id	sid	tid	token	lemma	pos
h415p844p19	1	1	this	this	DET
h415p844p19	1	2	thesis	thesis	NOUN
h415p844p19	1	3	proposes	propose	VERB
h415p844p19	1	4	two	two	NUM
h415p844p19	1	5	algorithms	algorithm	NOUN
h415p844p19	1	6	for	for	ADP
h415p844p19	1	7	data	data	NOUN
h415p844p19	1	8	generation	generation	NOUN
h415p844p19	1	9	and	and	CCONJ
h415p844p19	1	10	density	density	NOUN
h415p844p19	1	11	estimation	estimation	NOUN
h415p844p19	1	12	.	.	PUNCT
h415p844p19	2	1	the	the	DET
h415p844p19	2	2	algorithms	algorithm	NOUN
h415p844p19	2	3	use	use	VERB
h415p844p19	2	4	gradient	gradient	NOUN
h415p844p19	2	5	flow	flow	NOUN
h415p844p19	2	6	formulation	formulation	NOUN
h415p844p19	2	7	and	and	CCONJ
h415p844p19	2	8	neural	neural	ADJ
h415p844p19	2	9	-	-	PUNCT
h415p844p19	2	10	network	network	NOUN
h415p844p19	2	11	-	-	PUNCT
h415p844p19	2	12	based	base	VERB
h415p844p19	2	13	discretization	discretization	NOUN
h415p844p19	2	14	of	of	ADP
h415p844p19	2	15	the	the	DET
h415p844p19	2	16	gradient	gradient	NOUN
h415p844p19	2	17	flow	flow	NOUN
h415p844p19	2	18	.	.	PUNCT
h415p844p19	3	1	in	in	ADP
h415p844p19	3	2	our	our	PRON
h415p844p19	3	3	first	first	ADJ
h415p844p19	3	4	algorithm	algorithm	NOUN
h415p844p19	3	5	,	,	PUNCT
h415p844p19	3	6	we	we	PRON
h415p844p19	3	7	uniquely	uniquely	ADV
h415p844p19	3	8	propose	propose	VERB
h415p844p19	3	9	using	use	VERB
h415p844p19	3	10	maximum	maximum	ADJ
h415p844p19	3	11	mean	mean	NOUN
h415p844p19	3	12	discrepancy(mmd	discrepancy(mmd	PROPN
h415p844p19	3	13	)	)	PUNCT
h415p844p19	3	14	as	as	ADP
h415p844p19	3	15	an	an	DET
h415p844p19	3	16	energy	energy	NOUN
h415p844p19	3	17	functional	functional	ADJ
h415p844p19	3	18	to	to	PART
h415p844p19	3	19	measure	measure	VERB
h415p844p19	3	20	the	the	DET
h415p844p19	3	21	dissimilarity	dissimilarity	NOUN
h415p844p19	3	22	between	between	ADP
h415p844p19	3	23	distributions	distribution	NOUN
h415p844p19	3	24	.	.	PUNCT
h415p844p19	4	1	in	in	ADP
h415p844p19	4	2	our	our	PRON
h415p844p19	4	3	second	second	ADJ
h415p844p19	4	4	algorithm	algorithm	NOUN
h415p844p19	4	5	,	,	PUNCT
h415p844p19	4	6	we	we	PRON
h415p844p19	4	7	propose	propose	VERB
h415p844p19	4	8	the	the	DET
h415p844p19	4	9	neural	neural	ADJ
h415p844p19	4	10	network	network	NOUN
h415p844p19	4	11	-	-	PUNCT
h415p844p19	4	12	based	base	VERB
h415p844p19	4	13	algorithm	algorithm	PROPN
h415p844p19	4	14	that	that	PRON
h415p844p19	4	15	uses	use	VERB
h415p844p19	4	16	the	the	DET
h415p844p19	4	17	second	second	ADJ
h415p844p19	4	18	-	-	PUNCT
h415p844p19	4	19	order	order	NOUN
h415p844p19	4	20	backward	backward	ADJ
h415p844p19	4	21	differentiation	differentiation	NOUN
h415p844p19	4	22	formula(bdf2	formula(bdf2	SPACE
h415p844p19	4	23	)	)	PUNCT
h415p844p19	4	24	scheme	scheme	NOUN
h415p844p19	4	25	to	to	PART
h415p844p19	4	26	discretize	discretize	VERB
h415p844p19	4	27	in	in	ADP
h415p844p19	4	28	the	the	DET
h415p844p19	4	29	time	time	NOUN
h415p844p19	4	30	dimension	dimension	NOUN
h415p844p19	4	31	.	.	PUNCT
h415p844p19	5	1	meanwhile	meanwhile	ADV
h415p844p19	5	2	,	,	PUNCT
h415p844p19	5	3	the	the	DET
h415p844p19	5	4	importance	importance	NOUN
h415p844p19	5	5	of	of	ADP
h415p844p19	5	6	understanding	understanding	NOUN
h415p844p19	5	7	and	and	CCONJ
h415p844p19	5	8	leveraging	leverage	VERB
h415p844p19	5	9	the	the	DET
h415p844p19	5	10	underlying	underlying	ADJ
h415p844p19	5	11	geometric	geometric	ADJ
h415p844p19	5	12	structures	structure	NOUN
h415p844p19	5	13	in	in	ADP
h415p844p19	5	14	datasets	dataset	NOUN
h415p844p19	5	15	has	have	AUX
h415p844p19	5	16	been	be	AUX
h415p844p19	5	17	increasingly	increasingly	ADV
h415p844p19	5	18	recognized	recognize	VERB
h415p844p19	5	19	.	.	PUNCT
h415p844p19	6	1	the	the	DET
h415p844p19	6	2	other	other	ADJ
h415p844p19	6	3	topic	topic	NOUN
h415p844p19	6	4	of	of	ADP
h415p844p19	6	5	this	this	DET
h415p844p19	6	6	thesis	thesis	NOUN
h415p844p19	6	7	is	be	AUX
h415p844p19	6	8	utilizing	utilize	VERB
h415p844p19	6	9	intrinsic	intrinsic	ADJ
h415p844p19	6	10	geometric	geometric	ADJ
h415p844p19	6	11	information	information	NOUN
h415p844p19	6	12	to	to	PART
h415p844p19	6	13	enhance	enhance	VERB
h415p844p19	6	14	learning	learn	VERB
h415p844p19	6	15	.	.	PUNCT
h415p844p19	7	1	we	we	PRON
h415p844p19	7	2	propose	propose	VERB
h415p844p19	7	3	two	two	NUM
h415p844p19	7	4	algorithms	algorithm	NOUN
h415p844p19	7	5	to	to	PART
h415p844p19	7	6	extract	extract	VERB
h415p844p19	7	7	intrinsic	intrinsic	ADJ
h415p844p19	7	8	geometric	geometric	ADJ
h415p844p19	7	9	information	information	NOUN
h415p844p19	7	10	from	from	ADP
h415p844p19	7	11	the	the	DET
h415p844p19	7	12	data	datum	NOUN
h415p844p19	7	13	.	.	PUNCT
h415p844p19	8	1	the	the	DET
h415p844p19	8	2	first	first	ADJ
h415p844p19	8	3	algorithm	algorithm	PROPN
h415p844p19	8	4	is	be	AUX
h415p844p19	8	5	constructing	construct	VERB
h415p844p19	8	6	a	a	DET
h415p844p19	8	7	graph	graph	NOUN
h415p844p19	8	8	for	for	ADP
h415p844p19	8	9	points	point	NOUN
h415p844p19	8	10	in	in	ADP
h415p844p19	8	11	the	the	DET
h415p844p19	8	12	dataset	dataset	NOUN
h415p844p19	8	13	and	and	CCONJ
h415p844p19	8	14	applying	apply	VERB
h415p844p19	8	15	the	the	DET
h415p844p19	8	16	graph	graph	NOUN
h415p844p19	8	17	neural	neural	ADJ
h415p844p19	8	18	network	network	NOUN
h415p844p19	8	19	to	to	PART
h415p844p19	8	20	conduct	conduct	VERB
h415p844p19	8	21	the	the	DET
h415p844p19	8	22	learning	learning	NOUN
h415p844p19	8	23	tasks	task	NOUN
h415p844p19	8	24	like	like	ADP
h415p844p19	8	25	regression	regression	NOUN
h415p844p19	8	26	and	and	CCONJ
h415p844p19	8	27	classification	classification	NOUN
h415p844p19	8	28	.	.	PUNCT
h415p844p19	9	1	the	the	DET
h415p844p19	9	2	second	second	ADJ
h415p844p19	9	3	algorithm	algorithm	NOUN
h415p844p19	9	4	is	be	AUX
h415p844p19	9	5	manipulating	manipulate	VERB
h415p844p19	9	6	kernels	kernel	NOUN
h415p844p19	9	7	in	in	ADP
h415p844p19	9	8	gaussian	gaussian	ADJ
h415p844p19	9	9	processes	process	NOUN
h415p844p19	9	10	and	and	CCONJ
h415p844p19	9	11	using	use	VERB
h415p844p19	9	12	the	the	DET
h415p844p19	9	13	properties	property	NOUN
h415p844p19	9	14	of	of	ADP
h415p844p19	9	15	the	the	DET
h415p844p19	9	16	diffusion	diffusion	NOUN
h415p844p19	9	17	process	process	NOUN
h415p844p19	9	18	to	to	PART
h415p844p19	9	19	redefine	redefine	VERB
h415p844p19	9	20	metrics	metric	NOUN
h415p844p19	9	21	in	in	ADP
h415p844p19	9	22	the	the	DET
h415p844p19	9	23	dataset	dataset	NOUN
h415p844p19	9	24	and	and	CCONJ
h415p844p19	9	25	,	,	PUNCT
h415p844p19	9	26	hence	hence	ADV
h415p844p19	9	27	,	,	PUNCT
h415p844p19	9	28	extract	extract	VERB
h415p844p19	9	29	geometric	geometric	ADJ
h415p844p19	9	30	information	information	NOUN
h415p844p19	9	31	.	.	PUNCT
h415p844p19	10	1	then	then	ADV
h415p844p19	10	2	mcmc	mcmc	NOUN
h415p844p19	10	3	method	method	NOUN
h415p844p19	10	4	can	can	AUX
h415p844p19	10	5	be	be	AUX
h415p844p19	10	6	applied	apply	VERB
h415p844p19	10	7	to	to	PART
h415p844p19	10	8	obtain	obtain	VERB
h415p844p19	10	9	the	the	DET
h415p844p19	10	10	samples	sample	NOUN
h415p844p19	10	11	of	of	ADP
h415p844p19	10	12	parameters	parameter	NOUN
h415p844p19	10	13	.	.	PUNCT