3176	35	13	principle	principle	ADJ
ejpam-3176	35	14	component	component	NOUN
ejpam-3176	35	15	analysis	analysis	NOUN
ejpam-3176	35	16	(	(	PUNCT
ejpam-3176	35	17	croux	croux	NOUN
ejpam-3176	35	18	and	and	CCONJ
ejpam-3176	35	19	haesbroeck	haesbroeck	NOUN
ejpam-3176	36	1	[	[	X
ejpam-3176	36	2	7	7	NUM
ejpam-3176	36	3	]	]	PUNCT
ejpam-3176	36	4	;	;	PUNCT
ejpam-3176	36	5	hubert	hubert	PROPN
ejpam-3176	36	6	et	et	PROPN
ejpam-3176	36	7	al	al	PROPN
ejpam-3176	36	8	.	.	PUNCT
ejpam-3176	37	1	[	[	X
ejpam-3176	37	2	17	17	NUM
ejpam-3176	37	3	]	]	NUM
ejpam-3176	37	4	)	)	PUNCT
ejpam-3176	37	5	,	,	PUNCT
ejpam-3176	37	6	factor	factor	NOUN
ejpam-3176	37	7	analysis	analysis	NOUN
ejpam-3176	37	8	(	(	PUNCT
ejpam-3176	37	9	pison	pison	NOUN
ejpam-3176	37	10	et	et	PROPN
ejpam-3176	37	11	al	al	PROPN
ejpam-3176	37	12	.	.	PUNCT
ejpam-3176	38	1	[	[	X
ejpam-3176	38	2	22	22	NUM
ejpam-3176	38	3	]	]	SYM
ejpam-3176	38	4	)	)	PUNCT
ejpam-3176	38	5	,	,	PUNCT
ejpam-3176	38	6	classifications	classification	NOUN
ejpam-3176	38	7	,	,	PUNCT
ejpam-3176	38	8	and	and	CCONJ
ejpam-3176	38	9	clustering	cluster	VERB
ejpam-3176	38	10	techniques	technique	NOUN
ejpam-3176	38	11	.	.	PUNCT
ejpam-3176	39	1	in	in	ADP
ejpam-3176	39	2	a	a	DET
ejpam-3176	39	3	multivariate	multivariate	NOUN
ejpam-3176	39	4	time	time	NOUN
ejpam-3176	39	5	series	series	NOUN
ejpam-3176	39	6	,	,	PUNCT
ejpam-3176	39	7	croux	croux	VERB
ejpam-3176	39	8	et	et	PROPN
ejpam-3176	39	9	al	al	PROPN
ejpam-3176	39	10	.	.	PUNCT
ejpam-3176	40	1	[	[	X
ejpam-3176	40	2	6	6	NUM
ejpam-3176	40	3	]	]	PUNCT
ejpam-3176	40	4	proposed	propose	VERB
ejpam-3176	40	5	the	the	DET
ejpam-3176	40	6	robust	robust	ADJ
ejpam-3176	40	7	exponential	exponential	ADJ
ejpam-3176	40	8	smoothing	smoothing	NOUN
ejpam-3176	40	9	of	of	ADP
ejpam-3176	40	10	a	a	DET
ejpam-3176	40	11	multivariate	multivariate	NOUN
ejpam-3176	40	12	time	time	NOUN
ejpam-3176	40	13	series	series	NOUN
ejpam-3176	40	14	to	to	PART
ejpam-3176	40	15	enhance	enhance	VERB
ejpam-3176	40	16	the	the	DET
ejpam-3176	40	17	robustness	robustness	NOUN
ejpam-3176	40	18	of	of	ADP
ejpam-3176	40	19	estimates	estimate	NOUN
ejpam-3176	40	20	for	for	ADP
ejpam-3176	40	21	contaminated	contaminated	ADJ
ejpam-3176	40	22	data	datum	NOUN
ejpam-3176	40	23	.	.	PUNCT
ejpam-3176	41	1	different	different	ADJ
ejpam-3176	41	2	techniques	technique	NOUN
ejpam-3176	41	3	and	and	CCONJ
ejpam-3176	41	4	approaches	approach	NOUN
ejpam-3176	41	5	use	use	VERB
ejpam-3176	41	6	features	feature	NOUN
ejpam-3176	41	7	of	of	ADP
ejpam-3176	41	8	mcd	mcd	NOUN
ejpam-3176	41	9	to	to	PART
ejpam-3176	41	10	improve	improve	VERB
ejpam-3176	41	11	the	the	DET
ejpam-3176	41	12	robustness	robustness	NOUN
ejpam-3176	41	13	of	of	ADP
ejpam-3176	41	14	parameter	parameter	NOUN
ejpam-3176	41	15	estimation	estimation	NOUN
ejpam-3176	41	16	.	.	PUNCT
ejpam-3176	42	1	hubert	hubert	PROPN
ejpam-3176	42	2	and	and	CCONJ
ejpam-3176	42	3	rousseeuw	rousseeuw	NOUN
ejpam-3176	42	4	[	[	X
ejpam-3176	42	5	15	15	NUM
ejpam-3176	42	6	]	]	PUNCT
ejpam-3176	42	7	presented	present	VERB
ejpam-3176	42	8	a	a	DET
ejpam-3176	42	9	robust	robust	ADJ
ejpam-3176	42	10	regression	regression	NOUN
ejpam-3176	42	11	method	method	NOUN
ejpam-3176	42	12	for	for	ADP
ejpam-3176	42	13	continuous	continuous	ADJ
ejpam-3176	42	14	situations	situation	NOUN
ejpam-3176	42	15	and	and	CCONJ
ejpam-3176	42	16	binary	binary	ADJ
ejpam-3176	42	17	regression	regression	NOUN
ejpam-3176	42	18	.	.	PUNCT
ejpam-3176	43	1	croux	croux	PROPN
ejpam-3176	43	2	and	and	CCONJ
ejpam-3176	43	3	dehon	dehon	NOUN
ejpam-3176	43	4	[	[	X
ejpam-3176	43	5	4	4	X
ejpam-3176	43	6	]	]	PUNCT
ejpam-3176	43	7	used	use	VERB
ejpam-3176	43	8	robust	robust	ADJ
ejpam-3176	43	9	canonical	canonical	ADJ
ejpam-3176	43	10	correlation	correlation	NOUN
ejpam-3176	43	11	.	.	PUNCT
ejpam-3176	44	1	hubert	hubert	PROPN
ejpam-3176	44	2	and	and	CCONJ
ejpam-3176	44	3	branden	branden	VERB
ejpam-3176	45	1	[	[	X
ejpam-3176	45	2	13	13	NUM
ejpam-3176	45	3	]	]	PUNCT
ejpam-3176	45	4	introduced	introduce	VERB
ejpam-3176	45	5	robustified	robustifie	VERB
ejpam-3176	45	6	versions	version	NOUN
ejpam-3176	45	7	of	of	ADP
ejpam-3176	45	8	the	the	DET
ejpam-3176	45	9	simpls	simpls	PROPN
ejpam-3176	45	10	algorithm	algorithm	PROPN
ejpam-3176	45	11	.	.	PUNCT
ejpam-3176	46	1	a	a	DET
ejpam-3176	46	2	robust	robust	ADJ
ejpam-3176	46	3	multivariate	multivariate	NOUN
ejpam-3176	46	4	calibration	calibration	NOUN
ejpam-3176	46	5	model	model	NOUN
ejpam-3176	46	6	was	be	AUX
ejpam-3176	46	7	used	use	VERB
ejpam-3176	46	8	by	by	ADP
ejpam-3176	46	9	hubert	hubert	NOUN
ejpam-3176	46	10	and	and	CCONJ
ejpam-3176	46	11	verboven	verboven	VERB
ejpam-3176	47	1	[	[	X
ejpam-3176	47	2	19	19	NUM
ejpam-3176	47	3	]	]	PUNCT
ejpam-3176	47	4	,	,	PUNCT
ejpam-3176	47	5	and	and	CCONJ
ejpam-3176	47	6	a	a	DET
ejpam-3176	47	7	robust	robust	ADJ
ejpam-3176	47	8	error	error	NOUN
ejpam-3176	47	9	in	in	ADP
ejpam-3176	47	10	variable	variable	ADJ
ejpam-3176	47	11	regression	regression	NOUN
ejpam-3176	47	12	was	be	AUX
ejpam-3176	47	13	used	use	VERB
ejpam-3176	47	14	by	by	ADP
ejpam-3176	47	15	fekri	fekri	NOUN
ejpam-3176	47	16	and	and	CCONJ
ejpam-3176	47	17	ruiz	ruiz	NOUN
ejpam-3176	47	18	-	-	PUNCT
ejpam-3176	47	19	gazen	gazen	NOUN
ejpam-3176	48	1	[	[	X
ejpam-3176	48	2	9	9	NUM
ejpam-3176	48	3	]	]	PUNCT
ejpam-3176	48	4	.	.	PUNCT
ejpam-3176	49	1	the	the	DET
ejpam-3176	49	2	mcd	mcd	PROPN
ejpam-3176	49	3	algorithm	algorithm	NOUN
ejpam-3176	49	4	was	be	AUX
ejpam-3176	49	5	used	use	VERB
ejpam-3176	49	6	for	for	ADP
ejpam-3176	49	7	a	a	DET
ejpam-3176	49	8	genetic	genetic	ADJ
ejpam-3176	49	9	algorithm	algorithm	NOUN
ejpam-3176	49	10	by	by	ADP
ejpam-3176	49	11	wiegand	wiegand	PROPN
ejpam-3176	49	12	et	et	PROPN
ejpam-3176	49	13	al	al	PROPN
ejpam-3176	49	14	.	.	PUNCT
ejpam-3176	50	1	[	[	X
ejpam-3176	50	2	29	29	NUM
ejpam-3176	50	3	]	]	SYM
ejpam-3176	50	4	)	)	PUNCT
ejpam-3176	50	5	.	.	PUNCT
ejpam-3176	51	1	hubert	hubert	PROPN
ejpam-3176	51	2	et	et	PROPN
ejpam-3176	51	3	al	al	PROPN
ejpam-3176	51	4	.	.	PUNCT
ejpam-3176	52	1	[	[	X
ejpam-3176	52	2	18	18	NUM
ejpam-3176	52	3	]	]	PUNCT
ejpam-3176	52	4	used	use	VERB
ejpam-3176	52	5	a	a	DET
ejpam-3176	52	6	new	new	ADJ
ejpam-3176	52	7	estimator	estimator	NOUN
ejpam-3176	52	8	deterministic	deterministic	ADJ
ejpam-3176	52	9	algorithm	algorithm	NOUN
ejpam-3176	52	10	for	for	ADP
ejpam-3176	52	11	robust	robust	ADJ
ejpam-3176	52	12	location	location	NOUN
ejpam-3176	52	13	and	and	CCONJ
ejpam-3176	52	14	scatter	scatter	NOUN
ejpam-3176	52	15	,	,	PUNCT
ejpam-3176	52	16	called	call	VERB
ejpam-3176	52	17	deterministic	deterministic	ADJ
ejpam-3176	52	18	minimum	minimum	ADJ
ejpam-3176	52	19	covariance	covariance	NOUN
ejpam-3176	52	20	determinant	determinant	ADJ
ejpam-3176	52	21	(	(	PUNCT
ejpam-3176	52	22	detmcd	detmcd	NOUN
ejpam-3176	52	23	)	)	PUNCT
ejpam-3176	52	24	,	,	PUNCT
ejpam-3176	52	25	and	and	CCONJ
ejpam-3176	52	26	compared	compare	VERB
ejpam-3176	52	27	it	it	PRON
ejpam-3176	52	28	with	with	ADP
ejpam-3176	52	29	two	two	NUM
ejpam-3176	52	30	estimators	estimator	NOUN
ejpam-3176	52	31	,	,	PUNCT
ejpam-3176	52	32	namely	namely	ADV
ejpam-3176	52	33	,	,	PUNCT
ejpam-3176	52	34	fastmcd	fastmcd	NOUN
ejpam-3176	52	35	and	and	CCONJ
ejpam-3176	52	36	orthogonalized	orthogonalized	ADJ
ejpam-3176	52	37	gnanadesikankettenring	gnanadesikankettenre	VERB
ejpam-3176	52	38	(	(	PUNCT
ejpam-3176	52	39	ogk	ogk	NOUN
ejpam-3176	52	40	)	)	PUNCT
ejpam-3176	52	41	,	,	PUNCT
ejpam-3176	52	42	of	of	ADP
ejpam-3176	52	43	maronna	maronna	NOUN
ejpam-3176	52	44	and	and	CCONJ
ejpam-3176	52	45	zamar	zamar	PROPN
ejpam-3176	53	1	[	[	X
ejpam-3176	53	2	20	20	NUM
ejpam-3176	53	3	]	]	PUNCT
ejpam-3176	53	4	.	.	PUNCT
ejpam-3176	54	1	the	the	DET
ejpam-3176	54	2	new	new	ADJ
ejpam-3176	54	3	estimator	estimator	NOUN
ejpam-3176	54	4	uses	use	VERB
ejpam-3176	54	5	the	the	DET
ejpam-3176	54	6	same	same	ADJ
ejpam-3176	54	7	iteration	iteration	NOUN
ejpam-3176	54	8	as	as	ADP
ejpam-3176	54	9	fastmcd	fastmcd	NOUN
ejpam-3176	54	10	but	but	CCONJ
ejpam-3176	54	11	does	do	AUX
ejpam-3176	54	12	not	not	PART
ejpam-3176	54	13	draw	draw	VERB
ejpam-3176	54	14	random	random	ADJ
ejpam-3176	54	15	subsets	subset	NOUN
ejpam-3176	54	16	,	,	PUNCT
ejpam-3176	54	17	whereas	whereas	SCONJ
ejpam-3176	54	18	fastmcd	fastmcd	NOUN
ejpam-3176	54	19	draws	draw	VERB
ejpam-3176	54	20	random	random	ADJ
ejpam-3176	54	21	subsets	subset	NOUN
ejpam-3176	54	22	of	of	ADP
ejpam-3176	54	23	size	size	NOUN
ejpam-3176	54	24	p	p	PROPN
ejpam-3176	54	25	+	+	PROPN
ejpam-3176	54	26	1	1	NUM
ejpam-3176	54	27	and	and	CCONJ
ejpam-3176	54	28	is	be	AUX
ejpam-3176	54	29	required	require	VERB
ejpam-3176	54	30	to	to	PART
ejpam-3176	54	31	draw	draw	VERB
ejpam-3176	54	32	several	several	ADJ
ejpam-3176	54	33	times	time	NOUN
ejpam-3176	54	34	to	to	PART
ejpam-3176	54	35	obtain	obtain	VERB
ejpam-3176	54	36	at	at	ADV
ejpam-3176	54	37	least	least	ADV
ejpam-3176	54	38	one	one	NUM
ejpam-3176	54	39	subset	subset	NOUN
ejpam-3176	54	40	that	that	PRON
ejpam-3176	54	41	is	be	AUX
ejpam-3176	54	42	free	free	ADJ
ejpam-3176	54	43	from	from	ADP
ejpam-3176	54	44	outliers	outlier	NOUN
ejpam-3176	54	45	.	.	PUNCT
ejpam-3176	55	1	detmcd	detmcd	NOUN
ejpam-3176	55	2	exhibited	exhibit	VERB
ejpam-3176	55	3	better	well	ADJ
ejpam-3176	55	4	performance	performance	NOUN
ejpam-3176	55	5	and	and	CCONJ
ejpam-3176	55	6	was	be	AUX
ejpam-3176	55	7	faster	fast	ADJ
ejpam-3176	55	8	than	than	ADP
ejpam-3176	55	9	the	the	DET
ejpam-3176	55	10	other	other	ADJ
ejpam-3176	55	11	two	two	NUM
ejpam-3176	55	12	estimators	estimator	NOUN
ejpam-3176	55	13	in	in	ADP
ejpam-3176	55	14	estimating	estimate	VERB
ejpam-3176	55	15	location	location	NOUN
ejpam-3176	55	16	and	and	CCONJ
ejpam-3176	55	17	scatter	scatter	NOUN
ejpam-3176	55	18	matrices	matrix	NOUN
ejpam-3176	55	19	.	.	PUNCT
ejpam-3176	56	1	olive	olive	NOUN
ejpam-3176	56	2	and	and	CCONJ
ejpam-3176	56	3	hawkins	hawkin	NOUN
ejpam-3176	57	1	[	[	X
ejpam-3176	57	2	21	21	NUM
ejpam-3176	57	3	]	]	PUNCT
ejpam-3176	57	4	proposed	propose	VERB
ejpam-3176	57	5	an	an	DET
ejpam-3176	57	6	easy	easy	ADJ
ejpam-3176	57	7	method	method	NOUN
ejpam-3176	57	8	for	for	ADP
ejpam-3176	57	9	computing	compute	VERB
ejpam-3176	57	10	√	√	NUM
ejpam-3176	57	11	n	n	CCONJ
ejpam-3176	57	12	consistent	consistent	ADJ
ejpam-3176	57	13	outlier	outlier	ADJ
ejpam-3176	57	14	resistant	resistant	ADJ
ejpam-3176	57	15	estimators	estimator	NOUN
ejpam-3176	57	16	that	that	PRON
ejpam-3176	57	17	can	can	AUX
ejpam-3176	57	18	be	be	AUX
ejpam-3176	57	19	used	use	VERB
ejpam-3176	57	20	for	for	ADP
ejpam-3176	57	21	inference	inference	NOUN
ejpam-3176	57	22	and	and	CCONJ
ejpam-3176	57	23	adopted	adopt	VERB
ejpam-3176	57	24	numerous	numerous	ADJ
ejpam-3176	57	25	applications	application	NOUN
ejpam-3176	57	26	,	,	PUNCT
ejpam-3176	57	27	including	include	VERB
ejpam-3176	57	28	outlier	outlier	NOUN
ejpam-3176	57	29	detection	detection	NOUN
ejpam-3176	57	30	and	and	CCONJ
ejpam-3176	57	31	diagnostics	diagnostic	NOUN
ejpam-3176	57	32	,	,	PUNCT
ejpam-3176	57	33	to	to	PART
ejpam-3176	57	34	determine	determine	VERB
ejpam-3176	57	35	whether	whether	SCONJ
ejpam-3176	57	36	data	datum	NOUN
ejpam-3176	57	37	distribution	distribution	NOUN
ejpam-3176	57	38	is	be	AUX
ejpam-3176	57	39	elliptically	elliptically	ADV
ejpam-3176	57	40	contoured	contour	VERB
ejpam-3176	57	41	.	.	PUNCT
ejpam-3176	58	1	olive	olive	NOUN
ejpam-3176	58	2	and	and	CCONJ
ejpam-3176	58	3	ye	ye	PRON
ejpam-3176	59	1	[	[	X
ejpam-3176	59	2	30	30	NUM
ejpam-3176	59	3	]	]	PUNCT
ejpam-3176	59	4	used	use	VERB
ejpam-3176	59	5	three	three	NUM
ejpam-3176	59	6	robust	robust	ADJ
ejpam-3176	59	7	estimators	estimator	NOUN
ejpam-3176	59	8	of	of	ADP
ejpam-3176	59	9	multivariate	multivariate	NOUN
ejpam-3176	59	10	location	location	NOUN
ejpam-3176	59	11	and	and	CCONJ
ejpam-3176	59	12	dispersion	dispersion	NOUN
ejpam-3176	59	13	and	and	CCONJ
ejpam-3176	59	14	then	then	ADV
ejpam-3176	59	15	applied	apply	VERB
ejpam-3176	59	16	one	one	NUM
ejpam-3176	59	17	of	of	ADP
ejpam-3176	59	18	these	these	DET
ejpam-3176	59	19	estimators	estimator	NOUN
ejpam-3176	59	20	to	to	PART
ejpam-3176	59	21	create	create	VERB
ejpam-3176	59	22	a	a	DET
ejpam-3176	59	23	robust	robust	ADJ
ejpam-3176	59	24	method	method	NOUN
ejpam-3176	59	25	for	for	ADP
ejpam-3176	59	26	canonical	canonical	ADJ
ejpam-3176	59	27	correlation	correlation	NOUN
ejpam-3176	59	28	analysis	analysis	NOUN
ejpam-3176	59	29	.	.	PUNCT
ejpam-3176	60	1	one	one	NUM
ejpam-3176	60	2	of	of	ADP
ejpam-3176	60	3	these	these	DET
ejpam-3176	60	4	methods	method	NOUN
ejpam-3176	60	5	is	be	AUX
ejpam-3176	60	6	the	the	DET
ejpam-3176	60	7	fast	fast	ADV
ejpam-3176	60	8	consistent	consistent	ADJ
ejpam-3176	60	9	high	high	ADJ
ejpam-3176	60	10	breakdown	breakdown	NOUN
ejpam-3176	60	11	(	(	PUNCT
ejpam-3176	60	12	fch	fch	NOUN
ejpam-3176	60	13	)	)	PUNCT
ejpam-3176	60	14	estimator	estimator	NOUN
ejpam-3176	60	15	,	,	PUNCT
ejpam-3176	60	16	which	which	PRON
ejpam-3176	60	17	is	be	AUX
ejpam-3176	60	18	fast	fast	ADJ
ejpam-3176	60	19	,	,	PUNCT
ejpam-3176	60	20	consistent	consistent	ADJ
ejpam-3176	60	21	,	,	PUNCT
ejpam-3176	60	22	and	and	CCONJ
ejpam-3176	60	23	highly	highly	ADV
ejpam-3176	60	24	resistant	resistant	ADJ
ejpam-3176	60	25	to	to	ADP
ejpam-3176	60	26	outliers	outlier	NOUN
ejpam-3176	60	27	.	.	PUNCT
ejpam-3176	61	1	the	the	DET
ejpam-3176	61	2	current	current	ADJ
ejpam-3176	61	3	work	work	NOUN
ejpam-3176	61	4	aims	aim	VERB
ejpam-3176	61	5	to	to	PART
ejpam-3176	61	6	combine	combine	VERB
ejpam-3176	61	7	detmcd	detmcd	VERB
ejpam-3176	61	8	with	with	ADP
ejpam-3176	61	9	lda	lda	PROPN
ejpam-3176	61	10	and	and	CCONJ
ejpam-3176	61	11	compare	compare	VERB
ejpam-3176	61	12	it	it	PRON
ejpam-3176	61	13	with	with	ADP
ejpam-3176	61	14	the	the	DET
ejpam-3176	61	15	fastmcd	fastmcd	NOUN
ejpam-3176	61	16	,	,	PUNCT
ejpam-3176	61	17	fch	fch	PROPN
ejpam-3176	61	18	,	,	PUNCT
ejpam-3176	61	19	and	and	CCONJ
ejpam-3176	61	20	robust	robust	ADJ
ejpam-3176	61	21	fch	fch	NOUN
ejpam-3176	61	22	(	(	PUNCT
ejpam-3176	61	23	rfch	rfch	NOUN
ejpam-3176	61	24	)	)	PUNCT
ejpam-3176	61	25	algorithms	algorithm	NOUN
ejpam-3176	61	26	through	through	ADP
ejpam-3176	61	27	simulation	simulation	NOUN
ejpam-3176	61	28	and	and	CCONJ
ejpam-3176	61	29	actual	actual	ADJ
ejpam-3176	61	30	data	datum	NOUN
ejpam-3176	61	31	.	.	PUNCT
ejpam-3176	62	1	three	three	NUM
ejpam-3176	62	2	approaches	approach	NOUN
ejpam-3176	62	3	,	,	PUNCT
ejpam-3176	62	4	namely	namely	ADV
ejpam-3176	62	5	,	,	PUNCT
ejpam-3176	62	6	pooled	pooled	ADJ
ejpam-3176	62	7	covariance	covariance	NOUN
ejpam-3176	62	8	(	(	PUNCT
ejpam-3176	62	9	pcov	pcov	PROPN
ejpam-3176	62	10	)	)	PUNCT
ejpam-3176	62	11	,	,	PUNCT
ejpam-3176	62	12	pobs	pob	NOUN
ejpam-3176	62	13	,	,	PUNCT
ejpam-3176	62	14	and	and	CCONJ
ejpam-3176	62	15	minimum	minimum	ADJ
ejpam-3176	62	16	within	within	ADP
ejpam-3176	62	17	-	-	PUNCT
ejpam-3176	62	18	group	group	NOUN
ejpam-3176	62	19	covariance	covariance	NOUN
ejpam-3176	62	20	determinant	determinant	ADJ
ejpam-3176	62	21	(	(	PUNCT
ejpam-3176	62	22	mwcd	mwcd	NOUN
ejpam-3176	62	23	)	)	PUNCT
ejpam-3176	62	24	,	,	PUNCT
ejpam-3176	62	25	are	be	AUX
ejpam-3176	62	26	applied	apply	VERB
ejpam-3176	62	27	to	to	PART
ejpam-3176	62	28	improve	improve	VERB
ejpam-3176	62	29	the	the	DET
ejpam-3176	62	30	initial	initial	ADJ
ejpam-3176	62	31	covariance	covariance	NOUN
ejpam-3176	62	32	estimate	estimate	VERB
ejpam-3176	62	33	∑	∑	PROPN
ejpam-3176	62	34	0	0	NUM
ejpam-3176	62	35	for	for	ADP
ejpam-3176	62	36	all	all	DET
ejpam-3176	62	37	the	the	DET
ejpam-3176	62	38	estimators	estimator	NOUN
ejpam-3176	62	39	used	use	VERB
ejpam-3176	62	40	in	in	ADP
ejpam-3176	62	41	this	this	DET
ejpam-3176	62	42	study	study	NOUN
ejpam-3176	62	43	.	.	PUNCT
ejpam-3176	63	1	the	the	DET
ejpam-3176	63	2	performance	performance	NOUN
ejpam-3176	63	3	of	of	ADP
ejpam-3176	63	4	this	this	DET
ejpam-3176	63	5	lda	lda	PROPN
ejpam-3176	63	6	is	be	AUX
ejpam-3176	63	7	evaluated	evaluate	VERB
ejpam-3176	63	8	based	base	VERB
ejpam-3176	63	9	on	on	ADP
ejpam-3176	63	10	these	these	DET
ejpam-3176	63	11	estimators	estimator	NOUN
ejpam-3176	63	12	.	.	PUNCT
ejpam-3176	64	1	our	our	PRON
ejpam-3176	64	2	analysis	analysis	NOUN
ejpam-3176	64	3	indicates	indicate	VERB
ejpam-3176	64	4	that	that	SCONJ
ejpam-3176	64	5	detmcd	detmcd	NOUN
ejpam-3176	64	6	performs	perform	VERB
ejpam-3176	64	7	better	well	ADV
ejpam-3176	64	8	than	than	ADP
ejpam-3176	64	9	the	the	DET
ejpam-3176	64	10	other	other	ADJ
ejpam-3176	64	11	estimators	estimator	NOUN
ejpam-3176	64	12	for	for	ADP
ejpam-3176	64	13	the	the	DET
ejpam-3176	64	14	raw	raw	ADJ
ejpam-3176	64	15	and	and	CCONJ
ejpam-3176	64	16	reweighted	reweighted	ADJ
ejpam-3176	64	17	versions	version	NOUN
ejpam-3176	64	18	.	.	PUNCT
ejpam-3176	65	1	mufda	mufda	PROPN
ejpam-3176	65	2	j.	j.	PROPN
ejpam-3176	65	3	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	65	4	,	,	PUNCT
ejpam-3176	65	5	taha	taha	PROPN
ejpam-3176	65	6	radwan	radwan	PROPN
ejpam-3176	65	7	and	and	CCONJ
ejpam-3176	65	8	khalid	khalid	PROPN
ejpam-3176	65	9	abunawas	abunawas	PROPN
ejpam-3176	65	10	/	/	SYM
ejpam-3176	65	11	eur	eur	PROPN
ejpam-3176	65	12	.	.	PUNCT
ejpam-3176	66	1	j.	j.	PROPN
ejpam-3176	66	2	pure	pure	PROPN
ejpam-3176	66	3	appl	appl	PROPN
ejpam-3176	66	4	.	.	PROPN
ejpam-3176	66	5	math	math	PROPN
ejpam-3176	66	6	,	,	PUNCT
ejpam-3176	66	7	11	11	NUM
ejpam-3176	66	8	(	(	PUNCT
ejpam-3176	66	9	1	1	NUM
ejpam-3176	66	10	)	)	PUNCT
ejpam-3176	66	11	(	(	PUNCT
ejpam-3176	66	12	2018	2018	NUM
ejpam-3176	66	13	)	)	PUNCT
ejpam-3176	66	14	,	,	PUNCT
ejpam-3176	66	15	284	284	NUM
ejpam-3176	66	16	-	-	SYM
ejpam-3176	66	17	298	298	NUM
ejpam-3176	66	18	286	286	NUM
ejpam-3176	66	19	2	2	NUM
ejpam-3176	66	20	.	.	PUNCT
ejpam-3176	67	1	lda	lda	VERB
ejpam-3176	67	2	our	our	PRON
ejpam-3176	67	3	proposed	propose	VERB
ejpam-3176	67	4	datasets	dataset	NOUN
ejpam-3176	67	5	of	of	ADP
ejpam-3176	67	6	actual	actual	ADJ
ejpam-3176	67	7	and	and	CCONJ
ejpam-3176	67	8	generated	generate	VERB
ejpam-3176	67	9	data	datum	NOUN
ejpam-3176	67	10	p	p	NOUN
ejpam-3176	67	11	variables	variable	NOUN
ejpam-3176	67	12	measured	measure	VERB
ejpam-3176	67	13	in	in	ADP
ejpam-3176	67	14	n	n	ADP
ejpam-3176	68	1	observations	observation	NOUN
ejpam-3176	69	1	may	may	AUX
ejpam-3176	69	2	be	be	AUX
ejpam-3176	69	3	summarized	summarize	VERB
ejpam-3176	69	4	as	as	ADP
ejpam-3176	69	5	the	the	DET
ejpam-3176	69	6	n×	n×	PROPN
ejpam-3176	69	7	p	p	NOUN
ejpam-3176	69	8	matrix	matrix	NOUN
ejpam-3176	69	9	x	x	PUNCT
ejpam-3176	69	10	=	=	SYM
ejpam-3176	69	11	(	(	PUNCT
ejpam-3176	69	12	xij	xij	X
ejpam-3176	69	13	)	)	PUNCT
ejpam-3176	69	14	,	,	PUNCT
ejpam-3176	69	15	where	where	SCONJ
ejpam-3176	69	16	xij	xij	PROPN
ejpam-3176	69	17	denotes	denote	VERB
ejpam-3176	69	18	the	the	DET
ejpam-3176	69	19	expression	expression	NOUN
ejpam-3176	69	20	level	level	NOUN
ejpam-3176	69	21	of	of	ADP
ejpam-3176	69	22	p	p	NOUN
ejpam-3176	69	23	variables	variable	NOUN
ejpam-3176	69	24	in	in	ADP
ejpam-3176	69	25	observations	observation	NOUN
ejpam-3176	70	1	i	i	PRON
ejpam-3176	70	2	=	=	NOUN
ejpam-3176	70	3	1	1	NUM
ejpam-3176	70	4	,	,	PUNCT
ejpam-3176	70	5	2	2	NUM
ejpam-3176	70	6	,	,	PUNCT
ejpam-3176	70	7	.	.	PUNCT
ejpam-3176	70	8	.	.	PUNCT
ejpam-3176	70	9	.	.	PUNCT
ejpam-3176	71	1	,	,	PUNCT
ejpam-3176	71	2	nj	nj	PROPN
ejpam-3176	71	3	that	that	PRON
ejpam-3176	71	4	are	be	AUX
ejpam-3176	71	5	sampled	sample	VERB
ejpam-3176	71	6	from	from	ADP
ejpam-3176	71	7	l	l	PROPN
ejpam-3176	71	8	different	different	ADJ
ejpam-3176	71	9	populations	population	NOUN
ejpam-3176	71	10	π1	π1	NOUN
ejpam-3176	71	11	,	,	PUNCT
ejpam-3176	71	12	π2	π2	NOUN
ejpam-3176	71	13	,	,	PUNCT
ejpam-3176	71	14	.	.	PUNCT
ejpam-3176	71	15	.	.	PUNCT
ejpam-3176	72	1	.	.	PUNCT
ejpam-3176	73	1	,	,	PUNCT
ejpam-3176	73	2	πl	πl	X
ejpam-3176	73	3	.	.	PUNCT
ejpam-3176	74	1	in	in	ADP
ejpam-3176	74	2	the	the	DET
ejpam-3176	74	3	lda	lda	PROPN
ejpam-3176	74	4	setting	setting	NOUN
ejpam-3176	74	5	,	,	PUNCT
ejpam-3176	74	6	membership	membership	NOUN
ejpam-3176	74	7	probability	probability	NOUN
ejpam-3176	74	8	is	be	AUX
ejpam-3176	74	9	estimated	estimate	VERB
ejpam-3176	74	10	for	for	ADP
ejpam-3176	74	11	each	each	DET
ejpam-3176	74	12	observation	observation	NOUN
ejpam-3176	74	13	with	with	ADP
ejpam-3176	74	14	respect	respect	NOUN
ejpam-3176	74	15	to	to	ADP
ejpam-3176	74	16	the	the	DET
ejpam-3176	74	17	population	population	NOUN
ejpam-3176	74	18	.	.	PUNCT
ejpam-3176	75	1	the	the	DET
ejpam-3176	75	2	data	datum	NOUN
ejpam-3176	75	3	are	be	AUX
ejpam-3176	75	4	sampled	sample	VERB
ejpam-3176	75	5	from	from	ADP
ejpam-3176	75	6	l	l	NOUN
ejpam-3176	75	7	populations	population	NOUN
ejpam-3176	75	8	,	,	PUNCT
ejpam-3176	75	9	and	and	CCONJ
ejpam-3176	75	10	each	each	DET
ejpam-3176	75	11	population	population	NOUN
ejpam-3176	75	12	has	have	VERB
ejpam-3176	75	13	nj	nj	PROPN
ejpam-3176	75	14	observations	observation	NOUN
ejpam-3176	75	15	,	,	PUNCT
ejpam-3176	75	16	j	j	PROPN
ejpam-3176	75	17	=	=	SYM
ejpam-3176	75	18	1	1	NUM
ejpam-3176	75	19	,	,	PUNCT
ejpam-3176	75	20	2	2	NUM
ejpam-3176	75	21	,	,	PUNCT
ejpam-3176	75	22	.	.	PUNCT
ejpam-3176	75	23	.	.	PUNCT
ejpam-3176	76	1	.	.	PUNCT
ejpam-3176	77	1	,	,	PUNCT
ejpam-3176	77	2	l.	l.	PROPN
ejpam-3176	77	3	∑l	∑l	PROPN
ejpam-3176	78	1	j=1	j=1	PROPN
ejpam-3176	78	2	nj	nj	PROPN
ejpam-3176	78	3	=	=	SYM
ejpam-3176	79	1	n	n	NUM
ejpam-3176	79	2	observations	observation	NOUN
ejpam-3176	79	3	can	can	AUX
ejpam-3176	79	4	be	be	AUX
ejpam-3176	79	5	denoted	denote	VERB
ejpam-3176	79	6	by	by	ADP
ejpam-3176	79	7	{	{	PUNCT
ejpam-3176	79	8	xij	xij	PROPN
ejpam-3176	79	9	=	=	SYM
ejpam-3176	79	10	j	j	PROPN
ejpam-3176	79	11	=	=	SYM
ejpam-3176	79	12	1	1	NUM
ejpam-3176	79	13	,	,	PUNCT
ejpam-3176	79	14	2	2	NUM
ejpam-3176	79	15	,	,	PUNCT
ejpam-3176	79	16	.	.	PUNCT
ejpam-3176	79	17	.	.	PUNCT
ejpam-3176	79	18	.	.	PUNCT
ejpam-3176	80	1	,	,	PUNCT
ejpam-3176	80	2	l	l	NOUN
ejpam-3176	80	3	,	,	PUNCT
ejpam-3176	80	4	i	i	PRON
ejpam-3176	80	5	=	=	NOUN
ejpam-3176	80	6	1	1	NUM
ejpam-3176	80	7	,	,	PUNCT
ejpam-3176	80	8	2	2	NUM
ejpam-3176	80	9	,	,	PUNCT
ejpam-3176	80	10	.	.	PUNCT
ejpam-3176	80	11	.	.	PUNCT
ejpam-3176	80	12	.	.	PUNCT
ejpam-3176	80	13	,	,	PUNCT
ejpam-3176	80	14	n	n	CCONJ
ejpam-3176	80	15	}	}	PUNCT
ejpam-3176	80	16	.	.	PUNCT
ejpam-3176	81	1	lda	lda	PROPN
ejpam-3176	81	2	has	have	VERB
ejpam-3176	81	3	µj	µj	PROPN
ejpam-3176	81	4	,	,	PUNCT
ejpam-3176	81	5	σj	σj	ADJ
ejpam-3176	81	6	,	,	PUNCT
ejpam-3176	81	7	and	and	CCONJ
ejpam-3176	81	8	pj	pj	PROPN
ejpam-3176	81	9	.	.	PUNCT
ejpam-3176	82	1	µj	µj	PROPN
ejpam-3176	82	2	is	be	AUX
ejpam-3176	82	3	the	the	DET
ejpam-3176	82	4	mean	mean	NOUN
ejpam-3176	82	5	,	,	PUNCT
ejpam-3176	82	6	σj	σj	ADJ
ejpam-3176	82	7	is	be	AUX
ejpam-3176	82	8	the	the	DET
ejpam-3176	82	9	covariance	covariance	NOUN
ejpam-3176	82	10	matrix	matrix	NOUN
ejpam-3176	82	11	,	,	PUNCT
ejpam-3176	82	12	and	and	CCONJ
ejpam-3176	82	13	pj	pj	PROPN
ejpam-3176	82	14	is	be	AUX
ejpam-3176	82	15	the	the	DET
ejpam-3176	82	16	membership	membership	NOUN
ejpam-3176	82	17	probability	probability	NOUN
ejpam-3176	82	18	for	for	SCONJ
ejpam-3176	82	19	each	each	DET
ejpam-3176	82	20	population	population	NOUN
ejpam-3176	82	21	πj	πj	VERB
ejpam-3176	82	22	.	.	PUNCT
ejpam-3176	83	1	lda	lda	PROPN
ejpam-3176	83	2	assumes	assume	VERB
ejpam-3176	83	3	a	a	DET
ejpam-3176	83	4	common	common	ADJ
ejpam-3176	83	5	covariance	covariance	NOUN
ejpam-3176	83	6	matrix	matrix	NOUN
ejpam-3176	83	7	σ	σ	NOUN
ejpam-3176	83	8	;	;	PUNCT
ejpam-3176	83	9	all	all	DET
ejpam-3176	83	10	the	the	DET
ejpam-3176	83	11	parameters	parameter	NOUN
ejpam-3176	83	12	are	be	AUX
ejpam-3176	83	13	unknown	unknown	ADJ
ejpam-3176	83	14	in	in	ADP
ejpam-3176	83	15	practice	practice	NOUN
ejpam-3176	83	16	and	and	CCONJ
ejpam-3176	83	17	must	must	AUX
ejpam-3176	83	18	be	be	AUX
ejpam-3176	83	19	estimated	estimate	VERB
ejpam-3176	83	20	from	from	ADP
ejpam-3176	83	21	the	the	DET
ejpam-3176	83	22	sample	sample	NOUN
ejpam-3176	83	23	data	datum	NOUN
ejpam-3176	83	24	.	.	PUNCT
ejpam-3176	84	1	in	in	ADP
ejpam-3176	84	2	general	general	ADJ
ejpam-3176	84	3	,	,	PUNCT
ejpam-3176	84	4	lda	lda	PROPN
ejpam-3176	84	5	parameters	parameter	NOUN
ejpam-3176	84	6	are	be	AUX
ejpam-3176	84	7	estimated	estimate	VERB
ejpam-3176	84	8	empirically	empirically	ADV
ejpam-3176	84	9	,	,	PUNCT
ejpam-3176	84	10	which	which	PRON
ejpam-3176	84	11	leads	lead	VERB
ejpam-3176	84	12	to	to	ADP
ejpam-3176	84	13	inaccurate	inaccurate	ADJ
ejpam-3176	84	14	values	value	NOUN
ejpam-3176	84	15	because	because	SCONJ
ejpam-3176	84	16	lda	lda	PROPN
ejpam-3176	84	17	is	be	AUX
ejpam-3176	84	18	highly	highly	ADV
ejpam-3176	84	19	influenced	influence	VERB
ejpam-3176	84	20	by	by	ADP
ejpam-3176	84	21	outliers	outlier	NOUN
ejpam-3176	84	22	.	.	PUNCT
ejpam-3176	85	1	all	all	DET
ejpam-3176	85	2	the	the	DET
ejpam-3176	85	3	parameters	parameter	NOUN
ejpam-3176	85	4	must	must	AUX
ejpam-3176	85	5	be	be	AUX
ejpam-3176	85	6	estimated	estimate	VERB
ejpam-3176	85	7	based	base	VERB
ejpam-3176	85	8	on	on	ADP
ejpam-3176	85	9	robust	robust	ADJ
ejpam-3176	85	10	estimators	estimator	NOUN
ejpam-3176	85	11	to	to	PART
ejpam-3176	85	12	overcome	overcome	VERB
ejpam-3176	85	13	the	the	DET
ejpam-3176	85	14	outlier	outlier	ADJ
ejpam-3176	85	15	problem	problem	NOUN
ejpam-3176	85	16	,	,	PUNCT
ejpam-3176	85	17	thereby	thereby	ADV
ejpam-3176	85	18	requiring	require	VERB
ejpam-3176	85	19	high	high	ADJ
ejpam-3176	85	20	-	-	PUNCT
ejpam-3176	85	21	performance	performance	NOUN
ejpam-3176	85	22	robust	robust	ADJ
ejpam-3176	85	23	estimators	estimator	NOUN
ejpam-3176	85	24	.	.	PUNCT
ejpam-3176	86	1	the	the	DET
ejpam-3176	86	2	robust	robust	ADJ
ejpam-3176	86	3	lda	lda	PROPN
ejpam-3176	86	4	(	(	PUNCT
ejpam-3176	86	5	rlda	rlda	NOUN
ejpam-3176	86	6	)	)	PUNCT
ejpam-3176	86	7	rule	rule	NOUN
ejpam-3176	86	8	is	be	AUX
ejpam-3176	86	9	expressed	express	VERB
ejpam-3176	86	10	as	as	SCONJ
ejpam-3176	86	11	follows	follow	VERB
ejpam-3176	86	12	:	:	PUNCT
ejpam-3176	86	13	allocate	allocate	VERB
ejpam-3176	86	14	x	x	PART
ejpam-3176	86	15	to	to	PART
ejpam-3176	86	16	πj	πj	VERB
ejpam-3176	86	17	if	if	SCONJ
ejpam-3176	86	18	_	_	PUNCT
ejpam-3176	87	1	d	d	X
ejpam-3176	87	2	rl	rl	X
ejpam-3176	87	3	k	k	PROPN
ejpam-3176	87	4	(	(	PUNCT
ejpam-3176	87	5	x	x	X
ejpam-3176	87	6	)	)	PUNCT
ejpam-3176	87	7	>	>	PUNCT
ejpam-3176	88	1	_	_	PUNCT
ejpam-3176	89	1	d	d	X
ejpam-3176	89	2	rl	rl	PROPN
ejpam-3176	89	3	j	j	PROPN
ejpam-3176	89	4	(	(	PUNCT
ejpam-3176	89	5	x	x	NOUN
ejpam-3176	89	6	)	)	PUNCT
ejpam-3176	89	7	for	for	ADP
ejpam-3176	89	8	j	j	PROPN
ejpam-3176	89	9	=	=	SYM
ejpam-3176	89	10	1	1	NUM
ejpam-3176	89	11	,	,	PUNCT
ejpam-3176	89	12	2	2	NUM
ejpam-3176	89	13	,	,	PUNCT
ejpam-3176	89	14	.	.	PUNCT
ejpam-3176	89	15	.	.	PUNCT
ejpam-3176	89	16	.	.	PUNCT
ejpam-3176	90	1	,	,	PUNCT
ejpam-3176	90	2	g	g	PROPN
ejpam-3176	90	3	,	,	PUNCT
ejpam-3176	90	4	j	j	PROPN
ejpam-3176	90	5	6=	6=	PROPN
ejpam-3176	90	6	k	k	NOUN
ejpam-3176	90	7	with	with	ADP
ejpam-3176	90	8	_	_	PUNCT
ejpam-3176	91	1	d	d	X
ejpam-3176	91	2	rl	rl	PROPN
ejpam-3176	91	3	j	j	PROPN
ejpam-3176	91	4	(	(	PUNCT
ejpam-3176	91	5	x	x	X
ejpam-3176	91	6	)	)	PUNCT
ejpam-3176	91	7	=	=	PRON
ejpam-3176	91	8	µtjς	µtjς	VERB
ejpam-3176	91	9	−1x−	−1x−	ADJ
ejpam-3176	91	10	1	1	NUM
ejpam-3176	91	11	2	2	NUM
ejpam-3176	91	12	µtjς	µtjς	NOUN
ejpam-3176	91	13	−1µj	−1µj	X
ejpam-3176	91	14	+	+	X
ejpam-3176	91	15	ln(pj	ln(pj	PROPN
ejpam-3176	91	16	)	)	PUNCT
ejpam-3176	91	17	,	,	PUNCT
ejpam-3176	91	18	(	(	PUNCT
ejpam-3176	91	19	1	1	X
ejpam-3176	91	20	)	)	PUNCT
ejpam-3176	91	21	where	where	SCONJ
ejpam-3176	91	22	σ	σ	PROPN
ejpam-3176	91	23	is	be	AUX
ejpam-3176	91	24	the	the	DET
ejpam-3176	91	25	common	common	ADJ
ejpam-3176	91	26	covariance	covariance	NOUN
ejpam-3176	91	27	matrix	matrix	NOUN
ejpam-3176	91	28	with	with	ADP
ejpam-3176	91	29	mean	mean	PROPN
ejpam-3176	91	30	µj	µj	PROPN
ejpam-3176	91	31	and	and	CCONJ
ejpam-3176	91	32	prior	prior	ADJ
ejpam-3176	91	33	probability	probability	NOUN
ejpam-3176	91	34	pj	pj	PROPN
ejpam-3176	91	35	.	.	PUNCT
ejpam-3176	92	1	for	for	ADP
ejpam-3176	92	2	the	the	DET
ejpam-3176	92	3	estimates	estimate	NOUN
ejpam-3176	92	4	of	of	ADP
ejpam-3176	92	5	the	the	DET
ejpam-3176	92	6	membership	membership	NOUN
ejpam-3176	92	7	probability	probability	NOUN
ejpam-3176	92	8	pj	pj	PROPN
ejpam-3176	92	9	in	in	ADP
ejpam-3176	92	10	eq	eq	PROPN
ejpam-3176	92	11	.	.	PUNCT
ejpam-3176	93	1	(	(	PUNCT
ejpam-3176	93	2	1	1	NUM
ejpam-3176	93	3	)	)	PUNCT
ejpam-3176	93	4	,	,	PUNCT
ejpam-3176	93	5	we	we	PRON
ejpam-3176	93	6	discuss	discuss	VERB
ejpam-3176	93	7	two	two	NUM
ejpam-3176	93	8	well	well	ADV
ejpam-3176	93	9	-	-	PUNCT
ejpam-3176	93	10	known	know	VERB
ejpam-3176	93	11	choices	choice	NOUN
ejpam-3176	93	12	.	.	PUNCT
ejpam-3176	94	1	either	either	CCONJ
ejpam-3176	94	2	pj	pj	PROPN
ejpam-3176	94	3	is	be	AUX
ejpam-3176	94	4	considered	consider	VERB
ejpam-3176	94	5	constant	constant	ADJ
ejpam-3176	94	6	over	over	ADP
ejpam-3176	94	7	all	all	DET
ejpam-3176	94	8	populations	population	NOUN
ejpam-3176	94	9	,	,	PUNCT
ejpam-3176	94	10	thereby	thereby	ADV
ejpam-3176	94	11	yielding	yield	VERB
ejpam-3176	94	12	pj	pj	PROPN
ejpam-3176	94	13	=	=	SYM
ejpam-3176	94	14	1	1	NUM
ejpam-3176	94	15	/	/	SYM
ejpam-3176	94	16	l	l	NOUN
ejpam-3176	94	17	for	for	ADP
ejpam-3176	94	18	each	each	DET
ejpam-3176	94	19	j	j	NOUN
ejpam-3176	94	20	,	,	PUNCT
ejpam-3176	94	21	or	or	CCONJ
ejpam-3176	94	22	it	it	PRON
ejpam-3176	94	23	is	be	AUX
ejpam-3176	94	24	estimated	estimate	VERB
ejpam-3176	94	25	as	as	ADP
ejpam-3176	94	26	the	the	DET
ejpam-3176	94	27	relative	relative	ADJ
ejpam-3176	94	28	frequencies	frequency	NOUN
ejpam-3176	94	29	of	of	ADP
ejpam-3176	94	30	the	the	DET
ejpam-3176	94	31	observations	observation	NOUN
ejpam-3176	94	32	in	in	ADP
ejpam-3176	94	33	each	each	DET
ejpam-3176	94	34	group	group	NOUN
ejpam-3176	94	35	,	,	PUNCT
ejpam-3176	94	36	thereby	thereby	ADV
ejpam-3176	94	37	yielding	yield	VERB
ejpam-3176	94	38	pj	pj	PROPN
ejpam-3176	94	39	=	=	PROPN
ejpam-3176	94	40	nj	nj	PROPN
ejpam-3176	94	41	/	/	SYM
ejpam-3176	94	42	n.	n.	PROPN
ejpam-3176	94	43	if	if	SCONJ
ejpam-3176	94	44	ñj	ñj	NOUN
ejpam-3176	94	45	denotes	denote	VERB
ejpam-3176	94	46	the	the	DET
ejpam-3176	94	47	number	number	NOUN
ejpam-3176	94	48	of	of	ADP
ejpam-3176	94	49	non	non	NOUN
ejpam-3176	94	50	-	-	NOUN
ejpam-3176	94	51	outliers	outlier	NOUN
ejpam-3176	94	52	in	in	ADP
ejpam-3176	94	53	group	group	PROPN
ejpam-3176	94	54	j	j	PROPN
ejpam-3176	94	55	and	and	CCONJ
ejpam-3176	94	56	ñ	ñ	PROPN
ejpam-3176	94	57	=	=	PUNCT
ejpam-3176	94	58	∑l	∑l	ADJ
ejpam-3176	94	59	j=1	j=1	ADJ
ejpam-3176	94	60	ñj	ñj	NOUN
ejpam-3176	94	61	,	,	PUNCT
ejpam-3176	94	62	then	then	ADV
ejpam-3176	94	63	the	the	DET
ejpam-3176	94	64	membership	membership	NOUN
ejpam-3176	94	65	probability	probability	NOUN
ejpam-3176	94	66	is	be	AUX
ejpam-3176	94	67	robustly	robustly	ADV
ejpam-3176	94	68	estimated	estimate	VERB
ejpam-3176	94	69	as	as	SCONJ
ejpam-3176	94	70	follows	follow	VERB
ejpam-3176	94	71	:	:	PUNCT
ejpam-3176	95	1	_	_	PUNCT
ejpam-3176	95	2	p	p	X
ejpam-3176	95	3	rl	rl	ADP
ejpam-3176	95	4	j	j	PROPN
ejpam-3176	95	5	=	=	PROPN
ejpam-3176	95	6	ñj	ñj	PROPN
ejpam-3176	95	7	ñ	ñ	PROPN
ejpam-3176	95	8	.	.	PUNCT
ejpam-3176	96	1	(	(	PUNCT
ejpam-3176	96	2	2	2	X
ejpam-3176	96	3	)	)	PUNCT
ejpam-3176	96	4	as	as	SCONJ
ejpam-3176	96	5	previously	previously	ADV
ejpam-3176	96	6	mentioned	mention	VERB
ejpam-3176	96	7	,	,	PUNCT
ejpam-3176	96	8	the	the	DET
ejpam-3176	96	9	lda	lda	PROPN
ejpam-3176	96	10	parameters	parameter	NOUN
ejpam-3176	96	11	are	be	AUX
ejpam-3176	96	12	unknown	unknown	ADJ
ejpam-3176	96	13	and	and	CCONJ
ejpam-3176	96	14	have	have	VERB
ejpam-3176	96	15	to	to	PART
ejpam-3176	96	16	be	be	AUX
ejpam-3176	96	17	estimated	estimate	VERB
ejpam-3176	96	18	.	.	PUNCT
ejpam-3176	97	1	all	all	DET
ejpam-3176	97	2	the	the	DET
ejpam-3176	97	3	estimators	estimator	NOUN
ejpam-3176	97	4	will	will	AUX
ejpam-3176	97	5	be	be	AUX
ejpam-3176	97	6	used	use	VERB
ejpam-3176	97	7	to	to	PART
ejpam-3176	97	8	estimate	estimate	VERB
ejpam-3176	97	9	the	the	DET
ejpam-3176	97	10	lda	lda	PROPN
ejpam-3176	97	11	parameters	parameter	NOUN
ejpam-3176	97	12	to	to	PART
ejpam-3176	97	13	apply	apply	VERB
ejpam-3176	97	14	lda	lda	PROPN
ejpam-3176	97	15	to	to	ADP
ejpam-3176	97	16	the	the	DET
ejpam-3176	97	17	estimation	estimation	NOUN
ejpam-3176	97	18	of	of	ADP
ejpam-3176	97	19	the	the	DET
ejpam-3176	97	20	misclassification	misclassification	NOUN
ejpam-3176	97	21	probability	probability	NOUN
ejpam-3176	97	22	.	.	PUNCT
ejpam-3176	98	1	mufda	mufda	PROPN
ejpam-3176	98	2	j.	j.	PROPN
ejpam-3176	98	3	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	98	4	,	,	PUNCT
ejpam-3176	98	5	taha	taha	PROPN
ejpam-3176	98	6	radwan	radwan	PROPN
ejpam-3176	98	7	and	and	CCONJ
ejpam-3176	98	8	khalid	khalid	PROPN
ejpam-3176	98	9	abunawas	abunawas	PROPN
ejpam-3176	98	10	/	/	SYM
ejpam-3176	98	11	eur	eur	PROPN
ejpam-3176	98	12	.	.	PUNCT
ejpam-3176	99	1	j.	j.	PROPN
ejpam-3176	99	2	pure	pure	PROPN
ejpam-3176	99	3	appl	appl	PROPN
ejpam-3176	99	4	.	.	PROPN
ejpam-3176	99	5	math	math	PROPN
ejpam-3176	99	6	,	,	PUNCT
ejpam-3176	99	7	11	11	NUM
ejpam-3176	99	8	(	(	PUNCT
ejpam-3176	99	9	1	1	NUM
ejpam-3176	99	10	)	)	PUNCT
ejpam-3176	99	11	(	(	PUNCT
ejpam-3176	99	12	2018	2018	NUM
ejpam-3176	99	13	)	)	PUNCT
ejpam-3176	99	14	,	,	PUNCT
ejpam-3176	99	15	284	284	NUM
ejpam-3176	99	16	-	-	SYM
ejpam-3176	99	17	298	298	NUM
ejpam-3176	99	18	287	287	NUM
ejpam-3176	99	19	3	3	NUM
ejpam-3176	99	20	.	.	PUNCT
ejpam-3176	100	1	detmcd	detmcd	PROPN
ejpam-3176	100	2	estimator	estimator	NOUN
ejpam-3176	100	3	the	the	DET
ejpam-3176	100	4	detmcd	detmcd	PROPN
ejpam-3176	100	5	algorithm	algorithm	NOUN
ejpam-3176	100	6	starts	start	VERB
ejpam-3176	100	7	by	by	ADP
ejpam-3176	100	8	standardizing	standardize	VERB
ejpam-3176	100	9	data	datum	NOUN
ejpam-3176	100	10	to	to	PART
ejpam-3176	100	11	obtain	obtain	VERB
ejpam-3176	100	12	the	the	DET
ejpam-3176	100	13	standardized	standardized	ADJ
ejpam-3176	100	14	z.	z.	NOUN
ejpam-3176	100	15	each	each	DET
ejpam-3176	100	16	variable	variable	PROPN
ejpam-3176	100	17	xj	xj	PROPN
ejpam-3176	100	18	will	will	AUX
ejpam-3176	100	19	be	be	AUX
ejpam-3176	100	20	subtracted	subtract	VERB
ejpam-3176	100	21	from	from	ADP
ejpam-3176	100	22	the	the	DET
ejpam-3176	100	23	median	median	NOUN
ejpam-3176	100	24	and	and	CCONJ
ejpam-3176	100	25	divided	divide	VERB
ejpam-3176	100	26	by	by	ADP
ejpam-3176	100	27	the	the	DET
ejpam-3176	100	28	qn	qn	NOUN
ejpam-3176	100	29	scale	scale	NOUN
ejpam-3176	100	30	estimator	estimator	NOUN
ejpam-3176	100	31	(	(	PUNCT
ejpam-3176	100	32	rousseeuw	rousseeuw	NOUN
ejpam-3176	100	33	and	and	CCONJ
ejpam-3176	100	34	croux	croux	VERB
ejpam-3176	100	35	[	[	X
ejpam-3176	100	36	24	24	NUM
ejpam-3176	100	37	]	]	PUNCT
ejpam-3176	100	38	)	)	PUNCT
ejpam-3176	100	39	.	.	PUNCT
ejpam-3176	101	1	this	this	DET
ejpam-3176	101	2	standardization	standardization	NOUN
ejpam-3176	101	3	enables	enable	VERB
ejpam-3176	101	4	the	the	DET
ejpam-3176	101	5	equivariance	equivariance	NOUN
ejpam-3176	101	6	of	of	ADP
ejpam-3176	101	7	algorithm	algorithm	NOUN
ejpam-3176	101	8	location	location	NOUN
ejpam-3176	101	9	and	and	CCONJ
ejpam-3176	101	10	scale	scale	NOUN
ejpam-3176	101	11	.	.	PUNCT
ejpam-3176	102	1	the	the	DET
ejpam-3176	102	2	standardized	standardized	ADJ
ejpam-3176	102	3	dataset	dataset	NOUN
ejpam-3176	102	4	is	be	AUX
ejpam-3176	102	5	denoted	denote	VERB
ejpam-3176	102	6	by	by	ADP
ejpam-3176	102	7	the	the	DET
ejpam-3176	102	8	n×	n×	PROPN
ejpam-3176	102	9	p	p	NOUN
ejpam-3176	102	10	matrix	matrix	NOUN
ejpam-3176	102	11	z	z	NOUN
ejpam-3176	102	12	with	with	ADP
ejpam-3176	102	13	row	row	NOUN
ejpam-3176	102	14	zti	zti	NOUN
ejpam-3176	103	1	(	(	PUNCT
ejpam-3176	103	2	i	i	NOUN
ejpam-3176	103	3	=	=	NOUN
ejpam-3176	103	4	1	1	NUM
ejpam-3176	103	5	,	,	PUNCT
ejpam-3176	103	6	2	2	NUM
ejpam-3176	103	7	,	,	PUNCT
ejpam-3176	103	8	.	.	PUNCT
ejpam-3176	103	9	.	.	PUNCT
ejpam-3176	103	10	.	.	PUNCT
ejpam-3176	103	11	,	,	PUNCT
ejpam-3176	103	12	n	n	CCONJ
ejpam-3176	103	13	)	)	PUNCT
ejpam-3176	103	14	and	and	CCONJ
ejpam-3176	103	15	column	column	PROPN
ejpam-3176	103	16	zj	zj	PROPN
ejpam-3176	103	17	(	(	PUNCT
ejpam-3176	103	18	j	j	PROPN
ejpam-3176	103	19	=	=	SYM
ejpam-3176	103	20	1	1	NUM
ejpam-3176	103	21	,	,	PUNCT
ejpam-3176	103	22	2	2	NUM
ejpam-3176	103	23	,	,	PUNCT
ejpam-3176	103	24	.	.	PUNCT
ejpam-3176	103	25	.	.	PUNCT
ejpam-3176	103	26	.	.	PUNCT
ejpam-3176	104	1	,	,	PUNCT
ejpam-3176	104	2	p	p	NOUN
ejpam-3176	104	3	)	)	PUNCT
ejpam-3176	104	4	.	.	PUNCT
ejpam-3176	105	1	six	six	NUM
ejpam-3176	105	2	initial	initial	ADJ
ejpam-3176	105	3	estimates	estimate	NOUN
ejpam-3176	105	4	of	of	ADP
ejpam-3176	105	5	µk(z	µk(z	NOUN
ejpam-3176	105	6	)	)	PUNCT
ejpam-3176	105	7	and	and	CCONJ
ejpam-3176	105	8	σk(z	σk(z	NUM
ejpam-3176	105	9	)	)	PUNCT
ejpam-3176	105	10	,	,	PUNCT
ejpam-3176	105	11	where	where	SCONJ
ejpam-3176	105	12	(	(	PUNCT
ejpam-3176	105	13	k	k	NOUN
ejpam-3176	105	14	=	=	SYM
ejpam-3176	105	15	1	1	NUM
ejpam-3176	105	16	,	,	PUNCT
ejpam-3176	105	17	2	2	NUM
ejpam-3176	105	18	,	,	PUNCT
ejpam-3176	105	19	.	.	PUNCT
ejpam-3176	105	20	.	.	PUNCT
ejpam-3176	105	21	.	.	PUNCT
ejpam-3176	106	1	,	,	PUNCT
ejpam-3176	106	2	6	6	NUM
ejpam-3176	106	3	)	)	PUNCT
ejpam-3176	106	4	,	,	PUNCT
ejpam-3176	106	5	represent	represent	VERB
ejpam-3176	106	6	the	the	DET
ejpam-3176	106	7	mean	mean	NOUN
ejpam-3176	106	8	and	and	CCONJ
ejpam-3176	106	9	covariance	covariance	NOUN
ejpam-3176	106	10	matrix	matrix	NOUN
ejpam-3176	106	11	,	,	PUNCT
ejpam-3176	106	12	respectively	respectively	ADV
ejpam-3176	106	13	,	,	PUNCT
ejpam-3176	106	14	of	of	ADP
ejpam-3176	106	15	z.	z.	PROPN
ejpam-3176	106	16	each	each	DET
ejpam-3176	106	17	sk	sk	PROPN
ejpam-3176	106	18	estimator	estimator	NOUN
ejpam-3176	106	19	computes	compute	VERB
ejpam-3176	106	20	the	the	DET
ejpam-3176	106	21	covariance	covariance	NOUN
ejpam-3176	106	22	or	or	CCONJ
ejpam-3176	106	23	correlations	correlation	NOUN
ejpam-3176	106	24	of	of	ADP
ejpam-3176	106	25	matrix	matrix	NOUN
ejpam-3176	106	26	z.	z.	NOUN
ejpam-3176	106	27	4	4	NUM
ejpam-3176	106	28	.	.	NOUN
ejpam-3176	106	29	six	six	NUM
ejpam-3176	106	30	initial	initial	ADJ
ejpam-3176	106	31	scatter	scatter	NOUN
ejpam-3176	106	32	estimators	estimator	NOUN
ejpam-3176	106	33	(	(	PUNCT
ejpam-3176	106	34	i	i	NOUN
ejpam-3176	106	35	)	)	PUNCT
ejpam-3176	106	36	s1	s1	PROPN
ejpam-3176	106	37	is	be	AUX
ejpam-3176	106	38	computed	compute	VERB
ejpam-3176	106	39	by	by	ADP
ejpam-3176	106	40	the	the	DET
ejpam-3176	106	41	hyperbolic	hyperbolic	ADJ
ejpam-3176	106	42	tangent	tangent	NOUN
ejpam-3176	106	43	of	of	ADP
ejpam-3176	106	44	each	each	DET
ejpam-3176	106	45	column	column	NOUN
ejpam-3176	106	46	of	of	ADP
ejpam-3176	106	47	z	z	PROPN
ejpam-3176	106	48	,	,	PUNCT
ejpam-3176	106	49	yj	yj	PROPN
ejpam-3176	106	50	=	=	SYM
ejpam-3176	106	51	tanh(zj	tanh(zj	PROPN
ejpam-3176	106	52	)	)	PUNCT
ejpam-3176	106	53	for	for	ADP
ejpam-3176	106	54	j	j	PROPN
ejpam-3176	106	55	=	=	SYM
ejpam-3176	106	56	1	1	PROPN
ejpam-3176	106	57	,	,	PUNCT
ejpam-3176	106	58	.	.	PUNCT
ejpam-3176	106	59	.	.	PUNCT
ejpam-3176	107	1	.	.	PUNCT
ejpam-3176	108	1	,	,	PUNCT
ejpam-3176	109	1	p.	p.	NOUN
ejpam-3176	109	2	this	this	DET
ejpam-3176	109	3	bounded	bounded	ADJ
ejpam-3176	109	4	function	function	NOUN
ejpam-3176	109	5	reduces	reduce	VERB
ejpam-3176	109	6	the	the	DET
ejpam-3176	109	7	effect	effect	NOUN
ejpam-3176	109	8	of	of	ADP
ejpam-3176	109	9	large	large	ADJ
ejpam-3176	109	10	coordinatewise	coordinatewise	NOUN
ejpam-3176	109	11	outliers	outlier	NOUN
ejpam-3176	109	12	.	.	PUNCT
ejpam-3176	110	1	then	then	ADV
ejpam-3176	110	2	,	,	PUNCT
ejpam-3176	110	3	the	the	DET
ejpam-3176	110	4	classical	classical	ADJ
ejpam-3176	110	5	correlation	correlation	NOUN
ejpam-3176	110	6	matrix	matrix	NOUN
ejpam-3176	110	7	of	of	ADP
ejpam-3176	110	8	y	y	PROPN
ejpam-3176	110	9	is	be	AUX
ejpam-3176	110	10	computed	compute	VERB
ejpam-3176	110	11	to	to	PART
ejpam-3176	110	12	obtain	obtain	VERB
ejpam-3176	110	13	s1	s1	NOUN
ejpam-3176	110	14	=	=	SYM
ejpam-3176	110	15	corr(y	corr(y	ADJ
ejpam-3176	110	16	)	)	PUNCT
ejpam-3176	110	17	.	.	PUNCT
ejpam-3176	111	1	(	(	PUNCT
ejpam-3176	111	2	ii	ii	X
ejpam-3176	111	3	)	)	PUNCT
ejpam-3176	111	4	s2	s2	NOUN
ejpam-3176	111	5	is	be	AUX
ejpam-3176	111	6	computed	compute	VERB
ejpam-3176	111	7	by	by	ADP
ejpam-3176	111	8	determining	determine	VERB
ejpam-3176	111	9	rj	rj	PROPN
ejpam-3176	111	10	,	,	PUNCT
ejpam-3176	111	11	the	the	DET
ejpam-3176	111	12	rank	rank	NOUN
ejpam-3176	111	13	of	of	ADP
ejpam-3176	111	14	each	each	DET
ejpam-3176	111	15	column	column	NOUN
ejpam-3176	111	16	zj	zj	PROPN
ejpam-3176	111	17	.	.	PUNCT
ejpam-3176	112	1	then	then	ADV
ejpam-3176	112	2	,	,	PUNCT
ejpam-3176	112	3	s2	s2	PROPN
ejpam-3176	112	4	=	=	SYM
ejpam-3176	112	5	corr(r	corr(r	PROPN
ejpam-3176	112	6	)	)	PUNCT
ejpam-3176	112	7	,	,	PUNCT
ejpam-3176	112	8	which	which	PRON
ejpam-3176	112	9	is	be	AUX
ejpam-3176	112	10	the	the	DET
ejpam-3176	112	11	spearman	spearman	NOUN
ejpam-3176	112	12	correlation	correlation	NOUN
ejpam-3176	112	13	matrix	matrix	NOUN
ejpam-3176	112	14	of	of	ADP
ejpam-3176	112	15	z.	z.	PROPN
ejpam-3176	112	16	(	(	PUNCT
ejpam-3176	112	17	iii	iii	X
ejpam-3176	112	18	)	)	PUNCT
ejpam-3176	112	19	s3	s3	PROPN
ejpam-3176	112	20	is	be	AUX
ejpam-3176	112	21	the	the	DET
ejpam-3176	112	22	normal	normal	ADJ
ejpam-3176	112	23	score	score	NOUN
ejpam-3176	112	24	computed	compute	VERB
ejpam-3176	112	25	from	from	ADP
ejpam-3176	112	26	rj	rj	PROPN
ejpam-3176	112	27	;	;	PUNCT
ejpam-3176	112	28	that	that	ADV
ejpam-3176	112	29	is	is	ADV
ejpam-3176	112	30	,	,	PUNCT
ejpam-3176	112	31	tj	tj	NOUN
ejpam-3176	112	32	=	=	NOUN
ejpam-3176	112	33	φ−1((rj	φ−1((rj	NOUN
ejpam-3176	112	34	−	−	PROPN
ejpam-3176	113	1	1/3)/(n+	1/3)/(n+	NUM
ejpam-3176	113	2	1/3	1/3	NUM
ejpam-3176	113	3	)	)	PUNCT
ejpam-3176	113	4	)	)	PUNCT
ejpam-3176	113	5	,	,	PUNCT
ejpam-3176	113	6	where	where	SCONJ
ejpam-3176	113	7	φ	φ	PROPN
ejpam-3176	113	8	(	(	PUNCT
ejpam-3176	113	9	·	·	PUNCT
ejpam-3176	113	10	)	)	PUNCT
ejpam-3176	113	11	is	be	AUX
ejpam-3176	113	12	the	the	DET
ejpam-3176	113	13	normal	normal	ADJ
ejpam-3176	113	14	cumulative	cumulative	ADJ
ejpam-3176	113	15	distribution	distribution	NOUN
ejpam-3176	113	16	function	function	NOUN
ejpam-3176	113	17	.	.	PUNCT
ejpam-3176	114	1	then	then	ADV
ejpam-3176	114	2	,	,	PUNCT
ejpam-3176	114	3	s3	s3	PROPN
ejpam-3176	114	4	=	=	NOUN
ejpam-3176	114	5	corr(t	corr(t	NOUN
ejpam-3176	114	6	)	)	PUNCT
ejpam-3176	114	7	.	.	PUNCT
ejpam-3176	115	1	(	(	PUNCT
ejpam-3176	115	2	iv	iv	X
ejpam-3176	115	3	)	)	PUNCT
ejpam-3176	115	4	s4	s4	PROPN
ejpam-3176	115	5	is	be	AUX
ejpam-3176	115	6	the	the	DET
ejpam-3176	115	7	scatter	scatter	NOUN
ejpam-3176	115	8	estimator	estimator	NOUN
ejpam-3176	115	9	computed	compute	VERB
ejpam-3176	115	10	based	base	VERB
ejpam-3176	115	11	on	on	ADP
ejpam-3176	115	12	the	the	DET
ejpam-3176	115	13	spatial	spatial	ADJ
ejpam-3176	115	14	sign	sign	NOUN
ejpam-3176	115	15	covariance	covariance	NOUN
ejpam-3176	115	16	matrix	matrix	NOUN
ejpam-3176	115	17	(	(	PUNCT
ejpam-3176	115	18	visuri	visuri	PROPN
ejpam-3176	115	19	et	et	PROPN
ejpam-3176	115	20	al	al	PROPN
ejpam-3176	115	21	.	.	PUNCT
ejpam-3176	116	1	[	[	X
ejpam-3176	116	2	28	28	NUM
ejpam-3176	116	3	]	]	PUNCT
ejpam-3176	116	4	)	)	PUNCT
ejpam-3176	116	5	and	and	CCONJ
ejpam-3176	116	6	is	be	AUX
ejpam-3176	116	7	defined	define	VERB
ejpam-3176	116	8	as	as	ADP
ejpam-3176	116	9	ki	ki	PROPN
ejpam-3176	116	10	=	=	PROPN
ejpam-3176	116	11	zi/	zi/	X
ejpam-3176	116	12	‖zi‖	‖zi‖	NOUN
ejpam-3176	116	13	for	for	ADP
ejpam-3176	116	14	all	all	DET
ejpam-3176	116	15	i.	i.	NOUN
ejpam-3176	116	16	then	then	ADV
ejpam-3176	116	17	,	,	PUNCT
ejpam-3176	116	18	s4	s4	PROPN
ejpam-3176	116	19	=	=	SYM
ejpam-3176	116	20	(	(	PUNCT
ejpam-3176	116	21	1	1	NUM
ejpam-3176	116	22	/	/	SYM
ejpam-3176	116	23	n	n	CCONJ
ejpam-3176	116	24	)	)	PUNCT
ejpam-3176	117	1	∑n	∑n	PROPN
ejpam-3176	117	2	i=1	i=1	PROPN
ejpam-3176	117	3	kik	kik	PROPN
ejpam-3176	117	4	t	t	PROPN
ejpam-3176	117	5	i	i	PRON
ejpam-3176	117	6	.	.	PUNCT
ejpam-3176	118	1	(	(	PUNCT
ejpam-3176	118	2	v	v	NOUN
ejpam-3176	118	3	)	)	PUNCT
ejpam-3176	118	4	s5	s5	PROPN
ejpam-3176	118	5	is	be	AUX
ejpam-3176	118	6	the	the	DET
ejpam-3176	118	7	first	first	ADJ
ejpam-3176	118	8	step	step	NOUN
ejpam-3176	118	9	of	of	ADP
ejpam-3176	118	10	the	the	DET
ejpam-3176	118	11	bacon	bacon	NOUN
ejpam-3176	118	12	algorithm	algorithm	NOUN
ejpam-3176	118	13	(	(	PUNCT
ejpam-3176	118	14	billor	billor	NOUN
ejpam-3176	118	15	et	et	PROPN
ejpam-3176	118	16	al	al	PROPN
ejpam-3176	118	17	.	.	PUNCT
ejpam-3176	119	1	[	[	X
ejpam-3176	119	2	2	2	NUM
ejpam-3176	119	3	]	]	PUNCT
ejpam-3176	119	4	)	)	PUNCT
ejpam-3176	119	5	.	.	PUNCT
ejpam-3176	120	1	the	the	DET
ejpam-3176	120	2	{	{	PUNCT
ejpam-3176	120	3	n/2	n/2	NOUN
ejpam-3176	120	4	}	}	PUNCT
ejpam-3176	120	5	standardized	standardize	VERB
ejpam-3176	120	6	observation	observation	NOUN
ejpam-3176	120	7	zi	zi	NOUN
ejpam-3176	120	8	has	have	VERB
ejpam-3176	120	9	the	the	DET
ejpam-3176	120	10	smallest	small	ADJ
ejpam-3176	120	11	norm	norm	NOUN
ejpam-3176	120	12	and	and	CCONJ
ejpam-3176	120	13	is	be	AUX
ejpam-3176	120	14	used	use	VERB
ejpam-3176	120	15	to	to	PART
ejpam-3176	120	16	compute	compute	VERB
ejpam-3176	120	17	the	the	DET
ejpam-3176	120	18	mean	mean	NOUN
ejpam-3176	120	19	and	and	CCONJ
ejpam-3176	120	20	covariance	covariance	NOUN
ejpam-3176	120	21	matrix	matrix	NOUN
ejpam-3176	120	22	.	.	PUNCT
ejpam-3176	121	1	(	(	PUNCT
ejpam-3176	121	2	vi	vi	X
ejpam-3176	121	3	)	)	PUNCT
ejpam-3176	121	4	a	a	DET
ejpam-3176	121	5	scatter	scatter	NOUN
ejpam-3176	121	6	estimate	estimate	NOUN
ejpam-3176	121	7	is	be	AUX
ejpam-3176	121	8	the	the	DET
ejpam-3176	121	9	raw	raw	ADJ
ejpam-3176	121	10	version	version	NOUN
ejpam-3176	121	11	of	of	ADP
ejpam-3176	121	12	the	the	DET
ejpam-3176	121	13	ogk	ogk	NOUN
ejpam-3176	121	14	estimator	estimator	NOUN
ejpam-3176	121	15	.	.	PUNCT
ejpam-3176	122	1	for	for	ADP
ejpam-3176	122	2	m	m	PROPN
ejpam-3176	122	3	(	(	PUNCT
ejpam-3176	122	4	·	·	PUNCT
ejpam-3176	122	5	)	)	PUNCT
ejpam-3176	122	6	,	,	PUNCT
ejpam-3176	122	7	s	s	X
ejpam-3176	122	8	(	(	PUNCT
ejpam-3176	122	9	·	·	PUNCT
ejpam-3176	122	10	)	)	PUNCT
ejpam-3176	122	11	,	,	PUNCT
ejpam-3176	122	12	and	and	CCONJ
ejpam-3176	122	13	the	the	DET
ejpam-3176	122	14	median	median	NOUN
ejpam-3176	122	15	,	,	PUNCT
ejpam-3176	122	16	qn	qn	PROPN
ejpam-3176	122	17	is	be	AUX
ejpam-3176	122	18	used	use	VERB
ejpam-3176	122	19	for	for	ADP
ejpam-3176	122	20	simplicity	simplicity	NOUN
ejpam-3176	122	21	.	.	PUNCT
ejpam-3176	123	1	after	after	ADP
ejpam-3176	123	2	standardizing	standardize	VERB
ejpam-3176	123	3	the	the	DET
ejpam-3176	123	4	data	datum	NOUN
ejpam-3176	123	5	and	and	CCONJ
ejpam-3176	123	6	obtaining	obtain	VERB
ejpam-3176	123	7	c	c	NOUN
ejpam-3176	123	8	,	,	PUNCT
ejpam-3176	123	9	three	three	NUM
ejpam-3176	123	10	steps	step	NOUN
ejpam-3176	123	11	are	be	AUX
ejpam-3176	123	12	completed	complete	VERB
ejpam-3176	123	13	to	to	PART
ejpam-3176	123	14	obtain	obtain	VERB
ejpam-3176	123	15	the	the	DET
ejpam-3176	123	16	covariance	covariance	NOUN
ejpam-3176	123	17	and	and	CCONJ
ejpam-3176	123	18	mean	mean	ADV
ejpam-3176	123	19	of	of	ADP
ejpam-3176	123	20	detmcd	detmcd	NOUN
ejpam-3176	123	21	.	.	PUNCT
ejpam-3176	124	1	(	(	PUNCT
ejpam-3176	124	2	i	i	NOUN
ejpam-3176	124	3	)	)	PUNCT
ejpam-3176	124	4	the	the	DET
ejpam-3176	124	5	matrix	matrix	NOUN
ejpam-3176	124	6	e	e	NOUN
ejpam-3176	124	7	of	of	ADP
ejpam-3176	124	8	the	the	DET
ejpam-3176	124	9	eigenvectors	eigenvector	NOUN
ejpam-3176	124	10	of	of	ADP
ejpam-3176	124	11	sk	sk	PROPN
ejpam-3176	124	12	is	be	AUX
ejpam-3176	124	13	computed	compute	VERB
ejpam-3176	124	14	and	and	CCONJ
ejpam-3176	124	15	b	b	X
ejpam-3176	124	16	=	=	SYM
ejpam-3176	124	17	ze	ze	PROPN
ejpam-3176	124	18	is	be	AUX
ejpam-3176	124	19	applied	apply	VERB
ejpam-3176	124	20	.	.	PUNCT
ejpam-3176	125	1	(	(	PUNCT
ejpam-3176	125	2	ii	ii	NOUN
ejpam-3176	125	3	)	)	PUNCT
ejpam-3176	125	4	the	the	DET
ejpam-3176	125	5	center	center	NOUN
ejpam-3176	125	6	of	of	ADP
ejpam-3176	125	7	z	z	PROPN
ejpam-3176	125	8	is	be	AUX
ejpam-3176	125	9	estimated	estimate	VERB
ejpam-3176	125	10	using	use	VERB
ejpam-3176	125	11	σk(z	σk(z	NOUN
ejpam-3176	125	12	)	)	PUNCT
ejpam-3176	125	13	=	=	NOUN
ejpam-3176	125	14	elet	elet	NOUN
ejpam-3176	125	15	,	,	PUNCT
ejpam-3176	125	16	where	where	SCONJ
ejpam-3176	125	17	l	l	NOUN
ejpam-3176	125	18	=	=	SYM
ejpam-3176	125	19	diag(q2	diag(q2	NOUN
ejpam-3176	125	20	n(b1	n(b1	NOUN
ejpam-3176	125	21	)	)	PUNCT
ejpam-3176	125	22	,	,	PUNCT
ejpam-3176	125	23	.	.	PUNCT
ejpam-3176	125	24	.	.	PUNCT
ejpam-3176	125	25	.	.	PUNCT
ejpam-3176	126	1	,	,	PUNCT
ejpam-3176	126	2	q2	q2	PROPN
ejpam-3176	126	3	n(bp	n(bp	NUM
ejpam-3176	126	4	)	)	PUNCT
ejpam-3176	126	5	)	)	PUNCT
ejpam-3176	126	6	.	.	PUNCT
ejpam-3176	127	1	(	(	PUNCT
ejpam-3176	127	2	iii	iii	X
ejpam-3176	127	3	)	)	PUNCT
ejpam-3176	127	4	the	the	DET
ejpam-3176	127	5	covariance	covariance	NOUN
ejpam-3176	127	6	of	of	ADP
ejpam-3176	127	7	z	z	NOUN
ejpam-3176	127	8	is	be	AUX
ejpam-3176	127	9	estimated	estimate	VERB
ejpam-3176	127	10	using	use	VERB
ejpam-3176	127	11	sphere	sphere	NOUN
ejpam-3176	127	12	data	datum	NOUN
ejpam-3176	127	13	,	,	PUNCT
ejpam-3176	127	14	the	the	DET
ejpam-3176	127	15	coordinate	coordinate	NOUN
ejpam-3176	127	16	-	-	PUNCT
ejpam-3176	127	17	wise	wise	ADJ
ejpam-3176	127	18	median	median	NOUN
ejpam-3176	127	19	is	be	AUX
ejpam-3176	127	20	applied	apply	VERB
ejpam-3176	127	21	and	and	CCONJ
ejpam-3176	127	22	transformed	transform	VERB
ejpam-3176	127	23	back	back	ADV
ejpam-3176	127	24	,	,	PUNCT
ejpam-3176	127	25	µk(z	µk(z	ADV
ejpam-3176	127	26	)	)	PUNCT
ejpam-3176	127	27	=	=	SYM
ejpam-3176	127	28	σ	σ	PROPN
ejpam-3176	127	29	1/2	1/2	NUM
ejpam-3176	127	30	k	k	X
ejpam-3176	127	31	(	(	PUNCT
ejpam-3176	127	32	med(zς	med(zς	PROPN
ejpam-3176	127	33	1/2	1/2	NUM
ejpam-3176	127	34	k	k	NOUN
ejpam-3176	127	35	)	)	PUNCT
ejpam-3176	127	36	)	)	PUNCT
ejpam-3176	127	37	.	.	PUNCT
ejpam-3176	128	1	mufda	mufda	PROPN
ejpam-3176	128	2	j.	j.	PROPN
ejpam-3176	128	3	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	128	4	,	,	PUNCT
ejpam-3176	128	5	taha	taha	PROPN
ejpam-3176	128	6	radwan	radwan	PROPN
ejpam-3176	128	7	and	and	CCONJ
ejpam-3176	128	8	khalid	khalid	PROPN
ejpam-3176	128	9	abunawas	abunawas	PROPN
ejpam-3176	128	10	/	/	SYM
ejpam-3176	128	11	eur	eur	PROPN
ejpam-3176	128	12	.	.	PUNCT
ejpam-3176	129	1	j.	j.	PROPN
ejpam-3176	129	2	pure	pure	PROPN
ejpam-3176	129	3	appl	appl	PROPN
ejpam-3176	129	4	.	.	PROPN
ejpam-3176	129	5	math	math	PROPN
ejpam-3176	129	6	,	,	PUNCT
ejpam-3176	129	7	11	11	NUM
ejpam-3176	129	8	(	(	PUNCT
ejpam-3176	129	9	1	1	NUM
ejpam-3176	129	10	)	)	PUNCT
ejpam-3176	129	11	(	(	PUNCT
ejpam-3176	129	12	2018	2018	NUM
ejpam-3176	129	13	)	)	PUNCT
ejpam-3176	129	14	,	,	PUNCT
ejpam-3176	129	15	284	284	NUM
ejpam-3176	129	16	-	-	SYM
ejpam-3176	129	17	298	298	NUM
ejpam-3176	129	18	288	288	NUM
ejpam-3176	129	19	for	for	ADP
ejpam-3176	129	20	all	all	DET
ejpam-3176	129	21	the	the	DET
ejpam-3176	129	22	six	six	NUM
ejpam-3176	129	23	estimates	estimate	NOUN
ejpam-3176	129	24	sk	sk	ADP
ejpam-3176	129	25	,	,	PUNCT
ejpam-3176	129	26	(	(	PUNCT
ejpam-3176	129	27	µk(z),σk(z	µk(z),σk(z	NOUN
ejpam-3176	129	28	)	)	PUNCT
ejpam-3176	129	29	)	)	PUNCT
ejpam-3176	130	1	is	be	AUX
ejpam-3176	130	2	used	use	VERB
ejpam-3176	130	3	to	to	PART
ejpam-3176	130	4	compute	compute	VERB
ejpam-3176	130	5	the	the	DET
ejpam-3176	130	6	statistical	statistical	ADJ
ejpam-3176	130	7	distance	distance	NOUN
ejpam-3176	130	8	,	,	PUNCT
ejpam-3176	130	9	as	as	SCONJ
ejpam-3176	130	10	follows	follow	VERB
ejpam-3176	130	11	:	:	PUNCT
ejpam-3176	130	12	dik	dik	ADJ
ejpam-3176	130	13	=	=	NOUN
ejpam-3176	130	14	d(µk(z),σk(z	d(µk(z),σk(z	NOUN
ejpam-3176	130	15	)	)	PUNCT
ejpam-3176	130	16	)	)	PUNCT
ejpam-3176	130	17	.	.	PUNCT
ejpam-3176	131	1	(	(	PUNCT
ejpam-3176	131	2	3	3	X
ejpam-3176	131	3	)	)	PUNCT
ejpam-3176	131	4	for	for	ADP
ejpam-3176	131	5	the	the	DET
ejpam-3176	131	6	initial	initial	ADJ
ejpam-3176	131	7	estimate	estimate	NOUN
ejpam-3176	131	8	k	k	PROPN
ejpam-3176	131	9	,	,	PUNCT
ejpam-3176	131	10	h	h	NOUN
ejpam-3176	131	11	◦	◦	NOUN
ejpam-3176	131	12	=	=	PUNCT
ejpam-3176	132	1	[	[	X
ejpam-3176	132	2	n/2	n/2	X
ejpam-3176	132	3	]	]	PUNCT
ejpam-3176	132	4	observations	observation	NOUN
ejpam-3176	132	5	are	be	AUX
ejpam-3176	132	6	taken	take	VERB
ejpam-3176	132	7	with	with	ADP
ejpam-3176	132	8	the	the	DET
ejpam-3176	132	9	smallest	small	ADJ
ejpam-3176	132	10	dik	dik	NOUN
ejpam-3176	132	11	.	.	PUNCT
ejpam-3176	133	1	then	then	ADV
ejpam-3176	133	2	,	,	PUNCT
ejpam-3176	133	3	the	the	DET
ejpam-3176	133	4	statistical	statistical	ADJ
ejpam-3176	133	5	distances	distance	NOUN
ejpam-3176	133	6	d∗ik	d∗ik	PROPN
ejpam-3176	133	7	for	for	ADP
ejpam-3176	133	8	h	h	NOUN
ejpam-3176	133	9	◦	◦	NOUN
ejpam-3176	133	10	observations	observation	NOUN
ejpam-3176	133	11	are	be	AUX
ejpam-3176	133	12	computed	compute	VERB
ejpam-3176	133	13	.	.	PUNCT
ejpam-3176	134	1	all	all	DET
ejpam-3176	134	2	h	h	NOUN
ejpam-3176	134	3	observations	observation	NOUN
ejpam-3176	134	4	xi	xi	X
ejpam-3176	134	5	are	be	AUX
ejpam-3176	134	6	calculated	calculate	VERB
ejpam-3176	134	7	with	with	ADP
ejpam-3176	134	8	the	the	DET
ejpam-3176	134	9	smallest	small	ADJ
ejpam-3176	134	10	d∗ik	d∗ik	NOUN
ejpam-3176	134	11	for	for	ADP
ejpam-3176	134	12	all	all	DET
ejpam-3176	134	13	the	the	DET
ejpam-3176	134	14	six	six	NUM
ejpam-3176	134	15	estimates	estimate	NOUN
ejpam-3176	134	16	.	.	PUNCT
ejpam-3176	135	1	the	the	DET
ejpam-3176	135	2	final	final	ADJ
ejpam-3176	135	3	step	step	NOUN
ejpam-3176	135	4	is	be	AUX
ejpam-3176	135	5	the	the	DET
ejpam-3176	135	6	application	application	NOUN
ejpam-3176	135	7	of	of	ADP
ejpam-3176	135	8	the	the	DET
ejpam-3176	135	9	concentration	concentration	NOUN
ejpam-3176	135	10	step	step	NOUN
ejpam-3176	135	11	(	(	PUNCT
ejpam-3176	135	12	c	c	NOUN
ejpam-3176	135	13	-	-	PUNCT
ejpam-3176	135	14	step	step	NOUN
ejpam-3176	135	15	)	)	PUNCT
ejpam-3176	135	16	until	until	ADP
ejpam-3176	135	17	convergence	convergence	NOUN
ejpam-3176	135	18	.	.	PUNCT
ejpam-3176	136	1	the	the	DET
ejpam-3176	136	2	estimate	estimate	NOUN
ejpam-3176	136	3	with	with	ADP
ejpam-3176	136	4	the	the	DET
ejpam-3176	136	5	smallest	small	ADJ
ejpam-3176	136	6	determinant	determinant	NOUN
ejpam-3176	136	7	is	be	AUX
ejpam-3176	136	8	called	call	VERB
ejpam-3176	136	9	the	the	DET
ejpam-3176	136	10	raw	raw	ADJ
ejpam-3176	136	11	detmcd	detmcd	NOUN
ejpam-3176	136	12	.	.	PUNCT
ejpam-3176	137	1	the	the	DET
ejpam-3176	137	2	final	final	ADJ
ejpam-3176	137	3	detmcd	detmcd	NOUN
ejpam-3176	137	4	is	be	AUX
ejpam-3176	137	5	obtained	obtain	VERB
ejpam-3176	137	6	by	by	ADP
ejpam-3176	137	7	applying	apply	VERB
ejpam-3176	137	8	the	the	DET
ejpam-3176	137	9	reweighted	reweighte	VERB
ejpam-3176	137	10	fastmcd	fastmcd	NOUN
ejpam-3176	137	11	algorithm	algorithm	NOUN
ejpam-3176	137	12	.	.	PUNCT
ejpam-3176	138	1	as	as	SCONJ
ejpam-3176	138	2	previously	previously	ADV
ejpam-3176	138	3	mentioned	mention	VERB
ejpam-3176	138	4	,	,	PUNCT
ejpam-3176	138	5	rlda	rlda	NOUN
ejpam-3176	138	6	is	be	AUX
ejpam-3176	138	7	constructed	construct	VERB
ejpam-3176	138	8	based	base	VERB
ejpam-3176	138	9	on	on	ADP
ejpam-3176	138	10	the	the	DET
ejpam-3176	138	11	detmcd	detmcd	NOUN
ejpam-3176	138	12	algorithm	algorithm	NOUN
ejpam-3176	138	13	.	.	PUNCT
ejpam-3176	139	1	rlda	rlda	NOUN
ejpam-3176	139	2	is	be	AUX
ejpam-3176	139	3	derived	derive	VERB
ejpam-3176	139	4	by	by	ADP
ejpam-3176	139	5	inputting	inputte	VERB
ejpam-3176	139	6	the	the	DET
ejpam-3176	139	7	location	location	NOUN
ejpam-3176	139	8	and	and	CCONJ
ejpam-3176	139	9	scatter	scatter	NOUN
ejpam-3176	139	10	matrices	matrix	NOUN
ejpam-3176	139	11	obtained	obtain	VERB
ejpam-3176	139	12	based	base	VERB
ejpam-3176	139	13	on	on	ADP
ejpam-3176	139	14	the	the	DET
ejpam-3176	139	15	detmcd	detmcd	NOUN
ejpam-3176	139	16	algorithm	algorithm	NOUN
ejpam-3176	139	17	into	into	ADP
ejpam-3176	139	18	lda	lda	PROPN
ejpam-3176	139	19	,	,	PUNCT
ejpam-3176	139	20	as	as	SCONJ
ejpam-3176	139	21	follows	follow	VERB
ejpam-3176	139	22	:	:	PUNCT
ejpam-3176	139	23	_	_	PUNCT
ejpam-3176	140	1	d	d	X
ejpam-3176	140	2	detmcd	detmcd	PROPN
ejpam-3176	140	3	j	j	PROPN
ejpam-3176	140	4	(	(	PUNCT
ejpam-3176	140	5	x	x	X
ejpam-3176	140	6	)	)	PUNCT
ejpam-3176	140	7	=	=	PRON
ejpam-3176	140	8	µtjς	µtjς	VERB
ejpam-3176	140	9	−1x−	−1x−	ADJ
ejpam-3176	140	10	1	1	NUM
ejpam-3176	140	11	2	2	NUM
ejpam-3176	140	12	µtjς	µtjς	NOUN
ejpam-3176	140	13	−1µj	−1µj	X
ejpam-3176	140	14	+	+	X
ejpam-3176	140	15	ln(pj	ln(pj	PROPN
ejpam-3176	140	16	)	)	PUNCT
ejpam-3176	140	17	.	.	PUNCT
ejpam-3176	141	1	(	(	PUNCT
ejpam-3176	141	2	4	4	X
ejpam-3176	141	3	)	)	PUNCT
ejpam-3176	141	4	outliers	outlier	NOUN
ejpam-3176	141	5	in	in	ADP
ejpam-3176	141	6	the	the	DET
ejpam-3176	141	7	data	datum	NOUN
ejpam-3176	141	8	are	be	AUX
ejpam-3176	141	9	flagged	flag	VERB
ejpam-3176	141	10	to	to	PART
ejpam-3176	141	11	robustify	robustify	VERB
ejpam-3176	141	12	the	the	DET
ejpam-3176	141	13	location	location	NOUN
ejpam-3176	141	14	and	and	CCONJ
ejpam-3176	141	15	scatter	scatter	NOUN
ejpam-3176	141	16	matrices	matrix	NOUN
ejpam-3176	141	17	.	.	PUNCT
ejpam-3176	142	1	the	the	DET
ejpam-3176	142	2	robust	robust	ADJ
ejpam-3176	142	3	distance	distance	NOUN
ejpam-3176	142	4	for	for	ADP
ejpam-3176	142	5	each	each	DET
ejpam-3176	142	6	observation	observation	NOUN
ejpam-3176	142	7	xij	xij	PRON
ejpam-3176	142	8	is	be	AUX
ejpam-3176	142	9	computed	compute	VERB
ejpam-3176	142	10	from	from	ADP
ejpam-3176	142	11	the	the	DET
ejpam-3176	142	12	group	group	NOUN
ejpam-3176	142	13	πj	πj	VERB
ejpam-3176	142	14	to	to	PART
ejpam-3176	142	15	estimate	estimate	VERB
ejpam-3176	142	16	the	the	DET
ejpam-3176	142	17	membership	membership	NOUN
ejpam-3176	142	18	probability	probability	NOUN
ejpam-3176	142	19	,	,	PUNCT
ejpam-3176	142	20	as	as	SCONJ
ejpam-3176	142	21	follows	follow	VERB
ejpam-3176	142	22	:	:	PUNCT
ejpam-3176	143	1	rddetmcd	rddetmcd	NOUN
ejpam-3176	143	2	ij	ij	NOUN
ejpam-3176	143	3	=	=	NOUN
ejpam-3176	143	4	√	√	INTJ
ejpam-3176	143	5	(	(	PUNCT
ejpam-3176	143	6	xij	xij	NOUN
ejpam-3176	143	7	−	−	PROPN
ejpam-3176	143	8	µ̂j)tς−1(xij	µ̂j)tς−1(xij	ADV
ejpam-3176	143	9	−	−	PROPN
ejpam-3176	143	10	µ̂j	µ̂j	NOUN
ejpam-3176	143	11	)	)	PUNCT
ejpam-3176	143	12	.	.	PUNCT
ejpam-3176	144	1	(	(	PUNCT
ejpam-3176	144	2	5	5	NUM
ejpam-3176	144	3	)	)	PUNCT
ejpam-3176	144	4	then	then	ADV
ejpam-3176	144	5	,	,	PUNCT
ejpam-3176	144	6	xij	xij	PRON
ejpam-3176	144	7	is	be	AUX
ejpam-3176	144	8	considered	consider	VERB
ejpam-3176	144	9	the	the	DET
ejpam-3176	144	10	outlier	outlier	ADJ
ejpam-3176	144	11	observation	observation	NOUN
ejpam-3176	144	12	if	if	SCONJ
ejpam-3176	144	13	and	and	CCONJ
ejpam-3176	144	14	only	only	ADV
ejpam-3176	144	15	if	if	SCONJ
ejpam-3176	144	16	rdij	rdij	VERB
ejpam-3176	144	17	>	>	PUNCT
ejpam-3176	144	18	√	√	PUNCT
ejpam-3176	144	19	χ2	χ2	PROPN
ejpam-3176	144	20	p,0.975	p,0.975	PROPN
ejpam-3176	144	21	.	.	PUNCT
ejpam-3176	145	1	(	(	PUNCT
ejpam-3176	145	2	6	6	NUM
ejpam-3176	145	3	)	)	PUNCT
ejpam-3176	145	4	finally	finally	ADV
ejpam-3176	145	5	,	,	PUNCT
ejpam-3176	145	6	the	the	DET
ejpam-3176	145	7	membership	membership	NOUN
ejpam-3176	145	8	probability	probability	NOUN
ejpam-3176	145	9	_	_	NOUN
ejpam-3176	146	1	p	p	NOUN
ejpam-3176	146	2	detmcd	detmcd	PROPN
ejpam-3176	146	3	j	j	PROPN
ejpam-3176	146	4	can	can	AUX
ejpam-3176	146	5	be	be	AUX
ejpam-3176	146	6	obtained	obtain	VERB
ejpam-3176	146	7	using	use	VERB
ejpam-3176	146	8	formula	formula	NOUN
ejpam-3176	146	9	(	(	PUNCT
ejpam-3176	146	10	8)	8)	NUM
ejpam-3176	146	11	after	after	ADP
ejpam-3176	146	12	applying	apply	VERB
ejpam-3176	146	13	the	the	DET
ejpam-3176	146	14	detmcd	detmcd	NOUN
ejpam-3176	146	15	estimator	estimator	NOUN
ejpam-3176	146	16	that	that	PRON
ejpam-3176	146	17	is	be	AUX
ejpam-3176	146	18	defined	define	VERB
ejpam-3176	146	19	as	as	SCONJ
ejpam-3176	146	20	follows	follow	VERB
ejpam-3176	146	21	:	:	PUNCT
ejpam-3176	146	22	_	_	PUNCT
ejpam-3176	146	23	p	p	NOUN
ejpam-3176	146	24	detmcd	detmcd	PROPN
ejpam-3176	146	25	j	j	PROPN
ejpam-3176	146	26	=	=	PROPN
ejpam-3176	146	27	ñj	ñj	NOUN
ejpam-3176	146	28	ñ	ñ	PROPN
ejpam-3176	146	29	.	.	PUNCT
ejpam-3176	147	1	(	(	PUNCT
ejpam-3176	147	2	7	7	NUM
ejpam-3176	147	3	)	)	SYM
ejpam-3176	147	4	5	5	NUM
ejpam-3176	147	5	.	.	X
ejpam-3176	147	6	fastmcd	fastmcd	VERB
ejpam-3176	147	7	the	the	DET
ejpam-3176	147	8	main	main	ADJ
ejpam-3176	147	9	feature	feature	NOUN
ejpam-3176	147	10	of	of	ADP
ejpam-3176	147	11	the	the	DET
ejpam-3176	147	12	fastmcd	fastmcd	NOUN
ejpam-3176	147	13	algorithm	algorithm	NOUN
ejpam-3176	147	14	is	be	AUX
ejpam-3176	147	15	the	the	DET
ejpam-3176	147	16	c	c	NOUN
ejpam-3176	147	17	-	-	PUNCT
ejpam-3176	147	18	step	step	NOUN
ejpam-3176	147	19	,	,	PUNCT
ejpam-3176	147	20	where	where	SCONJ
ejpam-3176	147	21	det(σnew	det(σnew	NOUN
ejpam-3176	147	22	)	)	PUNCT
ejpam-3176	147	23	det(σold	det(σold	PRON
ejpam-3176	147	24	)	)	PUNCT
ejpam-3176	147	25	with	with	ADP
ejpam-3176	147	26	equality	equality	NOUN
ejpam-3176	147	27	if	if	SCONJ
ejpam-3176	147	28	det(σnew	det(σnew	NOUN
ejpam-3176	147	29	)	)	PUNCT
ejpam-3176	147	30	=	=	PUNCT
ejpam-3176	147	31	det(σold	det(σold	PRON
ejpam-3176	147	32	)	)	PUNCT
ejpam-3176	147	33	(	(	PUNCT
ejpam-3176	147	34	rousseeuw	rousseeuw	NOUN
ejpam-3176	147	35	and	and	CCONJ
ejpam-3176	147	36	driessen	driessen	NOUN
ejpam-3176	147	37	[	[	X
ejpam-3176	147	38	25	25	NUM
ejpam-3176	147	39	]	]	PUNCT
ejpam-3176	147	40	)	)	PUNCT
ejpam-3176	147	41	.	.	PUNCT
ejpam-3176	148	1	the	the	DET
ejpam-3176	148	2	application	application	NOUN
ejpam-3176	148	3	of	of	ADP
ejpam-3176	148	4	the	the	DET
ejpam-3176	148	5	c	c	NOUN
ejpam-3176	148	6	-	-	PUNCT
ejpam-3176	148	7	step	step	NOUN
ejpam-3176	148	8	will	will	AUX
ejpam-3176	148	9	yield	yield	VERB
ejpam-3176	148	10	the	the	DET
ejpam-3176	148	11	sequence	sequence	NOUN
ejpam-3176	148	12	of	of	ADP
ejpam-3176	148	13	determinants	determinant	NOUN
ejpam-3176	148	14	,	,	PUNCT
ejpam-3176	148	15	which	which	PRON
ejpam-3176	148	16	must	must	AUX
ejpam-3176	148	17	converge	converge	VERB
ejpam-3176	148	18	in	in	ADP
ejpam-3176	148	19	a	a	DET
ejpam-3176	148	20	finite	finite	ADJ
ejpam-3176	148	21	number	number	NOUN
ejpam-3176	148	22	of	of	ADP
ejpam-3176	148	23	steps	step	NOUN
ejpam-3176	148	24	.	.	PUNCT
ejpam-3176	149	1	the	the	DET
ejpam-3176	149	2	final	final	ADJ
ejpam-3176	149	3	iteration	iteration	NOUN
ejpam-3176	149	4	can	can	AUX
ejpam-3176	149	5	not	not	PART
ejpam-3176	149	6	be	be	AUX
ejpam-3176	149	7	guaranteed	guarantee	VERB
ejpam-3176	149	8	to	to	PART
ejpam-3176	149	9	be	be	AUX
ejpam-3176	149	10	the	the	DET
ejpam-3176	149	11	minimum	minimum	ADJ
ejpam-3176	149	12	value	value	NOUN
ejpam-3176	149	13	of	of	ADP
ejpam-3176	149	14	the	the	DET
ejpam-3176	149	15	mcd	mcd	NOUN
ejpam-3176	149	16	objective	objective	ADJ
ejpam-3176	149	17	function	function	NOUN
ejpam-3176	149	18	.	.	PUNCT
ejpam-3176	150	1	the	the	DET
ejpam-3176	150	2	fastmcd	fastmcd	NOUN
ejpam-3176	150	3	algorithm	algorithm	NOUN
ejpam-3176	150	4	applied	apply	VERB
ejpam-3176	150	5	two	two	NUM
ejpam-3176	150	6	c	c	NOUN
ejpam-3176	150	7	-	-	PUNCT
ejpam-3176	150	8	steps	step	NOUN
ejpam-3176	150	9	to	to	ADP
ejpam-3176	150	10	each	each	DET
ejpam-3176	150	11	initial	initial	ADJ
ejpam-3176	150	12	subset	subset	NOUN
ejpam-3176	150	13	,	,	PUNCT
ejpam-3176	150	14	and	and	CCONJ
ejpam-3176	150	15	only	only	ADV
ejpam-3176	150	16	ten	ten	NUM
ejpam-3176	150	17	subsets	subset	NOUN
ejpam-3176	150	18	with	with	ADP
ejpam-3176	150	19	the	the	DET
ejpam-3176	150	20	smallest	small	ADJ
ejpam-3176	150	21	determinant	determinant	ADJ
ejpam-3176	150	22	for	for	ADP
ejpam-3176	150	23	c	c	NOUN
ejpam-3176	150	24	-	-	PUNCT
ejpam-3176	150	25	step	step	NOUN
ejpam-3176	150	26	are	be	AUX
ejpam-3176	150	27	taken	take	VERB
ejpam-3176	150	28	until	until	ADP
ejpam-3176	150	29	initial	initial	ADJ
ejpam-3176	150	30	convergence	convergence	NOUN
ejpam-3176	150	31	.	.	PUNCT
ejpam-3176	151	1	three	three	NUM
ejpam-3176	151	2	approaches	approach	NOUN
ejpam-3176	151	3	will	will	AUX
ejpam-3176	151	4	be	be	AUX
ejpam-3176	151	5	used	use	VERB
ejpam-3176	151	6	with	with	ADP
ejpam-3176	151	7	the	the	DET
ejpam-3176	151	8	fastmcd	fastmcd	NOUN
ejpam-3176	151	9	algorithm	algorithm	NOUN
ejpam-3176	151	10	to	to	PART
ejpam-3176	151	11	estimate	estimate	VERB
ejpam-3176	151	12	the	the	DET
ejpam-3176	151	13	mufda	mufda	NOUN
ejpam-3176	151	14	j.	j.	PROPN
ejpam-3176	151	15	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	151	16	,	,	PUNCT
ejpam-3176	151	17	taha	taha	PROPN
ejpam-3176	151	18	radwan	radwan	PROPN
ejpam-3176	151	19	and	and	CCONJ
ejpam-3176	151	20	khalid	khalid	PROPN
ejpam-3176	151	21	abunawas	abunawas	PROPN
ejpam-3176	151	22	/	/	SYM
ejpam-3176	151	23	eur	eur	PROPN
ejpam-3176	151	24	.	.	PUNCT
ejpam-3176	152	1	j.	j.	PROPN
ejpam-3176	152	2	pure	pure	PROPN
ejpam-3176	152	3	appl	appl	PROPN
ejpam-3176	152	4	.	.	PROPN
ejpam-3176	152	5	math	math	PROPN
ejpam-3176	152	6	,	,	PUNCT
ejpam-3176	152	7	11	11	NUM
ejpam-3176	152	8	(	(	PUNCT
ejpam-3176	152	9	1	1	NUM
ejpam-3176	152	10	)	)	PUNCT
ejpam-3176	152	11	(	(	PUNCT
ejpam-3176	152	12	2018	2018	NUM
ejpam-3176	152	13	)	)	PUNCT
ejpam-3176	152	14	,	,	PUNCT
ejpam-3176	152	15	284	284	NUM
ejpam-3176	152	16	-	-	SYM
ejpam-3176	152	17	298	298	NUM
ejpam-3176	152	18	289	289	NUM
ejpam-3176	152	19	mean	mean	NOUN
ejpam-3176	152	20	and	and	CCONJ
ejpam-3176	152	21	covariance	covariance	NOUN
ejpam-3176	152	22	matrix	matrix	NOUN
ejpam-3176	152	23	.	.	PUNCT
ejpam-3176	153	1	the	the	DET
ejpam-3176	153	2	same	same	ADJ
ejpam-3176	153	3	approach	approach	NOUN
ejpam-3176	153	4	applied	apply	VERB
ejpam-3176	153	5	to	to	ADP
ejpam-3176	153	6	the	the	DET
ejpam-3176	153	7	detmcd	detmcd	ADJ
ejpam-3176	153	8	algorithm	algorithm	NOUN
ejpam-3176	153	9	to	to	PART
ejpam-3176	153	10	obtain	obtain	VERB
ejpam-3176	153	11	rlda	rlda	NOUN
ejpam-3176	153	12	,	,	PUNCT
ejpam-3176	153	13	and	and	CCONJ
ejpam-3176	153	14	prj	prj	NOUN
ejpam-3176	153	15	is	be	AUX
ejpam-3176	153	16	applied	apply	VERB
ejpam-3176	153	17	to	to	ADP
ejpam-3176	153	18	the	the	DET
ejpam-3176	153	19	fastmcd	fastmcd	NOUN
ejpam-3176	153	20	algorithm	algorithm	NOUN
ejpam-3176	153	21	,	,	PUNCT
ejpam-3176	153	22	which	which	PRON
ejpam-3176	153	23	is	be	AUX
ejpam-3176	153	24	defined	define	VERB
ejpam-3176	153	25	as	as	SCONJ
ejpam-3176	153	26	follows	follow	VERB
ejpam-3176	153	27	:	:	PUNCT
ejpam-3176	153	28	_	_	PUNCT
ejpam-3176	154	1	d	d	X
ejpam-3176	154	2	fastmcd	fastmcd	NOUN
ejpam-3176	154	3	j	j	PROPN
ejpam-3176	154	4	(	(	PUNCT
ejpam-3176	154	5	x	x	X
ejpam-3176	154	6	)	)	PUNCT
ejpam-3176	154	7	=	=	PRON
ejpam-3176	154	8	µtjς	µtjς	VERB
ejpam-3176	154	9	−1x−	−1x−	ADJ
ejpam-3176	154	10	1	1	NUM
ejpam-3176	154	11	2	2	NUM
ejpam-3176	154	12	µtjς	µtjς	NOUN
ejpam-3176	154	13	−1µj	−1µj	X
ejpam-3176	154	14	+	+	X
ejpam-3176	154	15	ln(pj	ln(pj	PROPN
ejpam-3176	154	16	)	)	PUNCT
ejpam-3176	154	17	.	.	PUNCT
ejpam-3176	155	1	(	(	PUNCT
ejpam-3176	155	2	8)	8)	NUM
ejpam-3176	155	3	the	the	DET
ejpam-3176	155	4	membership	membership	NOUN
ejpam-3176	155	5	probability	probability	NOUN
ejpam-3176	155	6	pj	pj	PROPN
ejpam-3176	155	7	of	of	ADP
ejpam-3176	155	8	the	the	DET
ejpam-3176	155	9	robust	robust	ADJ
ejpam-3176	155	10	distance	distance	NOUN
ejpam-3176	155	11	is	be	AUX
ejpam-3176	155	12	defined	define	VERB
ejpam-3176	155	13	as	as	SCONJ
ejpam-3176	155	14	follows	follow	VERB
ejpam-3176	155	15	:	:	PUNCT
ejpam-3176	155	16	rdfastmcd	rdfastmcd	NOUN
ejpam-3176	155	17	ij	ij	NOUN
ejpam-3176	155	18	=	=	NOUN
ejpam-3176	155	19	√	√	PROPN
ejpam-3176	155	20	(	(	PUNCT
ejpam-3176	155	21	xij	xij	NOUN
ejpam-3176	155	22	−	−	PROPN
ejpam-3176	155	23	µ̂j)tς−1(xij	µ̂j)tς−1(xij	ADV
ejpam-3176	155	24	−	−	PROPN
ejpam-3176	155	25	µ̂j	µ̂j	NOUN
ejpam-3176	155	26	)	)	PUNCT
ejpam-3176	155	27	.	.	PUNCT
ejpam-3176	156	1	(	(	PUNCT
ejpam-3176	156	2	9	9	X
ejpam-3176	156	3	)	)	PUNCT
ejpam-3176	156	4	if	if	SCONJ
ejpam-3176	156	5	xij	xij	PRON
ejpam-3176	156	6	is	be	AUX
ejpam-3176	156	7	used	use	VERB
ejpam-3176	156	8	to	to	PART
ejpam-3176	156	9	consider	consider	VERB
ejpam-3176	156	10	the	the	DET
ejpam-3176	156	11	outliers	outlier	NOUN
ejpam-3176	156	12	in	in	ADP
ejpam-3176	156	13	formula	formula	NOUN
ejpam-3176	156	14	(	(	PUNCT
ejpam-3176	156	15	6	6	NUM
ejpam-3176	156	16	)	)	PUNCT
ejpam-3176	156	17	,	,	PUNCT
ejpam-3176	156	18	then	then	ADV
ejpam-3176	156	19	the	the	DET
ejpam-3176	156	20	membership	membership	NOUN
ejpam-3176	156	21	probability	probability	NOUN
ejpam-3176	156	22	of	of	ADP
ejpam-3176	156	23	the	the	DET
ejpam-3176	156	24	fastmcd	fastmcd	NOUN
ejpam-3176	156	25	estimator	estimator	NOUN
ejpam-3176	156	26	is	be	AUX
ejpam-3176	156	27	expressed	express	VERB
ejpam-3176	156	28	as	as	SCONJ
ejpam-3176	156	29	follows	follow	VERB
ejpam-3176	156	30	:	:	PUNCT
ejpam-3176	157	1	_	_	PUNCT
ejpam-3176	157	2	p	p	PRON
ejpam-3176	157	3	fastmcd	fastmcd	NOUN
ejpam-3176	158	1	j	j	PROPN
ejpam-3176	158	2	=	=	PROPN
ejpam-3176	158	3	ñj	ñj	PROPN
ejpam-3176	158	4	ñ	ñ	PROPN
ejpam-3176	158	5	.	.	PUNCT
ejpam-3176	159	1	(	(	PUNCT
ejpam-3176	159	2	10	10	NUM
ejpam-3176	159	3	)	)	PUNCT
ejpam-3176	159	4	6	6	NUM
ejpam-3176	159	5	.	.	PUNCT
ejpam-3176	160	1	fch	fch	PROPN
ejpam-3176	160	2	the	the	DET
ejpam-3176	160	3	most	most	ADV
ejpam-3176	160	4	practical	practical	ADJ
ejpam-3176	160	5	estimators	estimator	NOUN
ejpam-3176	160	6	are	be	AUX
ejpam-3176	160	7	used	use	VERB
ejpam-3176	160	8	as	as	ADP
ejpam-3176	160	9	a	a	DET
ejpam-3176	160	10	sequence	sequence	NOUN
ejpam-3176	160	11	of	of	ADP
ejpam-3176	160	12	n	n	PRON
ejpam-3176	160	13	trial	trial	NOUN
ejpam-3176	160	14	fits	fit	VERB
ejpam-3176	160	15	called	call	VERB
ejpam-3176	160	16	initial	initial	ADJ
ejpam-3176	160	17	estimator	estimator	NOUN
ejpam-3176	160	18	,	,	PUNCT
ejpam-3176	160	19	(	(	PUNCT
ejpam-3176	160	20	µ1,σ1	µ1,σ1	PROPN
ejpam-3176	160	21	)	)	PUNCT
ejpam-3176	160	22	,	,	PUNCT
ejpam-3176	160	23	(	(	PUNCT
ejpam-3176	160	24	µ2,σ2	µ2,σ2	PROPN
ejpam-3176	160	25	)	)	PUNCT
ejpam-3176	160	26	,	,	PUNCT
ejpam-3176	160	27	.	.	PUNCT
ejpam-3176	160	28	.	.	PUNCT
ejpam-3176	160	29	.	.	PUNCT
ejpam-3176	161	1	,	,	PUNCT
ejpam-3176	161	2	(	(	PUNCT
ejpam-3176	161	3	µn	µn	NOUN
ejpam-3176	161	4	,	,	PUNCT
ejpam-3176	161	5	σn	σn	NOUN
ejpam-3176	161	6	)	)	PUNCT
ejpam-3176	161	7	.	.	PUNCT
ejpam-3176	162	1	the	the	DET
ejpam-3176	162	2	initial	initial	ADJ
ejpam-3176	162	3	estimator	estimator	NOUN
ejpam-3176	162	4	(	(	PUNCT
ejpam-3176	162	5	µi	µi	PROPN
ejpam-3176	162	6	,	,	PUNCT
ejpam-3176	162	7	σi	σi	NOUN
ejpam-3176	162	8	)	)	PUNCT
ejpam-3176	162	9	that	that	PRON
ejpam-3176	162	10	minimizes	minimize	VERB
ejpam-3176	162	11	the	the	DET
ejpam-3176	162	12	evaluation	evaluation	NOUN
ejpam-3176	162	13	criterion	criterion	NOUN
ejpam-3176	162	14	will	will	AUX
ejpam-3176	162	15	be	be	AUX
ejpam-3176	162	16	used	use	VERB
ejpam-3176	162	17	in	in	ADP
ejpam-3176	162	18	the	the	DET
ejpam-3176	162	19	final	final	ADJ
ejpam-3176	162	20	estimator	estimator	NOUN
ejpam-3176	162	21	.	.	PUNCT
ejpam-3176	163	1	the	the	DET
ejpam-3176	163	2	initial	initial	ADJ
ejpam-3176	163	3	estimator	estimator	NOUN
ejpam-3176	163	4	obtained	obtain	VERB
ejpam-3176	163	5	by	by	ADP
ejpam-3176	163	6	the	the	DET
ejpam-3176	163	7	generated	generate	VERB
ejpam-3176	163	8	trial	trial	NOUN
ejpam-3176	163	9	fits	fit	NOUN
ejpam-3176	163	10	is	be	AUX
ejpam-3176	163	11	called	call	VERB
ejpam-3176	163	12	start	start	NOUN
ejpam-3176	163	13	.	.	PUNCT
ejpam-3176	164	1	then	then	ADV
ejpam-3176	164	2	,	,	PUNCT
ejpam-3176	164	3	the	the	DET
ejpam-3176	164	4	c	c	NOUN
ejpam-3176	164	5	-	-	PUNCT
ejpam-3176	164	6	step	step	NOUN
ejpam-3176	164	7	technique	technique	NOUN
ejpam-3176	164	8	will	will	AUX
ejpam-3176	164	9	be	be	AUX
ejpam-3176	164	10	applied	apply	VERB
ejpam-3176	164	11	.	.	PUNCT
ejpam-3176	165	1	we	we	PRON
ejpam-3176	165	2	let	let	VERB
ejpam-3176	165	3	(	(	PUNCT
ejpam-3176	165	4	µ0,i	µ0,i	PROPN
ejpam-3176	165	5	,	,	PUNCT
ejpam-3176	165	6	σ0,i	σ0,i	PROPN
ejpam-3176	165	7	)	)	PUNCT
ejpam-3176	165	8	be	be	VERB
ejpam-3176	165	9	the	the	DET
ejpam-3176	165	10	ith	ith	PROPN
ejpam-3176	165	11	start	start	NOUN
ejpam-3176	165	12	and	and	CCONJ
ejpam-3176	165	13	all	all	DET
ejpam-3176	165	14	n	n	PRON
ejpam-3176	165	15	mahalanobis	mahalanobis	ADJ
ejpam-3176	165	16	distances	distance	NOUN
ejpam-3176	165	17	di(µ0,i	di(µ0,i	PROPN
ejpam-3176	165	18	,	,	PUNCT
ejpam-3176	165	19	σ0,i	σ0,i	PROPN
ejpam-3176	165	20	)	)	PUNCT
ejpam-3176	165	21	.	.	PUNCT
ejpam-3176	166	1	the	the	DET
ejpam-3176	166	2	classical	classical	ADJ
ejpam-3176	166	3	estimator	estimator	NOUN
ejpam-3176	166	4	(	(	PUNCT
ejpam-3176	166	5	µ1,i	µ1,i	PROPN
ejpam-3176	166	6	,	,	PUNCT
ejpam-3176	166	7	σ1,i	σ1,i	PROPN
ejpam-3176	166	8	)	)	PUNCT
ejpam-3176	166	9	is	be	AUX
ejpam-3176	166	10	computed	compute	VERB
ejpam-3176	166	11	from	from	ADP
ejpam-3176	166	12	cn	cn	PROPN
ejpam-3176	166	13	≈	≈	PROPN
ejpam-3176	166	14	n/2	n/2	PROPN
ejpam-3176	166	15	cases	case	NOUN
ejpam-3176	166	16	that	that	PRON
ejpam-3176	166	17	correspond	correspond	VERB
ejpam-3176	166	18	to	to	ADP
ejpam-3176	166	19	the	the	DET
ejpam-3176	166	20	smallest	small	ADJ
ejpam-3176	166	21	distance	distance	NOUN
ejpam-3176	166	22	.	.	PUNCT
ejpam-3176	167	1	we	we	PRON
ejpam-3176	167	2	continue	continue	VERB
ejpam-3176	167	3	the	the	DET
ejpam-3176	167	4	iteration	iteration	NOUN
ejpam-3176	167	5	for	for	ADP
ejpam-3176	167	6	k	k	PROPN
ejpam-3176	167	7	steps	step	NOUN
ejpam-3176	167	8	,	,	PUNCT
ejpam-3176	167	9	thereby	thereby	ADV
ejpam-3176	167	10	resulting	result	VERB
ejpam-3176	167	11	in	in	ADP
ejpam-3176	167	12	the	the	DET
ejpam-3176	167	13	following	follow	VERB
ejpam-3176	167	14	sequence	sequence	NOUN
ejpam-3176	167	15	:	:	PUNCT
ejpam-3176	167	16	(	(	PUNCT
ejpam-3176	167	17	µ0,j	µ0,j	INTJ
ejpam-3176	167	18	,	,	PUNCT
ejpam-3176	167	19	σ0,j)(µ1,j	σ0,j)(µ1,j	NOUN
ejpam-3176	167	20	,	,	PUNCT
ejpam-3176	167	21	σ1,j	σ1,j	NOUN
ejpam-3176	167	22	)	)	PUNCT
ejpam-3176	167	23	,	,	PUNCT
ejpam-3176	167	24	(	(	PUNCT
ejpam-3176	167	25	µ2,j	µ2,j	PROPN
ejpam-3176	167	26	,	,	PUNCT
ejpam-3176	167	27	σ2,j	σ2,j	PROPN
ejpam-3176	167	28	)	)	PUNCT
ejpam-3176	167	29	,	,	PUNCT
ejpam-3176	167	30	.	.	PUNCT
ejpam-3176	167	31	.	.	PUNCT
ejpam-3176	168	1	.	.	PUNCT
ejpam-3176	169	1	,	,	PUNCT
ejpam-3176	169	2	(	(	PUNCT
ejpam-3176	169	3	µk	µk	INTJ
ejpam-3176	169	4	,	,	PUNCT
ejpam-3176	169	5	j	j	PROPN
ejpam-3176	169	6	,	,	PUNCT
ejpam-3176	169	7	σn	σn	PROPN
ejpam-3176	169	8	,	,	PUNCT
ejpam-3176	169	9	k	k	NOUN
ejpam-3176	169	10	)	)	PUNCT
ejpam-3176	169	11	.	.	PUNCT
ejpam-3176	170	1	the	the	DET
ejpam-3176	170	2	values	value	NOUN
ejpam-3176	170	3	of	of	ADP
ejpam-3176	170	4	cn	cn	PROPN
ejpam-3176	170	5	and	and	CCONJ
ejpam-3176	170	6	k	k	PROPN
ejpam-3176	170	7	depend	depend	VERB
ejpam-3176	170	8	on	on	ADP
ejpam-3176	170	9	the	the	DET
ejpam-3176	170	10	c	c	NOUN
ejpam-3176	170	11	-	-	PUNCT
ejpam-3176	170	12	step	step	NOUN
ejpam-3176	170	13	estimator	estimator	NOUN
ejpam-3176	170	14	.	.	PUNCT
ejpam-3176	171	1	the	the	DET
ejpam-3176	171	2	value	value	NOUN
ejpam-3176	171	3	of	of	ADP
ejpam-3176	171	4	k	k	PROPN
ejpam-3176	171	5	in	in	ADP
ejpam-3176	171	6	the	the	DET
ejpam-3176	171	7	fastmcd	fastmcd	NOUN
ejpam-3176	171	8	estimator	estimator	NOUN
ejpam-3176	171	9	is	be	AUX
ejpam-3176	171	10	500	500	NUM
ejpam-3176	171	11	,	,	PUNCT
ejpam-3176	171	12	with	with	ADP
ejpam-3176	171	13	randomly	randomly	ADV
ejpam-3176	171	14	drawn	draw	VERB
ejpam-3176	171	15	elemental	elemental	ADJ
ejpam-3176	171	16	sets	set	NOUN
ejpam-3176	171	17	of	of	ADP
ejpam-3176	171	18	p+	p+	NOUN
ejpam-3176	171	19	1	1	NUM
ejpam-3176	171	20	cases	case	NOUN
ejpam-3176	171	21	as	as	ADP
ejpam-3176	171	22	the	the	DET
ejpam-3176	171	23	start	start	NOUN
ejpam-3176	171	24	.	.	PUNCT
ejpam-3176	172	1	the	the	DET
ejpam-3176	172	2	initial	initial	ADJ
ejpam-3176	172	3	estimator	estimator	NOUN
ejpam-3176	172	4	with	with	ADP
ejpam-3176	172	5	the	the	DET
ejpam-3176	172	6	smallest	small	ADJ
ejpam-3176	172	7	determinant	determinant	NOUN
ejpam-3176	172	8	is	be	AUX
ejpam-3176	172	9	used	use	VERB
ejpam-3176	172	10	for	for	ADP
ejpam-3176	172	11	the	the	DET
ejpam-3176	172	12	final	final	ADJ
ejpam-3176	172	13	estimator	estimator	NOUN
ejpam-3176	172	14	.	.	PUNCT
ejpam-3176	173	1	hawkins	hawkins	PROPN
ejpam-3176	173	2	and	and	CCONJ
ejpam-3176	173	3	olive	olive	NOUN
ejpam-3176	174	1	[	[	X
ejpam-3176	174	2	10	10	NUM
ejpam-3176	174	3	]	]	PUNCT
ejpam-3176	174	4	have	have	VERB
ejpam-3176	174	5	a	a	DET
ejpam-3176	174	6	similar	similar	ADJ
ejpam-3176	174	7	estimator	estimator	NOUN
ejpam-3176	174	8	.	.	PUNCT
ejpam-3176	175	1	the	the	DET
ejpam-3176	175	2	fch	fch	PROPN
ejpam-3176	175	3	estimator	estimator	NOUN
ejpam-3176	175	4	uses	use	VERB
ejpam-3176	175	5	two	two	NUM
ejpam-3176	175	6	estimators	estimator	NOUN
ejpam-3176	175	7	.	.	PUNCT
ejpam-3176	176	1	the	the	DET
ejpam-3176	176	2	first	first	ADJ
ejpam-3176	176	3	estimator	estimator	NOUN
ejpam-3176	176	4	is	be	AUX
ejpam-3176	176	5	the	the	DET
ejpam-3176	176	6	dgk	dgk	PROPN
ejpam-3176	176	7	estimator	estimator	NOUN
ejpam-3176	176	8	(	(	PUNCT
ejpam-3176	176	9	devlin	devlin	PROPN
ejpam-3176	176	10	,	,	PUNCT
ejpam-3176	176	11	gnanadesikan	gnanadesikan	NOUN
ejpam-3176	176	12	,	,	PUNCT
ejpam-3176	176	13	and	and	CCONJ
ejpam-3176	176	14	kettenring	kettenre	VERB
ejpam-3176	176	15	[	[	X
ejpam-3176	176	16	8	8	NUM
ejpam-3176	176	17	]	]	NUM
ejpam-3176	176	18	)	)	PUNCT
ejpam-3176	176	19	,	,	PUNCT
ejpam-3176	176	20	which	which	PRON
ejpam-3176	176	21	uses	use	VERB
ejpam-3176	176	22	the	the	DET
ejpam-3176	176	23	classical	classical	ADJ
ejpam-3176	176	24	estimator	estimator	NOUN
ejpam-3176	176	25	as	as	ADP
ejpam-3176	176	26	the	the	DET
ejpam-3176	176	27	start	start	NOUN
ejpam-3176	176	28	.	.	PUNCT
ejpam-3176	177	1	the	the	DET
ejpam-3176	177	2	second	second	ADJ
ejpam-3176	177	3	estimator	estimator	NOUN
ejpam-3176	177	4	is	be	AUX
ejpam-3176	177	5	the	the	DET
ejpam-3176	177	6	median	median	ADJ
ejpam-3176	177	7	ball	ball	NOUN
ejpam-3176	177	8	(	(	PUNCT
ejpam-3176	177	9	mb	mb	NOUN
ejpam-3176	177	10	)	)	PUNCT
ejpam-3176	177	11	estimator	estimator	NOUN
ejpam-3176	177	12	,	,	PUNCT
ejpam-3176	177	13	where	where	SCONJ
ejpam-3176	177	14	the	the	DET
ejpam-3176	177	15	classical	classical	ADJ
ejpam-3176	177	16	estimator	estimator	NOUN
ejpam-3176	177	17	is	be	AUX
ejpam-3176	177	18	computed	compute	VERB
ejpam-3176	177	19	from	from	ADP
ejpam-3176	177	20	cases	case	NOUN
ejpam-3176	177	21	with	with	ADP
ejpam-3176	177	22	di(med(x	di(med(x	NOUN
ejpam-3176	177	23	)	)	PUNCT
ejpam-3176	177	24	,	,	PUNCT
ejpam-3176	177	25	ip	ip	NOUN
ejpam-3176	177	26	)	)	PUNCT
ejpam-3176	177	27	≤med(di(med(x	≤med(di(med(x	PROPN
ejpam-3176	177	28	)	)	PUNCT
ejpam-3176	177	29	,	,	PUNCT
ejpam-3176	177	30	ip	ip	NOUN
ejpam-3176	177	31	)	)	PUNCT
ejpam-3176	177	32	)	)	PUNCT
ejpam-3176	177	33	as	as	ADP
ejpam-3176	177	34	the	the	DET
ejpam-3176	177	35	start	start	NOUN
ejpam-3176	177	36	,	,	PUNCT
ejpam-3176	177	37	and	and	CCONJ
ejpam-3176	177	38	med(x	med(x	NOUN
ejpam-3176	177	39	)	)	PUNCT
ejpam-3176	177	40	is	be	AUX
ejpam-3176	177	41	the	the	DET
ejpam-3176	177	42	coordinate	coordinate	NOUN
ejpam-3176	177	43	-	-	PUNCT
ejpam-3176	177	44	wise	wise	ADJ
ejpam-3176	177	45	median	median	NOUN
ejpam-3176	177	46	.	.	PUNCT
ejpam-3176	178	1	in	in	ADP
ejpam-3176	178	2	case	case	NOUN
ejpam-3176	178	3	the	the	DET
ejpam-3176	178	4	dgk	dgk	PROPN
ejpam-3176	178	5	location	location	NOUN
ejpam-3176	178	6	estimator	estimator	NOUN
ejpam-3176	178	7	obtains	obtain	VERB
ejpam-3176	178	8	a	a	DET
ejpam-3176	178	9	greater	great	ADJ
ejpam-3176	178	10	euclidean	euclidean	ADJ
ejpam-3176	178	11	distance	distance	NOUN
ejpam-3176	178	12	from	from	ADP
ejpam-3176	178	13	med(x	med(x	NOUN
ejpam-3176	178	14	)	)	PUNCT
ejpam-3176	178	15	than	than	ADP
ejpam-3176	178	16	half	half	NOUN
ejpam-3176	178	17	of	of	ADP
ejpam-3176	178	18	the	the	DET
ejpam-3176	178	19	data	datum	NOUN
ejpam-3176	178	20	,	,	PUNCT
ejpam-3176	178	21	then	then	ADV
ejpam-3176	178	22	fch	fch	PROPN
ejpam-3176	178	23	will	will	AUX
ejpam-3176	178	24	apply	apply	VERB
ejpam-3176	178	25	the	the	DET
ejpam-3176	178	26	median	median	ADJ
ejpam-3176	178	27	ball	ball	NOUN
ejpam-3176	178	28	estimator	estimator	NOUN
ejpam-3176	178	29	.	.	PUNCT
ejpam-3176	179	1	we	we	PRON
ejpam-3176	179	2	let	let	VERB
ejpam-3176	179	3	(	(	PUNCT
ejpam-3176	179	4	µ0,σ0	µ0,σ0	VERB
ejpam-3176	179	5	)	)	PUNCT
ejpam-3176	179	6	be	be	VERB
ejpam-3176	179	7	the	the	DET
ejpam-3176	179	8	initial	initial	ADJ
ejpam-3176	179	9	estimator	estimator	NOUN
ejpam-3176	179	10	used	use	VERB
ejpam-3176	179	11	.	.	PUNCT
ejpam-3176	180	1	then	then	ADV
ejpam-3176	180	2	,	,	PUNCT
ejpam-3176	180	3	the	the	DET
ejpam-3176	180	4	estimator	estimator	NOUN
ejpam-3176	180	5	(	(	PUNCT
ejpam-3176	180	6	µ,σ	µ,σ	NOUN
ejpam-3176	180	7	)	)	PUNCT
ejpam-3176	180	8	takes	take	VERB
ejpam-3176	180	9	µ0	µ0	NOUN
ejpam-3176	180	10	=	=	SYM
ejpam-3176	180	11	µ	µ	X
ejpam-3176	180	12	and	and	CCONJ
ejpam-3176	180	13	σ	σ	PROPN
ejpam-3176	180	14	=	=	PUNCT
ejpam-3176	180	15	med(d2	med(d2	PROPN
ejpam-3176	180	16	i	i	NOUN
ejpam-3176	180	17	(	(	PUNCT
ejpam-3176	180	18	µ0,σ0	µ0,σ0	PROPN
ejpam-3176	180	19	)	)	PUNCT
ejpam-3176	180	20	)	)	PUNCT
ejpam-3176	181	1	χ2	χ2	PROPN
ejpam-3176	181	2	p,0.5	p,0.5	PROPN
ejpam-3176	181	3	σ0	σ0	PROPN
ejpam-3176	181	4	,	,	PUNCT
ejpam-3176	181	5	where	where	SCONJ
ejpam-3176	181	6	χ2	χ2	PROPN
ejpam-3176	181	7	p,0.5	p,0.5	PROPN
ejpam-3176	181	8	is	be	AUX
ejpam-3176	181	9	the	the	DET
ejpam-3176	181	10	50th	50th	ADJ
ejpam-3176	181	11	percentile	percentile	NOUN
ejpam-3176	181	12	of	of	ADP
ejpam-3176	181	13	the	the	DET
ejpam-3176	181	14	chi	chi	ADJ
ejpam-3176	181	15	-	-	PUNCT
ejpam-3176	181	16	square	square	ADJ
ejpam-3176	181	17	distribution	distribution	NOUN
ejpam-3176	181	18	with	with	ADP
ejpam-3176	181	19	p	p	NOUN
ejpam-3176	181	20	degrees	degree	NOUN
ejpam-3176	181	21	of	of	ADP
ejpam-3176	181	22	freedom	freedom	NOUN
ejpam-3176	181	23	.	.	PUNCT
ejpam-3176	182	1	the	the	DET
ejpam-3176	182	2	rlda	rlda	NOUN
ejpam-3176	182	3	model	model	NOUN
ejpam-3176	182	4	is	be	AUX
ejpam-3176	182	5	obtained	obtain	VERB
ejpam-3176	182	6	using	use	VERB
ejpam-3176	182	7	the	the	DET
ejpam-3176	182	8	fch	fch	PROPN
ejpam-3176	182	9	estimator	estimator	NOUN
ejpam-3176	182	10	in	in	ADP
ejpam-3176	182	11	the	the	DET
ejpam-3176	182	12	raw	raw	ADJ
ejpam-3176	182	13	and	and	CCONJ
ejpam-3176	182	14	reweighted	reweighted	ADJ
ejpam-3176	182	15	versions	version	NOUN
ejpam-3176	182	16	mufda	mufda	PROPN
ejpam-3176	182	17	j.	j.	PROPN
ejpam-3176	182	18	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	182	19	,	,	PUNCT
ejpam-3176	182	20	taha	taha	PROPN
ejpam-3176	182	21	radwan	radwan	PROPN
ejpam-3176	182	22	and	and	CCONJ
ejpam-3176	182	23	khalid	khalid	PROPN
ejpam-3176	182	24	abunawas	abunawas	PROPN
ejpam-3176	182	25	/	/	SYM
ejpam-3176	182	26	eur	eur	PROPN
ejpam-3176	182	27	.	.	PUNCT
ejpam-3176	183	1	j.	j.	PROPN
ejpam-3176	183	2	pure	pure	PROPN
ejpam-3176	183	3	appl	appl	PROPN
ejpam-3176	183	4	.	.	PROPN
ejpam-3176	183	5	math	math	PROPN
ejpam-3176	183	6	,	,	PUNCT
ejpam-3176	183	7	11	11	NUM
ejpam-3176	183	8	(	(	PUNCT
ejpam-3176	183	9	1	1	NUM
ejpam-3176	183	10	)	)	PUNCT
ejpam-3176	183	11	(	(	PUNCT
ejpam-3176	183	12	2018	2018	NUM
ejpam-3176	183	13	)	)	PUNCT
ejpam-3176	183	14	,	,	PUNCT
ejpam-3176	183	15	284	284	NUM
ejpam-3176	183	16	-	-	SYM
ejpam-3176	183	17	298	298	NUM
ejpam-3176	183	18	290	290	NUM
ejpam-3176	183	19	and	and	CCONJ
ejpam-3176	183	20	is	be	AUX
ejpam-3176	183	21	expressed	express	VERB
ejpam-3176	183	22	as	as	SCONJ
ejpam-3176	183	23	follows	follow	VERB
ejpam-3176	183	24	:	:	PUNCT
ejpam-3176	183	25	_	_	PUNCT
ejpam-3176	184	1	d	d	X
ejpam-3176	184	2	fch	fch	PROPN
ejpam-3176	184	3	j	j	PROPN
ejpam-3176	184	4	(	(	PUNCT
ejpam-3176	184	5	x	x	X
ejpam-3176	184	6	)	)	PUNCT
ejpam-3176	184	7	=	=	PRON
ejpam-3176	184	8	µtjς	µtjς	VERB
ejpam-3176	184	9	−1x−	−1x−	ADJ
ejpam-3176	184	10	1	1	NUM
ejpam-3176	184	11	2	2	NUM
ejpam-3176	184	12	µtjς	µtjς	NOUN
ejpam-3176	184	13	−1µj	−1µj	X
ejpam-3176	184	14	+	+	X
ejpam-3176	184	15	ln(pj	ln(pj	PROPN
ejpam-3176	184	16	)	)	PUNCT
ejpam-3176	184	17	.	.	PUNCT
ejpam-3176	185	1	(	(	PUNCT
ejpam-3176	185	2	11	11	NUM
ejpam-3176	185	3	)	)	PUNCT
ejpam-3176	185	4	the	the	DET
ejpam-3176	185	5	membership	membership	NOUN
ejpam-3176	185	6	probability	probability	NOUN
ejpam-3176	185	7	prj	prj	NOUN
ejpam-3176	185	8	of	of	ADP
ejpam-3176	185	9	the	the	DET
ejpam-3176	185	10	robust	robust	ADJ
ejpam-3176	185	11	distance	distance	NOUN
ejpam-3176	185	12	is	be	AUX
ejpam-3176	185	13	defined	define	VERB
ejpam-3176	185	14	as	as	SCONJ
ejpam-3176	185	15	follows	follow	VERB
ejpam-3176	185	16	:	:	PUNCT
ejpam-3176	186	1	rdfch	rdfch	NOUN
ejpam-3176	186	2	ij	ij	NOUN
ejpam-3176	186	3	=	=	NOUN
ejpam-3176	186	4	√	√	PROPN
ejpam-3176	186	5	(	(	PUNCT
ejpam-3176	186	6	xij	xij	NOUN
ejpam-3176	186	7	−	−	PROPN
ejpam-3176	186	8	µ̂j)tς−1(xij	µ̂j)tς−1(xij	ADV
ejpam-3176	186	9	−	−	PROPN
ejpam-3176	186	10	µ̂j	µ̂j	NOUN
ejpam-3176	186	11	)	)	PUNCT
ejpam-3176	186	12	.	.	PUNCT
ejpam-3176	187	1	(	(	PUNCT
ejpam-3176	187	2	12	12	NUM
ejpam-3176	187	3	)	)	PUNCT
ejpam-3176	187	4	if	if	SCONJ
ejpam-3176	187	5	xij	xij	PRON
ejpam-3176	187	6	is	be	AUX
ejpam-3176	187	7	used	use	VERB
ejpam-3176	187	8	to	to	PART
ejpam-3176	187	9	consider	consider	VERB
ejpam-3176	187	10	the	the	DET
ejpam-3176	187	11	outliers	outlier	NOUN
ejpam-3176	187	12	in	in	ADP
ejpam-3176	187	13	formula	formula	NOUN
ejpam-3176	187	14	(	(	PUNCT
ejpam-3176	187	15	6	6	NUM
ejpam-3176	187	16	)	)	PUNCT
ejpam-3176	187	17	,	,	PUNCT
ejpam-3176	187	18	then	then	ADV
ejpam-3176	187	19	the	the	DET
ejpam-3176	187	20	membership	membership	NOUN
ejpam-3176	187	21	probability	probability	NOUN
ejpam-3176	187	22	of	of	ADP
ejpam-3176	187	23	the	the	DET
ejpam-3176	187	24	fch	fch	PROPN
ejpam-3176	187	25	estimator	estimator	NOUN
ejpam-3176	187	26	is	be	AUX
ejpam-3176	187	27	expressed	express	VERB
ejpam-3176	187	28	as	as	SCONJ
ejpam-3176	187	29	follows	follow	VERB
ejpam-3176	187	30	:	:	PUNCT
ejpam-3176	188	1	_	_	PUNCT
ejpam-3176	189	1	p	p	X
ejpam-3176	189	2	fch	fch	PROPN
ejpam-3176	189	3	j	j	PROPN
ejpam-3176	189	4	=	=	PROPN
ejpam-3176	189	5	ñj	ñj	PROPN
ejpam-3176	189	6	ñ	ñ	PROPN
ejpam-3176	189	7	.	.	PUNCT
ejpam-3176	190	1	(	(	PUNCT
ejpam-3176	190	2	13	13	NUM
ejpam-3176	190	3	)	)	PUNCT
ejpam-3176	190	4	7	7	NUM
ejpam-3176	190	5	.	.	PUNCT
ejpam-3176	190	6	simulation	simulation	NOUN
ejpam-3176	190	7	study	study	NOUN
ejpam-3176	190	8	in	in	ADP
ejpam-3176	190	9	this	this	DET
ejpam-3176	190	10	section	section	NOUN
ejpam-3176	190	11	,	,	PUNCT
ejpam-3176	190	12	different	different	ADJ
ejpam-3176	190	13	algorithms	algorithm	NOUN
ejpam-3176	190	14	are	be	AUX
ejpam-3176	190	15	applied	apply	VERB
ejpam-3176	190	16	to	to	PART
ejpam-3176	190	17	estimate	estimate	VERB
ejpam-3176	190	18	the	the	DET
ejpam-3176	190	19	lda	lda	PROPN
ejpam-3176	190	20	parameters	parameter	NOUN
ejpam-3176	190	21	using	use	VERB
ejpam-3176	190	22	small	small	ADJ
ejpam-3176	190	23	and	and	CCONJ
ejpam-3176	190	24	medium	medium	ADJ
ejpam-3176	190	25	datasets	dataset	NOUN
ejpam-3176	190	26	.	.	PUNCT
ejpam-3176	191	1	the	the	DET
ejpam-3176	191	2	simulation	simulation	NOUN
ejpam-3176	191	3	is	be	AUX
ejpam-3176	191	4	similar	similar	ADJ
ejpam-3176	191	5	to	to	ADP
ejpam-3176	191	6	that	that	PRON
ejpam-3176	191	7	of	of	ADP
ejpam-3176	191	8	he	he	PRON
ejpam-3176	191	9	and	and	CCONJ
ejpam-3176	191	10	fung	fung	PROPN
ejpam-3176	192	1	[	[	X
ejpam-3176	192	2	12	12	NUM
ejpam-3176	192	3	]	]	PUNCT
ejpam-3176	192	4	.	.	PUNCT
ejpam-3176	193	1	all	all	DET
ejpam-3176	193	2	the	the	DET
ejpam-3176	193	3	estimators	estimator	NOUN
ejpam-3176	193	4	are	be	AUX
ejpam-3176	193	5	used	use	VERB
ejpam-3176	193	6	in	in	ADP
ejpam-3176	193	7	the	the	DET
ejpam-3176	193	8	raw	raw	ADJ
ejpam-3176	193	9	and	and	CCONJ
ejpam-3176	193	10	reweighted	reweighte	VERB
ejpam-3176	193	11	versions	version	NOUN
ejpam-3176	193	12	to	to	PART
ejpam-3176	193	13	obtain	obtain	VERB
ejpam-3176	193	14	the	the	DET
ejpam-3176	193	15	initial	initial	ADJ
ejpam-3176	193	16	mean	mean	NOUN
ejpam-3176	193	17	and	and	CCONJ
ejpam-3176	193	18	covariance	covariance	NOUN
ejpam-3176	193	19	matrix	matrix	NOUN
ejpam-3176	193	20	,	,	PUNCT
ejpam-3176	193	21	that	that	ADV
ejpam-3176	193	22	is	is	ADV
ejpam-3176	193	23	,	,	PUNCT
ejpam-3176	193	24	µ0	µ0	NOUN
ejpam-3176	193	25	and	and	CCONJ
ejpam-3176	193	26	σ0	σ0	NOUN
ejpam-3176	193	27	,	,	PUNCT
ejpam-3176	193	28	respectively	respectively	ADV
ejpam-3176	193	29	.	.	PUNCT
ejpam-3176	194	1	this	this	DET
ejpam-3176	194	2	estimator	estimator	NOUN
ejpam-3176	194	3	will	will	AUX
ejpam-3176	194	4	yield	yield	VERB
ejpam-3176	194	5	a	a	DET
ejpam-3176	194	6	discriminate	discriminate	ADJ
ejpam-3176	194	7	rule	rule	NOUN
ejpam-3176	194	8	based	base	VERB
ejpam-3176	194	9	on	on	ADP
ejpam-3176	194	10	robust	robust	ADJ
ejpam-3176	194	11	drlj	drlj	ADJ
ejpam-3176	194	12	(	(	PUNCT
ejpam-3176	194	13	x	x	NOUN
ejpam-3176	194	14	,	,	PUNCT
ejpam-3176	194	15	µ0,σ0	µ0,σ0	PROPN
ejpam-3176	194	16	)	)	PUNCT
ejpam-3176	194	17	.	.	PUNCT
ejpam-3176	195	1	then	then	ADV
ejpam-3176	195	2	,	,	PUNCT
ejpam-3176	195	3	the	the	DET
ejpam-3176	195	4	reweighted	reweighte	VERB
ejpam-3176	195	5	version	version	NOUN
ejpam-3176	195	6	will	will	AUX
ejpam-3176	195	7	be	be	AUX
ejpam-3176	195	8	obtained	obtain	VERB
ejpam-3176	195	9	based	base	VERB
ejpam-3176	195	10	on	on	ADP
ejpam-3176	195	11	the	the	DET
ejpam-3176	195	12	robust	robust	ADJ
ejpam-3176	195	13	distances	distance	NOUN
ejpam-3176	195	14	(	(	PUNCT
ejpam-3176	195	15	rousseeuw	rousseeuw	NOUN
ejpam-3176	195	16	and	and	CCONJ
ejpam-3176	195	17	van	van	PROPN
ejpam-3176	195	18	zomeren	zomeren	PROPN
ejpam-3176	196	1	[	[	X
ejpam-3176	196	2	26	26	NUM
ejpam-3176	196	3	]	]	PUNCT
ejpam-3176	196	4	)	)	PUNCT
ejpam-3176	196	5	,	,	PUNCT
ejpam-3176	196	6	as	as	SCONJ
ejpam-3176	196	7	follows	follow	VERB
ejpam-3176	196	8	:	:	PUNCT
ejpam-3176	196	9	rdij	rdij	VERB
ejpam-3176	196	10	=	=	PUNCT
ejpam-3176	196	11	√	√	PROPN
ejpam-3176	196	12	(	(	PUNCT
ejpam-3176	196	13	xij	xij	NOUN
ejpam-3176	196	14	−	−	PROPN
ejpam-3176	196	15	µ̂j,0)tς−1	µ̂j,0)tς−1	NOUN
ejpam-3176	196	16	0	0	PUNCT
ejpam-3176	197	1	(	(	PUNCT
ejpam-3176	197	2	xij	xij	NOUN
ejpam-3176	197	3	−	−	PROPN
ejpam-3176	197	4	µ̂j,0	µ̂j,0	ADJ
ejpam-3176	197	5	)	)	PUNCT
ejpam-3176	197	6	.	.	PUNCT
ejpam-3176	198	1	(	(	PUNCT
ejpam-3176	198	2	14	14	NUM
ejpam-3176	198	3	)	)	PUNCT
ejpam-3176	198	4	for	for	ADP
ejpam-3176	198	5	each	each	DET
ejpam-3176	198	6	observation	observation	NOUN
ejpam-3176	198	7	in	in	ADP
ejpam-3176	198	8	group	group	PROPN
ejpam-3176	198	9	j	j	PROPN
ejpam-3176	198	10	,	,	PUNCT
ejpam-3176	198	11	wij	wij	PROPN
ejpam-3176	198	12	=	=	SYM
ejpam-3176	198	13	{	{	PUNCT
ejpam-3176	198	14	1	1	NUM
ejpam-3176	198	15	if	if	SCONJ
ejpam-3176	198	16	rdij	rdij	VERB
ejpam-3176	198	17	≤	≤	NUM
ejpam-3176	198	18	√	√	NUM
ejpam-3176	198	19	χ2	χ2	PROPN
ejpam-3176	198	20	p,0.975	p,0.975	PROPN
ejpam-3176	198	21	0	0	PUNCT
ejpam-3176	199	1	otherwise	otherwise	ADV
ejpam-3176	199	2	.	.	PUNCT
ejpam-3176	200	1	(	(	PUNCT
ejpam-3176	200	2	15	15	NUM
ejpam-3176	200	3	)	)	PUNCT
ejpam-3176	200	4	three	three	NUM
ejpam-3176	200	5	approaches	approach	NOUN
ejpam-3176	200	6	presented	present	VERB
ejpam-3176	200	7	by	by	ADP
ejpam-3176	200	8	hubert	hubert	PROPN
ejpam-3176	200	9	and	and	CCONJ
ejpam-3176	200	10	van	van	PROPN
ejpam-3176	200	11	driessen	driessen	PROPN
ejpam-3176	200	12	[	[	X
ejpam-3176	200	13	14	14	NUM
ejpam-3176	200	14	]	]	PUNCT
ejpam-3176	200	15	are	be	AUX
ejpam-3176	200	16	adopted	adopt	VERB
ejpam-3176	200	17	to	to	PART
ejpam-3176	200	18	estimate	estimate	VERB
ejpam-3176	200	19	the	the	DET
ejpam-3176	200	20	means	mean	NOUN
ejpam-3176	200	21	and	and	CCONJ
ejpam-3176	200	22	common	common	ADJ
ejpam-3176	200	23	covariance	covariance	NOUN
ejpam-3176	200	24	matrices	matrix	NOUN
ejpam-3176	200	25	for	for	ADP
ejpam-3176	200	26	all	all	DET
ejpam-3176	200	27	the	the	DET
ejpam-3176	200	28	groups	group	NOUN
ejpam-3176	200	29	with	with	ADP
ejpam-3176	200	30	raw	raw	ADJ
ejpam-3176	200	31	and	and	CCONJ
ejpam-3176	200	32	reweighted	reweighted	ADJ
ejpam-3176	200	33	versions	version	NOUN
ejpam-3176	200	34	.	.	PUNCT
ejpam-3176	201	1	the	the	DET
ejpam-3176	201	2	same	same	ADJ
ejpam-3176	201	3	approaches	approach	NOUN
ejpam-3176	201	4	have	have	AUX
ejpam-3176	201	5	been	be	AUX
ejpam-3176	201	6	used	use	VERB
ejpam-3176	201	7	to	to	PART
ejpam-3176	201	8	compare	compare	VERB
ejpam-3176	201	9	the	the	DET
ejpam-3176	201	10	robust	robust	ADJ
ejpam-3176	201	11	and	and	CCONJ
ejpam-3176	201	12	classical	classical	ADJ
ejpam-3176	201	13	lda	lda	NOUN
ejpam-3176	201	14	(	(	PUNCT
ejpam-3176	201	15	alrawashdeh	alrawashdeh	ADV
ejpam-3176	201	16	et	et	PROPN
ejpam-3176	201	17	al	al	PROPN
ejpam-3176	201	18	.	.	PUNCT
ejpam-3176	202	1	[	[	X
ejpam-3176	202	2	1	1	NUM
ejpam-3176	202	3	]	]	PUNCT
ejpam-3176	202	4	)	)	PUNCT
ejpam-3176	202	5	.	.	PUNCT
ejpam-3176	203	1	these	these	DET
ejpam-3176	203	2	approaches	approach	NOUN
ejpam-3176	203	3	are	be	AUX
ejpam-3176	203	4	also	also	ADV
ejpam-3176	203	5	applied	apply	VERB
ejpam-3176	203	6	to	to	ADP
ejpam-3176	203	7	the	the	DET
ejpam-3176	203	8	estimators	estimator	NOUN
ejpam-3176	203	9	of	of	ADP
ejpam-3176	203	10	the	the	DET
ejpam-3176	203	11	fastmcd	fastmcd	NOUN
ejpam-3176	203	12	,	,	PUNCT
ejpam-3176	203	13	detmcd	detmcd	NOUN
ejpam-3176	203	14	,	,	PUNCT
ejpam-3176	203	15	and	and	CCONJ
ejpam-3176	203	16	fch	fch	PROPN
ejpam-3176	203	17	algorithms	algorithm	NOUN
ejpam-3176	203	18	.	.	PUNCT
ejpam-3176	204	1	the	the	DET
ejpam-3176	204	2	first	first	ADJ
ejpam-3176	204	3	approach	approach	NOUN
ejpam-3176	204	4	is	be	AUX
ejpam-3176	204	5	direct	direct	ADJ
ejpam-3176	204	6	and	and	CCONJ
ejpam-3176	204	7	has	have	AUX
ejpam-3176	204	8	been	be	AUX
ejpam-3176	204	9	applied	apply	VERB
ejpam-3176	204	10	by	by	ADP
ejpam-3176	204	11	chork	chork	NOUN
ejpam-3176	204	12	and	and	CCONJ
ejpam-3176	204	13	rousseeuw	rousseeuw	NOUN
ejpam-3176	205	1	[	[	X
ejpam-3176	205	2	3	3	NUM
ejpam-3176	205	3	]	]	PUNCT
ejpam-3176	205	4	,	,	PUNCT
ejpam-3176	205	5	where	where	SCONJ
ejpam-3176	205	6	µj	µj	PROPN
ejpam-3176	205	7	and	and	CCONJ
ejpam-3176	205	8	σj	σj	NOUN
ejpam-3176	205	9	are	be	AUX
ejpam-3176	205	10	obtained	obtain	VERB
ejpam-3176	205	11	by	by	ADP
ejpam-3176	205	12	pooling	pool	VERB
ejpam-3176	205	13	the	the	DET
ejpam-3176	205	14	covariance	covariance	NOUN
ejpam-3176	205	15	matrix	matrix	NOUN
ejpam-3176	205	16	σj	σj	NOUN
ejpam-3176	205	17	,	,	PUNCT
ejpam-3176	205	18	algorithm	algorithm	NOUN
ejpam-3176	205	19	as	as	SCONJ
ejpam-3176	205	20	follows	follow	VERB
ejpam-3176	205	21	:	:	PUNCT
ejpam-3176	206	1	_	_	DET
ejpam-3176	206	2	σpcov	σpcov	NOUN
ejpam-3176	206	3	=	=	PUNCT
ejpam-3176	206	4	σl	σl	X
ejpam-3176	206	5	j=1nj	j=1nj	X
ejpam-3176	206	6	_	_	PUNCT
ejpam-3176	207	1	σjalgorithm∑l	σjalgorithm∑l	INTJ
ejpam-3176	207	2	j=1	j=1	PROPN
ejpam-3176	207	3	nj	nj	PROPN
ejpam-3176	207	4	.	.	PUNCT
ejpam-3176	208	1	(	(	PUNCT
ejpam-3176	208	2	16	16	X
ejpam-3176	208	3	)	)	PUNCT
ejpam-3176	208	4	mufda	mufda	PROPN
ejpam-3176	208	5	j.	j.	PROPN
ejpam-3176	208	6	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	208	7	,	,	PUNCT
ejpam-3176	208	8	taha	taha	PROPN
ejpam-3176	208	9	radwan	radwan	PROPN
ejpam-3176	208	10	and	and	CCONJ
ejpam-3176	208	11	khalid	khalid	PROPN
ejpam-3176	208	12	abunawas	abunawas	PROPN
ejpam-3176	208	13	/	/	SYM
ejpam-3176	208	14	eur	eur	PROPN
ejpam-3176	208	15	.	.	PUNCT
ejpam-3176	209	1	j.	j.	PROPN
ejpam-3176	209	2	pure	pure	PROPN
ejpam-3176	209	3	appl	appl	PROPN
ejpam-3176	209	4	.	.	PROPN
ejpam-3176	209	5	math	math	PROPN
ejpam-3176	209	6	,	,	PUNCT
ejpam-3176	209	7	11	11	NUM
ejpam-3176	209	8	(	(	PUNCT
ejpam-3176	209	9	1	1	NUM
ejpam-3176	209	10	)	)	PUNCT
ejpam-3176	209	11	(	(	PUNCT
ejpam-3176	209	12	2018	2018	NUM
ejpam-3176	209	13	)	)	PUNCT
ejpam-3176	209	14	,	,	PUNCT
ejpam-3176	209	15	284	284	NUM
ejpam-3176	209	16	-	-	SYM
ejpam-3176	209	17	298	298	NUM
ejpam-3176	209	18	291	291	NUM
ejpam-3176	209	19	this	this	DET
ejpam-3176	209	20	approach	approach	NOUN
ejpam-3176	209	21	will	will	AUX
ejpam-3176	209	22	be	be	AUX
ejpam-3176	209	23	denoted	denote	VERB
ejpam-3176	209	24	by	by	ADP
ejpam-3176	209	25	pcov	pcov	NOUN
ejpam-3176	209	26	for	for	ADP
ejpam-3176	209	27	the	the	DET
ejpam-3176	209	28	raw	raw	ADJ
ejpam-3176	209	29	version	version	NOUN
ejpam-3176	209	30	and	and	CCONJ
ejpam-3176	209	31	pcov	pcov	NOUN
ejpam-3176	209	32	-	-	PUNCT
ejpam-3176	209	33	w	w	NOUN
ejpam-3176	209	34	for	for	ADP
ejpam-3176	209	35	the	the	DET
ejpam-3176	209	36	reweighted	reweighted	ADJ
ejpam-3176	209	37	version	version	NOUN
ejpam-3176	209	38	.	.	PUNCT
ejpam-3176	210	1	for	for	ADP
ejpam-3176	210	2	the	the	DET
ejpam-3176	210	3	second	second	ADJ
ejpam-3176	210	4	approach	approach	NOUN
ejpam-3176	210	5	,	,	PUNCT
ejpam-3176	210	6	the	the	DET
ejpam-3176	210	7	concept	concept	NOUN
ejpam-3176	210	8	is	be	AUX
ejpam-3176	210	9	based	base	VERB
ejpam-3176	210	10	on	on	ADP
ejpam-3176	210	11	pooling	pool	VERB
ejpam-3176	210	12	the	the	DET
ejpam-3176	210	13	observations	observation	NOUN
ejpam-3176	210	14	instead	instead	ADV
ejpam-3176	210	15	of	of	ADP
ejpam-3176	210	16	the	the	DET
ejpam-3176	210	17	group	group	NOUN
ejpam-3176	210	18	covariance	covariance	NOUN
ejpam-3176	210	19	matrices	matrix	NOUN
ejpam-3176	210	20	.	.	PUNCT
ejpam-3176	211	1	this	this	DET
ejpam-3176	211	2	approach	approach	NOUN
ejpam-3176	211	3	was	be	AUX
ejpam-3176	211	4	proposed	propose	VERB
ejpam-3176	211	5	by	by	ADP
ejpam-3176	211	6	he	he	PRON
ejpam-3176	211	7	and	and	CCONJ
ejpam-3176	211	8	fung	fung	PROPN
ejpam-3176	212	1	[	[	X
ejpam-3176	212	2	12	12	NUM
ejpam-3176	212	3	]	]	X
ejpam-3176	212	4	,	,	PUNCT
ejpam-3176	212	5	who	who	PRON
ejpam-3176	212	6	used	use	VERB
ejpam-3176	212	7	an	an	DET
ejpam-3176	212	8	s	s	NOUN
ejpam-3176	212	9	-	-	NOUN
ejpam-3176	212	10	estimator	estimator	NOUN
ejpam-3176	212	11	,	,	PUNCT
ejpam-3176	212	12	and	and	CCONJ
ejpam-3176	212	13	adopted	adopt	VERB
ejpam-3176	212	14	by	by	ADP
ejpam-3176	212	15	hubert	hubert	PROPN
ejpam-3176	212	16	and	and	CCONJ
ejpam-3176	212	17	van	van	PROPN
ejpam-3176	212	18	driessen	driessen	PROPN
ejpam-3176	213	1	[	[	X
ejpam-3176	213	2	14	14	NUM
ejpam-3176	213	3	]	]	PUNCT
ejpam-3176	213	4	.	.	PUNCT
ejpam-3176	214	1	the	the	DET
ejpam-3176	214	2	number	number	NOUN
ejpam-3176	214	3	of	of	ADP
ejpam-3176	214	4	groups	group	NOUN
ejpam-3176	214	5	in	in	ADP
ejpam-3176	214	6	the	the	DET
ejpam-3176	214	7	simulation	simulation	NOUN
ejpam-3176	214	8	and	and	CCONJ
ejpam-3176	214	9	that	that	SCONJ
ejpam-3176	214	10	for	for	ADP
ejpam-3176	214	11	other	other	ADJ
ejpam-3176	214	12	groups	group	NOUN
ejpam-3176	214	13	will	will	AUX
ejpam-3176	214	14	follow	follow	VERB
ejpam-3176	214	15	the	the	DET
ejpam-3176	214	16	same	same	ADJ
ejpam-3176	214	17	pattern	pattern	NOUN
ejpam-3176	214	18	to	to	PART
ejpam-3176	214	19	simplify	simplify	VERB
ejpam-3176	214	20	the	the	DET
ejpam-3176	214	21	notation	notation	NOUN
ejpam-3176	214	22	of	of	ADP
ejpam-3176	214	23	the	the	DET
ejpam-3176	214	24	three	three	NUM
ejpam-3176	214	25	groups	group	NOUN
ejpam-3176	214	26	.	.	PUNCT
ejpam-3176	215	1	in	in	ADP
ejpam-3176	215	2	the	the	DET
ejpam-3176	215	3	three	three	NUM
ejpam-3176	215	4	samples	sample	NOUN
ejpam-3176	215	5	a	a	DET
ejpam-3176	215	6	=	=	SYM
ejpam-3176	215	7	(	(	PUNCT
ejpam-3176	215	8	a11	a11	PROPN
ejpam-3176	215	9	,	,	PUNCT
ejpam-3176	215	10	a21	a21	PROPN
ejpam-3176	215	11	,	,	PUNCT
ejpam-3176	215	12	.	.	PUNCT
ejpam-3176	215	13	.	.	PUNCT
ejpam-3176	215	14	.	.	PUNCT
ejpam-3176	216	1	,	,	PUNCT
ejpam-3176	216	2	an1,1	an1,1	NOUN
ejpam-3176	216	3	)	)	PUNCT
ejpam-3176	216	4	,	,	PUNCT
ejpam-3176	216	5	b	b	X
ejpam-3176	216	6	=	=	SYM
ejpam-3176	216	7	(	(	PUNCT
ejpam-3176	216	8	b12	b12	NOUN
ejpam-3176	216	9	,	,	PUNCT
ejpam-3176	216	10	b22	b22	PROPN
ejpam-3176	216	11	,	,	PUNCT
ejpam-3176	216	12	.	.	PUNCT
ejpam-3176	216	13	.	.	PUNCT
ejpam-3176	217	1	.	.	PUNCT
ejpam-3176	218	1	,	,	PUNCT
ejpam-3176	218	2	bn2,2	bn2,2	NOUN
ejpam-3176	218	3	)	)	PUNCT
ejpam-3176	218	4	,	,	PUNCT
ejpam-3176	218	5	and	and	CCONJ
ejpam-3176	218	6	c	c	X
ejpam-3176	218	7	=	=	SYM
ejpam-3176	218	8	(	(	PUNCT
ejpam-3176	218	9	c13	c13	PROPN
ejpam-3176	218	10	,	,	PUNCT
ejpam-3176	218	11	c23	c23	PROPN
ejpam-3176	218	12	,	,	PUNCT
ejpam-3176	218	13	.	.	PUNCT
ejpam-3176	218	14	.	.	PUNCT
ejpam-3176	218	15	.	.	PUNCT
ejpam-3176	219	1	,	,	PUNCT
ejpam-3176	219	2	cn3,3	cn3,3	NOUN
ejpam-3176	219	3	)	)	PUNCT
ejpam-3176	219	4	,	,	PUNCT
ejpam-3176	219	5	µa	µa	PROPN
ejpam-3176	219	6	,	,	PUNCT
ejpam-3176	219	7	µb	µb	VERB
ejpam-3176	219	8	,	,	PUNCT
ejpam-3176	219	9	and	and	CCONJ
ejpam-3176	219	10	µc	µc	PRON
ejpam-3176	219	11	are	be	AUX
ejpam-3176	219	12	the	the	DET
ejpam-3176	219	13	location	location	NOUN
ejpam-3176	219	14	estimators	estimator	NOUN
ejpam-3176	219	15	of	of	ADP
ejpam-3176	219	16	the	the	DET
ejpam-3176	219	17	populations	population	NOUN
ejpam-3176	219	18	in	in	ADP
ejpam-3176	219	19	the	the	DET
ejpam-3176	219	20	reweighted	reweighted	ADJ
ejpam-3176	219	21	fastmcd	fastmcd	NOUN
ejpam-3176	219	22	.	.	PUNCT
ejpam-3176	220	1	the	the	DET
ejpam-3176	220	2	pooled	pooled	ADJ
ejpam-3176	220	3	and	and	CCONJ
ejpam-3176	220	4	shifted	shift	VERB
ejpam-3176	220	5	observations	observation	NOUN
ejpam-3176	220	6	are	be	AUX
ejpam-3176	220	7	expressed	express	VERB
ejpam-3176	220	8	as	as	SCONJ
ejpam-3176	220	9	follows	follow	VERB
ejpam-3176	220	10	:	:	PUNCT
ejpam-3176	220	11	z	z	X
ejpam-3176	220	12	=	=	SYM
ejpam-3176	220	13	(	(	PUNCT
ejpam-3176	220	14	z1	z1	PROPN
ejpam-3176	220	15	,	,	PUNCT
ejpam-3176	220	16	z2	z2	PROPN
ejpam-3176	220	17	,	,	PUNCT
ejpam-3176	220	18	.	.	PUNCT
ejpam-3176	220	19	.	.	PUNCT
ejpam-3176	221	1	.	.	PUNCT
ejpam-3176	222	1	,	,	PUNCT
ejpam-3176	222	2	zn	zn	X
ejpam-3176	222	3	)	)	PUNCT
ejpam-3176	222	4	=	=	SYM
ejpam-3176	222	5	(	(	PUNCT
ejpam-3176	222	6	a11−µa	a11−µa	NOUN
ejpam-3176	222	7	,	,	PUNCT
ejpam-3176	222	8	a21−µa	a21−µa	ADP
ejpam-3176	222	9	,	,	PUNCT
ejpam-3176	222	10	.	.	PUNCT
ejpam-3176	222	11	.	.	PUNCT
ejpam-3176	222	12	.	.	PUNCT
ejpam-3176	223	1	,	,	PUNCT
ejpam-3176	223	2	an1,1−µa	an1,1−µa	INTJ
ejpam-3176	223	3	,	,	PUNCT
ejpam-3176	223	4	b12−µb	b12−µb	PROPN
ejpam-3176	223	5	,	,	PUNCT
ejpam-3176	223	6	b22−µb	b22−µb	NOUN
ejpam-3176	223	7	,	,	PUNCT
ejpam-3176	223	8	.	.	PUNCT
ejpam-3176	223	9	.	.	PUNCT
ejpam-3176	223	10	.	.	PUNCT
ejpam-3176	224	1	,	,	PUNCT
ejpam-3176	224	2	bn2,2−µb	bn2,2−µb	NOUN
ejpam-3176	224	3	,	,	PUNCT
ejpam-3176	224	4	c13	c13	PROPN
ejpam-3176	224	5	−	−	PROPN
ejpam-3176	224	6	µc	µc	PROPN
ejpam-3176	224	7	,	,	PUNCT
ejpam-3176	224	8	c23	c23	PROPN
ejpam-3176	224	9	−	−	PROPN
ejpam-3176	224	10	µc	µc	INTJ
ejpam-3176	224	11	,	,	PUNCT
ejpam-3176	224	12	.	.	PUNCT
ejpam-3176	224	13	.	.	PUNCT
ejpam-3176	224	14	.	.	PUNCT
ejpam-3176	225	1	,	,	PUNCT
ejpam-3176	225	2	cn3,3	cn3,3	NOUN
ejpam-3176	225	3	−	−	NOUN
ejpam-3176	225	4	µc	µc	PROPN
ejpam-3176	225	5	)	)	PUNCT
ejpam-3176	225	6	.	.	PUNCT
ejpam-3176	226	1	the	the	DET
ejpam-3176	226	2	covariance	covariance	NOUN
ejpam-3176	226	3	matrix	matrix	NOUN
ejpam-3176	226	4	σz	σz	X
ejpam-3176	226	5	is	be	AUX
ejpam-3176	226	6	estimated	estimate	VERB
ejpam-3176	226	7	as	as	SCONJ
ejpam-3176	226	8	the	the	DET
ejpam-3176	226	9	reweighted	reweighted	ADJ
ejpam-3176	226	10	fastmcd	fastmcd	NOUN
ejpam-3176	226	11	of	of	ADP
ejpam-3176	226	12	the	the	DET
ejpam-3176	226	13	scatter	scatter	NOUN
ejpam-3176	226	14	matrix	matrix	NOUN
ejpam-3176	226	15	of	of	ADP
ejpam-3176	226	16	z.	z.	PROPN
ejpam-3176	226	17	the	the	DET
ejpam-3176	226	18	location	location	NOUN
ejpam-3176	226	19	µz	µz	PROPN
ejpam-3176	226	20	is	be	AUX
ejpam-3176	226	21	estimated	estimate	VERB
ejpam-3176	226	22	by	by	ADP
ejpam-3176	226	23	the	the	DET
ejpam-3176	226	24	mcd	mcd	PROPN
ejpam-3176	226	25	estimator	estimator	NOUN
ejpam-3176	226	26	.	.	PUNCT
ejpam-3176	227	1	µz	µz	PROPN
ejpam-3176	227	2	is	be	AUX
ejpam-3176	227	3	used	use	VERB
ejpam-3176	227	4	to	to	PART
ejpam-3176	227	5	upgrade	upgrade	VERB
ejpam-3176	227	6	the	the	DET
ejpam-3176	227	7	locations	location	NOUN
ejpam-3176	227	8	of	of	ADP
ejpam-3176	227	9	µj	µj	PROPN
ejpam-3176	227	10	to	to	PART
ejpam-3176	227	11	obtain	obtain	VERB
ejpam-3176	227	12	_	_	PRON
ejpam-3176	227	13	µa=	µa=	NOUN
ejpam-3176	227	14	µa	µa	PROPN
ejpam-3176	227	15	+	+	PROPN
ejpam-3176	227	16	µz	µz	PROPN
ejpam-3176	227	17	,	,	PUNCT
ejpam-3176	227	18	_	_	PRON
ejpam-3176	227	19	µb=	µb=	NOUN
ejpam-3176	227	20	µb	µb	VERB
ejpam-3176	227	21	+	+	NUM
ejpam-3176	227	22	µz	µz	NOUN
ejpam-3176	227	23	,	,	PUNCT
ejpam-3176	227	24	and	and	CCONJ
ejpam-3176	227	25	_	_	PRON
ejpam-3176	227	26	µc=	µc=	NOUN
ejpam-3176	227	27	µc	µc	ADP
ejpam-3176	227	28	+	+	NUM
ejpam-3176	227	29	µz	µz	NOUN
ejpam-3176	227	30	.	.	PUNCT
ejpam-3176	228	1	the	the	DET
ejpam-3176	228	2	observations	observation	NOUN
ejpam-3176	228	3	in	in	ADP
ejpam-3176	228	4	this	this	DET
ejpam-3176	228	5	approach	approach	NOUN
ejpam-3176	228	6	are	be	AUX
ejpam-3176	228	7	pooled	pool	VERB
ejpam-3176	228	8	instead	instead	ADV
ejpam-3176	228	9	of	of	ADP
ejpam-3176	228	10	the	the	DET
ejpam-3176	228	11	covariance	covariance	NOUN
ejpam-3176	228	12	matrices	matrix	NOUN
ejpam-3176	228	13	.	.	PUNCT
ejpam-3176	229	1	hence	hence	ADV
ejpam-3176	229	2	,	,	PUNCT
ejpam-3176	229	3	rlda	rlda	NOUN
ejpam-3176	229	4	is	be	AUX
ejpam-3176	229	5	denoted	denote	VERB
ejpam-3176	229	6	by	by	ADP
ejpam-3176	229	7	pobs	pob	NOUN
ejpam-3176	229	8	for	for	ADP
ejpam-3176	229	9	the	the	DET
ejpam-3176	229	10	raw	raw	ADJ
ejpam-3176	229	11	version	version	NOUN
ejpam-3176	229	12	and	and	CCONJ
ejpam-3176	229	13	pobs	pobs	NOUN
ejpam-3176	229	14	-	-	PUNCT
ejpam-3176	229	15	w	w	NOUN
ejpam-3176	229	16	for	for	ADP
ejpam-3176	229	17	the	the	DET
ejpam-3176	229	18	reweighted	reweighted	ADJ
ejpam-3176	229	19	version	version	NOUN
ejpam-3176	229	20	.	.	PUNCT
ejpam-3176	230	1	the	the	DET
ejpam-3176	230	2	third	third	ADJ
ejpam-3176	230	3	method	method	NOUN
ejpam-3176	230	4	is	be	AUX
ejpam-3176	230	5	a	a	DET
ejpam-3176	230	6	combination	combination	NOUN
ejpam-3176	230	7	of	of	ADP
ejpam-3176	230	8	the	the	DET
ejpam-3176	230	9	two	two	NUM
ejpam-3176	230	10	previous	previous	ADJ
ejpam-3176	230	11	methods	method	NOUN
ejpam-3176	230	12	.	.	PUNCT
ejpam-3176	231	1	this	this	DET
ejpam-3176	231	2	method	method	NOUN
ejpam-3176	231	3	aims	aim	VERB
ejpam-3176	231	4	to	to	PART
ejpam-3176	231	5	derive	derive	VERB
ejpam-3176	231	6	a	a	DET
ejpam-3176	231	7	fast	fast	ADJ
ejpam-3176	231	8	approximation	approximation	NOUN
ejpam-3176	231	9	of	of	ADP
ejpam-3176	231	10	the	the	DET
ejpam-3176	231	11	mwcd	mwcd	NOUN
ejpam-3176	231	12	criterion	criterion	NOUN
ejpam-3176	231	13	(	(	PUNCT
ejpam-3176	231	14	hawkins	hawkin	NOUN
ejpam-3176	231	15	and	and	CCONJ
ejpam-3176	231	16	mclachlan	mclachlan	NOUN
ejpam-3176	231	17	[	[	X
ejpam-3176	231	18	11	11	NUM
ejpam-3176	231	19	]	]	PUNCT
ejpam-3176	231	20	)	)	PUNCT
ejpam-3176	231	21	.	.	PUNCT
ejpam-3176	232	1	instead	instead	ADV
ejpam-3176	232	2	of	of	ADP
ejpam-3176	232	3	performing	perform	VERB
ejpam-3176	232	4	the	the	DET
ejpam-3176	232	5	same	same	ADJ
ejpam-3176	232	6	adjustment	adjustment	NOUN
ejpam-3176	232	7	for	for	ADP
ejpam-3176	232	8	each	each	DET
ejpam-3176	232	9	group	group	NOUN
ejpam-3176	232	10	,	,	PUNCT
ejpam-3176	232	11	h	h	PROPN
ejpam-3176	232	12	is	be	AUX
ejpam-3176	232	13	identified	identify	VERB
ejpam-3176	232	14	from	from	ADP
ejpam-3176	232	15	all	all	DET
ejpam-3176	232	16	observations	observation	NOUN
ejpam-3176	232	17	with	with	ADP
ejpam-3176	232	18	set	set	ADJ
ejpam-3176	232	19	size	size	NOUN
ejpam-3176	232	20	n	n	CCONJ
ejpam-3176	232	21	,	,	PUNCT
ejpam-3176	232	22	where	where	SCONJ
ejpam-3176	232	23	the	the	DET
ejpam-3176	232	24	covariance	covariance	NOUN
ejpam-3176	232	25	matrix	matrix	NOUN
ejpam-3176	232	26	σh	σh	NOUN
ejpam-3176	232	27	of	of	ADP
ejpam-3176	232	28	h	h	NOUN
ejpam-3176	232	29	has	have	VERB
ejpam-3176	232	30	a	a	DET
ejpam-3176	232	31	minimal	minimal	ADJ
ejpam-3176	232	32	determinant	determinant	ADJ
ejpam-3176	232	33	.	.	PUNCT
ejpam-3176	233	1	then	then	ADV
ejpam-3176	233	2	,	,	PUNCT
ejpam-3176	233	3	the	the	DET
ejpam-3176	233	4	covariance	covariance	NOUN
ejpam-3176	233	5	matrix	matrix	NOUN
ejpam-3176	233	6	of	of	ADP
ejpam-3176	233	7	h	h	PROPN
ejpam-3176	233	8	(	(	PUNCT
ejpam-3176	233	9	h	h	NOUN
ejpam-3176	233	10	out	out	ADP
ejpam-3176	233	11	of	of	ADP
ejpam-3176	233	12	n	n	CCONJ
ejpam-3176	233	13	)	)	PUNCT
ejpam-3176	233	14	is	be	AUX
ejpam-3176	233	15	obtained	obtain	VERB
ejpam-3176	233	16	.	.	PUNCT
ejpam-3176	234	1	the	the	DET
ejpam-3176	234	2	approach	approach	NOUN
ejpam-3176	234	3	for	for	ADP
ejpam-3176	234	4	the	the	DET
ejpam-3176	234	5	three	three	NUM
ejpam-3176	234	6	groups	group	NOUN
ejpam-3176	234	7	is	be	AUX
ejpam-3176	234	8	as	as	SCONJ
ejpam-3176	234	9	follows	follow	VERB
ejpam-3176	234	10	.	.	PUNCT
ejpam-3176	235	1	(	(	PUNCT
ejpam-3176	235	2	i	i	NOUN
ejpam-3176	235	3	)	)	PUNCT
ejpam-3176	235	4	the	the	DET
ejpam-3176	235	5	canters	canter	NOUN
ejpam-3176	235	6	of	of	ADP
ejpam-3176	235	7	the	the	DET
ejpam-3176	235	8	groups	group	NOUN
ejpam-3176	235	9	are	be	AUX
ejpam-3176	235	10	estimated	estimate	VERB
ejpam-3176	235	11	.	.	PUNCT
ejpam-3176	236	1	(	(	PUNCT
ejpam-3176	236	2	ii	ii	X
ejpam-3176	236	3	)	)	PUNCT
ejpam-3176	236	4	the	the	DET
ejpam-3176	236	5	observations	observation	NOUN
ejpam-3176	236	6	are	be	AUX
ejpam-3176	236	7	shifted	shift	VERB
ejpam-3176	236	8	and	and	CCONJ
ejpam-3176	236	9	pooled	pool	VERB
ejpam-3176	236	10	to	to	PART
ejpam-3176	236	11	obtain	obtain	VERB
ejpam-3176	236	12	z	z	PROPN
ejpam-3176	236	13	’s	’s	NOUN
ejpam-3176	236	14	that	that	PRON
ejpam-3176	236	15	are	be	AUX
ejpam-3176	236	16	the	the	DET
ejpam-3176	236	17	same	same	ADJ
ejpam-3176	236	18	as	as	ADP
ejpam-3176	236	19	those	those	PRON
ejpam-3176	236	20	in	in	ADP
ejpam-3176	236	21	the	the	DET
ejpam-3176	236	22	second	second	ADJ
ejpam-3176	236	23	approach	approach	NOUN
ejpam-3176	236	24	using	use	VERB
ejpam-3176	236	25	the	the	DET
ejpam-3176	236	26	fastmcd	fastmcd	NOUN
ejpam-3176	236	27	estimator	estimator	NOUN
ejpam-3176	236	28	.	.	PUNCT
ejpam-3176	237	1	(	(	PUNCT
ejpam-3176	237	2	iii	iii	X
ejpam-3176	237	3	)	)	PUNCT
ejpam-3176	237	4	let	let	VERB
ejpam-3176	237	5	h	h	NOUN
ejpam-3176	237	6	be	be	AUX
ejpam-3176	237	7	h	h	NOUN
ejpam-3176	237	8	out	out	ADP
ejpam-3176	237	9	of	of	ADP
ejpam-3176	237	10	n	n	NUM
ejpam-3176	237	11	based	base	VERB
ejpam-3176	237	12	on	on	ADP
ejpam-3176	237	13	the	the	DET
ejpam-3176	237	14	minimized	minimized	ADJ
ejpam-3176	237	15	fastmcd	fastmcd	NOUN
ejpam-3176	237	16	estimator	estimator	NOUN
ejpam-3176	237	17	.	.	PUNCT
ejpam-3176	238	1	(	(	PUNCT
ejpam-3176	238	2	iv	iv	X
ejpam-3176	238	3	)	)	PUNCT
ejpam-3176	238	4	the	the	DET
ejpam-3176	238	5	subset	subset	NOUN
ejpam-3176	238	6	h	h	NOUN
ejpam-3176	238	7	is	be	AUX
ejpam-3176	238	8	partitioned	partition	VERB
ejpam-3176	238	9	into	into	ADP
ejpam-3176	238	10	ha	ha	INTJ
ejpam-3176	238	11	,	,	PUNCT
ejpam-3176	238	12	hb	hb	PROPN
ejpam-3176	238	13	,	,	PUNCT
ejpam-3176	238	14	and	and	CCONJ
ejpam-3176	238	15	hc	hc	PROPN
ejpam-3176	238	16	,	,	PUNCT
ejpam-3176	238	17	which	which	PRON
ejpam-3176	238	18	contain	contain	VERB
ejpam-3176	238	19	observations	observation	NOUN
ejpam-3176	238	20	from	from	ADP
ejpam-3176	238	21	a	a	DET
ejpam-3176	238	22	,	,	PUNCT
ejpam-3176	238	23	b	b	NOUN
ejpam-3176	238	24	,	,	PUNCT
ejpam-3176	238	25	and	and	CCONJ
ejpam-3176	238	26	c	c	NOUN
ejpam-3176	238	27	,	,	PUNCT
ejpam-3176	238	28	respectively	respectively	ADV
ejpam-3176	238	29	.	.	PUNCT
ejpam-3176	239	1	(	(	PUNCT
ejpam-3176	239	2	v	v	NOUN
ejpam-3176	239	3	)	)	PUNCT
ejpam-3176	239	4	the	the	DET
ejpam-3176	239	5	mean	mean	NOUN
ejpam-3176	239	6	of	of	ADP
ejpam-3176	239	7	all	all	DET
ejpam-3176	239	8	the	the	DET
ejpam-3176	239	9	groups	group	NOUN
ejpam-3176	239	10	is	be	AUX
ejpam-3176	239	11	estimated	estimate	VERB
ejpam-3176	239	12	as	as	ADP
ejpam-3176	239	13	µa	µa	PROPN
ejpam-3176	239	14	,	,	PUNCT
ejpam-3176	239	15	µb	µb	VERB
ejpam-3176	239	16	,	,	PUNCT
ejpam-3176	239	17	and	and	CCONJ
ejpam-3176	239	18	µc	µc	INTJ
ejpam-3176	239	19	.	.	PUNCT
ejpam-3176	240	1	in	in	ADP
ejpam-3176	240	2	this	this	DET
ejpam-3176	240	3	approach	approach	NOUN
ejpam-3176	240	4	,	,	PUNCT
ejpam-3176	240	5	rlda	rlda	NOUN
ejpam-3176	240	6	is	be	AUX
ejpam-3176	240	7	denoted	denote	VERB
ejpam-3176	240	8	based	base	VERB
ejpam-3176	240	9	on	on	ADP
ejpam-3176	240	10	the	the	DET
ejpam-3176	240	11	mwcd	mwcd	NOUN
ejpam-3176	240	12	for	for	ADP
ejpam-3176	240	13	the	the	DET
ejpam-3176	240	14	raw	raw	ADJ
ejpam-3176	240	15	version	version	NOUN
ejpam-3176	240	16	or	or	CCONJ
ejpam-3176	240	17	mwcdw	mwcdw	NOUN
ejpam-3176	240	18	for	for	ADP
ejpam-3176	240	19	the	the	DET
ejpam-3176	240	20	reweighted	reweighted	ADJ
ejpam-3176	240	21	version	version	NOUN
ejpam-3176	240	22	.	.	PUNCT
ejpam-3176	241	1	mufda	mufda	PROPN
ejpam-3176	241	2	j.	j.	PROPN
ejpam-3176	241	3	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	241	4	,	,	PUNCT
ejpam-3176	241	5	taha	taha	PROPN
ejpam-3176	241	6	radwan	radwan	PROPN
ejpam-3176	241	7	and	and	CCONJ
ejpam-3176	241	8	khalid	khalid	PROPN
ejpam-3176	241	9	abunawas	abunawas	PROPN
ejpam-3176	241	10	/	/	SYM
ejpam-3176	241	11	eur	eur	PROPN
ejpam-3176	241	12	.	.	PUNCT
ejpam-3176	242	1	j.	j.	PROPN
ejpam-3176	242	2	pure	pure	PROPN
ejpam-3176	242	3	appl	appl	PROPN
ejpam-3176	242	4	.	.	PROPN
ejpam-3176	242	5	math	math	PROPN
ejpam-3176	242	6	,	,	PUNCT
ejpam-3176	242	7	11	11	NUM
ejpam-3176	242	8	(	(	PUNCT
ejpam-3176	242	9	1	1	NUM
ejpam-3176	242	10	)	)	PUNCT
ejpam-3176	242	11	(	(	PUNCT
ejpam-3176	242	12	2018	2018	NUM
ejpam-3176	242	13	)	)	PUNCT
ejpam-3176	242	14	,	,	PUNCT
ejpam-3176	242	15	284	284	NUM
ejpam-3176	242	16	-	-	SYM
ejpam-3176	242	17	298	298	NUM
ejpam-3176	242	18	292	292	NUM
ejpam-3176	242	19	8	8	NUM
ejpam-3176	242	20	.	.	PUNCT
ejpam-3176	243	1	simulation	simulation	NOUN
ejpam-3176	243	2	result	result	VERB
ejpam-3176	243	3	through	through	ADP
ejpam-3176	243	4	the	the	DET
ejpam-3176	243	5	simulation	simulation	NOUN
ejpam-3176	243	6	study	study	NOUN
ejpam-3176	243	7	,	,	PUNCT
ejpam-3176	243	8	we	we	PRON
ejpam-3176	243	9	compare	compare	VERB
ejpam-3176	243	10	the	the	DET
ejpam-3176	243	11	performances	performance	NOUN
ejpam-3176	243	12	of	of	ADP
ejpam-3176	243	13	the	the	DET
ejpam-3176	243	14	three	three	NUM
ejpam-3176	243	15	approaches	approach	NOUN
ejpam-3176	243	16	in	in	ADP
ejpam-3176	243	17	estimating	estimate	VERB
ejpam-3176	243	18	the	the	DET
ejpam-3176	243	19	initial	initial	ADJ
ejpam-3176	243	20	values	value	NOUN
ejpam-3176	243	21	for	for	ADP
ejpam-3176	243	22	the	the	DET
ejpam-3176	243	23	mean	mean	NOUN
ejpam-3176	243	24	and	and	CCONJ
ejpam-3176	243	25	covariance	covariance	NOUN
ejpam-3176	243	26	matrix	matrix	NOUN
ejpam-3176	243	27	and	and	CCONJ
ejpam-3176	243	28	then	then	ADV
ejpam-3176	243	29	apply	apply	VERB
ejpam-3176	243	30	the	the	DET
ejpam-3176	243	31	obtained	obtain	VERB
ejpam-3176	243	32	values	value	NOUN
ejpam-3176	243	33	to	to	PART
ejpam-3176	243	34	fastmcd	fastmcd	VERB
ejpam-3176	243	35	,	,	PUNCT
ejpam-3176	243	36	detmcd	detmcd	NOUN
ejpam-3176	243	37	,	,	PUNCT
ejpam-3176	243	38	and	and	CCONJ
ejpam-3176	243	39	fhc	fhc	VERB
ejpam-3176	243	40	to	to	PART
ejpam-3176	243	41	obtain	obtain	VERB
ejpam-3176	243	42	the	the	DET
ejpam-3176	243	43	misclassification	misclassification	NOUN
ejpam-3176	243	44	probabilities	probability	NOUN
ejpam-3176	243	45	.	.	PUNCT
ejpam-3176	244	1	we	we	PRON
ejpam-3176	244	2	use	use	VERB
ejpam-3176	244	3	the	the	DET
ejpam-3176	244	4	raw	raw	ADJ
ejpam-3176	244	5	and	and	CCONJ
ejpam-3176	244	6	reweighted	reweighted	ADJ
ejpam-3176	244	7	versions	version	NOUN
ejpam-3176	244	8	for	for	ADP
ejpam-3176	244	9	all	all	DET
ejpam-3176	244	10	the	the	DET
ejpam-3176	244	11	algorithms	algorithm	NOUN
ejpam-3176	244	12	.	.	PUNCT
ejpam-3176	245	1	the	the	DET
ejpam-3176	245	2	rlda	rlda	NOUN
ejpam-3176	245	3	rule	rule	NOUN
ejpam-3176	245	4	is	be	AUX
ejpam-3176	245	5	applied	apply	VERB
ejpam-3176	245	6	using	use	VERB
ejpam-3176	245	7	settings	setting	NOUN
ejpam-3176	245	8	similar	similar	ADJ
ejpam-3176	245	9	to	to	ADP
ejpam-3176	245	10	those	those	PRON
ejpam-3176	245	11	in	in	ADP
ejpam-3176	245	12	the	the	DET
ejpam-3176	245	13	study	study	NOUN
ejpam-3176	245	14	of	of	ADP
ejpam-3176	245	15	he	he	PRON
ejpam-3176	245	16	and	and	CCONJ
ejpam-3176	245	17	fung	fung	PROPN
ejpam-3176	246	1	[	[	X
ejpam-3176	246	2	12	12	NUM
ejpam-3176	246	3	]	]	X
ejpam-3176	246	4	:	:	PUNCT
ejpam-3176	246	5	a⇒	a⇒	PROPN
ejpam-3176	246	6	π1	π1	NOUN
ejpam-3176	246	7	:	:	PUNCT
ejpam-3176	246	8	400n3(0	400n3(0	NUM
ejpam-3176	246	9	,	,	PUNCT
ejpam-3176	246	10	i	i	NOUN
ejpam-3176	246	11	)	)	PUNCT
ejpam-3176	247	1	π2	π2	ADV
ejpam-3176	247	2	:	:	PUNCT
ejpam-3176	247	3	400n3(1	400n3(1	X
ejpam-3176	247	4	,	,	PUNCT
ejpam-3176	247	5	i	i	NOUN
ejpam-3176	247	6	)	)	PUNCT
ejpam-3176	247	7	b	b	PROPN
ejpam-3176	247	8	⇒	⇒	NOUN
ejpam-3176	247	9	π1	π1	NOUN
ejpam-3176	247	10	:	:	PUNCT
ejpam-3176	247	11	400n3(0	400n3(0	NUM
ejpam-3176	247	12	,	,	PUNCT
ejpam-3176	247	13	i	i	NOUN
ejpam-3176	247	14	)	)	PUNCT
ejpam-3176	247	15	+	+	CCONJ
ejpam-3176	247	16	50n3(5	50n3(5	NUM
ejpam-3176	247	17	,	,	PUNCT
ejpam-3176	247	18	(	(	PUNCT
ejpam-3176	247	19	1/0.252)i	1/0.252)i	NUM
ejpam-3176	247	20	)	)	PUNCT
ejpam-3176	247	21	π2	π2	NOUN
ejpam-3176	247	22	:	:	PUNCT
ejpam-3176	247	23	400n3(1	400n3(1	X
ejpam-3176	247	24	,	,	PUNCT
ejpam-3176	247	25	i	i	NOUN
ejpam-3176	247	26	)	)	PUNCT
ejpam-3176	248	1	+	+	CCONJ
ejpam-3176	249	1	50n3(−4	50n3(−4	NUM
ejpam-3176	249	2	,	,	PUNCT
ejpam-3176	249	3	(	(	PUNCT
ejpam-3176	249	4	1/0.252)i	1/0.252)i	NUM
ejpam-3176	249	5	)	)	PUNCT
ejpam-3176	249	6	c	c	NOUN
ejpam-3176	249	7	⇒	⇒	PROPN
ejpam-3176	249	8	π1	π1	NOUN
ejpam-3176	249	9	:	:	PUNCT
ejpam-3176	249	10	80n3(0	80n3(0	NUM
ejpam-3176	249	11	,	,	PUNCT
ejpam-3176	249	12	i	i	NOUN
ejpam-3176	249	13	)	)	PUNCT
ejpam-3176	250	1	+	+	CCONJ
ejpam-3176	250	2	20n3((1	20n3((1	NUM
ejpam-3176	250	3	,	,	PUNCT
ejpam-3176	250	4	(	(	PUNCT
ejpam-3176	250	5	1/0.252)i	1/0.252)i	NUM
ejpam-3176	250	6	)	)	PUNCT
ejpam-3176	250	7	π2	π2	NOUN
ejpam-3176	250	8	:	:	PUNCT
ejpam-3176	250	9	80n3(1	80n3(1	NUM
ejpam-3176	250	10	,	,	PUNCT
ejpam-3176	250	11	i	i	PRON
ejpam-3176	250	12	)	)	PUNCT
ejpam-3176	251	1	+	+	CCONJ
ejpam-3176	251	2	20n3((−1	20n3((−1	NUM
ejpam-3176	251	3	,	,	PUNCT
ejpam-3176	251	4	(	(	PUNCT
ejpam-3176	251	5	1/0.252)i	1/0.252)i	NUM
ejpam-3176	251	6	)	)	PUNCT
ejpam-3176	251	7	d	d	NOUN
ejpam-3176	251	8	⇒	⇒	NOUN
ejpam-3176	251	9	π1	π1	NOUN
ejpam-3176	251	10	:	:	PUNCT
ejpam-3176	251	11	200n3(0	200n3(0	NUM
ejpam-3176	251	12	,	,	PUNCT
ejpam-3176	251	13	i	i	PRON
ejpam-3176	251	14	)	)	PUNCT
ejpam-3176	252	1	+	+	CCONJ
ejpam-3176	252	2	25n3((1	25n3((1	NUM
ejpam-3176	252	3	,	,	PUNCT
ejpam-3176	252	4	(	(	PUNCT
ejpam-3176	252	5	1/0.075)i	1/0.075)i	NUM
ejpam-3176	252	6	)	)	PUNCT
ejpam-3176	252	7	π2	π2	NOUN
ejpam-3176	252	8	:	:	PUNCT
ejpam-3176	252	9	200n3(1	200n3(1	NUM
ejpam-3176	252	10	,	,	PUNCT
ejpam-3176	252	11	i	i	PRON
ejpam-3176	252	12	)	)	PUNCT
ejpam-3176	252	13	+	+	CCONJ
ejpam-3176	252	14	25n3((−1	25n3((−1	NUM
ejpam-3176	252	15	,	,	PUNCT
ejpam-3176	252	16	(	(	PUNCT
ejpam-3176	252	17	1/0.075)i	1/0.075)i	NUM
ejpam-3176	252	18	)	)	PUNCT
ejpam-3176	252	19	e	e	NOUN
ejpam-3176	252	20	⇒	⇒	NOUN
ejpam-3176	252	21	π1	π1	NOUN
ejpam-3176	252	22	:	:	PUNCT
ejpam-3176	252	23	300n3(0	300n3(0	NUM
ejpam-3176	252	24	,	,	PUNCT
ejpam-3176	252	25	i	i	NOUN
ejpam-3176	252	26	)	)	PUNCT
ejpam-3176	252	27	+	+	NUM
ejpam-3176	252	28	10n3(5	10n3(5	NOUN
ejpam-3176	252	29	,	,	PUNCT
ejpam-3176	252	30	0.252i	0.252i	NOUN
ejpam-3176	252	31	)	)	PUNCT
ejpam-3176	252	32	π2	π2	NOUN
ejpam-3176	252	33	:	:	PUNCT
ejpam-3176	252	34	150n3(1	150n3(1	NUM
ejpam-3176	252	35	,	,	PUNCT
ejpam-3176	252	36	i	i	NOUN
ejpam-3176	252	37	)	)	PUNCT
ejpam-3176	252	38	+	+	X
ejpam-3176	253	1	5n3(−10	5n3(−10	NUM
ejpam-3176	253	2	,	,	PUNCT
ejpam-3176	253	3	0.252i	0.252i	NOUN
ejpam-3176	253	4	)	)	PUNCT
ejpam-3176	253	5	,	,	PUNCT
ejpam-3176	253	6	where	where	SCONJ
ejpam-3176	253	7	i	i	PRON
ejpam-3176	253	8	is	be	AUX
ejpam-3176	253	9	the	the	DET
ejpam-3176	253	10	3d	3d	NUM
ejpam-3176	253	11	identity	identity	NOUN
ejpam-3176	253	12	matrix	matrix	NOUN
ejpam-3176	253	13	,	,	PUNCT
ejpam-3176	253	14	the	the	DET
ejpam-3176	253	15	groups	group	NOUN
ejpam-3176	253	16	labeled	label	VERB
ejpam-3176	253	17	as	as	ADP
ejpam-3176	253	18	a	a	DET
ejpam-3176	253	19	,	,	PUNCT
ejpam-3176	253	20	b	b	NOUN
ejpam-3176	253	21	,	,	PUNCT
ejpam-3176	253	22	c	c	NOUN
ejpam-3176	253	23	,	,	PUNCT
ejpam-3176	253	24	d	d	NOUN
ejpam-3176	253	25	and	and	CCONJ
ejpam-3176	253	26	e	e	NOUN
ejpam-3176	253	27	,	,	PUNCT
ejpam-3176	253	28	each	each	DET
ejpam-3176	253	29	group	group	NOUN
ejpam-3176	253	30	has	have	VERB
ejpam-3176	253	31	to	to	ADP
ejpam-3176	253	32	different	different	ADJ
ejpam-3176	253	33	cases	case	NOUN
ejpam-3176	253	34	π1	π1	ADJ
ejpam-3176	253	35	andπ2	andπ2	PROPN
ejpam-3176	253	36	.	.	PUNCT
ejpam-3176	254	1	the	the	DET
ejpam-3176	254	2	membership	membership	NOUN
ejpam-3176	254	3	probability	probability	NOUN
ejpam-3176	254	4	is	be	AUX
ejpam-3176	254	5	calculated	calculate	VERB
ejpam-3176	254	6	based	base	VERB
ejpam-3176	254	7	on	on	ADP
ejpam-3176	254	8	formula	formula	NOUN
ejpam-3176	254	9	(	(	PUNCT
ejpam-3176	254	10	8)	8)	NUM
ejpam-3176	254	11	for	for	ADP
ejpam-3176	254	12	l	l	NOUN
ejpam-3176	254	13	in	in	ADP
ejpam-3176	254	14	consideration	consideration	NOUN
ejpam-3176	254	15	of	of	ADP
ejpam-3176	254	16	the	the	DET
ejpam-3176	254	17	two	two	NUM
ejpam-3176	254	18	algorithms	algorithm	NOUN
ejpam-3176	254	19	.	.	PUNCT
ejpam-3176	255	1	the	the	DET
ejpam-3176	255	2	classification	classification	NOUN
ejpam-3176	255	3	rule	rule	NOUN
ejpam-3176	255	4	(	(	PUNCT
ejpam-3176	255	5	eq	eq	NOUN
ejpam-3176	255	6	.	.	PUNCT
ejpam-3176	255	7	(	(	PUNCT
ejpam-3176	255	8	1	1	NUM
ejpam-3176	255	9	)	)	PUNCT
ejpam-3176	255	10	)	)	PUNCT
ejpam-3176	255	11	is	be	AUX
ejpam-3176	255	12	robustified	robustifie	VERB
ejpam-3176	255	13	as	as	ADP
ejpam-3176	255	14	the	the	DET
ejpam-3176	255	15	fisher	fisher	PROPN
ejpam-3176	255	16	discriminant	discriminant	PROPN
ejpam-3176	255	17	rule	rule	NOUN
ejpam-3176	255	18	,	,	PUNCT
ejpam-3176	255	19	which	which	PRON
ejpam-3176	255	20	expressed	express	VERB
ejpam-3176	255	21	as	as	SCONJ
ejpam-3176	255	22	follows	follow	VERB
ejpam-3176	255	23	:	:	PUNCT
ejpam-3176	255	24	x	x	SYM
ejpam-3176	255	25	∈	∈	NOUN
ejpam-3176	255	26	π1if	π1if	PUNCT
ejpam-3176	255	27	(	(	PUNCT
ejpam-3176	255	28	_	_	NOUN
ejpam-3176	256	1	µ1	µ1	ADJ
ejpam-3176	256	2	−	−	NOUN
ejpam-3176	256	3	_	_	PUNCT
ejpam-3176	257	1	µ2)tς−1(x−	µ2)tς−1(x−	ADP
ejpam-3176	257	2	(	(	PUNCT
ejpam-3176	257	3	_	_	NOUN
ejpam-3176	257	4	µ1	µ1	NOUN
ejpam-3176	257	5	−	−	NOUN
ejpam-3176	257	6	_	_	PUNCT
ejpam-3176	257	7	µ2)/2	µ2)/2	ADJ
ejpam-3176	257	8	)	)	PUNCT
ejpam-3176	257	9	>	>	X
ejpam-3176	257	10	0	0	PUNCT
ejpam-3176	257	11	(	(	PUNCT
ejpam-3176	257	12	17	17	NUM
ejpam-3176	257	13	)	)	PUNCT
ejpam-3176	257	14	and	and	CCONJ
ejpam-3176	257	15	x	x	PUNCT
ejpam-3176	257	16	∈	∈	NOUN
ejpam-3176	257	17	π2	π2	NOUN
ejpam-3176	257	18	otherwise	otherwise	ADV
ejpam-3176	257	19	.	.	PUNCT
ejpam-3176	258	1	figure	figure	VERB
ejpam-3176	258	2	1	1	NUM
ejpam-3176	258	3	presents	present	NOUN
ejpam-3176	258	4	only	only	ADV
ejpam-3176	258	5	the	the	DET
ejpam-3176	258	6	first	first	ADJ
ejpam-3176	258	7	100	100	NUM
ejpam-3176	258	8	observations	observation	NOUN
ejpam-3176	258	9	from	from	ADP
ejpam-3176	258	10	the	the	DET
ejpam-3176	258	11	groups	group	NOUN
ejpam-3176	258	12	of	of	ADP
ejpam-3176	258	13	case	case	NOUN
ejpam-3176	258	14	c.	c.	NOUN
ejpam-3176	258	15	the	the	DET
ejpam-3176	258	16	figure	figure	NOUN
ejpam-3176	258	17	shows	show	VERB
ejpam-3176	258	18	the	the	DET
ejpam-3176	258	19	scatter	scatter	NOUN
ejpam-3176	258	20	of	of	ADP
ejpam-3176	258	21	data	datum	NOUN
ejpam-3176	258	22	in	in	ADP
ejpam-3176	258	23	case	case	NOUN
ejpam-3176	258	24	c	c	NOUN
ejpam-3176	258	25	,	,	PUNCT
ejpam-3176	258	26	in	in	ADP
ejpam-3176	258	27	which	which	PRON
ejpam-3176	258	28	both	both	DET
ejpam-3176	258	29	groups	group	NOUN
ejpam-3176	258	30	have	have	VERB
ejpam-3176	258	31	outliers	outlier	NOUN
ejpam-3176	258	32	.	.	PUNCT
ejpam-3176	259	1	the	the	DET
ejpam-3176	259	2	effects	effect	NOUN
ejpam-3176	259	3	of	of	ADP
ejpam-3176	259	4	the	the	DET
ejpam-3176	259	5	outliers	outlier	NOUN
ejpam-3176	259	6	on	on	ADP
ejpam-3176	259	7	data	datum	NOUN
ejpam-3176	259	8	homogeneity	homogeneity	NOUN
ejpam-3176	259	9	and	and	CCONJ
ejpam-3176	259	10	on	on	ADP
ejpam-3176	259	11	the	the	DET
ejpam-3176	259	12	parameters	parameter	NOUN
ejpam-3176	259	13	of	of	ADP
ejpam-3176	259	14	the	the	DET
ejpam-3176	259	15	discriminate	discriminate	ADJ
ejpam-3176	259	16	rules	rule	NOUN
ejpam-3176	259	17	that	that	PRON
ejpam-3176	259	18	will	will	AUX
ejpam-3176	259	19	be	be	AUX
ejpam-3176	259	20	used	use	VERB
ejpam-3176	259	21	to	to	PART
ejpam-3176	259	22	estimate	estimate	VERB
ejpam-3176	259	23	values	value	NOUN
ejpam-3176	259	24	are	be	AUX
ejpam-3176	259	25	clearly	clearly	ADV
ejpam-3176	259	26	shown	show	VERB
ejpam-3176	259	27	.	.	PUNCT
ejpam-3176	260	1	graphs	graphs	PROPN
ejpam-3176	260	2	c1	c1	PROPN
ejpam-3176	260	3	and	and	CCONJ
ejpam-3176	260	4	c2	c2	PROPN
ejpam-3176	260	5	show	show	VERB
ejpam-3176	260	6	a	a	DET
ejpam-3176	260	7	cut	cut	NOUN
ejpam-3176	260	8	and	and	CCONJ
ejpam-3176	260	9	abnormal	abnormal	ADJ
ejpam-3176	260	10	spread	spread	NOUN
ejpam-3176	260	11	for	for	ADP
ejpam-3176	260	12	the	the	DET
ejpam-3176	260	13	observations	observation	NOUN
ejpam-3176	260	14	.	.	PUNCT
ejpam-3176	261	1	this	this	DET
ejpam-3176	261	2	study	study	NOUN
ejpam-3176	261	3	aims	aim	VERB
ejpam-3176	261	4	to	to	PART
ejpam-3176	261	5	compare	compare	VERB
ejpam-3176	261	6	detmcd	detmcd	NOUN
ejpam-3176	261	7	with	with	ADP
ejpam-3176	261	8	fastmcd	fastmcd	NOUN
ejpam-3176	261	9	and	and	CCONJ
ejpam-3176	261	10	fch	fch	PROPN
ejpam-3176	261	11	using	use	VERB
ejpam-3176	261	12	three	three	NUM
ejpam-3176	261	13	approaches	approach	NOUN
ejpam-3176	261	14	,	,	PUNCT
ejpam-3176	261	15	that	that	ADV
ejpam-3176	261	16	is	is	ADV
ejpam-3176	261	17	,	,	PUNCT
ejpam-3176	261	18	pcov	pcov	ADJ
ejpam-3176	261	19	,	,	PUNCT
ejpam-3176	261	20	pobs	pob	NOUN
ejpam-3176	261	21	,	,	PUNCT
ejpam-3176	261	22	and	and	CCONJ
ejpam-3176	261	23	mwcd	mwcd	NOUN
ejpam-3176	261	24	,	,	PUNCT
ejpam-3176	261	25	for	for	ADP
ejpam-3176	261	26	all	all	DET
ejpam-3176	261	27	algorithms	algorithm	NOUN
ejpam-3176	261	28	at	at	ADP
ejpam-3176	261	29	the	the	DET
ejpam-3176	261	30	raw	raw	ADJ
ejpam-3176	261	31	and	and	CCONJ
ejpam-3176	261	32	reweighted	reweighted	ADJ
ejpam-3176	261	33	versions	version	NOUN
ejpam-3176	261	34	.	.	PUNCT
ejpam-3176	262	1	as	as	SCONJ
ejpam-3176	262	2	shown	show	VERB
ejpam-3176	262	3	in	in	ADP
ejpam-3176	262	4	table	table	NOUN
ejpam-3176	262	5	1	1	NUM
ejpam-3176	262	6	and	and	CCONJ
ejpam-3176	262	7	table	table	NOUN
ejpam-3176	262	8	2	2	NUM
ejpam-3176	262	9	,	,	PUNCT
ejpam-3176	262	10	the	the	DET
ejpam-3176	262	11	detmcd	detmcd	PROPN
ejpam-3176	262	12	estimator	estimator	NOUN
ejpam-3176	262	13	clearly	clearly	ADV
ejpam-3176	262	14	increases	increase	VERB
ejpam-3176	262	15	the	the	DET
ejpam-3176	262	16	efficiency	efficiency	NOUN
ejpam-3176	262	17	of	of	ADP
ejpam-3176	262	18	the	the	DET
ejpam-3176	262	19	discriminate	discriminate	NOUN
ejpam-3176	262	20	rules	rule	NOUN
ejpam-3176	262	21	in	in	ADP
ejpam-3176	262	22	estimating	estimate	VERB
ejpam-3176	262	23	the	the	DET
ejpam-3176	262	24	error	error	NOUN
ejpam-3176	262	25	rate	rate	NOUN
ejpam-3176	262	26	of	of	ADP
ejpam-3176	262	27	classification	classification	NOUN
ejpam-3176	262	28	.	.	PUNCT
ejpam-3176	263	1	moreover	moreover	ADV
ejpam-3176	263	2	,	,	PUNCT
ejpam-3176	263	3	detmcd	detmcd	NOUN
ejpam-3176	263	4	performs	perform	VERB
ejpam-3176	263	5	better	well	ADV
ejpam-3176	263	6	than	than	ADP
ejpam-3176	263	7	the	the	DET
ejpam-3176	263	8	fastmcd	fastmcd	NOUN
ejpam-3176	263	9	and	and	CCONJ
ejpam-3176	263	10	fch	fch	NOUN
ejpam-3176	263	11	estimators	estimator	NOUN
ejpam-3176	263	12	for	for	ADP
ejpam-3176	263	13	the	the	DET
ejpam-3176	263	14	raw	raw	ADJ
ejpam-3176	263	15	and	and	CCONJ
ejpam-3176	263	16	reweighted	reweighted	ADJ
ejpam-3176	263	17	versions	version	NOUN
ejpam-3176	263	18	.	.	PUNCT
ejpam-3176	264	1	detmcd	detmcd	VERB
ejpam-3176	264	2	for	for	ADP
ejpam-3176	264	3	the	the	DET
ejpam-3176	264	4	raw	raw	ADJ
ejpam-3176	264	5	and	and	CCONJ
ejpam-3176	264	6	reweighted	reweighted	ADJ
ejpam-3176	264	7	versions	version	NOUN
ejpam-3176	264	8	are	be	AUX
ejpam-3176	264	9	more	more	ADV
ejpam-3176	264	10	accurate	accurate	ADJ
ejpam-3176	264	11	in	in	ADP
ejpam-3176	264	12	estimating	estimate	VERB
ejpam-3176	264	13	the	the	DET
ejpam-3176	264	14	misclassification	misclassification	NOUN
ejpam-3176	264	15	probabilities	probability	NOUN
ejpam-3176	264	16	of	of	ADP
ejpam-3176	264	17	data	datum	NOUN
ejpam-3176	264	18	compared	compare	VERB
ejpam-3176	264	19	with	with	ADP
ejpam-3176	264	20	the	the	DET
ejpam-3176	264	21	other	other	ADJ
ejpam-3176	264	22	two	two	NUM
ejpam-3176	264	23	estimators	estimator	NOUN
ejpam-3176	264	24	.	.	PUNCT
ejpam-3176	265	1	the	the	DET
ejpam-3176	265	2	second	second	ADJ
ejpam-3176	265	3	part	part	NOUN
ejpam-3176	265	4	of	of	ADP
ejpam-3176	265	5	this	this	DET
ejpam-3176	265	6	comparative	comparative	ADJ
ejpam-3176	265	7	study	study	NOUN
ejpam-3176	265	8	will	will	AUX
ejpam-3176	265	9	use	use	VERB
ejpam-3176	265	10	the	the	DET
ejpam-3176	265	11	three	three	NUM
ejpam-3176	265	12	approaches	approach	NOUN
ejpam-3176	265	13	for	for	ADP
ejpam-3176	265	14	the	the	DET
ejpam-3176	265	15	three	three	NUM
ejpam-3176	265	16	estimators	estimator	NOUN
ejpam-3176	265	17	.	.	PUNCT
ejpam-3176	266	1	the	the	DET
ejpam-3176	266	2	pcov	pcov	PROPN
ejpam-3176	266	3	approach	approach	NOUN
ejpam-3176	266	4	obtains	obtain	VERB
ejpam-3176	266	5	the	the	DET
ejpam-3176	266	6	best	well	ADV
ejpam-3176	266	7	estimated	estimate	VERB
ejpam-3176	266	8	values	value	NOUN
ejpam-3176	266	9	for	for	ADP
ejpam-3176	266	10	mp	mp	PROPN
ejpam-3176	266	11	under	under	ADP
ejpam-3176	266	12	the	the	DET
ejpam-3176	266	13	raw	raw	ADJ
ejpam-3176	266	14	version	version	NOUN
ejpam-3176	266	15	,	,	PUNCT
ejpam-3176	266	16	whereas	whereas	SCONJ
ejpam-3176	266	17	the	the	DET
ejpam-3176	266	18	pobs	pobs	NOUN
ejpam-3176	266	19	approach	approach	NOUN
ejpam-3176	266	20	achieves	achieve	VERB
ejpam-3176	266	21	the	the	DET
ejpam-3176	266	22	best	good	ADJ
ejpam-3176	266	23	result	result	NOUN
ejpam-3176	266	24	for	for	ADP
ejpam-3176	266	25	the	the	DET
ejpam-3176	266	26	reweighted	reweighted	ADJ
ejpam-3176	266	27	version	version	NOUN
ejpam-3176	266	28	.	.	PUNCT
ejpam-3176	267	1	detmcd	detmcd	NOUN
ejpam-3176	267	2	increases	increase	VERB
ejpam-3176	267	3	mufda	mufda	PROPN
ejpam-3176	267	4	j.	j.	PROPN
ejpam-3176	267	5	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	267	6	,	,	PUNCT
ejpam-3176	267	7	taha	taha	PROPN
ejpam-3176	267	8	radwan	radwan	PROPN
ejpam-3176	267	9	and	and	CCONJ
ejpam-3176	267	10	khalid	khalid	PROPN
ejpam-3176	267	11	abunawas	abunawas	PROPN
ejpam-3176	267	12	/	/	SYM
ejpam-3176	267	13	eur	eur	PROPN
ejpam-3176	267	14	.	.	PUNCT
ejpam-3176	268	1	j.	j.	PROPN
ejpam-3176	268	2	pure	pure	PROPN
ejpam-3176	268	3	appl	appl	PROPN
ejpam-3176	268	4	.	.	PROPN
ejpam-3176	268	5	math	math	PROPN
ejpam-3176	268	6	,	,	PUNCT
ejpam-3176	268	7	11	11	NUM
ejpam-3176	268	8	(	(	PUNCT
ejpam-3176	268	9	1	1	NUM
ejpam-3176	268	10	)	)	PUNCT
ejpam-3176	268	11	(	(	PUNCT
ejpam-3176	268	12	2018	2018	NUM
ejpam-3176	268	13	)	)	PUNCT
ejpam-3176	268	14	,	,	PUNCT
ejpam-3176	268	15	284	284	NUM
ejpam-3176	268	16	-	-	SYM
ejpam-3176	268	17	298	298	NUM
ejpam-3176	268	18	293	293	NUM
ejpam-3176	268	19	figure	figure	NOUN
ejpam-3176	268	20	1	1	NUM
ejpam-3176	268	21	:	:	PUNCT
ejpam-3176	268	22	observations	observation	NOUN
ejpam-3176	268	23	number	number	NOUN
ejpam-3176	268	24	group	group	NOUN
ejpam-3176	268	25	of	of	ADP
ejpam-3176	268	26	c1	c1	PROPN
ejpam-3176	268	27	and	and	CCONJ
ejpam-3176	268	28	c2	c2	VERB
ejpam-3176	268	29	the	the	DET
ejpam-3176	268	30	efficiency	efficiency	NOUN
ejpam-3176	268	31	of	of	ADP
ejpam-3176	268	32	the	the	DET
ejpam-3176	268	33	discriminate	discriminate	ADJ
ejpam-3176	268	34	estimates	estimate	NOUN
ejpam-3176	268	35	compared	compare	VERB
ejpam-3176	268	36	with	with	ADP
ejpam-3176	268	37	the	the	DET
ejpam-3176	268	38	other	other	ADJ
ejpam-3176	268	39	two	two	NUM
ejpam-3176	268	40	estimators	estimator	NOUN
ejpam-3176	268	41	.	.	PUNCT
ejpam-3176	269	1	in	in	ADP
ejpam-3176	269	2	the	the	DET
ejpam-3176	269	3	two	two	NUM
ejpam-3176	269	4	versions	version	NOUN
ejpam-3176	269	5	,	,	PUNCT
ejpam-3176	269	6	the	the	DET
ejpam-3176	269	7	detmcd	detmcd	NOUN
ejpam-3176	269	8	algorithm	algorithm	NOUN
ejpam-3176	269	9	obtains	obtain	VERB
ejpam-3176	269	10	more	more	ADV
ejpam-3176	269	11	accurate	accurate	ADJ
ejpam-3176	269	12	values	value	NOUN
ejpam-3176	269	13	for	for	ADP
ejpam-3176	269	14	mp	mp	PROPN
ejpam-3176	269	15	.	.	PUNCT
ejpam-3176	270	1	the	the	DET
ejpam-3176	270	2	detmcd	detmcd	PROPN
ejpam-3176	270	3	estimator	estimator	NOUN
ejpam-3176	270	4	values	value	NOUN
ejpam-3176	270	5	are	be	AUX
ejpam-3176	270	6	more	more	ADV
ejpam-3176	270	7	accurate	accurate	ADJ
ejpam-3176	270	8	and	and	CCONJ
ejpam-3176	270	9	less	less	ADJ
ejpam-3176	270	10	than	than	ADP
ejpam-3176	270	11	10	10	NUM
ejpam-3176	270	12	%	%	NOUN
ejpam-3176	270	13	for	for	ADP
ejpam-3176	270	14	all	all	DET
ejpam-3176	270	15	the	the	DET
ejpam-3176	270	16	cases	case	NOUN
ejpam-3176	270	17	in	in	ADP
ejpam-3176	270	18	the	the	DET
ejpam-3176	270	19	simulation	simulation	NOUN
ejpam-3176	270	20	.	.	PUNCT
ejpam-3176	271	1	the	the	DET
ejpam-3176	271	2	data	datum	NOUN
ejpam-3176	271	3	in	in	ADP
ejpam-3176	271	4	case	case	NOUN
ejpam-3176	271	5	a	a	DET
ejpam-3176	271	6	contain	contain	VERB
ejpam-3176	271	7	an	an	DET
ejpam-3176	271	8	uncontaminated	uncontaminated	ADJ
ejpam-3176	271	9	dataset	dataset	NOUN
ejpam-3176	271	10	and	and	CCONJ
ejpam-3176	271	11	the	the	DET
ejpam-3176	271	12	estimated	estimate	VERB
ejpam-3176	271	13	values	value	NOUN
ejpam-3176	271	14	are	be	AUX
ejpam-3176	271	15	comparable	comparable	ADJ
ejpam-3176	271	16	for	for	ADP
ejpam-3176	271	17	both	both	DET
ejpam-3176	271	18	versions	version	NOUN
ejpam-3176	271	19	performance	performance	NOUN
ejpam-3176	271	20	groups	group	NOUN
ejpam-3176	271	21	,	,	PUNCT
ejpam-3176	271	22	except	except	SCONJ
ejpam-3176	271	23	the	the	DET
ejpam-3176	271	24	mwcd	mwcd	NOUN
ejpam-3176	271	25	for	for	ADP
ejpam-3176	271	26	fastmcd	fastmcd	NOUN
ejpam-3176	271	27	and	and	CCONJ
ejpam-3176	271	28	rfch	rfch	NOUN
ejpam-3176	271	29	.	.	PUNCT
ejpam-3176	272	1	the	the	DET
ejpam-3176	272	2	datasets	dataset	NOUN
ejpam-3176	272	3	for	for	ADP
ejpam-3176	272	4	cases	case	NOUN
ejpam-3176	272	5	b	b	NOUN
ejpam-3176	272	6	,	,	PUNCT
ejpam-3176	272	7	c	c	NOUN
ejpam-3176	272	8	,	,	PUNCT
ejpam-3176	272	9	d	d	NOUN
ejpam-3176	272	10	,	,	PUNCT
ejpam-3176	272	11	and	and	CCONJ
ejpam-3176	272	12	e	e	NOUN
ejpam-3176	272	13	are	be	AUX
ejpam-3176	272	14	generated	generate	VERB
ejpam-3176	272	15	from	from	ADP
ejpam-3176	272	16	contaminated	contaminate	VERB
ejpam-3176	272	17	datasets	dataset	NOUN
ejpam-3176	272	18	with	with	ADP
ejpam-3176	272	19	different	different	ADJ
ejpam-3176	272	20	percentages	percentage	NOUN
ejpam-3176	272	21	of	of	ADP
ejpam-3176	272	22	outliers	outlier	NOUN
ejpam-3176	272	23	for	for	ADP
ejpam-3176	272	24	each	each	DET
ejpam-3176	272	25	case	case	NOUN
ejpam-3176	272	26	and	and	CCONJ
ejpam-3176	272	27	group	group	NOUN
ejpam-3176	272	28	.	.	PUNCT
ejpam-3176	273	1	groups	group	NOUN
ejpam-3176	273	2	a	a	PRON
ejpam-3176	273	3	and	and	CCONJ
ejpam-3176	273	4	b	b	NOUN
ejpam-3176	273	5	are	be	AUX
ejpam-3176	273	6	generated	generate	VERB
ejpam-3176	273	7	with	with	ADP
ejpam-3176	273	8	the	the	DET
ejpam-3176	273	9	same	same	ADJ
ejpam-3176	273	10	number	number	NOUN
ejpam-3176	273	11	of	of	ADP
ejpam-3176	273	12	observations	observation	NOUN
ejpam-3176	273	13	.	.	PUNCT
ejpam-3176	274	1	case	case	NOUN
ejpam-3176	274	2	b	b	NOUN
ejpam-3176	274	3	is	be	AUX
ejpam-3176	274	4	generated	generate	VERB
ejpam-3176	274	5	with	with	ADP
ejpam-3176	274	6	25	25	NUM
ejpam-3176	274	7	%	%	NOUN
ejpam-3176	274	8	outlier	outlier	NOUN
ejpam-3176	274	9	observations	observation	NOUN
ejpam-3176	274	10	.	.	PUNCT
ejpam-3176	275	1	the	the	DET
ejpam-3176	275	2	outliers	outlier	NOUN
ejpam-3176	275	3	influence	influence	VERB
ejpam-3176	275	4	the	the	DET
ejpam-3176	275	5	estimator	estimator	NOUN
ejpam-3176	275	6	rules	rule	NOUN
ejpam-3176	275	7	,	,	PUNCT
ejpam-3176	275	8	where	where	SCONJ
ejpam-3176	275	9	the	the	DET
ejpam-3176	275	10	estimated	estimate	VERB
ejpam-3176	275	11	values	value	NOUN
ejpam-3176	275	12	of	of	ADP
ejpam-3176	275	13	case	case	NOUN
ejpam-3176	275	14	b	b	NOUN
ejpam-3176	275	15	are	be	AUX
ejpam-3176	275	16	0.0751	0.0751	NUM
ejpam-3176	275	17	and	and	CCONJ
ejpam-3176	275	18	0.0559	0.0559	NUM
ejpam-3176	275	19	for	for	ADP
ejpam-3176	275	20	the	the	DET
ejpam-3176	275	21	raw	raw	ADJ
ejpam-3176	275	22	and	and	CCONJ
ejpam-3176	275	23	reweighted	reweighted	ADJ
ejpam-3176	275	24	versions	version	NOUN
ejpam-3176	275	25	,	,	PUNCT
ejpam-3176	275	26	respectively	respectively	ADV
ejpam-3176	275	27	,	,	PUNCT
ejpam-3176	275	28	table	table	NOUN
ejpam-3176	275	29	1	1	NUM
ejpam-3176	275	30	:	:	PUNCT
ejpam-3176	275	31	mean	mean	VERB
ejpam-3176	275	32	µ	µ	NOUN
ejpam-3176	275	33	for	for	ADP
ejpam-3176	275	34	the	the	DET
ejpam-3176	275	35	misclassification	misclassification	NOUN
ejpam-3176	275	36	probabilities	probability	NOUN
ejpam-3176	275	37	of	of	ADP
ejpam-3176	275	38	the	the	DET
ejpam-3176	275	39	rlda	rlda	NOUN
ejpam-3176	275	40	rules	rule	NOUN
ejpam-3176	275	41	for	for	ADP
ejpam-3176	275	42	500	500	NUM
ejpam-3176	275	43	replications	replication	NOUN
ejpam-3176	275	44	based	base	VERB
ejpam-3176	275	45	on	on	ADP
ejpam-3176	275	46	the	the	DET
ejpam-3176	275	47	detmcd	detmcd	NOUN
ejpam-3176	275	48	,	,	PUNCT
ejpam-3176	275	49	fastmcd	fastmcd	NOUN
ejpam-3176	275	50	,	,	PUNCT
ejpam-3176	275	51	and	and	CCONJ
ejpam-3176	275	52	fch	fch	PROPN
ejpam-3176	275	53	algorithms	algorithm	NOUN
ejpam-3176	275	54	for	for	ADP
ejpam-3176	275	55	the	the	DET
ejpam-3176	275	56	raw	raw	ADJ
ejpam-3176	275	57	version	version	NOUN
ejpam-3176	275	58	algorithm	algorithm	NOUN
ejpam-3176	275	59	fastmcd	fastmcd	NOUN
ejpam-3176	275	60	detmcd	detmcd	PROPN
ejpam-3176	275	61	fch	fch	PROPN
ejpam-3176	275	62	version	version	PROPN
ejpam-3176	275	63	raw	raw	ADJ
ejpam-3176	275	64	raw	raw	ADJ
ejpam-3176	275	65	raw	raw	ADJ
ejpam-3176	275	66	approach	approach	NOUN
ejpam-3176	275	67	pcov	pcov	VERB
ejpam-3176	275	68	pobs	pob	NOUN
ejpam-3176	275	69	mwcd	mwcd	NOUN
ejpam-3176	275	70	pcov	pcov	VERB
ejpam-3176	275	71	pobs	pob	NOUN
ejpam-3176	275	72	mwcd	mwcd	NOUN
ejpam-3176	275	73	pcov	pcov	VERB
ejpam-3176	275	74	pobs	pob	NOUN
ejpam-3176	275	75	mwcd	mwcd	PROPN
ejpam-3176	275	76	group	group	PROPN
ejpam-3176	275	77	a	a	DET
ejpam-3176	275	78	0.1967	0.1967	NUM
ejpam-3176	275	79	0.1951	0.1951	NUM
ejpam-3176	275	80	0.2662	0.2662	NUM
ejpam-3176	275	81	0.0103	0.0103	NUM
ejpam-3176	275	82	0.0211	0.0211	NUM
ejpam-3176	275	83	0.0915	0.0915	NUM
ejpam-3176	275	84	0.2151	0.2151	NUM
ejpam-3176	275	85	0.1764	0.1764	NUM
ejpam-3176	275	86	0.2235	0.2235	NUM
ejpam-3176	275	87	group	group	NOUN
ejpam-3176	275	88	b	b	PROPN
ejpam-3176	275	89	0.3145	0.3145	NUM
ejpam-3176	275	90	0.3135	0.3135	NUM
ejpam-3176	275	91	0.1314	0.1314	NUM
ejpam-3176	275	92	0.0883	0.0883	NUM
ejpam-3176	275	93	0.0751	0.0751	NUM
ejpam-3176	275	94	0.1214	0.1214	NUM
ejpam-3176	275	95	0.2945	0.2945	NUM
ejpam-3176	275	96	0.3167	0.3167	NUM
ejpam-3176	275	97	0.1736	0.1736	NUM
ejpam-3176	275	98	group	group	NOUN
ejpam-3176	275	99	c	c	NOUN
ejpam-3176	275	100	0.3651	0.3651	NUM
ejpam-3176	275	101	0.3626	0.3626	NUM
ejpam-3176	275	102	0.493	0.493	NUM
ejpam-3176	275	103	0.0998	0.0998	NUM
ejpam-3176	275	104	0.102	0.102	NUM
ejpam-3176	275	105	0.1155	0.1155	NUM
ejpam-3176	275	106	0.3476	0.3476	NUM
ejpam-3176	275	107	0.3678	0.3678	NUM
ejpam-3176	275	108	0.3987	0.3987	NUM
ejpam-3176	275	109	group	group	NOUN
ejpam-3176	275	110	d	d	NOUN
ejpam-3176	275	111	0.2785	0.2785	NUM
ejpam-3176	275	112	0.2767	0.2767	NUM
ejpam-3176	275	113	0.512	0.512	NUM
ejpam-3176	275	114	0.0963	0.0963	NUM
ejpam-3176	275	115	0.0998	0.0998	NUM
ejpam-3176	275	116	0.1089	0.1089	NUM
ejpam-3176	275	117	0.3164	0.3164	NUM
ejpam-3176	275	118	0.3023	0.3023	NUM
ejpam-3176	275	119	0.4728	0.4728	NUM
ejpam-3176	275	120	group	group	NOUN
ejpam-3176	275	121	e	e	PROPN
ejpam-3176	275	122	0.2965	0.2965	NUM
ejpam-3176	275	123	0.2944	0.2944	NUM
ejpam-3176	275	124	0.3678	0.3678	NUM
ejpam-3176	275	125	0.0247	0.0247	NUM
ejpam-3176	275	126	0.0499	0.0499	NUM
ejpam-3176	275	127	0.1169	0.1169	NUM
ejpam-3176	275	128	0.2993	0.2993	NUM
ejpam-3176	275	129	0.3284	0.3284	NUM
ejpam-3176	275	130	0.3786	0.3786	NUM
ejpam-3176	275	131	mufda	mufda	NOUN
ejpam-3176	275	132	j.	j.	PROPN
ejpam-3176	275	133	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	275	134	,	,	PUNCT
ejpam-3176	275	135	taha	taha	PROPN
ejpam-3176	275	136	radwan	radwan	PROPN
ejpam-3176	275	137	and	and	CCONJ
ejpam-3176	275	138	khalid	khalid	PROPN
ejpam-3176	275	139	abunawas	abunawas	PROPN
ejpam-3176	275	140	/	/	SYM
ejpam-3176	275	141	eur	eur	PROPN
ejpam-3176	275	142	.	.	PUNCT
ejpam-3176	276	1	j.	j.	PROPN
ejpam-3176	276	2	pure	pure	PROPN
ejpam-3176	276	3	appl	appl	PROPN
ejpam-3176	276	4	.	.	PROPN
ejpam-3176	276	5	math	math	PROPN
ejpam-3176	276	6	,	,	PUNCT
ejpam-3176	276	7	11	11	NUM
ejpam-3176	276	8	(	(	PUNCT
ejpam-3176	276	9	1	1	NUM
ejpam-3176	276	10	)	)	PUNCT
ejpam-3176	276	11	(	(	PUNCT
ejpam-3176	276	12	2018	2018	NUM
ejpam-3176	276	13	)	)	PUNCT
ejpam-3176	276	14	,	,	PUNCT
ejpam-3176	276	15	284	284	NUM
ejpam-3176	276	16	-	-	SYM
ejpam-3176	276	17	298	298	NUM
ejpam-3176	276	18	294	294	NUM
ejpam-3176	276	19	table	table	NOUN
ejpam-3176	276	20	2	2	NUM
ejpam-3176	276	21	:	:	PUNCT
ejpam-3176	276	22	mean	mean	VERB
ejpam-3176	276	23	µ	µ	NOUN
ejpam-3176	276	24	for	for	ADP
ejpam-3176	276	25	the	the	DET
ejpam-3176	276	26	misclassification	misclassification	NOUN
ejpam-3176	276	27	probabilities	probability	NOUN
ejpam-3176	276	28	of	of	ADP
ejpam-3176	276	29	the	the	DET
ejpam-3176	276	30	rlda	rlda	NOUN
ejpam-3176	276	31	rules	rule	NOUN
ejpam-3176	276	32	for	for	ADP
ejpam-3176	276	33	500	500	NUM
ejpam-3176	276	34	replications	replication	NOUN
ejpam-3176	276	35	based	base	VERB
ejpam-3176	276	36	on	on	ADP
ejpam-3176	276	37	the	the	DET
ejpam-3176	276	38	detmcd	detmcd	NOUN
ejpam-3176	276	39	,	,	PUNCT
ejpam-3176	276	40	fastmcd	fastmcd	NOUN
ejpam-3176	276	41	,	,	PUNCT
ejpam-3176	276	42	and	and	CCONJ
ejpam-3176	276	43	fch	fch	PROPN
ejpam-3176	276	44	algorithms	algorithm	NOUN
ejpam-3176	276	45	for	for	ADP
ejpam-3176	276	46	the	the	DET
ejpam-3176	276	47	reweighted	reweighte	VERB
ejpam-3176	276	48	version	version	NOUN
ejpam-3176	276	49	algorithm	algorithm	NOUN
ejpam-3176	276	50	fastmcd	fastmcd	NOUN
ejpam-3176	276	51	detmcd	detmcd	PROPN
ejpam-3176	276	52	rfch	rfch	NOUN
ejpam-3176	276	53	version	version	NOUN
ejpam-3176	276	54	reweighted	reweighte	VERB
ejpam-3176	276	55	reweighted	reweighte	VERB
ejpam-3176	276	56	reweighted	reweighte	VERB
ejpam-3176	276	57	approach	approach	NOUN
ejpam-3176	276	58	pcovw	pcovw	PROPN
ejpam-3176	276	59	pobsw	pobsw	NOUN
ejpam-3176	276	60	mwcdw	mwcdw	NOUN
ejpam-3176	276	61	pcovw	pcovw	VERB
ejpam-3176	276	62	pobsw	pobsw	ADJ
ejpam-3176	276	63	mwcdw	mwcdw	NOUN
ejpam-3176	276	64	pcovw	pcovw	VERB
ejpam-3176	276	65	pobsw	pobsw	PROPN
ejpam-3176	276	66	mwcdw	mwcdw	PROPN
ejpam-3176	276	67	group	group	NOUN
ejpam-3176	276	68	a	a	DET
ejpam-3176	276	69	0.1954	0.1954	NUM
ejpam-3176	276	70	0.1956	0.1956	NUM
ejpam-3176	276	71	0.3937	0.3937	NUM
ejpam-3176	276	72	0.0197	0.0197	NUM
ejpam-3176	276	73	0.0089	0.0089	NUM
ejpam-3176	276	74	0.0923	0.0923	NUM
ejpam-3176	276	75	0.2084	0.2084	NUM
ejpam-3176	276	76	0.1934	0.1934	NUM
ejpam-3176	276	77	0.3826	0.3826	NUM
ejpam-3176	276	78	group	group	NOUN
ejpam-3176	276	79	b	b	PROPN
ejpam-3176	277	1	0.3188	0.3188	NUM
ejpam-3176	277	2	0.3192	0.3192	NUM
ejpam-3176	277	3	0.3227	0.3227	NUM
ejpam-3176	277	4	0.0751	0.0751	NUM
ejpam-3176	277	5	0.0559	0.0559	NUM
ejpam-3176	277	6	0.1246	0.1246	NUM
ejpam-3176	277	7	0.3385	0.3385	NUM
ejpam-3176	277	8	0.3248	0.3248	NUM
ejpam-3176	277	9	0.3354	0.3354	NUM
ejpam-3176	277	10	group	group	NOUN
ejpam-3176	277	11	c	c	NOUN
ejpam-3176	277	12	0.3703	0.3703	NUM
ejpam-3176	277	13	0.3713	0.3713	NUM
ejpam-3176	277	14	0.5351	0.5351	NUM
ejpam-3176	277	15	0.0835	0.0835	NUM
ejpam-3176	277	16	0.076	0.076	NUM
ejpam-3176	277	17	0.1185	0.1185	NUM
ejpam-3176	277	18	0.4073	0.4073	NUM
ejpam-3176	277	19	0.3943	0.3943	NUM
ejpam-3176	277	20	0.5324	0.5324	NUM
ejpam-3176	277	21	group	group	NOUN
ejpam-3176	277	22	d	d	PROPN
ejpam-3176	277	23	0.2805	0.2805	NUM
ejpam-3176	277	24	0.2812	0.2812	NUM
ejpam-3176	277	25	0.5405	0.5405	NUM
ejpam-3176	277	26	0.082	0.082	NUM
ejpam-3176	277	27	0.0782	0.0782	NUM
ejpam-3176	277	28	0.1073	0.1073	NUM
ejpam-3176	277	29	0.2854	0.2854	NUM
ejpam-3176	277	30	0.2793	0.2793	NUM
ejpam-3176	277	31	0.5523	0.5523	NUM
ejpam-3176	277	32	group	group	NOUN
ejpam-3176	277	33	e	e	NOUN
ejpam-3176	277	34	0.2907	0.2907	NUM
ejpam-3176	277	35	0.2908	0.2908	NUM
ejpam-3176	277	36	0.3699	0.3699	NUM
ejpam-3176	277	37	0.0237	0.0237	NUM
ejpam-3176	277	38	0.0232	0.0232	NUM
ejpam-3176	277	39	0.1202	0.1202	NUM
ejpam-3176	277	40	0.3054	0.3054	NUM
ejpam-3176	277	41	0.3094	0.3094	NUM
ejpam-3176	277	42	0.3893	0.3893	NUM
ejpam-3176	277	43	based	base	VERB
ejpam-3176	277	44	on	on	ADP
ejpam-3176	277	45	detmcd	detmcd	NOUN
ejpam-3176	277	46	.	.	PUNCT
ejpam-3176	278	1	the	the	DET
ejpam-3176	278	2	estimated	estimate	VERB
ejpam-3176	278	3	values	value	NOUN
ejpam-3176	278	4	for	for	ADP
ejpam-3176	278	5	cases	case	NOUN
ejpam-3176	278	6	a	a	DET
ejpam-3176	278	7	,	,	PUNCT
ejpam-3176	278	8	b	b	NOUN
ejpam-3176	278	9	,	,	PUNCT
ejpam-3176	278	10	c	c	NOUN
ejpam-3176	278	11	,	,	PUNCT
ejpam-3176	278	12	d	d	NOUN
ejpam-3176	278	13	,	,	PUNCT
ejpam-3176	278	14	and	and	CCONJ
ejpam-3176	278	15	e	e	NOUN
ejpam-3176	278	16	are	be	AUX
ejpam-3176	278	17	comparable	comparable	ADJ
ejpam-3176	278	18	,	,	PUNCT
ejpam-3176	278	19	except	except	SCONJ
ejpam-3176	278	20	for	for	ADP
ejpam-3176	278	21	case	case	NOUN
ejpam-3176	278	22	a	a	PRON
ejpam-3176	278	23	,	,	PUNCT
ejpam-3176	278	24	which	which	PRON
ejpam-3176	278	25	is	be	AUX
ejpam-3176	278	26	determined	determine	VERB
ejpam-3176	278	27	to	to	PART
ejpam-3176	278	28	have	have	VERB
ejpam-3176	278	29	the	the	DET
ejpam-3176	278	30	best	good	ADJ
ejpam-3176	278	31	value	value	NOUN
ejpam-3176	278	32	of	of	ADP
ejpam-3176	278	33	0.0089	0.0089	NUM
ejpam-3176	278	34	for	for	ADP
ejpam-3176	278	35	the	the	DET
ejpam-3176	278	36	reweighted	reweighte	VERB
ejpam-3176	278	37	version	version	NOUN
ejpam-3176	278	38	using	use	VERB
ejpam-3176	278	39	the	the	DET
ejpam-3176	278	40	pobs	pobs	NOUN
ejpam-3176	278	41	approach	approach	NOUN
ejpam-3176	278	42	.	.	PUNCT
ejpam-3176	279	1	from	from	ADP
ejpam-3176	279	2	all	all	DET
ejpam-3176	279	3	the	the	DET
ejpam-3176	279	4	contaminated	contaminate	VERB
ejpam-3176	279	5	datasets	dataset	NOUN
ejpam-3176	279	6	and	and	CCONJ
ejpam-3176	279	7	cases	case	NOUN
ejpam-3176	279	8	,	,	PUNCT
ejpam-3176	279	9	e	e	X
ejpam-3176	279	10	is	be	AUX
ejpam-3176	279	11	the	the	DET
ejpam-3176	279	12	most	most	ADV
ejpam-3176	279	13	accurate	accurate	ADJ
ejpam-3176	279	14	because	because	SCONJ
ejpam-3176	279	15	it	it	PRON
ejpam-3176	279	16	is	be	AUX
ejpam-3176	279	17	close	close	ADJ
ejpam-3176	279	18	to	to	ADP
ejpam-3176	279	19	the	the	DET
ejpam-3176	279	20	uncontaminated	uncontaminated	ADJ
ejpam-3176	279	21	dataset	dataset	NOUN
ejpam-3176	279	22	in	in	ADP
ejpam-3176	279	23	case	case	NOUN
ejpam-3176	279	24	a	a	PRON
ejpam-3176	279	25	,	,	PUNCT
ejpam-3176	279	26	with	with	ADP
ejpam-3176	279	27	a	a	DET
ejpam-3176	279	28	difference	difference	NOUN
ejpam-3176	279	29	of	of	ADP
ejpam-3176	279	30	approximately	approximately	ADV
ejpam-3176	279	31	0.0144	0.0144	NUM
ejpam-3176	279	32	and	and	CCONJ
ejpam-3176	279	33	0.0143	0.0143	NUM
ejpam-3176	279	34	for	for	ADP
ejpam-3176	279	35	the	the	DET
ejpam-3176	279	36	raw	raw	ADJ
ejpam-3176	279	37	and	and	CCONJ
ejpam-3176	279	38	reweighted	reweighted	ADJ
ejpam-3176	279	39	versions	version	NOUN
ejpam-3176	279	40	,	,	PUNCT
ejpam-3176	279	41	respectively	respectively	ADV
ejpam-3176	279	42	.	.	PUNCT
ejpam-3176	280	1	by	by	ADP
ejpam-3176	280	2	contrast	contrast	NOUN
ejpam-3176	280	3	,	,	PUNCT
ejpam-3176	280	4	the	the	DET
ejpam-3176	280	5	mwcd	mwcd	NOUN
ejpam-3176	280	6	approach	approach	NOUN
ejpam-3176	280	7	performs	perform	VERB
ejpam-3176	280	8	poorly	poorly	ADV
ejpam-3176	280	9	with	with	ADP
ejpam-3176	280	10	the	the	DET
ejpam-3176	280	11	detmcd	detmcd	NOUN
ejpam-3176	280	12	algorithm	algorithm	NOUN
ejpam-3176	280	13	,	,	PUNCT
ejpam-3176	280	14	whereas	whereas	SCONJ
ejpam-3176	280	15	the	the	DET
ejpam-3176	280	16	other	other	ADJ
ejpam-3176	280	17	two	two	NUM
ejpam-3176	280	18	approaches	approach	NOUN
ejpam-3176	280	19	perform	perform	VERB
ejpam-3176	280	20	better	well	ADV
ejpam-3176	280	21	in	in	ADP
ejpam-3176	280	22	both	both	DET
ejpam-3176	280	23	versions	version	NOUN
ejpam-3176	280	24	.	.	PUNCT
ejpam-3176	281	1	the	the	DET
ejpam-3176	281	2	results	result	NOUN
ejpam-3176	281	3	of	of	ADP
ejpam-3176	281	4	fch	fch	PROPN
ejpam-3176	281	5	exhibit	exhibit	VERB
ejpam-3176	281	6	better	well	ADJ
ejpam-3176	281	7	performance	performance	NOUN
ejpam-3176	281	8	in	in	ADP
ejpam-3176	281	9	case	case	NOUN
ejpam-3176	281	10	a	a	PRON
ejpam-3176	281	11	,	,	PUNCT
ejpam-3176	281	12	which	which	PRON
ejpam-3176	281	13	indicates	indicate	VERB
ejpam-3176	281	14	better	well	ADJ
ejpam-3176	281	15	performance	performance	NOUN
ejpam-3176	281	16	at	at	ADP
ejpam-3176	281	17	the	the	DET
ejpam-3176	281	18	uncontaminated	uncontaminated	ADJ
ejpam-3176	281	19	dataset	dataset	NOUN
ejpam-3176	281	20	in	in	ADP
ejpam-3176	281	21	the	the	DET
ejpam-3176	281	22	raw	raw	ADJ
ejpam-3176	281	23	version	version	NOUN
ejpam-3176	281	24	,	,	PUNCT
ejpam-3176	281	25	but	but	CCONJ
ejpam-3176	281	26	fastmcd	fastmcd	NOUN
ejpam-3176	281	27	was	be	AUX
ejpam-3176	281	28	better	well	ADJ
ejpam-3176	281	29	at	at	ADP
ejpam-3176	281	30	the	the	DET
ejpam-3176	281	31	reweighted	reweighte	VERB
ejpam-3176	281	32	version	version	NOUN
ejpam-3176	281	33	for	for	ADP
ejpam-3176	281	34	all	all	DET
ejpam-3176	281	35	the	the	DET
ejpam-3176	281	36	approaches	approach	NOUN
ejpam-3176	281	37	.	.	PUNCT
ejpam-3176	282	1	by	by	ADP
ejpam-3176	282	2	contrast	contrast	NOUN
ejpam-3176	282	3	,	,	PUNCT
ejpam-3176	282	4	the	the	DET
ejpam-3176	282	5	performances	performance	NOUN
ejpam-3176	282	6	of	of	ADP
ejpam-3176	282	7	the	the	DET
ejpam-3176	282	8	pcov	pcov	NOUN
ejpam-3176	282	9	and	and	CCONJ
ejpam-3176	282	10	pobs	pob	VERB
ejpam-3176	282	11	approaches	approach	NOUN
ejpam-3176	282	12	are	be	AUX
ejpam-3176	282	13	approximately	approximately	ADV
ejpam-3176	282	14	close	close	ADJ
ejpam-3176	282	15	to	to	ADP
ejpam-3176	282	16	each	each	DET
ejpam-3176	282	17	other	other	ADJ
ejpam-3176	282	18	at	at	ADP
ejpam-3176	282	19	the	the	DET
ejpam-3176	282	20	fastmcd	fastmcd	NOUN
ejpam-3176	282	21	and	and	CCONJ
ejpam-3176	282	22	fch	fch	NOUN
ejpam-3176	282	23	estimators	estimator	NOUN
ejpam-3176	282	24	for	for	ADP
ejpam-3176	282	25	the	the	DET
ejpam-3176	282	26	raw	raw	ADJ
ejpam-3176	282	27	and	and	CCONJ
ejpam-3176	282	28	reweighted	reweighted	ADJ
ejpam-3176	282	29	versions	version	NOUN
ejpam-3176	282	30	,	,	PUNCT
ejpam-3176	282	31	but	but	CCONJ
ejpam-3176	282	32	mwcd	mwcd	NOUN
ejpam-3176	282	33	performs	perform	VERB
ejpam-3176	282	34	poorly	poorly	ADV
ejpam-3176	282	35	compared	compare	VERB
ejpam-3176	282	36	with	with	ADP
ejpam-3176	282	37	the	the	DET
ejpam-3176	282	38	other	other	ADJ
ejpam-3176	282	39	approaches	approach	NOUN
ejpam-3176	282	40	.	.	PUNCT
ejpam-3176	283	1	figure	figure	VERB
ejpam-3176	283	2	2	2	NUM
ejpam-3176	283	3	:	:	SYM
ejpam-3176	283	4	100	100	NUM
ejpam-3176	283	5	replications	replication	NOUN
ejpam-3176	283	6	for	for	ADP
ejpam-3176	283	7	det	det	PROPN
ejpam-3176	283	8	-	-	PUNCT
ejpam-3176	283	9	mcd	mcd	PROPN
ejpam-3176	283	10	estimator	estimator	NOUN
ejpam-3176	283	11	of	of	ADP
ejpam-3176	283	12	group	group	PROPN
ejpam-3176	283	13	b	b	PROPN
ejpam-3176	283	14	mufda	mufda	PROPN
ejpam-3176	283	15	j.	j.	PROPN
ejpam-3176	283	16	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	283	17	,	,	PUNCT
ejpam-3176	283	18	taha	taha	PROPN
ejpam-3176	283	19	radwan	radwan	PROPN
ejpam-3176	283	20	and	and	CCONJ
ejpam-3176	283	21	khalid	khalid	PROPN
ejpam-3176	283	22	abunawas	abunawas	PROPN
ejpam-3176	283	23	/	/	SYM
ejpam-3176	283	24	eur	eur	PROPN
ejpam-3176	283	25	.	.	PUNCT
ejpam-3176	284	1	j.	j.	PROPN
ejpam-3176	284	2	pure	pure	PROPN
ejpam-3176	284	3	appl	appl	PROPN
ejpam-3176	284	4	.	.	PROPN
ejpam-3176	284	5	math	math	PROPN
ejpam-3176	284	6	,	,	PUNCT
ejpam-3176	284	7	11	11	NUM
ejpam-3176	284	8	(	(	PUNCT
ejpam-3176	284	9	1	1	NUM
ejpam-3176	284	10	)	)	PUNCT
ejpam-3176	284	11	(	(	PUNCT
ejpam-3176	284	12	2018	2018	NUM
ejpam-3176	284	13	)	)	PUNCT
ejpam-3176	284	14	,	,	PUNCT
ejpam-3176	284	15	284	284	NUM
ejpam-3176	284	16	-	-	SYM
ejpam-3176	284	17	298	298	NUM
ejpam-3176	284	18	295	295	NUM
ejpam-3176	284	19	table	table	NOUN
ejpam-3176	284	20	3	3	NUM
ejpam-3176	284	21	:	:	PUNCT
ejpam-3176	284	22	misclassification	misclassification	NOUN
ejpam-3176	284	23	probability	probability	NOUN
ejpam-3176	284	24	estimates	estimate	NOUN
ejpam-3176	284	25	for	for	ADP
ejpam-3176	284	26	detmcd	detmcd	NOUN
ejpam-3176	284	27	,	,	PUNCT
ejpam-3176	284	28	fastmcd	fastmcd	NOUN
ejpam-3176	284	29	,	,	PUNCT
ejpam-3176	284	30	and	and	CCONJ
ejpam-3176	284	31	fch	fch	PROPN
ejpam-3176	284	32	for	for	ADP
ejpam-3176	284	33	the	the	DET
ejpam-3176	284	34	raw	raw	ADJ
ejpam-3176	284	35	version	version	NOUN
ejpam-3176	284	36	rlda	rlda	NOUN
ejpam-3176	284	37	rules	rule	NOUN
ejpam-3176	284	38	based	base	VERB
ejpam-3176	284	39	on	on	ADP
ejpam-3176	284	40	actual	actual	ADJ
ejpam-3176	284	41	data	datum	NOUN
ejpam-3176	284	42	(	(	PUNCT
ejpam-3176	284	43	financial	financial	ADJ
ejpam-3176	284	44	ratios	ratio	NOUN
ejpam-3176	284	45	for	for	ADP
ejpam-3176	284	46	islamic	islamic	ADJ
ejpam-3176	284	47	and	and	CCONJ
ejpam-3176	284	48	conventional	conventional	ADJ
ejpam-3176	284	49	banks	bank	NOUN
ejpam-3176	284	50	in	in	ADP
ejpam-3176	284	51	malaysia	malaysia	PROPN
ejpam-3176	284	52	)	)	PUNCT
ejpam-3176	284	53	algorithm	algorithm	PROPN
ejpam-3176	284	54	fastmcd	fastmcd	NOUN
ejpam-3176	284	55	detmcd	detmcd	PROPN
ejpam-3176	284	56	fch	fch	PROPN
ejpam-3176	284	57	version	version	PROPN
ejpam-3176	284	58	raw	raw	ADJ
ejpam-3176	284	59	approach	approach	NOUN
ejpam-3176	284	60	pcov	pcov	VERB
ejpam-3176	284	61	pobs	pob	NOUN
ejpam-3176	284	62	mwcd	mwcd	NOUN
ejpam-3176	284	63	pcov	pcov	VERB
ejpam-3176	284	64	pobs	pob	NOUN
ejpam-3176	284	65	mwcd	mwcd	NOUN
ejpam-3176	284	66	pcov	pcov	VERB
ejpam-3176	284	67	pobs	pob	NOUN
ejpam-3176	284	68	mwcd	mwcd	PROPN
ejpam-3176	284	69	mp	mp	PROPN
ejpam-3176	284	70	0.214	0.214	NUM
ejpam-3176	284	71	0.1845	0.1845	NUM
ejpam-3176	284	72	0.0352	0.0352	NUM
ejpam-3176	284	73	0.0182	0.0182	NUM
ejpam-3176	284	74	0.06	0.06	NUM
ejpam-3176	284	75	0.2227	0.2227	NUM
ejpam-3176	284	76	0.217	0.217	NUM
ejpam-3176	284	77	0.1954	0.1954	NUM
ejpam-3176	284	78	0.0932	0.0932	NUM
ejpam-3176	284	79	table	table	NOUN
ejpam-3176	284	80	4	4	NUM
ejpam-3176	284	81	:	:	PUNCT
ejpam-3176	284	82	misclassification	misclassification	NOUN
ejpam-3176	284	83	probability	probability	NOUN
ejpam-3176	284	84	estimates	estimate	NOUN
ejpam-3176	284	85	for	for	ADP
ejpam-3176	284	86	detmcd	detmcd	NOUN
ejpam-3176	284	87	,	,	PUNCT
ejpam-3176	284	88	fastmcd	fastmcd	NOUN
ejpam-3176	284	89	,	,	PUNCT
ejpam-3176	284	90	and	and	CCONJ
ejpam-3176	284	91	fch	fch	PROPN
ejpam-3176	284	92	for	for	ADP
ejpam-3176	284	93	the	the	DET
ejpam-3176	284	94	reweighted	reweighte	VERB
ejpam-3176	284	95	version	version	NOUN
ejpam-3176	284	96	rlda	rlda	NOUN
ejpam-3176	284	97	rules	rule	NOUN
ejpam-3176	284	98	based	base	VERB
ejpam-3176	284	99	on	on	ADP
ejpam-3176	284	100	actual	actual	ADJ
ejpam-3176	284	101	data	datum	NOUN
ejpam-3176	284	102	(	(	PUNCT
ejpam-3176	284	103	financial	financial	ADJ
ejpam-3176	284	104	ratios	ratio	NOUN
ejpam-3176	284	105	for	for	ADP
ejpam-3176	284	106	islamic	islamic	ADJ
ejpam-3176	284	107	and	and	CCONJ
ejpam-3176	284	108	conventional	conventional	ADJ
ejpam-3176	284	109	banks	bank	NOUN
ejpam-3176	284	110	in	in	ADP
ejpam-3176	284	111	malaysia	malaysia	PROPN
ejpam-3176	284	112	)	)	PUNCT
ejpam-3176	284	113	algorithm	algorithm	PROPN
ejpam-3176	284	114	fastmcd	fastmcd	NOUN
ejpam-3176	284	115	detmcd	detmcd	PROPN
ejpam-3176	284	116	rfch	rfch	NOUN
ejpam-3176	284	117	version	version	NOUN
ejpam-3176	284	118	reweighted	reweighte	VERB
ejpam-3176	284	119	approach	approach	NOUN
ejpam-3176	284	120	pcovw	pcovw	PROPN
ejpam-3176	284	121	pobsw	pobsw	NOUN
ejpam-3176	284	122	mwcdw	mwcdw	NOUN
ejpam-3176	284	123	pcovw	pcovw	VERB
ejpam-3176	284	124	pobsw	pobsw	ADJ
ejpam-3176	284	125	mwcdw	mwcdw	NOUN
ejpam-3176	284	126	pcovw	pcovw	VERB
ejpam-3176	284	127	pobsw	pobsw	ADJ
ejpam-3176	284	128	mwcdw	mwcdw	NOUN
ejpam-3176	284	129	mp	mp	PROPN
ejpam-3176	284	130	0.1914	0.1914	NUM
ejpam-3176	284	131	0.1745	0.1745	NUM
ejpam-3176	284	132	0.0302	0.0302	NUM
ejpam-3176	285	1	0.0153	0.0153	NUM
ejpam-3176	285	2	0.004	0.004	NUM
ejpam-3176	285	3	0.2246	0.2246	NUM
ejpam-3176	285	4	0.1923	0.1923	NUM
ejpam-3176	285	5	0.1976	0.1976	NUM
ejpam-3176	285	6	0.0763	0.0763	NUM
ejpam-3176	285	7	figure	figure	NOUN
ejpam-3176	285	8	2	2	NUM
ejpam-3176	285	9	presents	present	VERB
ejpam-3176	285	10	the	the	DET
ejpam-3176	285	11	replications	replication	NOUN
ejpam-3176	285	12	for	for	ADP
ejpam-3176	285	13	the	the	DET
ejpam-3176	285	14	first	first	ADJ
ejpam-3176	285	15	100	100	NUM
ejpam-3176	285	16	times	time	NOUN
ejpam-3176	285	17	out	out	ADP
ejpam-3176	285	18	of	of	ADP
ejpam-3176	285	19	the	the	DET
ejpam-3176	285	20	500	500	NUM
ejpam-3176	285	21	replications	replication	NOUN
ejpam-3176	285	22	.	.	PUNCT
ejpam-3176	286	1	the	the	DET
ejpam-3176	286	2	figure	figure	NOUN
ejpam-3176	286	3	describes	describe	VERB
ejpam-3176	286	4	the	the	DET
ejpam-3176	286	5	values	value	NOUN
ejpam-3176	286	6	of	of	ADP
ejpam-3176	286	7	the	the	DET
ejpam-3176	286	8	misclassification	misclassification	NOUN
ejpam-3176	286	9	probabilities	probability	NOUN
ejpam-3176	286	10	of	of	ADP
ejpam-3176	286	11	the	the	DET
ejpam-3176	286	12	discriminate	discriminate	ADJ
ejpam-3176	286	13	rules	rule	NOUN
ejpam-3176	286	14	based	base	VERB
ejpam-3176	286	15	on	on	ADP
ejpam-3176	286	16	the	the	DET
ejpam-3176	286	17	detmcd	detmcd	NOUN
ejpam-3176	286	18	estimator	estimator	NOUN
ejpam-3176	286	19	.	.	PUNCT
ejpam-3176	287	1	the	the	DET
ejpam-3176	287	2	disparity	disparity	NOUN
ejpam-3176	287	3	of	of	ADP
ejpam-3176	287	4	the	the	DET
ejpam-3176	287	5	misclassification	misclassification	NOUN
ejpam-3176	287	6	values	value	NOUN
ejpam-3176	287	7	is	be	AUX
ejpam-3176	287	8	shown	show	VERB
ejpam-3176	287	9	,	,	PUNCT
ejpam-3176	287	10	particularly	particularly	ADV
ejpam-3176	287	11	for	for	ADP
ejpam-3176	287	12	the	the	DET
ejpam-3176	287	13	raw	raw	ADJ
ejpam-3176	287	14	version	version	NOUN
ejpam-3176	287	15	,	,	PUNCT
ejpam-3176	287	16	where	where	SCONJ
ejpam-3176	287	17	the	the	DET
ejpam-3176	287	18	difference	difference	NOUN
ejpam-3176	287	19	between	between	ADP
ejpam-3176	287	20	the	the	DET
ejpam-3176	287	21	minimum	minimum	ADJ
ejpam-3176	287	22	and	and	CCONJ
ejpam-3176	287	23	maximum	maximum	ADJ
ejpam-3176	287	24	values	value	NOUN
ejpam-3176	287	25	is	be	AUX
ejpam-3176	287	26	considerable	considerable	ADJ
ejpam-3176	287	27	compared	compare	VERB
ejpam-3176	287	28	with	with	ADP
ejpam-3176	287	29	that	that	PRON
ejpam-3176	287	30	for	for	ADP
ejpam-3176	287	31	the	the	DET
ejpam-3176	287	32	reweighted	reweighted	ADJ
ejpam-3176	287	33	version	version	NOUN
ejpam-3176	287	34	.	.	PUNCT
ejpam-3176	288	1	in	in	ADP
ejpam-3176	288	2	terms	term	NOUN
ejpam-3176	288	3	of	of	ADP
ejpam-3176	288	4	the	the	DET
ejpam-3176	288	5	accuracy	accuracy	NOUN
ejpam-3176	288	6	of	of	ADP
ejpam-3176	288	7	the	the	DET
ejpam-3176	288	8	versions	version	NOUN
ejpam-3176	288	9	of	of	ADP
ejpam-3176	288	10	the	the	DET
ejpam-3176	288	11	detmcd	detmcd	NOUN
ejpam-3176	288	12	estimator	estimator	NOUN
ejpam-3176	288	13	,	,	PUNCT
ejpam-3176	288	14	the	the	DET
ejpam-3176	288	15	reweighted	reweighte	VERB
ejpam-3176	288	16	version	version	NOUN
ejpam-3176	288	17	is	be	AUX
ejpam-3176	288	18	more	more	ADV
ejpam-3176	288	19	accurate	accurate	ADJ
ejpam-3176	288	20	and	and	CCONJ
ejpam-3176	288	21	efficient	efficient	ADJ
ejpam-3176	288	22	than	than	ADP
ejpam-3176	288	23	the	the	DET
ejpam-3176	288	24	raw	raw	ADJ
ejpam-3176	288	25	version	version	NOUN
ejpam-3176	288	26	.	.	PUNCT
ejpam-3176	289	1	9	9	X
ejpam-3176	289	2	.	.	NOUN
ejpam-3176	289	3	example	example	NOUN
ejpam-3176	289	4	of	of	ADP
ejpam-3176	289	5	actual	actual	ADJ
ejpam-3176	289	6	data	datum	NOUN
ejpam-3176	289	7	:	:	PUNCT
ejpam-3176	289	8	islamic	islamic	ADJ
ejpam-3176	289	9	and	and	CCONJ
ejpam-3176	289	10	conventional	conventional	ADJ
ejpam-3176	289	11	banks	bank	NOUN
ejpam-3176	289	12	in	in	ADP
ejpam-3176	289	13	malaysia	malaysia	PROPN
ejpam-3176	289	14	the	the	DET
ejpam-3176	289	15	financial	financial	ADJ
ejpam-3176	289	16	ratios	ratio	NOUN
ejpam-3176	289	17	of	of	ADP
ejpam-3176	289	18	islamic	islamic	ADJ
ejpam-3176	289	19	and	and	CCONJ
ejpam-3176	289	20	conventional	conventional	ADJ
ejpam-3176	289	21	banks	bank	NOUN
ejpam-3176	289	22	in	in	ADP
ejpam-3176	289	23	malaysia	malaysia	PROPN
ejpam-3176	289	24	are	be	AUX
ejpam-3176	289	25	used	use	VERB
ejpam-3176	289	26	as	as	ADP
ejpam-3176	289	27	actual	actual	ADJ
ejpam-3176	289	28	data	datum	NOUN
ejpam-3176	289	29	.	.	PUNCT
ejpam-3176	290	1	a	a	DET
ejpam-3176	290	2	total	total	NOUN
ejpam-3176	290	3	of	of	ADP
ejpam-3176	290	4	271	271	NUM
ejpam-3176	290	5	observations	observation	NOUN
ejpam-3176	290	6	from	from	ADP
ejpam-3176	290	7	banks	bank	NOUN
ejpam-3176	290	8	for	for	ADP
ejpam-3176	290	9	the	the	DET
ejpam-3176	290	10	period	period	NOUN
ejpam-3176	290	11	of	of	ADP
ejpam-3176	290	12	2003–2011	2003–2011	NUM
ejpam-3176	290	13	are	be	AUX
ejpam-3176	290	14	used	use	VERB
ejpam-3176	290	15	,	,	PUNCT
ejpam-3176	290	16	where	where	SCONJ
ejpam-3176	290	17	96	96	NUM
ejpam-3176	290	18	observations	observation	NOUN
ejpam-3176	290	19	are	be	AUX
ejpam-3176	290	20	from	from	ADP
ejpam-3176	290	21	islamic	islamic	ADJ
ejpam-3176	290	22	banks	bank	NOUN
ejpam-3176	290	23	and	and	CCONJ
ejpam-3176	290	24	175	175	NUM
ejpam-3176	290	25	observations	observation	NOUN
ejpam-3176	290	26	are	be	AUX
ejpam-3176	290	27	from	from	ADP
ejpam-3176	290	28	conventional	conventional	ADJ
ejpam-3176	290	29	banks	bank	NOUN
ejpam-3176	290	30	.	.	PUNCT
ejpam-3176	291	1	the	the	DET
ejpam-3176	291	2	dataset	dataset	NOUN
ejpam-3176	291	3	has	have	VERB
ejpam-3176	291	4	23	23	NUM
ejpam-3176	291	5	financial	financial	ADJ
ejpam-3176	291	6	ratios	ratio	NOUN
ejpam-3176	291	7	(	(	PUNCT
ejpam-3176	291	8	variables	variable	NOUN
ejpam-3176	291	9	)	)	PUNCT
ejpam-3176	291	10	.	.	PUNCT
ejpam-3176	292	1	the	the	DET
ejpam-3176	292	2	data	datum	NOUN
ejpam-3176	292	3	were	be	AUX
ejpam-3176	292	4	collected	collect	VERB
ejpam-3176	292	5	from	from	ADP
ejpam-3176	292	6	the	the	DET
ejpam-3176	292	7	bankscope	bankscope	NOUN
ejpam-3176	292	8	database	database	NOUN
ejpam-3176	292	9	,	,	PUNCT
ejpam-3176	292	10	which	which	PRON
ejpam-3176	292	11	converts	convert	VERB
ejpam-3176	292	12	financial	financial	ADJ
ejpam-3176	292	13	data	datum	NOUN
ejpam-3176	292	14	according	accord	VERB
ejpam-3176	292	15	to	to	ADP
ejpam-3176	292	16	common	common	ADJ
ejpam-3176	292	17	international	international	ADJ
ejpam-3176	292	18	standards	standard	NOUN
ejpam-3176	292	19	to	to	PART
ejpam-3176	292	20	facilitate	facilitate	VERB
ejpam-3176	292	21	comparisons	comparison	NOUN
ejpam-3176	292	22	.	.	PUNCT
ejpam-3176	293	1	rlda	rlda	NOUN
ejpam-3176	293	2	is	be	AUX
ejpam-3176	293	3	applied	apply	VERB
ejpam-3176	293	4	using	use	VERB
ejpam-3176	293	5	the	the	DET
ejpam-3176	293	6	three	three	NUM
ejpam-3176	293	7	estimators	estimator	NOUN
ejpam-3176	293	8	.	.	PUNCT
ejpam-3176	294	1	all	all	DET
ejpam-3176	294	2	the	the	DET
ejpam-3176	294	3	estimators	estimator	NOUN
ejpam-3176	294	4	have	have	AUX
ejpam-3176	294	5	been	be	AUX
ejpam-3176	294	6	used	use	VERB
ejpam-3176	294	7	with	with	ADP
ejpam-3176	294	8	the	the	DET
ejpam-3176	294	9	three	three	NUM
ejpam-3176	294	10	approaches	approach	NOUN
ejpam-3176	294	11	described	describe	VERB
ejpam-3176	294	12	in	in	ADP
ejpam-3176	294	13	the	the	DET
ejpam-3176	294	14	previous	previous	ADJ
ejpam-3176	294	15	section	section	NOUN
ejpam-3176	294	16	.	.	PUNCT
ejpam-3176	295	1	the	the	DET
ejpam-3176	295	2	results	result	NOUN
ejpam-3176	295	3	are	be	AUX
ejpam-3176	295	4	presented	present	VERB
ejpam-3176	295	5	in	in	ADP
ejpam-3176	295	6	table	table	NOUN
ejpam-3176	295	7	3	3	NUM
ejpam-3176	295	8	and	and	CCONJ
ejpam-3176	295	9	table	table	NOUN
ejpam-3176	295	10	4	4	NUM
ejpam-3176	295	11	for	for	ADP
ejpam-3176	295	12	the	the	DET
ejpam-3176	295	13	raw	raw	ADJ
ejpam-3176	295	14	and	and	CCONJ
ejpam-3176	295	15	reweighted	reweighted	ADJ
ejpam-3176	295	16	versions	version	NOUN
ejpam-3176	295	17	,	,	PUNCT
ejpam-3176	295	18	respectively	respectively	ADV
ejpam-3176	295	19	.	.	PUNCT
ejpam-3176	296	1	table	table	NOUN
ejpam-3176	296	2	3	3	NUM
ejpam-3176	296	3	and	and	CCONJ
ejpam-3176	296	4	table	table	NOUN
ejpam-3176	296	5	4	4	NUM
ejpam-3176	296	6	show	show	VERB
ejpam-3176	296	7	the	the	DET
ejpam-3176	296	8	misclassification	misclassification	NOUN
ejpam-3176	296	9	probabilities	probability	NOUN
ejpam-3176	296	10	for	for	ADP
ejpam-3176	296	11	the	the	DET
ejpam-3176	296	12	two	two	NUM
ejpam-3176	296	13	types	type	NOUN
ejpam-3176	296	14	of	of	ADP
ejpam-3176	296	15	banks	bank	NOUN
ejpam-3176	296	16	in	in	ADP
ejpam-3176	296	17	malaysia	malaysia	PROPN
ejpam-3176	296	18	.	.	PUNCT
ejpam-3176	297	1	the	the	DET
ejpam-3176	297	2	detmcd	detmcd	PROPN
ejpam-3176	297	3	estimator	estimator	NOUN
ejpam-3176	297	4	outperforms	outperform	VERB
ejpam-3176	297	5	the	the	DET
ejpam-3176	297	6	fastmcd	fastmcd	NOUN
ejpam-3176	297	7	and	and	CCONJ
ejpam-3176	297	8	fch	fch	NOUN
ejpam-3176	297	9	estimators	estimator	NOUN
ejpam-3176	297	10	by	by	ADP
ejpam-3176	297	11	a	a	DET
ejpam-3176	297	12	significant	significant	ADJ
ejpam-3176	297	13	margin	margin	NOUN
ejpam-3176	297	14	.	.	PUNCT
ejpam-3176	298	1	detmcd	detmcd	NOUN
ejpam-3176	298	2	yields	yield	NOUN
ejpam-3176	298	3	more	more	ADV
ejpam-3176	298	4	accurate	accurate	ADJ
ejpam-3176	298	5	results	result	NOUN
ejpam-3176	298	6	for	for	ADP
ejpam-3176	298	7	the	the	DET
ejpam-3176	298	8	actual	actual	ADJ
ejpam-3176	298	9	data	datum	NOUN
ejpam-3176	298	10	,	,	PUNCT
ejpam-3176	298	11	as	as	SCONJ
ejpam-3176	298	12	expected	expect	VERB
ejpam-3176	298	13	in	in	ADP
ejpam-3176	298	14	the	the	DET
ejpam-3176	298	15	simulation	simulation	NOUN
ejpam-3176	298	16	study	study	NOUN
ejpam-3176	298	17	.	.	PUNCT
ejpam-3176	299	1	however	however	ADV
ejpam-3176	299	2	,	,	PUNCT
ejpam-3176	299	3	the	the	DET
ejpam-3176	299	4	reweighted	reweighte	VERB
ejpam-3176	299	5	version	version	NOUN
ejpam-3176	299	6	confirms	confirm	VERB
ejpam-3176	299	7	the	the	DET
ejpam-3176	299	8	high	high	ADJ
ejpam-3176	299	9	performance	performance	NOUN
ejpam-3176	299	10	in	in	ADP
ejpam-3176	299	11	mp	mp	PROPN
ejpam-3176	299	12	estimation	estimation	NOUN
ejpam-3176	299	13	.	.	PUNCT
ejpam-3176	300	1	when	when	SCONJ
ejpam-3176	300	2	the	the	DET
ejpam-3176	300	3	three	three	NUM
ejpam-3176	300	4	approaches	approach	NOUN
ejpam-3176	300	5	are	be	AUX
ejpam-3176	300	6	compared	compare	VERB
ejpam-3176	300	7	under	under	ADP
ejpam-3176	300	8	the	the	DET
ejpam-3176	300	9	two	two	NUM
ejpam-3176	300	10	algorithms	algorithm	NOUN
ejpam-3176	300	11	,	,	PUNCT
ejpam-3176	300	12	the	the	DET
ejpam-3176	300	13	pcov	pcov	NOUN
ejpam-3176	300	14	and	and	CCONJ
ejpam-3176	300	15	pobs	pob	VERB
ejpam-3176	300	16	approaches	approach	NOUN
ejpam-3176	300	17	perform	perform	VERB
ejpam-3176	300	18	better	well	ADV
ejpam-3176	300	19	with	with	ADP
ejpam-3176	300	20	detmcd	detmcd	NOUN
ejpam-3176	300	21	than	than	ADP
ejpam-3176	300	22	the	the	DET
ejpam-3176	300	23	other	other	ADJ
ejpam-3176	300	24	estimators	estimator	NOUN
ejpam-3176	300	25	and	and	CCONJ
ejpam-3176	300	26	increase	increase	VERB
ejpam-3176	300	27	the	the	DET
ejpam-3176	300	28	accurate	accurate	ADJ
ejpam-3176	300	29	estimation	estimation	NOUN
ejpam-3176	300	30	of	of	ADP
ejpam-3176	300	31	mp	mp	PROPN
ejpam-3176	300	32	.	.	PROPN
ejpam-3176	300	33	mwcd	mwcd	PROPN
ejpam-3176	300	34	performs	perform	VERB
ejpam-3176	300	35	better	well	ADV
ejpam-3176	300	36	with	with	ADP
ejpam-3176	300	37	fastmcd	fastmcd	NOUN
ejpam-3176	300	38	and	and	CCONJ
ejpam-3176	300	39	fch	fch	PROPN
ejpam-3176	300	40	than	than	ADP
ejpam-3176	300	41	detmcd	detmcd	PROPN
ejpam-3176	300	42	and	and	CCONJ
ejpam-3176	300	43	makes	make	VERB
ejpam-3176	300	44	the	the	DET
ejpam-3176	300	45	discriminate	discriminate	NOUN
ejpam-3176	300	46	rules	rule	NOUN
ejpam-3176	300	47	more	more	ADV
ejpam-3176	300	48	efficient	efficient	ADJ
ejpam-3176	300	49	and	and	CCONJ
ejpam-3176	300	50	accurate	accurate	ADJ
ejpam-3176	300	51	in	in	ADP
ejpam-3176	300	52	estimating	estimate	VERB
ejpam-3176	300	53	the	the	DET
ejpam-3176	300	54	misclassification	misclassification	NOUN
ejpam-3176	300	55	probabilities	probability	NOUN
ejpam-3176	300	56	of	of	ADP
ejpam-3176	300	57	islamic	islamic	ADJ
ejpam-3176	300	58	and	and	CCONJ
ejpam-3176	300	59	conventional	conventional	ADJ
ejpam-3176	300	60	banks	bank	NOUN
ejpam-3176	300	61	.	.	PUNCT
ejpam-3176	301	1	references	reference	NOUN
ejpam-3176	301	2	296	296	NUM
ejpam-3176	301	3	10	10	NUM
ejpam-3176	301	4	.	.	PUNCT
ejpam-3176	302	1	conclusion	conclusion	NOUN
ejpam-3176	302	2	in	in	ADP
ejpam-3176	302	3	this	this	DET
ejpam-3176	302	4	study	study	NOUN
ejpam-3176	302	5	,	,	PUNCT
ejpam-3176	302	6	we	we	PRON
ejpam-3176	302	7	investigated	investigate	VERB
ejpam-3176	302	8	the	the	DET
ejpam-3176	302	9	difference	difference	NOUN
ejpam-3176	302	10	in	in	ADP
ejpam-3176	302	11	efficiency	efficiency	NOUN
ejpam-3176	302	12	among	among	ADP
ejpam-3176	302	13	three	three	NUM
ejpam-3176	302	14	estimators	estimator	NOUN
ejpam-3176	302	15	,	,	PUNCT
ejpam-3176	302	16	namely	namely	ADV
ejpam-3176	302	17	,	,	PUNCT
ejpam-3176	302	18	detmcd	detmcd	NOUN
ejpam-3176	302	19	,	,	PUNCT
ejpam-3176	302	20	fastmcd	fastmcd	NOUN
ejpam-3176	302	21	,	,	PUNCT
ejpam-3176	302	22	and	and	CCONJ
ejpam-3176	302	23	fch	fch	PROPN
ejpam-3176	302	24	,	,	PUNCT
ejpam-3176	302	25	of	of	ADP
ejpam-3176	302	26	location	location	NOUN
ejpam-3176	302	27	and	and	CCONJ
ejpam-3176	302	28	scatter	scatter	NOUN
ejpam-3176	302	29	for	for	ADP
ejpam-3176	302	30	rlda	rlda	NOUN
ejpam-3176	302	31	,	,	PUNCT
ejpam-3176	302	32	in	in	ADP
ejpam-3176	302	33	which	which	PRON
ejpam-3176	302	34	the	the	DET
ejpam-3176	302	35	groups	group	NOUN
ejpam-3176	302	36	have	have	VERB
ejpam-3176	302	37	a	a	DET
ejpam-3176	302	38	common	common	ADJ
ejpam-3176	302	39	covariance	covariance	NOUN
ejpam-3176	302	40	matrix	matrix	NOUN
ejpam-3176	302	41	.	.	PUNCT
ejpam-3176	303	1	detmcd	detmcd	PROPN
ejpam-3176	303	2	,	,	PUNCT
ejpam-3176	303	3	fch	fch	PROPN
ejpam-3176	303	4	,	,	PUNCT
ejpam-3176	303	5	and	and	CCONJ
ejpam-3176	303	6	fastmcd	fastmcd	NOUN
ejpam-3176	303	7	are	be	AUX
ejpam-3176	303	8	compared	compare	VERB
ejpam-3176	303	9	based	base	VERB
ejpam-3176	303	10	on	on	ADP
ejpam-3176	303	11	three	three	NUM
ejpam-3176	303	12	approaches	approach	NOUN
ejpam-3176	303	13	to	to	PART
ejpam-3176	303	14	estimate	estimate	VERB
ejpam-3176	303	15	the	the	DET
ejpam-3176	303	16	common	common	ADJ
ejpam-3176	303	17	scatter	scatter	NOUN
ejpam-3176	303	18	matrix	matrix	NOUN
ejpam-3176	303	19	.	.	PUNCT
ejpam-3176	304	1	membership	membership	NOUN
ejpam-3176	304	2	probabilities	probability	NOUN
ejpam-3176	304	3	in	in	ADP
ejpam-3176	304	4	a	a	DET
ejpam-3176	304	5	robust	robust	ADJ
ejpam-3176	304	6	structure	structure	NOUN
ejpam-3176	304	7	are	be	AUX
ejpam-3176	304	8	estimated	estimate	VERB
ejpam-3176	304	9	by	by	ADP
ejpam-3176	304	10	considering	consider	VERB
ejpam-3176	304	11	only	only	ADJ
ejpam-3176	304	12	observations	observation	NOUN
ejpam-3176	304	13	of	of	ADP
ejpam-3176	304	14	non	non	NOUN
ejpam-3176	304	15	-	-	NOUN
ejpam-3176	304	16	outliers	outlier	NOUN
ejpam-3176	304	17	.	.	PUNCT
ejpam-3176	305	1	then	then	ADV
ejpam-3176	305	2	,	,	PUNCT
ejpam-3176	305	3	misclassification	misclassification	NOUN
ejpam-3176	305	4	probabilities	probability	NOUN
ejpam-3176	305	5	for	for	ADP
ejpam-3176	305	6	the	the	DET
ejpam-3176	305	7	data	datum	NOUN
ejpam-3176	305	8	are	be	AUX
ejpam-3176	305	9	obtained	obtain	VERB
ejpam-3176	305	10	and	and	CCONJ
ejpam-3176	305	11	set	set	VERB
ejpam-3176	305	12	into	into	ADP
ejpam-3176	305	13	two	two	NUM
ejpam-3176	305	14	groups	group	NOUN
ejpam-3176	305	15	of	of	ADP
ejpam-3176	305	16	datasets	dataset	NOUN
ejpam-3176	305	17	.	.	PUNCT
ejpam-3176	306	1	the	the	DET
ejpam-3176	306	2	results	result	NOUN
ejpam-3176	306	3	of	of	ADP
ejpam-3176	306	4	the	the	DET
ejpam-3176	306	5	simulation	simulation	NOUN
ejpam-3176	306	6	clearly	clearly	ADV
ejpam-3176	306	7	showed	show	VERB
ejpam-3176	306	8	how	how	SCONJ
ejpam-3176	306	9	the	the	DET
ejpam-3176	306	10	robust	robust	ADJ
ejpam-3176	306	11	structure	structure	NOUN
ejpam-3176	306	12	was	be	AUX
ejpam-3176	306	13	better	well	ADJ
ejpam-3176	306	14	and	and	CCONJ
ejpam-3176	306	15	how	how	SCONJ
ejpam-3176	306	16	the	the	DET
ejpam-3176	306	17	detmcd	detmcd	NOUN
ejpam-3176	306	18	algorithm	algorithm	NOUN
ejpam-3176	306	19	for	for	ADP
ejpam-3176	306	20	the	the	DET
ejpam-3176	306	21	robust	robust	ADJ
ejpam-3176	306	22	and	and	CCONJ
ejpam-3176	306	23	non	non	ADJ
ejpam-3176	306	24	-	-	ADJ
ejpam-3176	306	25	robust	robust	ADJ
ejpam-3176	306	26	structures	structure	NOUN
ejpam-3176	306	27	was	be	AUX
ejpam-3176	306	28	better	well	ADJ
ejpam-3176	306	29	than	than	ADP
ejpam-3176	306	30	the	the	DET
ejpam-3176	306	31	fastmcd	fastmcd	NOUN
ejpam-3176	306	32	and	and	CCONJ
ejpam-3176	306	33	fch	fch	PROPN
ejpam-3176	306	34	algorithms	algorithm	NOUN
ejpam-3176	306	35	.	.	PUNCT
ejpam-3176	307	1	for	for	ADP
ejpam-3176	307	2	the	the	DET
ejpam-3176	307	3	robust	robust	ADJ
ejpam-3176	307	4	structure	structure	NOUN
ejpam-3176	307	5	,	,	PUNCT
ejpam-3176	307	6	detmcd	detmcd	NOUN
ejpam-3176	307	7	performed	perform	VERB
ejpam-3176	307	8	better	well	ADV
ejpam-3176	307	9	and	and	CCONJ
ejpam-3176	307	10	was	be	AUX
ejpam-3176	307	11	unaffected	unaffected	ADJ
ejpam-3176	307	12	by	by	ADP
ejpam-3176	307	13	outliers	outlier	NOUN
ejpam-3176	307	14	.	.	PUNCT
ejpam-3176	308	1	the	the	DET
ejpam-3176	308	2	detmcd	detmcd	PROPN
ejpam-3176	308	3	algorithm	algorithm	NOUN
ejpam-3176	308	4	achieved	achieve	VERB
ejpam-3176	308	5	high	high	ADJ
ejpam-3176	308	6	efficiency	efficiency	NOUN
ejpam-3176	308	7	for	for	ADP
ejpam-3176	308	8	rlda	rlda	NOUN
ejpam-3176	308	9	and	and	CCONJ
ejpam-3176	308	10	was	be	AUX
ejpam-3176	308	11	more	more	ADV
ejpam-3176	308	12	accurate	accurate	ADJ
ejpam-3176	308	13	than	than	ADP
ejpam-3176	308	14	the	the	DET
ejpam-3176	308	15	fastmcd	fastmcd	NOUN
ejpam-3176	308	16	and	and	CCONJ
ejpam-3176	308	17	fch	fch	PROPN
ejpam-3176	308	18	algorithms	algorithm	NOUN
ejpam-3176	308	19	.	.	PUNCT
ejpam-3176	309	1	we	we	PRON
ejpam-3176	309	2	applied	apply	VERB
ejpam-3176	309	3	the	the	DET
ejpam-3176	309	4	rlda	rlda	NOUN
ejpam-3176	309	5	rules	rule	NOUN
ejpam-3176	309	6	on	on	ADP
ejpam-3176	309	7	actual	actual	ADJ
ejpam-3176	309	8	datasets	dataset	NOUN
ejpam-3176	309	9	based	base	VERB
ejpam-3176	309	10	on	on	ADP
ejpam-3176	309	11	the	the	DET
ejpam-3176	309	12	two	two	NUM
ejpam-3176	309	13	algorithms	algorithm	NOUN
ejpam-3176	309	14	.	.	PUNCT
ejpam-3176	310	1	the	the	DET
ejpam-3176	310	2	detmcd	detmcd	PROPN
ejpam-3176	310	3	algorithm	algorithm	NOUN
ejpam-3176	310	4	performed	perform	VERB
ejpam-3176	310	5	well	well	ADV
ejpam-3176	310	6	compared	compare	VERB
ejpam-3176	310	7	with	with	ADP
ejpam-3176	310	8	the	the	DET
ejpam-3176	310	9	fastmcd	fastmcd	NOUN
ejpam-3176	310	10	and	and	CCONJ
ejpam-3176	310	11	fch	fch	PROPN
ejpam-3176	310	12	algorithms	algorithm	NOUN
ejpam-3176	310	13	;	;	PUNCT
ejpam-3176	310	14	rlda	rlda	NOUN
ejpam-3176	310	15	with	with	ADP
ejpam-3176	310	16	detmcd	detmcd	NOUN
ejpam-3176	310	17	achieved	achieve	VERB
ejpam-3176	310	18	the	the	DET
ejpam-3176	310	19	highest	high	ADJ
ejpam-3176	310	20	efficiency	efficiency	NOUN
ejpam-3176	310	21	.	.	PUNCT
ejpam-3176	311	1	thus	thus	ADV
ejpam-3176	311	2	,	,	PUNCT
ejpam-3176	311	3	the	the	DET
ejpam-3176	311	4	detmcd	detmcd	PROPN
ejpam-3176	311	5	algorithm	algorithm	NOUN
ejpam-3176	311	6	increased	increase	VERB
ejpam-3176	311	7	the	the	DET
ejpam-3176	311	8	accuracy	accuracy	NOUN
ejpam-3176	311	9	and	and	CCONJ
ejpam-3176	311	10	performance	performance	NOUN
ejpam-3176	311	11	of	of	ADP
ejpam-3176	311	12	the	the	DET
ejpam-3176	311	13	lda	lda	PROPN
ejpam-3176	311	14	model	model	NOUN
ejpam-3176	311	15	,	,	PUNCT
ejpam-3176	311	16	which	which	PRON
ejpam-3176	311	17	indicates	indicate	VERB
ejpam-3176	311	18	more	more	ADJ
ejpam-3176	311	19	advantages	advantage	NOUN
ejpam-3176	311	20	to	to	PART
ejpam-3176	311	21	utilize	utilize	VERB
ejpam-3176	311	22	the	the	DET
ejpam-3176	311	23	model	model	NOUN
ejpam-3176	311	24	with	with	ADP
ejpam-3176	311	25	highly	highly	ADV
ejpam-3176	311	26	robust	robust	ADJ
ejpam-3176	311	27	estimations	estimation	NOUN
ejpam-3176	311	28	.	.	PUNCT
ejpam-3176	312	1	on	on	ADP
ejpam-3176	312	2	the	the	DET
ejpam-3176	312	3	basis	basis	NOUN
ejpam-3176	312	4	of	of	ADP
ejpam-3176	312	5	the	the	DET
ejpam-3176	312	6	results	result	NOUN
ejpam-3176	312	7	of	of	ADP
ejpam-3176	312	8	the	the	DET
ejpam-3176	312	9	simulation	simulation	NOUN
ejpam-3176	312	10	and	and	CCONJ
ejpam-3176	312	11	actual	actual	ADJ
ejpam-3176	312	12	data	datum	NOUN
ejpam-3176	312	13	,	,	PUNCT
ejpam-3176	312	14	the	the	DET
ejpam-3176	312	15	detmcd	detmcd	NOUN
ejpam-3176	312	16	algorithm	algorithm	NOUN
ejpam-3176	312	17	can	can	AUX
ejpam-3176	312	18	be	be	AUX
ejpam-3176	312	19	used	use	VERB
ejpam-3176	312	20	with	with	ADP
ejpam-3176	312	21	rlda	rlda	NOUN
ejpam-3176	312	22	in	in	ADP
ejpam-3176	312	23	financial	financial	ADJ
ejpam-3176	312	24	research	research	NOUN
ejpam-3176	312	25	to	to	PART
ejpam-3176	312	26	predict	predict	VERB
ejpam-3176	312	27	firm	firm	ADJ
ejpam-3176	312	28	failure	failure	NOUN
ejpam-3176	312	29	,	,	PUNCT
ejpam-3176	312	30	bankruptcy	bankruptcy	NOUN
ejpam-3176	312	31	,	,	PUNCT
ejpam-3176	312	32	and	and	CCONJ
ejpam-3176	312	33	company	company	NOUN
ejpam-3176	312	34	distress	distress	NOUN
ejpam-3176	312	35	.	.	PUNCT
ejpam-3176	313	1	it	it	PRON
ejpam-3176	313	2	also	also	ADV
ejpam-3176	313	3	has	have	VERB
ejpam-3176	313	4	applications	application	NOUN
ejpam-3176	313	5	in	in	ADP
ejpam-3176	313	6	other	other	ADJ
ejpam-3176	313	7	fields	field	NOUN
ejpam-3176	313	8	,	,	PUNCT
ejpam-3176	313	9	e.g.	e.g.	ADV
ejpam-3176	313	10	,	,	PUNCT
ejpam-3176	313	11	recognitions	recognition	NOUN
ejpam-3176	313	12	.	.	PUNCT
ejpam-3176	314	1	acknowledgements	acknowledgement	NOUN
ejpam-3176	314	2	the	the	DET
ejpam-3176	314	3	authors	author	NOUN
ejpam-3176	314	4	gratefully	gratefully	ADV
ejpam-3176	314	5	acknowledge	acknowledge	VERB
ejpam-3176	314	6	qassim	qassim	PROPN
ejpam-3176	314	7	university	university	PROPN
ejpam-3176	314	8	,	,	PUNCT
ejpam-3176	314	9	represented	represent	VERB
ejpam-3176	314	10	by	by	ADP
ejpam-3176	314	11	the	the	DET
ejpam-3176	314	12	deanship	deanship	NOUN
ejpam-3176	314	13	of	of	ADP
ejpam-3176	314	14	scientific	scientific	NOUN
ejpam-3176	314	15	on	on	ADP
ejpam-3176	314	16	the	the	DET
ejpam-3176	314	17	material	material	NOUN
ejpam-3176	314	18	support	support	NOUN
ejpam-3176	314	19	for	for	ADP
ejpam-3176	314	20	this	this	DET
ejpam-3176	314	21	research	research	NOUN
ejpam-3176	314	22	under	under	ADP
ejpam-3176	314	23	the	the	DET
ejpam-3176	314	24	number	number	NOUN
ejpam-3176	314	25	(	(	PUNCT
ejpam-3176	314	26	1475	1475	NUM
ejpam-3176	314	27	)	)	PUNCT
ejpam-3176	314	28	during	during	ADP
ejpam-3176	314	29	the	the	DET
ejpam-3176	314	30	academic	academic	ADJ
ejpam-3176	314	31	year	year	NOUN
ejpam-3176	314	32	1437	1437	NUM
ejpam-3176	314	33	ah/2016	ah/2016	PROPN
ejpam-3176	314	34	ad	ad	NOUN
ejpam-3176	314	35	.	.	PUNCT
ejpam-3176	315	1	references	reference	NOUN
ejpam-3176	315	2	[	[	X
ejpam-3176	315	3	1	1	NUM
ejpam-3176	315	4	]	]	PUNCT
ejpam-3176	315	5	mufda	mufda	PROPN
ejpam-3176	315	6	jameel	jameel	PROPN
ejpam-3176	315	7	alrawashdeh	alrawashdeh	PROPN
ejpam-3176	315	8	,	,	PUNCT
ejpam-3176	315	9	shamsul	shamsul	PROPN
ejpam-3176	315	10	rijal	rijal	PROPN
ejpam-3176	315	11	muhammad	muhammad	PROPN
ejpam-3176	315	12	sabri	sabri	PROPN
ejpam-3176	315	13	,	,	PUNCT
ejpam-3176	315	14	and	and	CCONJ
ejpam-3176	315	15	mohd	mohd	PROPN
ejpam-3176	315	16	tahir	tahir	PROPN
ejpam-3176	315	17	ismail	ismail	PROPN
ejpam-3176	315	18	.	.	PUNCT
ejpam-3176	316	1	robust	robust	ADJ
ejpam-3176	316	2	linear	linear	ADJ
ejpam-3176	316	3	discriminant	discriminant	ADJ
ejpam-3176	316	4	analysis	analysis	NOUN
ejpam-3176	316	5	with	with	ADP
ejpam-3176	316	6	financial	financial	ADJ
ejpam-3176	316	7	ratios	ratio	NOUN
ejpam-3176	316	8	in	in	ADP
ejpam-3176	316	9	special	special	ADJ
ejpam-3176	316	10	interval	interval	NOUN
ejpam-3176	316	11	.	.	PUNCT
ejpam-3176	317	1	applied	apply	VERB
ejpam-3176	317	2	mathematical	mathematical	ADJ
ejpam-3176	317	3	sciences	science	NOUN
ejpam-3176	317	4	,	,	PUNCT
ejpam-3176	317	5	6(121):6021–6034	6(121):6021–6034	NOUN
ejpam-3176	317	6	,	,	PUNCT
ejpam-3176	317	7	2012	2012	NUM
ejpam-3176	317	8	.	.	PUNCT
ejpam-3176	318	1	[	[	X
ejpam-3176	318	2	2	2	NUM
ejpam-3176	318	3	]	]	X
ejpam-3176	318	4	billor	billor	NOUN
ejpam-3176	318	5	,	,	PUNCT
ejpam-3176	318	6	nedret	nedret	PROPN
ejpam-3176	318	7	,	,	PUNCT
ejpam-3176	318	8	ali	ali	PROPN
ejpam-3176	318	9	s	s	PROPN
ejpam-3176	318	10	hadi	hadi	NOUN
ejpam-3176	318	11	,	,	PUNCT
ejpam-3176	318	12	and	and	CCONJ
ejpam-3176	318	13	paul	paul	PROPN
ejpam-3176	318	14	f	f	PROPN
ejpam-3176	318	15	velleman	velleman	PROPN
ejpam-3176	318	16	.	.	PUNCT
ejpam-3176	319	1	bacon	bacon	NOUN
ejpam-3176	319	2	:	:	PUNCT
ejpam-3176	319	3	blocked	block	VERB
ejpam-3176	319	4	adaptive	adaptive	ADJ
ejpam-3176	319	5	computationally	computationally	ADV
ejpam-3176	319	6	efficient	efficient	ADJ
ejpam-3176	319	7	outlier	outlier	NOUN
ejpam-3176	319	8	nominators	nominator	NOUN
ejpam-3176	319	9	.	.	PUNCT
ejpam-3176	320	1	computational	computational	ADJ
ejpam-3176	320	2	statistics	statistic	NOUN
ejpam-3176	320	3	and	and	CCONJ
ejpam-3176	320	4	data	datum	NOUN
ejpam-3176	320	5	analysis	analysis	NOUN
ejpam-3176	320	6	,	,	PUNCT
ejpam-3176	320	7	34(3):279–298	34(3):279–298	NUM
ejpam-3176	320	8	,	,	PUNCT
ejpam-3176	320	9	2000	2000	NUM
ejpam-3176	320	10	.	.	PUNCT
ejpam-3176	321	1	[	[	X
ejpam-3176	321	2	3	3	NUM
ejpam-3176	321	3	]	]	X
ejpam-3176	321	4	chork	chork	NOUN
ejpam-3176	321	5	,	,	PUNCT
ejpam-3176	321	6	cy	cy	PROPN
ejpam-3176	321	7	,	,	PUNCT
ejpam-3176	321	8	and	and	CCONJ
ejpam-3176	321	9	peter	peter	PROPN
ejpam-3176	321	10	j	j	PROPN
ejpam-3176	321	11	rousseeuw	rousseeuw	PROPN
ejpam-3176	321	12	.	.	PUNCT
ejpam-3176	322	1	integrating	integrate	VERB
ejpam-3176	322	2	a	a	DET
ejpam-3176	322	3	high	high	ADJ
ejpam-3176	322	4	-	-	PUNCT
ejpam-3176	322	5	breakdown	breakdown	NOUN
ejpam-3176	322	6	option	option	NOUN
ejpam-3176	322	7	into	into	ADP
ejpam-3176	322	8	discriminant	discriminant	ADJ
ejpam-3176	322	9	analysis	analysis	NOUN
ejpam-3176	322	10	in	in	ADP
ejpam-3176	322	11	exploration	exploration	NOUN
ejpam-3176	322	12	geochemistry	geochemistry	NOUN
ejpam-3176	322	13	.	.	PUNCT
ejpam-3176	323	1	journal	journal	NOUN
ejpam-3176	323	2	of	of	ADP
ejpam-3176	323	3	geochemical	geochemical	ADJ
ejpam-3176	323	4	exploration	exploration	NOUN
ejpam-3176	323	5	,	,	PUNCT
ejpam-3176	323	6	43(3):191–203	43(3):191–203	PROPN
ejpam-3176	323	7	,	,	PUNCT
ejpam-3176	323	8	1992	1992	NUM
ejpam-3176	323	9	.	.	PUNCT
ejpam-3176	324	1	references	reference	NOUN
ejpam-3176	324	2	297	297	NUM
ejpam-3176	324	3	[	[	X
ejpam-3176	324	4	4	4	NUM
ejpam-3176	324	5	]	]	X
ejpam-3176	324	6	croux	croux	NOUN
ejpam-3176	324	7	,	,	PUNCT
ejpam-3176	324	8	christophe	christophe	PROPN
ejpam-3176	324	9	,	,	PUNCT
ejpam-3176	324	10	and	and	CCONJ
ejpam-3176	324	11	catherine	catherine	PROPN
ejpam-3176	324	12	dehon	dehon	PROPN
ejpam-3176	324	13	.	.	PUNCT
ejpam-3176	325	1	analyse	analyse	NOUN
ejpam-3176	325	2	canonique	canonique	NOUN
ejpam-3176	325	3	base	base	PROPN
ejpam-3176	325	4	sur	sur	PROPN
ejpam-3176	325	5	des	des	PROPN
ejpam-3176	325	6	estimateurs	estimateur	VERB
ejpam-3176	325	7	robustes	robuste	VERB
ejpam-3176	325	8	de	de	X
ejpam-3176	325	9	la	la	X
ejpam-3176	325	10	matrice	matrice	X
ejpam-3176	325	11	de	de	PROPN
ejpam-3176	325	12	covariance	covariance	PROPN
ejpam-3176	325	13	.	.	PUNCT
ejpam-3176	326	1	revue	revue	NOUN
ejpam-3176	326	2	de	de	X
ejpam-3176	326	3	statistique	statistique	PROPN
ejpam-3176	326	4	applique	applique	NOUN
ejpam-3176	326	5	,	,	PUNCT
ejpam-3176	326	6	50(2):5–26	50(2):5–26	NOUN
ejpam-3176	326	7	,	,	PUNCT
ejpam-3176	326	8	2002	2002	NUM
ejpam-3176	326	9	.	.	PUNCT
ejpam-3176	327	1	[	[	X
ejpam-3176	327	2	5	5	NUM
ejpam-3176	327	3	]	]	X
ejpam-3176	327	4	croux	croux	NOUN
ejpam-3176	327	5	,	,	PUNCT
ejpam-3176	327	6	christophe	christophe	PROPN
ejpam-3176	327	7	,	,	PUNCT
ejpam-3176	327	8	peter	peter	PROPN
ejpam-3176	327	9	filzmoser	filzmoser	PROPN
ejpam-3176	327	10	,	,	PUNCT
ejpam-3176	327	11	and	and	CCONJ
ejpam-3176	327	12	kristel	kristel	PROPN
ejpam-3176	327	13	joossens	joossens	PROPN
ejpam-3176	327	14	.	.	PUNCT
ejpam-3176	328	1	classification	classification	NOUN
ejpam-3176	328	2	efficiencies	efficiency	NOUN
ejpam-3176	328	3	for	for	ADP
ejpam-3176	328	4	robust	robust	ADJ
ejpam-3176	328	5	linear	linear	ADJ
ejpam-3176	328	6	discriminant	discriminant	ADJ
ejpam-3176	328	7	analysis	analysis	NOUN
ejpam-3176	328	8	.	.	PUNCT
ejpam-3176	329	1	statistica	statistica	PROPN
ejpam-3176	329	2	sinica	sinica	PROPN
ejpam-3176	329	3	,	,	PUNCT
ejpam-3176	329	4	18(2):581–599	18(2):581–599	NUM
ejpam-3176	329	5	,	,	PUNCT
ejpam-3176	329	6	2008	2008	NUM
ejpam-3176	329	7	.	.	PUNCT
ejpam-3176	330	1	[	[	X
ejpam-3176	330	2	6	6	NUM
ejpam-3176	330	3	]	]	PUNCT
ejpam-3176	330	4	croux	croux	NOUN
ejpam-3176	330	5	,	,	PUNCT
ejpam-3176	330	6	christophe	christophe	PROPN
ejpam-3176	330	7	,	,	PUNCT
ejpam-3176	330	8	sarah	sarah	PROPN
ejpam-3176	330	9	gelper	gelper	NOUN
ejpam-3176	330	10	,	,	PUNCT
ejpam-3176	330	11	and	and	CCONJ
ejpam-3176	330	12	koen	koen	PROPN
ejpam-3176	330	13	mahieu	mahieu	PROPN
ejpam-3176	330	14	.	.	PUNCT
ejpam-3176	331	1	robust	robust	ADJ
ejpam-3176	331	2	exponential	exponential	ADJ
ejpam-3176	331	3	smoothing	smoothing	NOUN
ejpam-3176	331	4	of	of	ADP
ejpam-3176	331	5	multivariate	multivariate	NOUN
ejpam-3176	331	6	time	time	NOUN
ejpam-3176	331	7	series	series	NOUN
ejpam-3176	331	8	.	.	PUNCT
ejpam-3176	332	1	computational	computational	ADJ
ejpam-3176	332	2	statistics	statistic	NOUN
ejpam-3176	332	3	and	and	CCONJ
ejpam-3176	332	4	data	datum	NOUN
ejpam-3176	332	5	analysis	analysis	NOUN
ejpam-3176	332	6	,	,	PUNCT
ejpam-3176	332	7	54(12):2999	54(12):2999	NUM
ejpam-3176	332	8	–	–	PUNCT
ejpam-3176	332	9	3006	3006	NUM
ejpam-3176	332	10	,	,	PUNCT
ejpam-3176	332	11	2010	2010	NUM
ejpam-3176	332	12	.	.	PUNCT
ejpam-3176	333	1	[	[	X
ejpam-3176	333	2	7	7	NUM
ejpam-3176	333	3	]	]	X
ejpam-3176	333	4	croux	croux	NOUN
ejpam-3176	333	5	,	,	PUNCT
ejpam-3176	333	6	christophe	christophe	NOUN
ejpam-3176	333	7	,	,	PUNCT
ejpam-3176	333	8	and	and	CCONJ
ejpam-3176	333	9	gentiane	gentiane	ADJ
ejpam-3176	333	10	haesbroeck	haesbroeck	PROPN
ejpam-3176	333	11	.	.	PUNCT
ejpam-3176	334	1	principal	principal	ADJ
ejpam-3176	334	2	component	component	NOUN
ejpam-3176	334	3	analysis	analysis	NOUN
ejpam-3176	334	4	based	base	VERB
ejpam-3176	334	5	on	on	ADP
ejpam-3176	334	6	robust	robust	ADJ
ejpam-3176	334	7	estimators	estimator	NOUN
ejpam-3176	334	8	of	of	ADP
ejpam-3176	334	9	the	the	DET
ejpam-3176	334	10	covariance	covariance	NOUN
ejpam-3176	334	11	or	or	CCONJ
ejpam-3176	334	12	correlation	correlation	NOUN
ejpam-3176	334	13	matrix	matrix	NOUN
ejpam-3176	334	14	:	:	PUNCT
ejpam-3176	334	15	influence	influence	NOUN
ejpam-3176	334	16	functions	function	NOUN
ejpam-3176	334	17	and	and	CCONJ
ejpam-3176	334	18	efficiencies	efficiency	NOUN
ejpam-3176	334	19	.	.	PUNCT
ejpam-3176	335	1	biometrika	biometrika	PROPN
ejpam-3176	335	2	,	,	PUNCT
ejpam-3176	335	3	87(3):603–618	87(3):603–618	PROPN
ejpam-3176	335	4	,	,	PUNCT
ejpam-3176	335	5	2000	2000	NUM
ejpam-3176	335	6	.	.	PUNCT
ejpam-3176	336	1	[	[	X
ejpam-3176	336	2	8	8	NUM
ejpam-3176	336	3	]	]	X
ejpam-3176	336	4	devlin	devlin	NOUN
ejpam-3176	336	5	,	,	PUNCT
ejpam-3176	336	6	s.	s.	PROPN
ejpam-3176	336	7	j.	j.	PROPN
ejpam-3176	336	8	,	,	PUNCT
ejpam-3176	336	9	gnanadesikan	gnanadesikan	PROPN
ejpam-3176	336	10	,	,	PUNCT
ejpam-3176	336	11	r.	r.	PROPN
ejpam-3176	336	12	,	,	PUNCT
ejpam-3176	336	13	kettenring	kettenring	NOUN
ejpam-3176	336	14	,	,	PUNCT
ejpam-3176	336	15	and	and	CCONJ
ejpam-3176	336	16	j.	j.	PROPN
ejpam-3176	336	17	r.	r.	PROPN
ejpam-3176	336	18	robust	robust	PROPN
ejpam-3176	336	19	estimation	estimation	NOUN
ejpam-3176	336	20	of	of	ADP
ejpam-3176	336	21	dispersion	dispersion	NOUN
ejpam-3176	336	22	matrices	matrix	NOUN
ejpam-3176	336	23	and	and	CCONJ
ejpam-3176	336	24	principal	principal	ADJ
ejpam-3176	336	25	components	component	NOUN
ejpam-3176	336	26	.	.	PUNCT
ejpam-3176	337	1	journal	journal	NOUN
ejpam-3176	337	2	of	of	ADP
ejpam-3176	337	3	the	the	DET
ejpam-3176	337	4	american	american	PROPN
ejpam-3176	337	5	statistical	statistical	PROPN
ejpam-3176	337	6	association	association	NOUN
ejpam-3176	337	7	,	,	PUNCT
ejpam-3176	337	8	76(374):354–362	76(374):354–362	PROPN
ejpam-3176	337	9	,	,	PUNCT
ejpam-3176	337	10	1981	1981	NUM
ejpam-3176	337	11	.	.	PUNCT
ejpam-3176	338	1	[	[	X
ejpam-3176	338	2	9	9	NUM
ejpam-3176	338	3	]	]	SYM
ejpam-3176	338	4	fekri	fekri	NOUN
ejpam-3176	338	5	,	,	PUNCT
ejpam-3176	338	6	m	m	PROPN
ejpam-3176	338	7	,	,	PUNCT
ejpam-3176	338	8	and	and	CCONJ
ejpam-3176	338	9	anne	anne	PROPN
ejpam-3176	338	10	ruiz	ruiz	NOUN
ejpam-3176	338	11	-	-	PUNCT
ejpam-3176	338	12	gazen	gazen	NOUN
ejpam-3176	338	13	.	.	PUNCT
ejpam-3176	339	1	robust	robust	ADJ
ejpam-3176	339	2	weighted	weight	VERB
ejpam-3176	339	3	orthogonal	orthogonal	ADJ
ejpam-3176	339	4	regression	regression	NOUN
ejpam-3176	339	5	in	in	ADP
ejpam-3176	339	6	the	the	DET
ejpam-3176	339	7	errorsin	errorsin	NOUN
ejpam-3176	339	8	-	-	PUNCT
ejpam-3176	339	9	variables	variable	NOUN
ejpam-3176	339	10	model	model	NOUN
ejpam-3176	339	11	.	.	PUNCT
ejpam-3176	340	1	journal	journal	PROPN
ejpam-3176	340	2	of	of	ADP
ejpam-3176	340	3	the	the	DET
ejpam-3176	340	4	american	american	PROPN
ejpam-3176	340	5	statistical	statistical	PROPN
ejpam-3176	340	6	association	association	PROPN
ejpam-3176	340	7	,	,	PUNCT
ejpam-3176	340	8	88(1):89–108	88(1):89–108	NUM
ejpam-3176	340	9	,	,	PUNCT
ejpam-3176	340	10	2004	2004	NUM
ejpam-3176	340	11	.	.	PUNCT
ejpam-3176	341	1	[	[	X
ejpam-3176	341	2	10	10	NUM
ejpam-3176	341	3	]	]	X
ejpam-3176	341	4	hawkins	hawkin	NOUN
ejpam-3176	341	5	,	,	PUNCT
ejpam-3176	341	6	d.m	d.m	PROPN
ejpam-3176	341	7	.	.	PROPN
ejpam-3176	341	8	,	,	PUNCT
ejpam-3176	341	9	olive	olive	NOUN
ejpam-3176	341	10	,	,	PUNCT
ejpam-3176	341	11	and	and	CCONJ
ejpam-3176	341	12	d.j	d.j	PROPN
ejpam-3176	341	13	.	.	PROPN
ejpam-3176	341	14	improved	improve	VERB
ejpam-3176	341	15	feasible	feasible	ADJ
ejpam-3176	341	16	solution	solution	NOUN
ejpam-3176	341	17	algorithms	algorithm	NOUN
ejpam-3176	341	18	for	for	ADP
ejpam-3176	341	19	high	high	ADJ
ejpam-3176	341	20	breakdown	breakdown	ADJ
ejpam-3176	341	21	estimation	estimation	NOUN
ejpam-3176	341	22	.	.	PUNCT
ejpam-3176	342	1	computational	computational	ADJ
ejpam-3176	342	2	statistics	statistic	NOUN
ejpam-3176	342	3	and	and	CCONJ
ejpam-3176	342	4	data	datum	NOUN
ejpam-3176	342	5	analysis	analysis	NOUN
ejpam-3176	342	6	,	,	PUNCT
ejpam-3176	342	7	30:1–11	30:1–11	NUM
ejpam-3176	342	8	,	,	PUNCT
ejpam-3176	342	9	1999	1999	NUM
ejpam-3176	342	10	.	.	PUNCT
ejpam-3176	343	1	[	[	X
ejpam-3176	343	2	11	11	NUM
ejpam-3176	343	3	]	]	SYM
ejpam-3176	343	4	hawkins	hawkin	NOUN
ejpam-3176	343	5	,	,	PUNCT
ejpam-3176	343	6	douglas	douglas	PROPN
ejpam-3176	343	7	m	m	PROPN
ejpam-3176	343	8	,	,	PUNCT
ejpam-3176	343	9	and	and	CCONJ
ejpam-3176	343	10	geoffrey	geoffrey	PROPN
ejpam-3176	343	11	j	j	PROPN
ejpam-3176	343	12	mclachlan	mclachlan	PROPN
ejpam-3176	343	13	.	.	PUNCT
ejpam-3176	343	14	high	high	ADJ
ejpam-3176	343	15	-	-	PUNCT
ejpam-3176	343	16	breakdown	breakdown	NOUN
ejpam-3176	343	17	linear	linear	ADJ
ejpam-3176	343	18	discriminant	discriminant	ADJ
ejpam-3176	343	19	analysis	analysis	NOUN
ejpam-3176	343	20	.	.	PUNCT
ejpam-3176	344	1	journal	journal	NOUN
ejpam-3176	344	2	of	of	ADP
ejpam-3176	344	3	the	the	DET
ejpam-3176	344	4	american	american	PROPN
ejpam-3176	344	5	statistical	statistical	PROPN
ejpam-3176	344	6	association	association	PROPN
ejpam-3176	344	7	,	,	PUNCT
ejpam-3176	344	8	92(437):136–143	92(437):136–143	PROPN
ejpam-3176	344	9	,	,	PUNCT
ejpam-3176	344	10	1997	1997	NUM
ejpam-3176	344	11	.	.	PUNCT
ejpam-3176	345	1	[	[	X
ejpam-3176	345	2	12	12	NUM
ejpam-3176	345	3	]	]	PUNCT
ejpam-3176	345	4	he	he	PRON
ejpam-3176	345	5	,	,	PUNCT
ejpam-3176	345	6	xuming	xume	VERB
ejpam-3176	345	7	,	,	PUNCT
ejpam-3176	345	8	and	and	CCONJ
ejpam-3176	345	9	wing	wing	NOUN
ejpam-3176	345	10	k	k	PROPN
ejpam-3176	345	11	fung	fung	PROPN
ejpam-3176	345	12	.	.	PUNCT
ejpam-3176	346	1	high	high	ADJ
ejpam-3176	346	2	breakdown	breakdown	ADJ
ejpam-3176	346	3	estimation	estimation	NOUN
ejpam-3176	346	4	for	for	ADP
ejpam-3176	346	5	multiple	multiple	ADJ
ejpam-3176	346	6	populations	population	NOUN
ejpam-3176	346	7	with	with	ADP
ejpam-3176	346	8	applications	application	NOUN
ejpam-3176	346	9	to	to	PART
ejpam-3176	346	10	discriminant	discriminant	VERB
ejpam-3176	346	11	analysis	analysis	NOUN
ejpam-3176	346	12	.	.	PUNCT
ejpam-3176	347	1	journal	journal	NOUN
ejpam-3176	347	2	of	of	ADP
ejpam-3176	347	3	multivariate	multivariate	NOUN
ejpam-3176	347	4	analysis	analysis	NOUN
ejpam-3176	347	5	,	,	PUNCT
ejpam-3176	347	6	72(2):151–162	72(2):151–162	NOUN
ejpam-3176	347	7	,	,	PUNCT
ejpam-3176	347	8	2000	2000	NUM
ejpam-3176	347	9	.	.	PUNCT
ejpam-3176	348	1	[	[	X
ejpam-3176	348	2	13	13	NUM
ejpam-3176	348	3	]	]	X
ejpam-3176	348	4	hubert	hubert	PROPN
ejpam-3176	348	5	,	,	PUNCT
ejpam-3176	348	6	mia	mia	PROPN
ejpam-3176	348	7	,	,	PUNCT
ejpam-3176	348	8	and	and	CCONJ
ejpam-3176	348	9	k	k	PROPN
ejpam-3176	348	10	vanden	vanden	PROPN
ejpam-3176	348	11	branden	branden	PROPN
ejpam-3176	348	12	.	.	PUNCT
ejpam-3176	349	1	robust	robust	ADJ
ejpam-3176	349	2	methods	method	NOUN
ejpam-3176	349	3	for	for	ADP
ejpam-3176	349	4	partial	partial	ADJ
ejpam-3176	349	5	least	least	ADJ
ejpam-3176	349	6	squares	square	NOUN
ejpam-3176	349	7	regression	regression	NOUN
ejpam-3176	349	8	.	.	PUNCT
ejpam-3176	350	1	journal	journal	PROPN
ejpam-3176	350	2	of	of	ADP
ejpam-3176	350	3	chemometrics	chemometric	NOUN
ejpam-3176	350	4	,	,	PUNCT
ejpam-3176	350	5	17(10):537–549	17(10):537–549	NUM
ejpam-3176	350	6	,	,	PUNCT
ejpam-3176	350	7	2003	2003	NUM
ejpam-3176	350	8	.	.	PUNCT
ejpam-3176	351	1	[	[	X
ejpam-3176	351	2	14	14	NUM
ejpam-3176	351	3	]	]	X
ejpam-3176	351	4	hubert	hubert	PROPN
ejpam-3176	351	5	,	,	PUNCT
ejpam-3176	351	6	mia	mia	PROPN
ejpam-3176	351	7	,	,	PUNCT
ejpam-3176	351	8	and	and	CCONJ
ejpam-3176	351	9	katrien	katrien	PROPN
ejpam-3176	351	10	van	van	PROPN
ejpam-3176	351	11	driessen	driessen	PROPN
ejpam-3176	351	12	.	.	PUNCT
ejpam-3176	352	1	fast	fast	ADJ
ejpam-3176	352	2	and	and	CCONJ
ejpam-3176	352	3	robust	robust	ADJ
ejpam-3176	352	4	discriminant	discriminant	ADJ
ejpam-3176	352	5	analysis	analysis	NOUN
ejpam-3176	352	6	.	.	PUNCT
ejpam-3176	353	1	computational	computational	ADJ
ejpam-3176	353	2	statistics	statistic	NOUN
ejpam-3176	353	3	and	and	CCONJ
ejpam-3176	353	4	data	datum	NOUN
ejpam-3176	353	5	analysis	analysis	NOUN
ejpam-3176	353	6	,	,	PUNCT
ejpam-3176	353	7	45(2):301–320	45(2):301–320	NOUN
ejpam-3176	353	8	,	,	PUNCT
ejpam-3176	353	9	2004	2004	NUM
ejpam-3176	353	10	.	.	PUNCT
ejpam-3176	354	1	[	[	X
ejpam-3176	354	2	15	15	NUM
ejpam-3176	354	3	]	]	X
ejpam-3176	354	4	hubert	hubert	PROPN
ejpam-3176	354	5	,	,	PUNCT
ejpam-3176	354	6	mia	mia	PROPN
ejpam-3176	354	7	,	,	PUNCT
ejpam-3176	354	8	and	and	CCONJ
ejpam-3176	354	9	peter	peter	PROPN
ejpam-3176	354	10	j	j	PROPN
ejpam-3176	354	11	rousseeuw	rousseeuw	PROPN
ejpam-3176	354	12	.	.	PUNCT
ejpam-3176	355	1	robust	robust	ADJ
ejpam-3176	355	2	regression	regression	NOUN
ejpam-3176	355	3	with	with	ADP
ejpam-3176	355	4	both	both	CCONJ
ejpam-3176	355	5	continuous	continuous	ADJ
ejpam-3176	355	6	and	and	CCONJ
ejpam-3176	355	7	binary	binary	ADJ
ejpam-3176	355	8	regressors	regressor	NOUN
ejpam-3176	355	9	.	.	PUNCT
ejpam-3176	356	1	journal	journal	NOUN
ejpam-3176	356	2	of	of	ADP
ejpam-3176	356	3	statistical	statistical	ADJ
ejpam-3176	356	4	planning	planning	NOUN
ejpam-3176	356	5	and	and	CCONJ
ejpam-3176	356	6	inference	inference	NOUN
ejpam-3176	356	7	,	,	PUNCT
ejpam-3176	356	8	57(1):153–163	57(1):153–163	NUM
ejpam-3176	356	9	,	,	PUNCT
ejpam-3176	356	10	1997	1997	NUM
ejpam-3176	356	11	.	.	PUNCT
ejpam-3176	357	1	[	[	X
ejpam-3176	357	2	16	16	NUM
ejpam-3176	357	3	]	]	X
ejpam-3176	357	4	hubert	hubert	PROPN
ejpam-3176	357	5	,	,	PUNCT
ejpam-3176	357	6	mia	mia	PROPN
ejpam-3176	357	7	,	,	PUNCT
ejpam-3176	357	8	peter	peter	PROPN
ejpam-3176	357	9	j	j	PROPN
ejpam-3176	357	10	rousseeuw	rousseeuw	PROPN
ejpam-3176	357	11	,	,	PUNCT
ejpam-3176	357	12	and	and	CCONJ
ejpam-3176	357	13	stefan	stefan	PROPN
ejpam-3176	357	14	van	van	PROPN
ejpam-3176	357	15	aelst	aelst	PROPN
ejpam-3176	357	16	.	.	PUNCT
ejpam-3176	358	1	high	high	ADJ
ejpam-3176	358	2	-	-	PUNCT
ejpam-3176	358	3	breakdown	breakdown	NOUN
ejpam-3176	358	4	robust	robust	ADJ
ejpam-3176	358	5	multivariate	multivariate	NOUN
ejpam-3176	358	6	methods	method	NOUN
ejpam-3176	358	7	.	.	PUNCT
ejpam-3176	359	1	statistical	statistical	ADJ
ejpam-3176	359	2	science	science	NOUN
ejpam-3176	359	3	,	,	PUNCT
ejpam-3176	359	4	23(1):92–119	23(1):92–119	NUM
ejpam-3176	359	5	,	,	PUNCT
ejpam-3176	359	6	2008	2008	NUM
ejpam-3176	359	7	.	.	PUNCT
ejpam-3176	360	1	[	[	X
ejpam-3176	360	2	17	17	NUM
ejpam-3176	360	3	]	]	X
ejpam-3176	360	4	hubert	hubert	PROPN
ejpam-3176	360	5	,	,	PUNCT
ejpam-3176	360	6	mia	mia	PROPN
ejpam-3176	360	7	,	,	PUNCT
ejpam-3176	360	8	peter	peter	PROPN
ejpam-3176	360	9	j	j	PROPN
ejpam-3176	360	10	rousseeuw	rousseeuw	PROPN
ejpam-3176	360	11	,	,	PUNCT
ejpam-3176	360	12	and	and	CCONJ
ejpam-3176	360	13	karlien	karlien	PROPN
ejpam-3176	360	14	vanden	vanden	PROPN
ejpam-3176	360	15	branden	branden	PROPN
ejpam-3176	360	16	.	.	PUNCT
ejpam-3176	361	1	robpca	robpca	PROPN
ejpam-3176	361	2	:	:	PUNCT
ejpam-3176	361	3	a	a	DET
ejpam-3176	361	4	new	new	ADJ
ejpam-3176	361	5	approach	approach	NOUN
ejpam-3176	361	6	to	to	ADP
ejpam-3176	361	7	robust	robust	ADJ
ejpam-3176	361	8	principal	principal	ADJ
ejpam-3176	361	9	component	component	NOUN
ejpam-3176	361	10	analysis	analysis	NOUN
ejpam-3176	361	11	.	.	PUNCT
ejpam-3176	362	1	technometrics	technometric	NOUN
ejpam-3176	362	2	,	,	PUNCT
ejpam-3176	362	3	47(1):64–79	47(1):64–79	NUM
ejpam-3176	362	4	,	,	PUNCT
ejpam-3176	362	5	2005	2005	NUM
ejpam-3176	362	6	.	.	PUNCT
ejpam-3176	363	1	references	reference	NOUN
ejpam-3176	363	2	298	298	NUM
ejpam-3176	364	1	[	[	X
ejpam-3176	364	2	18	18	NUM
ejpam-3176	364	3	]	]	X
ejpam-3176	364	4	hubert	hubert	PROPN
ejpam-3176	364	5	,	,	PUNCT
ejpam-3176	364	6	mia	mia	PROPN
ejpam-3176	364	7	,	,	PUNCT
ejpam-3176	364	8	peter	peter	PROPN
ejpam-3176	364	9	j	j	PROPN
ejpam-3176	364	10	rousseeuw	rousseeuw	PROPN
ejpam-3176	364	11	,	,	PUNCT
ejpam-3176	364	12	and	and	CCONJ
ejpam-3176	364	13	tim	tim	PROPN
ejpam-3176	364	14	verdonck	verdonck	NOUN
ejpam-3176	364	15	.	.	PUNCT
ejpam-3176	365	1	a	a	DET
ejpam-3176	365	2	deterministic	deterministic	ADJ
ejpam-3176	365	3	algorithm	algorithm	NOUN
ejpam-3176	365	4	for	for	ADP
ejpam-3176	365	5	robust	robust	ADJ
ejpam-3176	365	6	location	location	NOUN
ejpam-3176	365	7	and	and	CCONJ
ejpam-3176	365	8	scatter	scatter	NOUN
ejpam-3176	365	9	.	.	PUNCT
ejpam-3176	366	1	journal	journal	PROPN
ejpam-3176	366	2	of	of	ADP
ejpam-3176	366	3	computational	computational	ADJ
ejpam-3176	366	4	and	and	CCONJ
ejpam-3176	366	5	graphical	graphical	ADJ
ejpam-3176	366	6	statistics	statistic	NOUN
ejpam-3176	366	7	,	,	PUNCT
ejpam-3176	366	8	21(3):618–637	21(3):618–637	PROPN
ejpam-3176	366	9	,	,	PUNCT
ejpam-3176	366	10	2012	2012	NUM
ejpam-3176	366	11	.	.	PUNCT
ejpam-3176	367	1	[	[	X
ejpam-3176	367	2	19	19	NUM
ejpam-3176	367	3	]	]	X
ejpam-3176	367	4	hubert	hubert	PROPN
ejpam-3176	367	5	,	,	PUNCT
ejpam-3176	367	6	mia	mia	PROPN
ejpam-3176	367	7	,	,	PUNCT
ejpam-3176	367	8	and	and	CCONJ
ejpam-3176	367	9	sabine	sabine	PROPN
ejpam-3176	367	10	verboven	verboven	PROPN
ejpam-3176	367	11	.	.	PUNCT
ejpam-3176	368	1	a	a	DET
ejpam-3176	368	2	robust	robust	ADJ
ejpam-3176	368	3	pcr	pcr	ADJ
ejpam-3176	368	4	method	method	NOUN
ejpam-3176	368	5	for	for	ADP
ejpam-3176	368	6	highdimensional	highdimensional	ADJ
ejpam-3176	368	7	regressors	regressor	NOUN
ejpam-3176	368	8	.	.	PUNCT
ejpam-3176	369	1	journal	journal	PROPN
ejpam-3176	369	2	of	of	ADP
ejpam-3176	369	3	chemometrics	chemometric	NOUN
ejpam-3176	369	4	,	,	PUNCT
ejpam-3176	369	5	17(89):438–452	17(89):438–452	NUM
ejpam-3176	369	6	,	,	PUNCT
ejpam-3176	369	7	2003	2003	NUM
ejpam-3176	369	8	.	.	PUNCT
ejpam-3176	370	1	[	[	X
ejpam-3176	370	2	20	20	NUM
ejpam-3176	370	3	]	]	PUNCT
ejpam-3176	370	4	maronna	maronna	NOUN
ejpam-3176	370	5	,	,	PUNCT
ejpam-3176	370	6	ricardo	ricardo	PROPN
ejpam-3176	370	7	a	a	X
ejpam-3176	370	8	,	,	PUNCT
ejpam-3176	370	9	and	and	CCONJ
ejpam-3176	370	10	ruben	ruben	PROPN
ejpam-3176	370	11	h	h	PROPN
ejpam-3176	370	12	zamar	zamar	PROPN
ejpam-3176	370	13	.	.	PUNCT
ejpam-3176	371	1	robust	robust	ADJ
ejpam-3176	371	2	estimates	estimate	NOUN
ejpam-3176	371	3	of	of	ADP
ejpam-3176	371	4	location	location	NOUN
ejpam-3176	371	5	and	and	CCONJ
ejpam-3176	371	6	dispersion	dispersion	NOUN
ejpam-3176	371	7	for	for	ADP
ejpam-3176	371	8	high	high	ADJ
ejpam-3176	371	9	-	-	PUNCT
ejpam-3176	371	10	dimensional	dimensional	ADJ
ejpam-3176	371	11	datasets	dataset	NOUN
ejpam-3176	371	12	.	.	PUNCT
ejpam-3176	372	1	technometrics	technometric	NOUN
ejpam-3176	372	2	,	,	PUNCT
ejpam-3176	372	3	44(4):307–317	44(4):307–317	PROPN
ejpam-3176	372	4	,	,	PUNCT
ejpam-3176	372	5	2002	2002	NUM
ejpam-3176	372	6	.	.	PUNCT
ejpam-3176	373	1	[	[	X
ejpam-3176	373	2	21	21	NUM
ejpam-3176	373	3	]	]	X
ejpam-3176	373	4	david	david	PROPN
ejpam-3176	373	5	j.	j.	PROPN
ejpam-3176	373	6	olive	olive	PROPN
ejpam-3176	373	7	and	and	CCONJ
ejpam-3176	373	8	douglas	douglas	PROPN
ejpam-3176	373	9	m.	m.	PROPN
ejpam-3176	373	10	hawkins	hawkins	PROPN
ejpam-3176	373	11	.	.	PUNCT
ejpam-3176	374	1	robust	robust	ADJ
ejpam-3176	374	2	multivariate	multivariate	NOUN
ejpam-3176	374	3	location	location	NOUN
ejpam-3176	374	4	and	and	CCONJ
ejpam-3176	374	5	dispersion	dispersion	NOUN
ejpam-3176	374	6	.	.	PUNCT
ejpam-3176	375	1	southern	southern	ADJ
ejpam-3176	375	2	illinois	illinois	PROPN
ejpam-3176	375	3	university	university	PROPN
ejpam-3176	375	4	and	and	CCONJ
ejpam-3176	375	5	university	university	PROPN
ejpam-3176	375	6	of	of	ADP
ejpam-3176	375	7	minnesota	minnesota	PROPN
ejpam-3176	375	8	,	,	PUNCT
ejpam-3176	375	9	2010	2010	NUM
ejpam-3176	375	10	.	.	PUNCT
ejpam-3176	376	1	[	[	X
ejpam-3176	376	2	22	22	NUM
ejpam-3176	376	3	]	]	X
ejpam-3176	376	4	pison	pison	NOUN
ejpam-3176	376	5	,	,	PUNCT
ejpam-3176	376	6	greet	greet	VERB
ejpam-3176	376	7	,	,	PUNCT
ejpam-3176	376	8	and	and	CCONJ
ejpam-3176	376	9	et	et	PROPN
ejpam-3176	376	10	al	al	PROPN
ejpam-3176	376	11	.	.	PROPN
ejpam-3176	376	12	robust	robust	ADJ
ejpam-3176	376	13	factor	factor	NOUN
ejpam-3176	376	14	analysis	analysis	NOUN
ejpam-3176	376	15	.	.	PUNCT
ejpam-3176	377	1	journal	journal	NOUN
ejpam-3176	377	2	of	of	ADP
ejpam-3176	377	3	multivariate	multivariate	NOUN
ejpam-3176	377	4	analysis	analysis	NOUN
ejpam-3176	377	5	,	,	PUNCT
ejpam-3176	377	6	84(1):145–172	84(1):145–172	NUM
ejpam-3176	377	7	,	,	PUNCT
ejpam-3176	377	8	2003	2003	NUM
ejpam-3176	377	9	.	.	PUNCT
ejpam-3176	378	1	[	[	X
ejpam-3176	378	2	23	23	NUM
ejpam-3176	378	3	]	]	SYM
ejpam-3176	378	4	rousseeuw	rousseeuw	NOUN
ejpam-3176	378	5	and	and	CCONJ
ejpam-3176	378	6	peter	peter	PROPN
ejpam-3176	378	7	j.	j.	PROPN
ejpam-3176	378	8	least	least	ADJ
ejpam-3176	378	9	median	median	PROPN
ejpam-3176	378	10	of	of	ADP
ejpam-3176	378	11	squares	square	NOUN
ejpam-3176	378	12	regression	regression	PROPN
ejpam-3176	378	13	.	.	PUNCT
ejpam-3176	379	1	journal	journal	NOUN
ejpam-3176	379	2	of	of	ADP
ejpam-3176	379	3	the	the	DET
ejpam-3176	379	4	american	american	PROPN
ejpam-3176	379	5	statistical	statistical	PROPN
ejpam-3176	379	6	association	association	PROPN
ejpam-3176	379	7	,	,	PUNCT
ejpam-3176	379	8	79(388):871–880	79(388):871–880	PROPN
ejpam-3176	379	9	,	,	PUNCT
ejpam-3176	379	10	1984	1984	NUM
ejpam-3176	379	11	.	.	PUNCT
ejpam-3176	380	1	[	[	X
ejpam-3176	380	2	24	24	NUM
ejpam-3176	380	3	]	]	X
ejpam-3176	380	4	rousseeuw	rousseeuw	PROPN
ejpam-3176	380	5	,	,	PUNCT
ejpam-3176	380	6	peter	peter	PROPN
ejpam-3176	380	7	j	j	PROPN
ejpam-3176	380	8	,	,	PUNCT
ejpam-3176	380	9	and	and	CCONJ
ejpam-3176	380	10	christophe	christophe	PROPN
ejpam-3176	380	11	croux	croux	VERB
ejpam-3176	380	12	.	.	PUNCT
ejpam-3176	381	1	alternatives	alternative	NOUN
ejpam-3176	381	2	to	to	ADP
ejpam-3176	381	3	the	the	DET
ejpam-3176	381	4	median	median	ADJ
ejpam-3176	381	5	absolute	absolute	ADJ
ejpam-3176	381	6	deviation	deviation	NOUN
ejpam-3176	381	7	.	.	PUNCT
ejpam-3176	382	1	journal	journal	NOUN
ejpam-3176	382	2	of	of	ADP
ejpam-3176	382	3	the	the	DET
ejpam-3176	382	4	american	american	PROPN
ejpam-3176	382	5	statistical	statistical	PROPN
ejpam-3176	382	6	association	association	PROPN
ejpam-3176	382	7	,	,	PUNCT
ejpam-3176	382	8	88(424):1273–1283	88(424):1273–1283	PROPN
ejpam-3176	382	9	,	,	PUNCT
ejpam-3176	382	10	1993	1993	NUM
ejpam-3176	382	11	.	.	PUNCT
ejpam-3176	383	1	[	[	X
ejpam-3176	383	2	25	25	NUM
ejpam-3176	383	3	]	]	X
ejpam-3176	383	4	rousseeuw	rousseeuw	PROPN
ejpam-3176	383	5	,	,	PUNCT
ejpam-3176	383	6	peter	peter	PROPN
ejpam-3176	383	7	j	j	PROPN
ejpam-3176	383	8	,	,	PUNCT
ejpam-3176	383	9	and	and	CCONJ
ejpam-3176	383	10	katrien	katrien	PROPN
ejpam-3176	383	11	van	van	PROPN
ejpam-3176	383	12	driessen	driessen	PROPN
ejpam-3176	383	13	.	.	PUNCT
ejpam-3176	384	1	a	a	DET
ejpam-3176	384	2	fast	fast	ADJ
ejpam-3176	384	3	algorithm	algorithm	NOUN
ejpam-3176	384	4	for	for	ADP
ejpam-3176	384	5	the	the	DET
ejpam-3176	384	6	minimum	minimum	ADJ
ejpam-3176	384	7	covariance	covariance	NOUN
ejpam-3176	384	8	determinant	determinant	ADJ
ejpam-3176	384	9	estimator	estimator	NOUN
ejpam-3176	384	10	.	.	PUNCT
ejpam-3176	385	1	technometrics	technometric	NOUN
ejpam-3176	385	2	,	,	PUNCT
ejpam-3176	385	3	41(3):212–223	41(3):212–223	PROPN
ejpam-3176	385	4	,	,	PUNCT
ejpam-3176	385	5	1999	1999	NUM
ejpam-3176	385	6	.	.	PUNCT
ejpam-3176	386	1	[	[	X
ejpam-3176	386	2	26	26	NUM
ejpam-3176	386	3	]	]	X
ejpam-3176	386	4	rousseeuw	rousseeuw	PROPN
ejpam-3176	386	5	,	,	PUNCT
ejpam-3176	386	6	peter	peter	PROPN
ejpam-3176	386	7	j	j	PROPN
ejpam-3176	386	8	,	,	PUNCT
ejpam-3176	386	9	and	and	CCONJ
ejpam-3176	386	10	bert	bert	PROPN
ejpam-3176	386	11	c	c	PROPN
ejpam-3176	386	12	van	van	PROPN
ejpam-3176	386	13	zomeren	zomeren	PROPN
ejpam-3176	386	14	.	.	PUNCT
ejpam-3176	387	1	unmasking	unmask	VERB
ejpam-3176	387	2	multivariate	multivariate	NOUN
ejpam-3176	387	3	outliers	outlier	NOUN
ejpam-3176	387	4	and	and	CCONJ
ejpam-3176	387	5	leverage	leverage	NOUN
ejpam-3176	387	6	points	point	NOUN
ejpam-3176	387	7	.	.	PUNCT
ejpam-3176	388	1	journal	journal	NOUN
ejpam-3176	388	2	of	of	ADP
ejpam-3176	388	3	the	the	DET
ejpam-3176	388	4	american	american	PROPN
ejpam-3176	388	5	statistical	statistical	PROPN
ejpam-3176	388	6	association	association	PROPN
ejpam-3176	388	7	,	,	PUNCT
ejpam-3176	388	8	85(411):633–639	85(411):633–639	PROPN
ejpam-3176	388	9	,	,	PUNCT
ejpam-3176	388	10	1990	1990	NUM
ejpam-3176	388	11	.	.	PUNCT
ejpam-3176	389	1	[	[	X
ejpam-3176	389	2	27	27	NUM
ejpam-3176	389	3	]	]	X
ejpam-3176	389	4	todorov	todorov	NOUN
ejpam-3176	389	5	and	and	CCONJ
ejpam-3176	389	6	valentin	valentin	NOUN
ejpam-3176	389	7	.	.	PUNCT
ejpam-3176	389	8	robust	robust	ADJ
ejpam-3176	389	9	selection	selection	NOUN
ejpam-3176	389	10	of	of	ADP
ejpam-3176	389	11	variables	variable	NOUN
ejpam-3176	389	12	in	in	ADP
ejpam-3176	389	13	linear	linear	ADJ
ejpam-3176	389	14	discriminant	discriminant	ADJ
ejpam-3176	389	15	analysis	analysis	NOUN
ejpam-3176	389	16	.	.	PUNCT
ejpam-3176	390	1	statistical	statistical	ADJ
ejpam-3176	390	2	methods	method	NOUN
ejpam-3176	390	3	and	and	CCONJ
ejpam-3176	390	4	applications	application	NOUN
ejpam-3176	390	5	,	,	PUNCT
ejpam-3176	390	6	15(3):395–407	15(3):395–407	PROPN
ejpam-3176	390	7	,	,	PUNCT
ejpam-3176	390	8	2007	2007	NUM
ejpam-3176	390	9	.	.	PUNCT
ejpam-3176	391	1	[	[	X
ejpam-3176	391	2	28	28	NUM
ejpam-3176	391	3	]	]	X
ejpam-3176	391	4	s.	s.	PROPN
ejpam-3176	391	5	visuri	visuri	PROPN
ejpam-3176	391	6	,	,	PUNCT
ejpam-3176	391	7	h.	h.	PROPN
ejpam-3176	391	8	oja	oja	PROPN
ejpam-3176	391	9	,	,	PUNCT
ejpam-3176	391	10	and	and	CCONJ
ejpam-3176	392	1	v.	v.	ADP
ejpam-3176	392	2	koivunen	koivunen	PROPN
ejpam-3176	392	3	.	.	PUNCT
ejpam-3176	393	1	sign	sign	NOUN
ejpam-3176	393	2	and	and	CCONJ
ejpam-3176	393	3	rank	rank	NOUN
ejpam-3176	393	4	covariane	covariane	PROPN
ejpam-3176	393	5	matrices	matrix	NOUN
ejpam-3176	393	6	.	.	PUNCT
ejpam-3176	394	1	j.	j.	PROPN
ejpam-3176	394	2	statist	statist	PROPN
ejpam-3176	394	3	.	.	PUNCT
ejpam-3176	395	1	plann	plann	PROPN
ejpam-3176	395	2	.	.	PUNCT
ejpam-3176	396	1	inference	inference	NOUN
ejpam-3176	396	2	,	,	PUNCT
ejpam-3176	396	3	91:557575	91:557575	NUM
ejpam-3176	396	4	,	,	PUNCT
ejpam-3176	396	5	2000	2000	NUM
ejpam-3176	396	6	.	.	PUNCT
ejpam-3176	397	1	[	[	X
ejpam-3176	397	2	29	29	NUM
ejpam-3176	397	3	]	]	SYM
ejpam-3176	397	4	wiegand	wiegand	NOUN
ejpam-3176	397	5	,	,	PUNCT
ejpam-3176	397	6	patrick	patrick	PROPN
ejpam-3176	397	7	,	,	PUNCT
ejpam-3176	397	8	randy	randy	ADJ
ejpam-3176	397	9	pell	pell	NOUN
ejpam-3176	397	10	,	,	PUNCT
ejpam-3176	397	11	and	and	CCONJ
ejpam-3176	397	12	enric	enric	PROPN
ejpam-3176	397	13	comas	coma	NOUN
ejpam-3176	397	14	.	.	PUNCT
ejpam-3176	398	1	simultaneous	simultaneous	ADJ
ejpam-3176	398	2	variable	variable	ADJ
ejpam-3176	398	3	selection	selection	NOUN
ejpam-3176	398	4	and	and	CCONJ
ejpam-3176	398	5	outlier	outlier	NOUN
ejpam-3176	398	6	detection	detection	NOUN
ejpam-3176	398	7	using	use	VERB
ejpam-3176	398	8	a	a	DET
ejpam-3176	398	9	robust	robust	ADJ
ejpam-3176	398	10	genetic	genetic	ADJ
ejpam-3176	398	11	algorithm	algorithm	NOUN
ejpam-3176	398	12	.	.	PUNCT
ejpam-3176	399	1	chemometrics	chemometric	NOUN
ejpam-3176	399	2	and	and	CCONJ
ejpam-3176	399	3	intelligent	intelligent	ADJ
ejpam-3176	399	4	laboratory	laboratory	NOUN
ejpam-3176	399	5	systems	system	NOUN
ejpam-3176	399	6	,	,	PUNCT
ejpam-3176	399	7	98(2):108–114	98(2):108–114	NUM
ejpam-3176	399	8	,	,	PUNCT
ejpam-3176	399	9	2009	2009	NUM
ejpam-3176	399	10	.	.	PUNCT
ejpam-3176	400	1	[	[	X
ejpam-3176	400	2	30	30	NUM
ejpam-3176	400	3	]	]	X
ejpam-3176	400	4	jianfeng	jianfeng	PROPN
ejpam-3176	400	5	zhang	zhang	PROPN
ejpam-3176	400	6	,	,	PUNCT
ejpam-3176	400	7	david	david	PROPN
ejpam-3176	400	8	j.	j.	PROPN
ejpam-3176	400	9	olive	olive	PROPN
ejpam-3176	400	10	,	,	PUNCT
ejpam-3176	400	11	and	and	CCONJ
ejpam-3176	400	12	ping	ping	NOUN
ejpam-3176	400	13	ye	ye	PROPN
ejpam-3176	400	14	.	.	PROPN
ejpam-3176	400	15	robust	robust	ADJ
ejpam-3176	400	16	covariance	covariance	NOUN
ejpam-3176	400	17	matrix	matrix	NOUN
ejpam-3176	400	18	estimation	estimation	NOUN
ejpam-3176	400	19	with	with	ADP
ejpam-3176	400	20	canonical	canonical	ADJ
ejpam-3176	400	21	correlation	correlation	NOUN
ejpam-3176	400	22	analysis	analysis	NOUN
ejpam-3176	400	23	.	.	PUNCT
ejpam-3176	401	1	international	international	ADJ
ejpam-3176	401	2	journal	journal	PROPN
ejpam-3176	401	3	of	of	ADP
ejpam-3176	401	4	statistics	statistic	NOUN
ejpam-3176	401	5	and	and	CCONJ
ejpam-3176	401	6	probability	probability	NOUN
ejpam-3176	401	7	,	,	PUNCT
ejpam-3176	401	8	1(2):119–136	1(2):119–136	NUM
ejpam-3176	401	9	,	,	PUNCT
ejpam-3176	401	10	2012	2012	NUM
ejpam-3176	401	11	.	.	PUNCT