id sid tid token lemma pos ajst-19323 1 1 academic academic ADJ ajst-19323 1 2 journal journal NOUN ajst-19323 1 3 of of ADP ajst-19323 1 4 science science NOUN ajst-19323 1 5 and and CCONJ ajst-19323 1 6 technology technology NOUN ajst-19323 1 7 issn issn NOUN ajst-19323 1 8 : : PUNCT ajst-19323 1 9 2771 2771 NUM ajst-19323 1 10 - - SYM ajst-19323 1 11 3032 3032 NUM ajst-19323 1 12 | | NOUN ajst-19323 1 13 vol vol NOUN ajst-19323 1 14 . . PROPN ajst-19323 2 1 10 10 NUM ajst-19323 2 2 , , PUNCT ajst-19323 2 3 no no INTJ ajst-19323 2 4 . . NOUN ajst-19323 2 5 1 1 NUM ajst-19323 2 6 , , PUNCT ajst-19323 2 7 2024 2024 NUM ajst-19323 2 8 364 364 NUM ajst-19323 2 9 multi‐label multi‐label NOUN ajst-19323 2 10 feature feature NOUN ajst-19323 2 11 selection selection NOUN ajst-19323 2 12 based base VERB ajst-19323 2 13 on on ADP ajst-19323 2 14 label‐specific label‐specific ADJ ajst-19323 2 15 features feature NOUN ajst-19323 2 16 and and CCONJ ajst-19323 2 17 manifold manifold ADJ ajst-19323 2 18 learning learning NOUN ajst-19323 2 19 wanzhu wanzhu PROPN ajst-19323 2 20 wang wang PROPN ajst-19323 2 21 , , PUNCT ajst-19323 2 22 yong yong PROPN ajst-19323 2 23 liu liu PROPN ajst-19323 2 24 henan henan PROPN ajst-19323 2 25 university university PROPN ajst-19323 2 26 of of ADP ajst-19323 2 27 science science NOUN ajst-19323 2 28 and and CCONJ ajst-19323 2 29 technology technology NOUN ajst-19323 2 30 , , PUNCT ajst-19323 2 31 471000 471000 NUM ajst-19323 2 32 , , PUNCT ajst-19323 2 33 luoyang luoyang PROPN ajst-19323 2 34 , , PUNCT ajst-19323 2 35 china china PROPN ajst-19323 2 36 abstract abstract NOUN ajst-19323 2 37 : : PUNCT ajst-19323 2 38 each each DET ajst-19323 2 39 instance instance NOUN ajst-19323 2 40 in in ADP ajst-19323 2 41 multi multi ADJ ajst-19323 2 42 - - ADJ ajst-19323 2 43 label label ADJ ajst-19323 2 44 data datum NOUN ajst-19323 2 45 is be AUX ajst-19323 2 46 associated associate VERB ajst-19323 2 47 with with ADP ajst-19323 2 48 multiple multiple ADJ ajst-19323 2 49 labels label NOUN ajst-19323 2 50 , , PUNCT ajst-19323 2 51 and and CCONJ ajst-19323 2 52 there there PRON ajst-19323 2 53 are be VERB ajst-19323 2 54 irrelevant irrelevant ADJ ajst-19323 2 55 or or CCONJ ajst-19323 2 56 redundant redundant ADJ ajst-19323 2 57 features feature NOUN ajst-19323 2 58 in in ADP ajst-19323 2 59 its its PRON ajst-19323 2 60 feature feature NOUN ajst-19323 2 61 space space NOUN ajst-19323 2 62 , , PUNCT ajst-19323 2 63 which which PRON ajst-19323 2 64 leads lead VERB ajst-19323 2 65 to to ADP ajst-19323 2 66 the the DET ajst-19323 2 67 performance performance NOUN ajst-19323 2 68 degradation degradation NOUN ajst-19323 2 69 of of ADP ajst-19323 2 70 multi multi ADJ ajst-19323 2 71 - - ADJ ajst-19323 2 72 label label ADJ ajst-19323 2 73 learning learning NOUN ajst-19323 2 74 algorithms algorithm NOUN ajst-19323 2 75 . . PUNCT ajst-19323 3 1 multi multi ADJ ajst-19323 3 2 - - ADJ ajst-19323 3 3 label label ADJ ajst-19323 3 4 feature feature NOUN ajst-19323 3 5 selection selection NOUN ajst-19323 3 6 selects select NOUN ajst-19323 3 7 representative representative ADJ ajst-19323 3 8 features feature NOUN ajst-19323 3 9 from from ADP ajst-19323 3 10 the the DET ajst-19323 3 11 feature feature NOUN ajst-19323 3 12 space space NOUN ajst-19323 3 13 to to PART ajst-19323 3 14 improve improve VERB ajst-19323 3 15 the the DET ajst-19323 3 16 accuracy accuracy NOUN ajst-19323 3 17 of of ADP ajst-19323 3 18 the the DET ajst-19323 3 19 model model NOUN ajst-19323 3 20 . . PUNCT ajst-19323 4 1 due due ADP ajst-19323 4 2 to to ADP ajst-19323 4 3 the the DET ajst-19323 4 4 high high ADJ ajst-19323 4 5 cost cost NOUN ajst-19323 4 6 of of ADP ajst-19323 4 7 labels label NOUN ajst-19323 4 8 and and CCONJ ajst-19323 4 9 the the DET ajst-19323 4 10 difficulty difficulty NOUN ajst-19323 4 11 of of ADP ajst-19323 4 12 data data NOUN ajst-19323 4 13 collection collection NOUN ajst-19323 4 14 , , PUNCT ajst-19323 4 15 there there PRON ajst-19323 4 16 will will AUX ajst-19323 4 17 be be AUX ajst-19323 4 18 some some DET ajst-19323 4 19 missing miss VERB ajst-19323 4 20 labels label NOUN ajst-19323 4 21 in in ADP ajst-19323 4 22 the the DET ajst-19323 4 23 data data NOUN ajst-19323 4 24 set set VERB ajst-19323 4 25 , , PUNCT ajst-19323 4 26 which which PRON ajst-19323 4 27 affects affect VERB ajst-19323 4 28 the the DET ajst-19323 4 29 accuracy accuracy NOUN ajst-19323 4 30 of of ADP ajst-19323 4 31 feature feature NOUN ajst-19323 4 32 selection selection NOUN ajst-19323 4 33 . . PUNCT ajst-19323 5 1 to to PART ajst-19323 5 2 solve solve VERB ajst-19323 5 3 this this DET ajst-19323 5 4 problem problem NOUN ajst-19323 5 5 , , PUNCT ajst-19323 5 6 a a DET ajst-19323 5 7 multi multi ADJ ajst-19323 5 8 - - ADJ ajst-19323 5 9 label label ADJ ajst-19323 5 10 feature feature NOUN ajst-19323 5 11 selection selection NOUN ajst-19323 5 12 algorithm algorithm NOUN ajst-19323 5 13 based base VERB ajst-19323 5 14 on on ADP ajst-19323 5 15 label label NOUN ajst-19323 5 16 - - PUNCT ajst-19323 5 17 specific specific ADJ ajst-19323 5 18 features feature NOUN ajst-19323 5 19 and and CCONJ ajst-19323 5 20 manifold manifold ADJ ajst-19323 5 21 learning learning NOUN ajst-19323 5 22 is be AUX ajst-19323 5 23 proposed propose VERB ajst-19323 5 24 . . PUNCT ajst-19323 6 1 the the DET ajst-19323 6 2 algorithm algorithm NOUN ajst-19323 6 3 uses use VERB ajst-19323 6 4 the the DET ajst-19323 6 5 linear linear ADJ ajst-19323 6 6 relationship relationship NOUN ajst-19323 6 7 between between ADP ajst-19323 6 8 the the DET ajst-19323 6 9 features feature NOUN ajst-19323 6 10 and and CCONJ ajst-19323 6 11 labels label NOUN ajst-19323 6 12 in in ADP ajst-19323 6 13 known know VERB ajst-19323 6 14 label label NOUN ajst-19323 6 15 samples sample NOUN ajst-19323 6 16 to to PART ajst-19323 6 17 build build VERB ajst-19323 6 18 a a DET ajst-19323 6 19 linear linear ADJ ajst-19323 6 20 regression regression NOUN ajst-19323 6 21 model model NOUN ajst-19323 6 22 for for ADP ajst-19323 6 23 learning learn VERB ajst-19323 6 24 label label NOUN ajst-19323 6 25 - - PUNCT ajst-19323 6 26 specific specific ADJ ajst-19323 6 27 features feature NOUN ajst-19323 6 28 . . PUNCT ajst-19323 7 1 by by ADP ajst-19323 7 2 using use VERB ajst-19323 7 3 the the DET ajst-19323 7 4 nonlinear nonlinear ADJ ajst-19323 7 5 relation relation NOUN ajst-19323 7 6 between between ADP ajst-19323 7 7 instances instance NOUN ajst-19323 7 8 and and CCONJ ajst-19323 7 9 the the DET ajst-19323 7 10 nonlinear nonlinear ADJ ajst-19323 7 11 relation relation NOUN ajst-19323 7 12 between between ADP ajst-19323 7 13 features feature NOUN ajst-19323 7 14 , , PUNCT ajst-19323 7 15 we we PRON ajst-19323 7 16 can can AUX ajst-19323 7 17 precisely precisely ADV ajst-19323 7 18 learn learn VERB ajst-19323 7 19 the the DET ajst-19323 7 20 label label NOUN ajst-19323 7 21 - - PUNCT ajst-19323 7 22 specific specific ADJ ajst-19323 7 23 features feature NOUN ajst-19323 7 24 . . PUNCT ajst-19323 8 1 we we PRON ajst-19323 8 2 use use VERB ajst-19323 8 3 the the DET ajst-19323 8 4 laplacian laplacian ADJ ajst-19323 8 5 feature feature NOUN ajst-19323 8 6 mapping mapping NOUN ajst-19323 8 7 method method NOUN ajst-19323 8 8 to to PART ajst-19323 8 9 construct construct VERB ajst-19323 8 10 the the DET ajst-19323 8 11 instance instance NOUN ajst-19323 8 12 manifold manifold ADJ ajst-19323 8 13 model model NOUN ajst-19323 8 14 and and CCONJ ajst-19323 8 15 the the DET ajst-19323 8 16 feature feature NOUN ajst-19323 8 17 manifold manifold ADJ ajst-19323 8 18 model model NOUN ajst-19323 8 19 , , PUNCT ajst-19323 8 20 which which PRON ajst-19323 8 21 are be AUX ajst-19323 8 22 also also ADV ajst-19323 8 23 used use VERB ajst-19323 8 24 as as ADP ajst-19323 8 25 the the DET ajst-19323 8 26 regular regular ADJ ajst-19323 8 27 term term NOUN ajst-19323 8 28 constraint constraint NOUN ajst-19323 8 29 weight weight NOUN ajst-19323 8 30 matrix matrix NOUN ajst-19323 8 31 . . PUNCT ajst-19323 9 1 the the DET ajst-19323 9 2 final final ADJ ajst-19323 9 3 model model NOUN ajst-19323 9 4 can can AUX ajst-19323 9 5 not not PART ajst-19323 9 6 only only ADV ajst-19323 9 7 complete complete VERB ajst-19323 9 8 the the DET ajst-19323 9 9 missing miss VERB ajst-19323 9 10 labels label NOUN ajst-19323 9 11 , , PUNCT ajst-19323 9 12 but but CCONJ ajst-19323 9 13 also also ADV ajst-19323 9 14 select select VERB ajst-19323 9 15 sparse sparse ADJ ajst-19323 9 16 and and CCONJ ajst-19323 9 17 representative representative ADJ ajst-19323 9 18 features feature NOUN ajst-19323 9 19 . . PUNCT ajst-19323 10 1 the the DET ajst-19323 10 2 feature feature NOUN ajst-19323 10 3 selection selection NOUN ajst-19323 10 4 is be AUX ajst-19323 10 5 carried carry VERB ajst-19323 10 6 out out ADP ajst-19323 10 7 by by ADP ajst-19323 10 8 analyzing analyze VERB ajst-19323 10 9 the the DET ajst-19323 10 10 weight weight NOUN ajst-19323 10 11 of of ADP ajst-19323 10 12 the the DET ajst-19323 10 13 feature feature NOUN ajst-19323 10 14 given give VERB ajst-19323 10 15 by by ADP ajst-19323 10 16 the the DET ajst-19323 10 17 final final ADJ ajst-19323 10 18 model model NOUN ajst-19323 10 19 . . PUNCT ajst-19323 11 1 experiments experiment NOUN ajst-19323 11 2 were be AUX ajst-19323 11 3 conducted conduct VERB ajst-19323 11 4 to to PART ajst-19323 11 5 verify verify VERB ajst-19323 11 6 the the DET ajst-19323 11 7 effectiveness effectiveness NOUN ajst-19323 11 8 of of ADP ajst-19323 11 9 the the DET ajst-19323 11 10 proposed propose VERB ajst-19323 11 11 algorithm algorithm NOUN ajst-19323 11 12 under under ADP ajst-19323 11 13 different different ADJ ajst-19323 11 14 label label NOUN ajst-19323 11 15 deletion deletion NOUN ajst-19323 11 16 rates rate NOUN ajst-19323 11 17 on on ADP ajst-19323 11 18 four four NUM ajst-19323 11 19 evaluation evaluation NOUN ajst-19323 11 20 indexes index NOUN ajst-19323 11 21 . . PUNCT ajst-19323 12 1 keywords keyword NOUN ajst-19323 12 2 : : PUNCT ajst-19323 12 3 multi multi ADJ ajst-19323 12 4 - - ADJ ajst-19323 12 5 label label ADJ ajst-19323 12 6 learning learning NOUN ajst-19323 12 7 ; ; PUNCT ajst-19323 12 8 manifold manifold ADJ ajst-19323 12 9 learning learning NOUN ajst-19323 12 10 ; ; PUNCT ajst-19323 12 11 label label NOUN ajst-19323 12 12 - - PUNCT ajst-19323 12 13 specific specific ADJ ajst-19323 12 14 features feature NOUN ajst-19323 12 15 ; ; PUNCT ajst-19323 12 16 feature feature NOUN ajst-19323 12 17 selection selection NOUN ajst-19323 12 18 . . PUNCT ajst-19323 13 1 1 1 X ajst-19323 13 2 . . X ajst-19323 13 3 introduction introduction NOUN ajst-19323 13 4 multi multi ADJ ajst-19323 13 5 - - ADJ ajst-19323 13 6 label label ADJ ajst-19323 13 7 learning learning NOUN ajst-19323 13 8 is be AUX ajst-19323 13 9 characterized characterize VERB ajst-19323 13 10 by by ADP ajst-19323 13 11 high high ADJ ajst-19323 13 12 efficiency efficiency NOUN ajst-19323 13 13 and and CCONJ ajst-19323 13 14 flexibility flexibility NOUN ajst-19323 13 15 , , PUNCT ajst-19323 13 16 and and CCONJ ajst-19323 13 17 has have AUX ajst-19323 13 18 shown show VERB ajst-19323 13 19 wide wide ADJ ajst-19323 13 20 application application NOUN ajst-19323 13 21 prospects prospect NOUN ajst-19323 13 22 in in ADP ajst-19323 13 23 many many ADJ ajst-19323 13 24 application application NOUN ajst-19323 13 25 fields field NOUN ajst-19323 13 26 such such ADJ ajst-19323 13 27 as as ADP ajst-19323 13 28 machine machine NOUN ajst-19323 13 29 learning learning NOUN ajst-19323 13 30 , , PUNCT ajst-19323 13 31 data datum NOUN ajst-19323 13 32 mining mining NOUN ajst-19323 13 33 and and CCONJ ajst-19323 13 34 pattern pattern NOUN ajst-19323 13 35 recognition recognition NOUN ajst-19323 14 1 [ [ X ajst-19323 14 2 1 1 NUM ajst-19323 14 3 - - SYM ajst-19323 14 4 3 3 NUM ajst-19323 14 5 ] ] PUNCT ajst-19323 14 6 , , PUNCT ajst-19323 14 7 and and CCONJ ajst-19323 14 8 is be AUX ajst-19323 14 9 also also ADV ajst-19323 14 10 widely widely ADV ajst-19323 14 11 used use VERB ajst-19323 14 12 in in ADP ajst-19323 14 13 various various ADJ ajst-19323 14 14 practical practical ADJ ajst-19323 14 15 tasks task NOUN ajst-19323 14 16 , , PUNCT ajst-19323 14 17 such such ADJ ajst-19323 14 18 as as ADP ajst-19323 14 19 text text NOUN ajst-19323 14 20 classification classification NOUN ajst-19323 14 21 , , PUNCT ajst-19323 14 22 music music NOUN ajst-19323 14 23 classification classification NOUN ajst-19323 14 24 and and CCONJ ajst-19323 14 25 gene gene NOUN ajst-19323 14 26 function function NOUN ajst-19323 14 27 prediction prediction NOUN ajst-19323 14 28 . . PUNCT ajst-19323 15 1 for for ADP ajst-19323 15 2 example example NOUN ajst-19323 15 3 , , PUNCT ajst-19323 15 4 in in ADP ajst-19323 15 5 the the DET ajst-19323 15 6 text text NOUN ajst-19323 15 7 category category NOUN ajst-19323 15 8 , , PUNCT ajst-19323 15 9 a a DET ajst-19323 15 10 news news NOUN ajst-19323 15 11 story story NOUN ajst-19323 15 12 can can AUX ajst-19323 15 13 be be AUX ajst-19323 15 14 classified classify VERB ajst-19323 15 15 as as ADP ajst-19323 15 16 both both CCONJ ajst-19323 15 17 social social ADJ ajst-19323 15 18 and and CCONJ ajst-19323 15 19 technological technological ADJ ajst-19323 15 20 ; ; PUNCT ajst-19323 15 21 in in ADP ajst-19323 15 22 the the DET ajst-19323 15 23 music music NOUN ajst-19323 15 24 category category NOUN ajst-19323 15 25 , , PUNCT ajst-19323 15 26 a a DET ajst-19323 15 27 piece piece NOUN ajst-19323 15 28 of of ADP ajst-19323 15 29 music music NOUN ajst-19323 15 30 may may AUX ajst-19323 15 31 belong belong VERB ajst-19323 15 32 to to ADP ajst-19323 15 33 both both PRON ajst-19323 15 34 light light ADJ ajst-19323 15 35 music music NOUN ajst-19323 15 36 and and CCONJ ajst-19323 15 37 english english ADJ ajst-19323 15 38 songs song NOUN ajst-19323 15 39 . . PUNCT ajst-19323 16 1 multiple multiple ADJ ajst-19323 16 2 concepts concept NOUN ajst-19323 16 3 can can AUX ajst-19323 16 4 be be AUX ajst-19323 16 5 expressed express VERB ajst-19323 16 6 by by ADP ajst-19323 16 7 labeling label VERB ajst-19323 16 8 each each DET ajst-19323 16 9 instance instance NOUN ajst-19323 16 10 differently differently ADV ajst-19323 16 11 . . PUNCT ajst-19323 17 1 however however ADV ajst-19323 17 2 , , PUNCT ajst-19323 17 3 in in ADP ajst-19323 17 4 such such ADJ ajst-19323 17 5 cases case NOUN ajst-19323 17 6 where where SCONJ ajst-19323 17 7 multi multi ADJ ajst-19323 17 8 - - ADJ ajst-19323 17 9 label label ADJ ajst-19323 17 10 classification classification NOUN ajst-19323 17 11 is be AUX ajst-19323 17 12 required require VERB ajst-19323 17 13 , , PUNCT ajst-19323 17 14 attaching attach VERB ajst-19323 17 15 only only ADV ajst-19323 17 16 one one NUM ajst-19323 17 17 label label NOUN ajst-19323 17 18 to to ADP ajst-19323 17 19 each each DET ajst-19323 17 20 instance instance NOUN ajst-19323 17 21 does do AUX ajst-19323 17 22 not not PART ajst-19323 17 23 express express VERB ajst-19323 17 24 the the DET ajst-19323 17 25 underlying underlie VERB ajst-19323 17 26 complex complex ADJ ajst-19323 17 27 semantics semantic NOUN ajst-19323 17 28 . . PUNCT ajst-19323 18 1 therefore therefore ADV ajst-19323 18 2 , , PUNCT ajst-19323 18 3 multi multi ADJ ajst-19323 18 4 - - ADJ ajst-19323 18 5 label label ADJ ajst-19323 18 6 learning learning NOUN ajst-19323 18 7 is be AUX ajst-19323 18 8 achieved achieve VERB ajst-19323 18 9 by by ADP ajst-19323 18 10 assigning assign VERB ajst-19323 18 11 multiple multiple ADJ ajst-19323 18 12 labels label NOUN ajst-19323 18 13 to to ADP ajst-19323 18 14 an an DET ajst-19323 18 15 instance instance NOUN ajst-19323 18 16 at at ADP ajst-19323 18 17 the the DET ajst-19323 18 18 same same ADJ ajst-19323 18 19 time time NOUN ajst-19323 18 20 . . PUNCT ajst-19323 19 1 like like ADP ajst-19323 19 2 single single ADJ ajst-19323 19 3 - - PUNCT ajst-19323 19 4 label label NOUN ajst-19323 19 5 learning learning NOUN ajst-19323 19 6 , , PUNCT ajst-19323 19 7 the the DET ajst-19323 19 8 feature feature NOUN ajst-19323 19 9 space space NOUN ajst-19323 19 10 of of ADP ajst-19323 19 11 multi multi ADJ ajst-19323 19 12 - - ADJ ajst-19323 19 13 label label ADJ ajst-19323 19 14 learning learning NOUN ajst-19323 19 15 is be AUX ajst-19323 19 16 often often ADV ajst-19323 19 17 a a DET ajst-19323 19 18 set set NOUN ajst-19323 19 19 containing contain VERB ajst-19323 19 20 a a DET ajst-19323 19 21 large large ADJ ajst-19323 19 22 number number NOUN ajst-19323 19 23 of of ADP ajst-19323 19 24 features feature NOUN ajst-19323 19 25 , , PUNCT ajst-19323 19 26 which which PRON ajst-19323 19 27 may may AUX ajst-19323 19 28 have have VERB ajst-19323 19 29 up up ADP ajst-19323 19 30 to to PART ajst-19323 19 31 thousands thousand NOUN ajst-19323 19 32 of of ADP ajst-19323 19 33 dimensions dimension NOUN ajst-19323 19 34 [ [ X ajst-19323 19 35 4 4 NUM ajst-19323 19 36 ] ] PUNCT ajst-19323 19 37 . . PUNCT ajst-19323 20 1 in in ADP ajst-19323 20 2 text text NOUN ajst-19323 20 3 classification classification NOUN ajst-19323 20 4 , , PUNCT ajst-19323 20 5 the the DET ajst-19323 20 6 complex complex ADJ ajst-19323 20 7 semantic semantic ADJ ajst-19323 20 8 information information NOUN ajst-19323 20 9 of of ADP ajst-19323 20 10 a a DET ajst-19323 20 11 text text NOUN ajst-19323 20 12 can can AUX ajst-19323 20 13 be be AUX ajst-19323 20 14 represented represent VERB ajst-19323 20 15 in in ADP ajst-19323 20 16 detail detail NOUN ajst-19323 20 17 by by ADP ajst-19323 20 18 thousands thousand NOUN ajst-19323 20 19 of of ADP ajst-19323 20 20 features feature NOUN ajst-19323 20 21 . . PUNCT ajst-19323 21 1 by by ADP ajst-19323 21 2 extracting extract VERB ajst-19323 21 3 and and CCONJ ajst-19323 21 4 utilizing utilize VERB ajst-19323 21 5 these these DET ajst-19323 21 6 features feature NOUN ajst-19323 21 7 effectively effectively ADV ajst-19323 21 8 , , PUNCT ajst-19323 21 9 we we PRON ajst-19323 21 10 can can AUX ajst-19323 21 11 understand understand VERB ajst-19323 21 12 the the DET ajst-19323 21 13 intrinsic intrinsic ADJ ajst-19323 21 14 meaning meaning NOUN ajst-19323 21 15 and and CCONJ ajst-19323 21 16 context context NOUN ajst-19323 21 17 of of ADP ajst-19323 21 18 the the DET ajst-19323 21 19 text text NOUN ajst-19323 21 20 more more ADV ajst-19323 21 21 deeply deeply ADV ajst-19323 21 22 . . PUNCT ajst-19323 22 1 however however ADV ajst-19323 22 2 , , PUNCT ajst-19323 22 3 in in ADP ajst-19323 22 4 the the DET ajst-19323 22 5 feature feature NOUN ajst-19323 22 6 set set NOUN ajst-19323 22 7 , , PUNCT ajst-19323 22 8 most most ADJ ajst-19323 22 9 of of ADP ajst-19323 22 10 the the DET ajst-19323 22 11 features feature NOUN ajst-19323 22 12 often often ADV ajst-19323 22 13 present present VERB ajst-19323 22 14 redundant redundant ADJ ajst-19323 22 15 or or CCONJ ajst-19323 22 16 irrelevant irrelevant ADJ ajst-19323 22 17 features feature NOUN ajst-19323 22 18 . . PUNCT ajst-19323 23 1 these these DET ajst-19323 23 2 redundant redundant ADJ ajst-19323 23 3 and and CCONJ ajst-19323 23 4 uncorrelated uncorrelated ADJ ajst-19323 23 5 features feature NOUN ajst-19323 23 6 not not PART ajst-19323 23 7 only only ADV ajst-19323 23 8 increase increase VERB ajst-19323 23 9 the the DET ajst-19323 23 10 computational computational ADJ ajst-19323 23 11 complexity complexity NOUN ajst-19323 23 12 , , PUNCT ajst-19323 23 13 but but CCONJ ajst-19323 23 14 also also ADV ajst-19323 23 15 may may AUX ajst-19323 23 16 negatively negatively ADV ajst-19323 23 17 affect affect VERB ajst-19323 23 18 the the DET ajst-19323 23 19 performance performance NOUN ajst-19323 23 20 of of ADP ajst-19323 23 21 multi multi ADJ ajst-19323 23 22 - - ADJ ajst-19323 23 23 label label ADJ ajst-19323 23 24 learning learning NOUN ajst-19323 23 25 algorithms algorithm NOUN ajst-19323 23 26 , , PUNCT ajst-19323 23 27 and and CCONJ ajst-19323 23 28 also also ADV ajst-19323 23 29 lead lead VERB ajst-19323 23 30 to to ADP ajst-19323 23 31 overfitting overfitte VERB ajst-19323 23 32 and and CCONJ ajst-19323 23 33 other other ADJ ajst-19323 23 34 problems problem NOUN ajst-19323 23 35 . . PUNCT ajst-19323 24 1 in in ADP ajst-19323 24 2 order order NOUN ajst-19323 24 3 to to PART ajst-19323 24 4 solve solve VERB ajst-19323 24 5 this this DET ajst-19323 24 6 problem problem NOUN ajst-19323 24 7 , , PUNCT ajst-19323 24 8 researchers researcher NOUN ajst-19323 24 9 proposed propose VERB ajst-19323 24 10 a a DET ajst-19323 24 11 multi multi ADJ ajst-19323 24 12 - - ADJ ajst-19323 24 13 label label ADJ ajst-19323 24 14 feature feature NOUN ajst-19323 24 15 selection selection NOUN ajst-19323 24 16 algorithm algorithm NOUN ajst-19323 24 17 , , PUNCT ajst-19323 24 18 which which PRON ajst-19323 24 19 is be AUX ajst-19323 24 20 mainly mainly ADV ajst-19323 24 21 used use VERB ajst-19323 24 22 for for ADP ajst-19323 24 23 dimensionality dimensionality NOUN ajst-19323 24 24 reduction reduction NOUN ajst-19323 24 25 of of ADP ajst-19323 24 26 multilabel multilabel NOUN ajst-19323 24 27 data datum NOUN ajst-19323 24 28 . . PUNCT ajst-19323 25 1 literature literature NOUN ajst-19323 26 1 [ [ X ajst-19323 26 2 5 5 NUM ajst-19323 26 3 ] ] PUNCT ajst-19323 26 4 proposed propose VERB ajst-19323 26 5 a a DET ajst-19323 26 6 filter filter NOUN ajst-19323 26 7 - - PUNCT ajst-19323 26 8 based base VERB ajst-19323 26 9 feature feature NOUN ajst-19323 26 10 selection selection NOUN ajst-19323 26 11 method method NOUN ajst-19323 26 12 for for ADP ajst-19323 26 13 multi multi ADJ ajst-19323 26 14 - - ADJ ajst-19323 26 15 label label ADJ ajst-19323 26 16 classification classification NOUN ajst-19323 26 17 . . PUNCT ajst-19323 27 1 based base VERB ajst-19323 27 2 on on ADP ajst-19323 27 3 random random ADJ ajst-19323 27 4 forest forest NOUN ajst-19323 27 5 as as ADP ajst-19323 27 6 a a DET ajst-19323 27 7 learner learner NOUN ajst-19323 27 8 , , PUNCT ajst-19323 27 9 a a DET ajst-19323 27 10 filter filter NOUN ajst-19323 27 11 feature feature NOUN ajst-19323 27 12 selection selection NOUN ajst-19323 27 13 method method NOUN ajst-19323 27 14 based base VERB ajst-19323 27 15 on on ADP ajst-19323 27 16 statistical statistical ADJ ajst-19323 27 17 integration integration NOUN ajst-19323 27 18 method method NOUN ajst-19323 27 19 and and CCONJ ajst-19323 27 20 label label NOUN ajst-19323 27 21 space space NOUN ajst-19323 27 22 division division NOUN ajst-19323 27 23 based base VERB ajst-19323 27 24 on on ADP ajst-19323 27 25 integration integration NOUN ajst-19323 27 26 method method NOUN ajst-19323 27 27 were be AUX ajst-19323 27 28 used use VERB ajst-19323 27 29 . . PUNCT ajst-19323 28 1 literature literature NOUN ajst-19323 29 1 [ [ X ajst-19323 29 2 6 6 NUM ajst-19323 29 3 ] ] PUNCT ajst-19323 29 4 proposes propose VERB ajst-19323 29 5 a a DET ajst-19323 29 6 multi multi ADJ ajst-19323 29 7 - - ADJ ajst-19323 29 8 label label ADJ ajst-19323 29 9 feature feature NOUN ajst-19323 29 10 selection selection NOUN ajst-19323 29 11 based base VERB ajst-19323 29 12 on on ADP ajst-19323 29 13 coarsegrained coarsegraine VERB ajst-19323 29 14 balls ball NOUN ajst-19323 29 15 and and CCONJ ajst-19323 29 16 label label NOUN ajst-19323 29 17 distribution distribution NOUN ajst-19323 29 18 . . PUNCT ajst-19323 30 1 by by ADP ajst-19323 30 2 using use VERB ajst-19323 30 3 the the DET ajst-19323 30 4 pellet pellet NOUN ajst-19323 30 5 computing computing NOUN ajst-19323 30 6 model model NOUN ajst-19323 30 7 and and CCONJ ajst-19323 30 8 its its PRON ajst-19323 30 9 proposed propose VERB ajst-19323 30 10 label label NOUN ajst-19323 30 11 enhancement enhancement NOUN ajst-19323 30 12 method method NOUN ajst-19323 30 13 , , PUNCT ajst-19323 30 14 logical logical ADJ ajst-19323 30 15 labels label NOUN ajst-19323 30 16 are be AUX ajst-19323 30 17 transformed transform VERB ajst-19323 30 18 into into ADP ajst-19323 30 19 label label NOUN ajst-19323 30 20 distribution distribution NOUN ajst-19323 30 21 by by ADP ajst-19323 30 22 exploring explore VERB ajst-19323 30 23 the the DET ajst-19323 30 24 similarities similarity NOUN ajst-19323 30 25 between between ADP ajst-19323 30 26 instances instance NOUN ajst-19323 30 27 and and CCONJ ajst-19323 30 28 coarsegrained coarsegraine VERB ajst-19323 30 29 balls ball NOUN ajst-19323 30 30 . . PUNCT ajst-19323 31 1 the the DET ajst-19323 31 2 method method NOUN ajst-19323 31 3 measures measure VERB ajst-19323 31 4 feature feature NOUN ajst-19323 31 5 significance significance NOUN ajst-19323 31 6 by by ADP ajst-19323 31 7 the the DET ajst-19323 31 8 consistency consistency NOUN ajst-19323 31 9 of of ADP ajst-19323 31 10 labels label NOUN ajst-19323 31 11 in in ADP ajst-19323 31 12 samples sample NOUN ajst-19323 31 13 within within ADP ajst-19323 31 14 the the DET ajst-19323 31 15 same same ADJ ajst-19323 31 16 information information NOUN ajst-19323 31 17 granularity granularity NOUN ajst-19323 31 18 . . PUNCT ajst-19323 32 1 based base VERB ajst-19323 32 2 on on ADP ajst-19323 32 3 mutual mutual ADJ ajst-19323 32 4 information information NOUN ajst-19323 32 5 theory theory NOUN ajst-19323 32 6 , , PUNCT ajst-19323 32 7 literature literature NOUN ajst-19323 32 8 [ [ X ajst-19323 32 9 7 7 NUM ajst-19323 32 10 ] ] PUNCT ajst-19323 32 11 designed design VERB ajst-19323 32 12 a a DET ajst-19323 32 13 new new ADJ ajst-19323 32 14 multi multi ADJ ajst-19323 32 15 - - ADJ ajst-19323 32 16 label label ADJ ajst-19323 32 17 feature feature NOUN ajst-19323 32 18 selection selection NOUN ajst-19323 32 19 method method NOUN ajst-19323 32 20 combining combine VERB ajst-19323 32 21 dynamic dynamic ADJ ajst-19323 32 22 correlation correlation NOUN ajst-19323 32 23 change change NOUN ajst-19323 32 24 , , PUNCT ajst-19323 32 25 label label NOUN ajst-19323 32 26 redundancy redundancy NOUN ajst-19323 32 27 and and CCONJ ajst-19323 32 28 interactive interactive ADJ ajst-19323 32 29 information information NOUN ajst-19323 32 30 . . PUNCT ajst-19323 33 1 in in ADP ajst-19323 33 2 the the DET ajst-19323 33 3 model model NOUN ajst-19323 33 4 training training NOUN ajst-19323 33 5 of of ADP ajst-19323 33 6 multi multi ADJ ajst-19323 33 7 - - ADJ ajst-19323 33 8 label label ADJ ajst-19323 33 9 feature feature NOUN ajst-19323 33 10 selection selection NOUN ajst-19323 33 11 algorithm algorithm NOUN ajst-19323 33 12 , , PUNCT ajst-19323 33 13 the the DET ajst-19323 33 14 correlation correlation NOUN ajst-19323 33 15 between between ADP ajst-19323 33 16 labels label NOUN ajst-19323 33 17 is be AUX ajst-19323 33 18 a a DET ajst-19323 33 19 key key ADJ ajst-19323 33 20 information information NOUN ajst-19323 33 21 , , PUNCT ajst-19323 33 22 which which PRON ajst-19323 33 23 can can AUX ajst-19323 33 24 be be AUX ajst-19323 33 25 effectively effectively ADV ajst-19323 33 26 used use VERB ajst-19323 33 27 to to PART ajst-19323 33 28 optimize optimize VERB ajst-19323 33 29 the the DET ajst-19323 33 30 feature feature NOUN ajst-19323 33 31 selection selection NOUN ajst-19323 33 32 process process NOUN ajst-19323 33 33 . . PUNCT ajst-19323 34 1 by by ADP ajst-19323 34 2 considering consider VERB ajst-19323 34 3 the the DET ajst-19323 34 4 correlation correlation NOUN ajst-19323 34 5 between between ADP ajst-19323 34 6 labels label NOUN ajst-19323 34 7 , , PUNCT ajst-19323 34 8 the the DET ajst-19323 34 9 algorithm algorithm NOUN ajst-19323 34 10 can can AUX ajst-19323 34 11 more more ADV ajst-19323 34 12 accurately accurately ADV ajst-19323 34 13 capture capture VERB ajst-19323 34 14 the the DET ajst-19323 34 15 correlation correlation NOUN ajst-19323 34 16 and and CCONJ ajst-19323 34 17 dependency dependency NOUN ajst-19323 34 18 between between ADP ajst-19323 34 19 different different ADJ ajst-19323 34 20 labels label NOUN ajst-19323 34 21 , , PUNCT ajst-19323 34 22 so so SCONJ ajst-19323 34 23 as as SCONJ ajst-19323 34 24 to to PART ajst-19323 34 25 select select VERB ajst-19323 34 26 the the DET ajst-19323 34 27 features feature NOUN ajst-19323 34 28 that that PRON ajst-19323 34 29 are be AUX ajst-19323 34 30 closely closely ADV ajst-19323 34 31 related relate VERB ajst-19323 34 32 to to ADP ajst-19323 34 33 multiple multiple ADJ ajst-19323 34 34 labels label NOUN ajst-19323 34 35 . . PUNCT ajst-19323 35 1 literature literature NOUN ajst-19323 36 1 [ [ X ajst-19323 36 2 8 8 NUM ajst-19323 36 3 ] ] PUNCT ajst-19323 36 4 uses use VERB ajst-19323 36 5 the the DET ajst-19323 36 6 co co NOUN ajst-19323 36 7 - - NOUN ajst-19323 36 8 occurrence occurrence ADJ ajst-19323 36 9 relationship relationship NOUN ajst-19323 36 10 between between ADP ajst-19323 36 11 labels label NOUN ajst-19323 36 12 to to PART ajst-19323 36 13 evaluate evaluate VERB ajst-19323 36 14 the the DET ajst-19323 36 15 similarity similarity NOUN ajst-19323 36 16 relationship relationship NOUN ajst-19323 36 17 between between ADP ajst-19323 36 18 samples sample NOUN ajst-19323 36 19 . . PUNCT ajst-19323 37 1 according accord VERB ajst-19323 37 2 to to ADP ajst-19323 37 3 literature literature NOUN ajst-19323 37 4 [ [ X ajst-19323 37 5 9 9 NUM ajst-19323 37 6 ] ] PUNCT ajst-19323 37 7 , , PUNCT ajst-19323 37 8 label label NOUN ajst-19323 37 9 correlation correlation NOUN ajst-19323 37 10 and and CCONJ ajst-19323 37 11 label label NOUN ajst-19323 37 12 - - PUNCT ajst-19323 37 13 specific specific ADJ ajst-19323 37 14 features feature NOUN ajst-19323 37 15 are be AUX ajst-19323 37 16 two two NUM ajst-19323 37 17 important important ADJ ajst-19323 37 18 features feature NOUN ajst-19323 37 19 of of ADP ajst-19323 37 20 multi multi ADJ ajst-19323 37 21 - - ADJ ajst-19323 37 22 label label ADJ ajst-19323 37 23 learning learning NOUN ajst-19323 37 24 . . PUNCT ajst-19323 38 1 the the DET ajst-19323 38 2 global global ADJ ajst-19323 38 3 and and CCONJ ajst-19323 38 4 local local ADJ ajst-19323 38 5 label label NOUN ajst-19323 38 6 correlation correlation NOUN ajst-19323 38 7 are be AUX ajst-19323 38 8 calculated calculate VERB ajst-19323 38 9 by by ADP ajst-19323 38 10 label label NOUN ajst-19323 38 11 co co NOUN ajst-19323 38 12 - - NOUN ajst-19323 38 13 occurrence occurrence ADJ ajst-19323 38 14 and and CCONJ ajst-19323 38 15 neighborhood neighborhood NOUN ajst-19323 38 16 information information NOUN ajst-19323 38 17 respectively respectively ADV ajst-19323 38 18 , , PUNCT ajst-19323 38 19 and and CCONJ ajst-19323 38 20 the the DET ajst-19323 38 21 label label NOUN ajst-19323 38 22 - - PUNCT ajst-19323 38 23 specific specific ADJ ajst-19323 38 24 features feature NOUN ajst-19323 38 25 is be AUX ajst-19323 38 26 learned learn VERB ajst-19323 38 27 by by ADP ajst-19323 38 28 l_1 l_1 PROPN ajst-19323 38 29 norm norm PROPN ajst-19323 38 30 . . PUNCT ajst-19323 39 1 literature literature NOUN ajst-19323 40 1 [ [ X ajst-19323 40 2 10 10 NUM ajst-19323 40 3 ] ] PUNCT ajst-19323 40 4 considers consider VERB ajst-19323 40 5 the the DET ajst-19323 40 6 influence influence NOUN ajst-19323 40 7 of of ADP ajst-19323 40 8 two two NUM ajst-19323 40 9 types type NOUN ajst-19323 40 10 of of ADP ajst-19323 40 11 label label NOUN ajst-19323 40 12 dependencies dependency NOUN ajst-19323 40 13 on on ADP ajst-19323 40 14 the the DET ajst-19323 40 15 algorithm algorithm NOUN ajst-19323 40 16 . . PUNCT ajst-19323 41 1 this this DET ajst-19323 41 2 method method NOUN ajst-19323 41 3 considers consider VERB ajst-19323 41 4 that that SCONJ ajst-19323 41 5 second second ADJ ajst-19323 41 6 - - PUNCT ajst-19323 41 7 order order NOUN ajst-19323 41 8 label label NOUN ajst-19323 41 9 dependencies dependency NOUN ajst-19323 41 10 and and CCONJ ajst-19323 41 11 higher high ADJ ajst-19323 41 12 - - PUNCT ajst-19323 41 13 order order NOUN ajst-19323 41 14 label label NOUN ajst-19323 41 15 dependencies dependency NOUN ajst-19323 41 16 are be AUX ajst-19323 41 17 both both ADV ajst-19323 41 18 important important ADJ ajst-19323 41 19 and and CCONJ ajst-19323 41 20 complementary complementary ADJ ajst-19323 41 21 to to PART ajst-19323 41 22 capture capture VERB ajst-19323 41 23 label label NOUN ajst-19323 41 24 information information NOUN ajst-19323 41 25 , , PUNCT ajst-19323 41 26 so so CCONJ ajst-19323 41 27 second second ADJ ajst-19323 41 28 - - PUNCT ajst-19323 41 29 order order NOUN ajst-19323 41 30 label label NOUN ajst-19323 41 31 dependencies dependency NOUN ajst-19323 41 32 and and CCONJ ajst-19323 41 33 higher high ADJ ajst-19323 41 34 - - PUNCT ajst-19323 41 35 order order NOUN ajst-19323 41 36 label label NOUN ajst-19323 41 37 dependencies dependency NOUN ajst-19323 41 38 are be AUX ajst-19323 41 39 considered consider VERB ajst-19323 41 40 at at ADP ajst-19323 41 41 the the DET ajst-19323 41 42 same same ADJ ajst-19323 41 43 time time NOUN ajst-19323 41 44 . . PUNCT ajst-19323 42 1 compared compare VERB ajst-19323 42 2 with with ADP ajst-19323 42 3 traditional traditional ADJ ajst-19323 42 4 machine machine NOUN ajst-19323 42 5 learning learning NOUN ajst-19323 42 6 , , PUNCT ajst-19323 42 7 multi multi ADJ ajst-19323 42 8 - - ADJ ajst-19323 42 9 label label ADJ ajst-19323 42 10 labeling labeling NOUN ajst-19323 42 11 is be AUX ajst-19323 42 12 very very ADV ajst-19323 42 13 difficult difficult ADJ ajst-19323 42 14 . . PUNCT ajst-19323 43 1 with with ADP ajst-19323 43 2 the the DET ajst-19323 43 3 increasing increase VERB ajst-19323 43 4 complexity complexity NOUN ajst-19323 43 5 of of ADP ajst-19323 43 6 multi multi ADJ ajst-19323 43 7 - - ADJ ajst-19323 43 8 label label ADJ ajst-19323 43 9 tasks task NOUN ajst-19323 43 10 , , PUNCT ajst-19323 43 11 there there PRON ajst-19323 43 12 are be VERB ajst-19323 43 13 also also ADV ajst-19323 43 14 many many ADJ ajst-19323 43 15 challenges challenge NOUN ajst-19323 43 16 , , PUNCT ajst-19323 43 17 among among ADP ajst-19323 43 18 which which PRON ajst-19323 43 19 the the DET ajst-19323 43 20 increase increase NOUN ajst-19323 43 21 of of ADP ajst-19323 43 22 sample sample NOUN ajst-19323 43 23 dimension dimension NOUN ajst-19323 43 24 and and CCONJ ajst-19323 43 25 data datum NOUN ajst-19323 43 26 volume volume NOUN ajst-19323 43 27 will will AUX ajst-19323 43 28 significantly significantly ADV ajst-19323 43 29 affect affect VERB ajst-19323 43 30 the the DET ajst-19323 43 31 labeling labeling NOUN ajst-19323 43 32 cost cost NOUN ajst-19323 43 33 . . PUNCT ajst-19323 44 1 first first ADV ajst-19323 44 2 , , PUNCT ajst-19323 44 3 because because SCONJ ajst-19323 44 4 of of ADP ajst-19323 44 5 the the DET ajst-19323 44 6 365 365 NUM ajst-19323 44 7 large large ADJ ajst-19323 44 8 number number NOUN ajst-19323 44 9 of of ADP ajst-19323 44 10 sample sample NOUN ajst-19323 44 11 instances instance NOUN ajst-19323 44 12 , , PUNCT ajst-19323 44 13 the the DET ajst-19323 44 14 process process NOUN ajst-19323 44 15 of of ADP ajst-19323 44 16 labeling label VERB ajst-19323 44 17 them they PRON ajst-19323 44 18 one one NUM ajst-19323 44 19 by by ADP ajst-19323 44 20 one one NUM ajst-19323 44 21 is be AUX ajst-19323 44 22 time time NOUN ajst-19323 44 23 - - PUNCT ajst-19323 44 24 consuming consume VERB ajst-19323 44 25 and and CCONJ ajst-19323 44 26 labor labor NOUN ajst-19323 44 27 - - PUNCT ajst-19323 44 28 intensive intensive ADJ ajst-19323 44 29 . . PUNCT ajst-19323 45 1 therefore therefore ADV ajst-19323 45 2 , , PUNCT ajst-19323 45 3 in in ADP ajst-19323 45 4 the the DET ajst-19323 45 5 labeling labeling NOUN ajst-19323 45 6 process process NOUN ajst-19323 45 7 , , PUNCT ajst-19323 45 8 manual manual ADJ ajst-19323 45 9 labeling labeling NOUN ajst-19323 45 10 tends tend VERB ajst-19323 45 11 to to PART ajst-19323 45 12 select select VERB ajst-19323 45 13 labels label NOUN ajst-19323 45 14 that that PRON ajst-19323 45 15 are be AUX ajst-19323 45 16 more more ADV ajst-19323 45 17 important important ADJ ajst-19323 45 18 to to PART ajst-19323 45 19 label label VERB ajst-19323 45 20 and and CCONJ ajst-19323 45 21 discard discard VERB ajst-19323 45 22 other other ADJ ajst-19323 45 23 labels label NOUN ajst-19323 45 24 . . PUNCT ajst-19323 46 1 the the DET ajst-19323 46 2 second second ADJ ajst-19323 46 3 is be AUX ajst-19323 46 4 that that SCONJ ajst-19323 46 5 human human ADJ ajst-19323 46 6 taggers tagger NOUN ajst-19323 46 7 can can AUX ajst-19323 46 8 only only ADV ajst-19323 46 9 label label VERB ajst-19323 46 10 what what PRON ajst-19323 46 11 they they PRON ajst-19323 46 12 know know VERB ajst-19323 46 13 according accord VERB ajst-19323 46 14 to to ADP ajst-19323 46 15 cognitive cognitive ADJ ajst-19323 46 16 differences difference NOUN ajst-19323 46 17 . . PUNCT ajst-19323 47 1 for for ADP ajst-19323 47 2 example example NOUN ajst-19323 47 3 , , PUNCT ajst-19323 47 4 an an DET ajst-19323 47 5 article article NOUN ajst-19323 47 6 is be AUX ajst-19323 47 7 labeled label VERB ajst-19323 47 8 machine machine NOUN ajst-19323 47 9 learning learning NOUN ajst-19323 47 10 , , PUNCT ajst-19323 47 11 data datum NOUN ajst-19323 47 12 mining mining NOUN ajst-19323 47 13 , , PUNCT ajst-19323 47 14 and and CCONJ ajst-19323 47 15 artificial artificial ADJ ajst-19323 47 16 neural neural ADJ ajst-19323 47 17 networks network NOUN ajst-19323 47 18 , , PUNCT ajst-19323 47 19 and and CCONJ ajst-19323 47 20 the the DET ajst-19323 47 21 human human ADJ ajst-19323 47 22 tagger tagger NOUN ajst-19323 47 23 does do AUX ajst-19323 47 24 not not PART ajst-19323 47 25 know know VERB ajst-19323 47 26 about about ADP ajst-19323 47 27 the the DET ajst-19323 47 28 category category NOUN ajst-19323 47 29 of of ADP ajst-19323 47 30 artificial artificial ADJ ajst-19323 47 31 neural neural ADJ ajst-19323 47 32 networks network NOUN ajst-19323 47 33 and and CCONJ ajst-19323 47 34 only only ADV ajst-19323 47 35 labels label VERB ajst-19323 47 36 the the DET ajst-19323 47 37 article article NOUN ajst-19323 47 38 with with ADP ajst-19323 47 39 the the DET ajst-19323 47 40 remaining remain VERB ajst-19323 47 41 two two NUM ajst-19323 47 42 labels label NOUN ajst-19323 47 43 . . PUNCT ajst-19323 48 1 a a DET ajst-19323 48 2 variety variety NOUN ajst-19323 48 3 of of ADP ajst-19323 48 4 circumstances circumstance NOUN ajst-19323 48 5 have have AUX ajst-19323 48 6 led lead VERB ajst-19323 48 7 to to ADP ajst-19323 48 8 the the DET ajst-19323 48 9 creation creation NOUN ajst-19323 48 10 of of ADP ajst-19323 48 11 missing miss VERB ajst-19323 48 12 labels label NOUN ajst-19323 48 13 . . PUNCT ajst-19323 49 1 however however ADV ajst-19323 49 2 , , PUNCT ajst-19323 49 3 the the DET ajst-19323 49 4 accuracy accuracy NOUN ajst-19323 49 5 of of ADP ajst-19323 49 6 the the DET ajst-19323 49 7 model model NOUN ajst-19323 49 8 or or CCONJ ajst-19323 49 9 the the DET ajst-19323 49 10 efficiency efficiency NOUN ajst-19323 49 11 of of ADP ajst-19323 49 12 classification classification NOUN ajst-19323 49 13 can can AUX ajst-19323 49 14 be be AUX ajst-19323 49 15 affected affect VERB ajst-19323 49 16 by by ADP ajst-19323 49 17 the the DET ajst-19323 49 18 multi multi ADJ ajst-19323 49 19 - - ADJ ajst-19323 49 20 label label ADJ ajst-19323 49 21 feature feature NOUN ajst-19323 49 22 selection selection NOUN ajst-19323 49 23 or or CCONJ ajst-19323 49 24 multi multi ADJ ajst-19323 49 25 - - ADJ ajst-19323 49 26 label label ADJ ajst-19323 49 27 classification classification NOUN ajst-19323 49 28 on on ADP ajst-19323 49 29 the the DET ajst-19323 49 30 multi multi ADJ ajst-19323 49 31 - - ADJ ajst-19323 49 32 label label ADJ ajst-19323 49 33 data datum NOUN ajst-19323 49 34 set set VERB ajst-19323 49 35 with with ADP ajst-19323 49 36 missing miss VERB ajst-19323 49 37 labels label NOUN ajst-19323 49 38 . . PUNCT ajst-19323 50 1 in in ADP ajst-19323 50 2 order order NOUN ajst-19323 50 3 to to PART ajst-19323 50 4 solve solve VERB ajst-19323 50 5 the the DET ajst-19323 50 6 impact impact NOUN ajst-19323 50 7 of of ADP ajst-19323 50 8 missing miss VERB ajst-19323 50 9 labels label NOUN ajst-19323 50 10 , , PUNCT ajst-19323 50 11 many many ADJ ajst-19323 50 12 multi multi ADJ ajst-19323 50 13 - - ADJ ajst-19323 50 14 label label ADJ ajst-19323 50 15 feature feature NOUN ajst-19323 50 16 selection selection NOUN ajst-19323 50 17 algorithms algorithm NOUN ajst-19323 50 18 have have AUX ajst-19323 50 19 been be AUX ajst-19323 50 20 proposed propose VERB ajst-19323 50 21 to to PART ajst-19323 50 22 deal deal VERB ajst-19323 50 23 with with ADP ajst-19323 50 24 missing miss VERB ajst-19323 50 25 labels label NOUN ajst-19323 50 26 based base VERB ajst-19323 50 27 on on ADP ajst-19323 50 28 theories theory NOUN ajst-19323 50 29 such such ADJ ajst-19323 50 30 as as ADP ajst-19323 50 31 multi multi ADJ ajst-19323 50 32 - - ADJ ajst-19323 50 33 label label ADJ ajst-19323 50 34 information information NOUN ajst-19323 50 35 entropy entropy NOUN ajst-19323 51 1 [ [ X ajst-19323 51 2 11 11 NUM ajst-19323 51 3 ] ] PUNCT ajst-19323 51 4 , , PUNCT ajst-19323 51 5 linear linear ADJ ajst-19323 51 6 regression regression NOUN ajst-19323 51 7 [ [ X ajst-19323 51 8 12 12 NUM ajst-19323 51 9 ] ] PUNCT ajst-19323 51 10 and and CCONJ ajst-19323 51 11 label label NOUN ajst-19323 51 12 correlation correlation NOUN ajst-19323 51 13 . . PUNCT ajst-19323 52 1 he he PRON ajst-19323 52 2 et et PROPN ajst-19323 52 3 al al PROPN ajst-19323 52 4 . . PUNCT ajst-19323 53 1 [ [ X ajst-19323 53 2 13 13 NUM ajst-19323 53 3 ] ] PUNCT ajst-19323 53 4 simultaneously simultaneously ADV ajst-19323 53 5 considered consider VERB ajst-19323 53 6 label label NOUN ajst-19323 53 7 relevance relevance NOUN ajst-19323 53 8 , , PUNCT ajst-19323 53 9 missing miss VERB ajst-19323 53 10 labels label NOUN ajst-19323 53 11 , , PUNCT ajst-19323 53 12 and and CCONJ ajst-19323 53 13 feature feature NOUN ajst-19323 53 14 selection selection NOUN ajst-19323 53 15 through through ADP ajst-19323 53 16 a a DET ajst-19323 53 17 multi multi ADJ ajst-19323 53 18 - - ADJ ajst-19323 53 19 classification classification ADJ ajst-19323 53 20 framework framework NOUN ajst-19323 53 21 . . PUNCT ajst-19323 54 1 literature literature NOUN ajst-19323 54 2 [ [ X ajst-19323 54 3 14 14 NUM ajst-19323 54 4 ] ] PUNCT ajst-19323 54 5 proposes propose VERB ajst-19323 54 6 an an DET ajst-19323 54 7 embedded embed VERB ajst-19323 54 8 packaging packaging NOUN ajst-19323 54 9 approach approach NOUN ajst-19323 54 10 that that PRON ajst-19323 54 11 combines combine VERB ajst-19323 54 12 the the DET ajst-19323 54 13 generation generation NOUN ajst-19323 54 14 of of ADP ajst-19323 54 15 label label NOUN ajst-19323 54 16 - - PUNCT ajst-19323 54 17 specific specific ADJ ajst-19323 54 18 features feature NOUN ajst-19323 54 19 with with ADP ajst-19323 54 20 subsequent subsequent ADJ ajst-19323 54 21 model model NOUN ajst-19323 54 22 induction induction NOUN ajst-19323 54 23 to to PART ajst-19323 54 24 solve solve VERB ajst-19323 54 25 the the DET ajst-19323 54 26 multi multi ADJ ajst-19323 54 27 - - ADJ ajst-19323 54 28 label label ADJ ajst-19323 54 29 classification classification NOUN ajst-19323 54 30 problem problem NOUN ajst-19323 54 31 of of ADP ajst-19323 54 32 missing miss VERB ajst-19323 54 33 labels label NOUN ajst-19323 54 34 . . PUNCT ajst-19323 55 1 at at ADP ajst-19323 55 2 the the DET ajst-19323 55 3 same same ADJ ajst-19323 55 4 time time NOUN ajst-19323 55 5 , , PUNCT ajst-19323 55 6 it it PRON ajst-19323 55 7 is be AUX ajst-19323 55 8 more more ADV ajst-19323 55 9 efficient efficient ADJ ajst-19323 55 10 to to PART ajst-19323 55 11 use use VERB ajst-19323 55 12 the the DET ajst-19323 55 13 specific specific ADJ ajst-19323 55 14 features feature NOUN ajst-19323 55 15 of of ADP ajst-19323 55 16 each each DET ajst-19323 55 17 label label NOUN ajst-19323 55 18 than than SCONJ ajst-19323 55 19 to to PART ajst-19323 55 20 use use VERB ajst-19323 55 21 all all DET ajst-19323 55 22 the the DET ajst-19323 55 23 features feature NOUN ajst-19323 55 24 to to PART ajst-19323 55 25 classify classify VERB ajst-19323 55 26 . . PUNCT ajst-19323 56 1 label label NOUN ajst-19323 56 2 - - PUNCT ajst-19323 56 3 specific specific ADJ ajst-19323 56 4 features feature NOUN ajst-19323 56 5 , , PUNCT ajst-19323 56 6 refer refer VERB ajst-19323 56 7 to to ADP ajst-19323 56 8 the the DET ajst-19323 56 9 subset subset NOUN ajst-19323 56 10 of of ADP ajst-19323 56 11 features feature NOUN ajst-19323 56 12 most most ADV ajst-19323 56 13 closely closely ADV ajst-19323 56 14 associated associate VERB ajst-19323 56 15 with with ADP ajst-19323 56 16 each each DET ajst-19323 56 17 label label NOUN ajst-19323 56 18 . . PUNCT ajst-19323 57 1 in in ADP ajst-19323 57 2 the the DET ajst-19323 57 3 multi multi ADJ ajst-19323 57 4 - - ADJ ajst-19323 57 5 label label ADJ ajst-19323 57 6 learning learn VERB ajst-19323 57 7 scenario scenario NOUN ajst-19323 57 8 , , PUNCT ajst-19323 57 9 the the DET ajst-19323 57 10 use use NOUN ajst-19323 57 11 of of ADP ajst-19323 57 12 label label NOUN ajst-19323 57 13 - - PUNCT ajst-19323 57 14 specific specific ADJ ajst-19323 57 15 features feature NOUN ajst-19323 57 16 for for ADP ajst-19323 57 17 feature feature NOUN ajst-19323 57 18 selection selection NOUN ajst-19323 57 19 can can AUX ajst-19323 57 20 reveal reveal VERB ajst-19323 57 21 the the DET ajst-19323 57 22 deep deep ADJ ajst-19323 57 23 correlation correlation NOUN ajst-19323 57 24 between between ADP ajst-19323 57 25 features feature NOUN ajst-19323 57 26 and and CCONJ ajst-19323 57 27 labels label NOUN ajst-19323 57 28 more more ADV ajst-19323 57 29 efficiently efficiently ADV ajst-19323 57 30 . . PUNCT ajst-19323 58 1 at at ADP ajst-19323 58 2 the the DET ajst-19323 58 3 same same ADJ ajst-19323 58 4 time time NOUN ajst-19323 58 5 , , PUNCT ajst-19323 58 6 from from ADP ajst-19323 58 7 the the DET ajst-19323 58 8 point point NOUN ajst-19323 58 9 of of ADP ajst-19323 58 10 view view NOUN ajst-19323 58 11 of of ADP ajst-19323 58 12 the the DET ajst-19323 58 13 nonlinear nonlinear ADJ ajst-19323 58 14 relationship relationship NOUN ajst-19323 58 15 between between ADP ajst-19323 58 16 data datum NOUN ajst-19323 58 17 , , PUNCT ajst-19323 58 18 consider consider VERB ajst-19323 58 19 the the DET ajst-19323 58 20 nonlinear nonlinear ADJ ajst-19323 58 21 relationship relationship NOUN ajst-19323 58 22 between between ADP ajst-19323 58 23 instances instance NOUN ajst-19323 58 24 and and CCONJ ajst-19323 58 25 the the DET ajst-19323 58 26 nonlinear nonlinear ADJ ajst-19323 58 27 relationship relationship NOUN ajst-19323 58 28 between between ADP ajst-19323 58 29 features feature NOUN ajst-19323 58 30 , , PUNCT ajst-19323 58 31 and and CCONJ ajst-19323 58 32 study study VERB ajst-19323 58 33 their their PRON ajst-19323 58 34 influence influence NOUN ajst-19323 58 35 on on ADP ajst-19323 58 36 the the DET ajst-19323 58 37 relationship relationship NOUN ajst-19323 58 38 between between ADP ajst-19323 58 39 labels label NOUN ajst-19323 58 40 and and CCONJ ajst-19323 58 41 features feature NOUN ajst-19323 58 42 . . PUNCT ajst-19323 59 1 based base VERB ajst-19323 59 2 on on ADP ajst-19323 59 3 the the DET ajst-19323 59 4 above above ADJ ajst-19323 59 5 considerations consideration NOUN ajst-19323 59 6 , , PUNCT ajst-19323 59 7 this this DET ajst-19323 59 8 paper paper NOUN ajst-19323 59 9 proposes propose VERB ajst-19323 59 10 a a DET ajst-19323 59 11 multi multi ADJ ajst-19323 59 12 - - ADJ ajst-19323 59 13 label label ADJ ajst-19323 59 14 feature feature NOUN ajst-19323 59 15 selection selection NOUN ajst-19323 59 16 algorithm algorithm NOUN ajst-19323 59 17 ( ( PUNCT ajst-19323 59 18 ml ml NOUN ajst-19323 59 19 - - PUNCT ajst-19323 59 20 lsf lsf NOUN ajst-19323 59 21 ) ) PUNCT ajst-19323 59 22 based base VERB ajst-19323 59 23 on on ADP ajst-19323 59 24 label label NOUN ajst-19323 59 25 - - PUNCT ajst-19323 59 26 specific specific ADJ ajst-19323 59 27 features feature NOUN ajst-19323 59 28 and and CCONJ ajst-19323 59 29 manifold manifold ADJ ajst-19323 59 30 learning learning NOUN ajst-19323 59 31 that that PRON ajst-19323 59 32 can can AUX ajst-19323 59 33 deal deal VERB ajst-19323 59 34 with with ADP ajst-19323 59 35 missing miss VERB ajst-19323 59 36 labels label NOUN ajst-19323 59 37 . . PUNCT ajst-19323 60 1 2 2 X ajst-19323 60 2 . . NUM ajst-19323 60 3 proposed propose VERB ajst-19323 60 4 models model NOUN ajst-19323 60 5 in in ADP ajst-19323 60 6 multi multi ADJ ajst-19323 60 7 - - ADJ ajst-19323 60 8 label label ADJ ajst-19323 60 9 learning learning NOUN ajst-19323 60 10 , , PUNCT ajst-19323 60 11 𝑋 𝑋 PROPN ajst-19323 60 12 π‘₯ π‘₯ PROPN ajst-19323 60 13 ， ， PROPN ajst-19323 60 14 π‘₯ π‘₯ X ajst-19323 60 15 ， ， PROPN ajst-19323 60 16 … … PUNCT ajst-19323 60 17 ， ， ADJ ajst-19323 60 18 π‘₯ π‘₯ PROPN ajst-19323 60 19 ∈ ∈ PROPN ajst-19323 60 20 𝑅 𝑅 PROPN ajst-19323 60 21 represents represent VERB ajst-19323 60 22 a a DET ajst-19323 60 23 multi multi ADJ ajst-19323 60 24 - - ADJ ajst-19323 60 25 label label ADJ ajst-19323 60 26 data datum NOUN ajst-19323 60 27 set set VERB ajst-19323 60 28 with with ADP ajst-19323 60 29 n n ADP ajst-19323 60 30 instances instance NOUN ajst-19323 60 31 , , PUNCT ajst-19323 60 32 where where SCONJ ajst-19323 60 33 d d NOUN ajst-19323 60 34 represents represent VERB ajst-19323 60 35 the the DET ajst-19323 60 36 feature feature NOUN ajst-19323 60 37 dimension dimension NOUN ajst-19323 60 38 ; ; PUNCT ajst-19323 60 39 π‘Œ π‘Œ PROPN ajst-19323 60 40 𝑦 𝑦 PRON ajst-19323 60 41 ， ， PROPN ajst-19323 60 42 𝑦 𝑦 NOUN ajst-19323 60 43 ， ， PROPN ajst-19323 60 44 … … PUNCT ajst-19323 60 45 ， ， ADJ ajst-19323 60 46 𝑦 𝑦 PROPN ajst-19323 60 47 ∈ ∈ PROPN ajst-19323 60 48 𝑅 𝑅 PROPN ajst-19323 60 49 represents represent VERB ajst-19323 60 50 the the DET ajst-19323 60 51 set set NOUN ajst-19323 60 52 of of ADP ajst-19323 60 53 labels label NOUN ajst-19323 60 54 corresponding correspond VERB ajst-19323 60 55 to to ADP ajst-19323 60 56 n n NUM ajst-19323 60 57 instances instance NOUN ajst-19323 60 58 , , PUNCT ajst-19323 60 59 where where SCONJ ajst-19323 60 60 l l NOUN ajst-19323 60 61 represents represent VERB ajst-19323 60 62 the the DET ajst-19323 60 63 number number NOUN ajst-19323 60 64 of of ADP ajst-19323 60 65 labels label NOUN ajst-19323 60 66 . . PUNCT ajst-19323 61 1 the the DET ajst-19323 61 2 label label NOUN ajst-19323 61 3 values value NOUN ajst-19323 61 4 in in ADP ajst-19323 61 5 set set NOUN ajst-19323 61 6 y y PROPN ajst-19323 61 7 are be AUX ajst-19323 61 8 set set VERB ajst-19323 61 9 to to ADP ajst-19323 61 10 1 1 NUM ajst-19323 61 11 , , PUNCT ajst-19323 61 12 0 0 NUM ajst-19323 61 13 , , PUNCT ajst-19323 61 14 and and CCONJ ajst-19323 61 15 -1.when -1.when VERB ajst-19323 61 16 the the DET ajst-19323 61 17 label label NOUN ajst-19323 61 18 value value NOUN ajst-19323 61 19 is be AUX ajst-19323 61 20 1 1 NUM ajst-19323 61 21 , , PUNCT ajst-19323 61 22 the the DET ajst-19323 61 23 instance instance NOUN ajst-19323 61 24 has have VERB ajst-19323 61 25 the the DET ajst-19323 61 26 label label NOUN ajst-19323 61 27 . . PUNCT ajst-19323 62 1 a a DET ajst-19323 62 2 label label NOUN ajst-19323 62 3 value value NOUN ajst-19323 62 4 of of ADP ajst-19323 62 5 -1 -1 PUNCT ajst-19323 62 6 indicates indicate VERB ajst-19323 62 7 that that SCONJ ajst-19323 62 8 the the DET ajst-19323 62 9 instance instance NOUN ajst-19323 62 10 does do AUX ajst-19323 62 11 not not PART ajst-19323 62 12 have have VERB ajst-19323 62 13 the the DET ajst-19323 62 14 label label NOUN ajst-19323 62 15 . . PUNCT ajst-19323 63 1 when when SCONJ ajst-19323 63 2 the the DET ajst-19323 63 3 label label NOUN ajst-19323 63 4 value value NOUN ajst-19323 63 5 is be AUX ajst-19323 63 6 0 0 NUM ajst-19323 63 7 , , PUNCT ajst-19323 63 8 the the DET ajst-19323 63 9 instance instance NOUN ajst-19323 63 10 is be AUX ajst-19323 63 11 missing miss VERB ajst-19323 63 12 the the DET ajst-19323 63 13 label label NOUN ajst-19323 63 14 . . PUNCT ajst-19323 64 1 2.1 2.1 NUM ajst-19323 64 2 . . PUNCT ajst-19323 65 1 build build VERB ajst-19323 65 2 a a DET ajst-19323 65 3 label label NOUN ajst-19323 65 4 - - PUNCT ajst-19323 65 5 specific specific ADJ ajst-19323 65 6 feature feature NOUN ajst-19323 65 7 model model NOUN ajst-19323 65 8 in in ADP ajst-19323 65 9 a a DET ajst-19323 65 10 multi multi ADJ ajst-19323 65 11 - - ADJ ajst-19323 65 12 label label ADJ ajst-19323 65 13 classification classification NOUN ajst-19323 65 14 task task NOUN ajst-19323 65 15 , , PUNCT ajst-19323 65 16 each each DET ajst-19323 65 17 label label NOUN ajst-19323 65 18 in in ADP ajst-19323 65 19 the the DET ajst-19323 65 20 label label NOUN ajst-19323 65 21 space space NOUN ajst-19323 65 22 has have VERB ajst-19323 65 23 the the DET ajst-19323 65 24 most most ADV ajst-19323 65 25 relevant relevant ADJ ajst-19323 65 26 and and CCONJ ajst-19323 65 27 discriminating discriminate VERB ajst-19323 65 28 features feature NOUN ajst-19323 65 29 for for ADP ajst-19323 65 30 the the DET ajst-19323 65 31 label label NOUN ajst-19323 65 32 , , PUNCT ajst-19323 65 33 which which PRON ajst-19323 65 34 are be AUX ajst-19323 65 35 called call VERB ajst-19323 65 36 label label NOUN ajst-19323 65 37 - - PUNCT ajst-19323 65 38 specific specific ADJ ajst-19323 65 39 feature feature NOUN ajst-19323 65 40 . . PUNCT ajst-19323 66 1 first first ADV ajst-19323 66 2 , , PUNCT ajst-19323 66 3 the the DET ajst-19323 66 4 basic basic ADJ ajst-19323 66 5 label label NOUN ajst-19323 66 6 - - PUNCT ajst-19323 66 7 specific specific ADJ ajst-19323 66 8 feature feature NOUN ajst-19323 66 9 model model NOUN ajst-19323 66 10 is be AUX ajst-19323 66 11 constructed construct VERB ajst-19323 66 12 by by ADP ajst-19323 66 13 linear linear PROPN ajst-19323 66 14 regression regression NOUN ajst-19323 66 15 method method NOUN ajst-19323 66 16 . . PUNCT ajst-19323 67 1 the the DET ajst-19323 67 2 purpose purpose NOUN ajst-19323 67 3 of of ADP ajst-19323 67 4 adding add VERB ajst-19323 67 5 𝑙 𝑙 DET ajst-19323 67 6 norm norm NOUN ajst-19323 67 7 is be AUX ajst-19323 67 8 to to PART ajst-19323 67 9 ensure ensure VERB ajst-19323 67 10 that that SCONJ ajst-19323 67 11 the the DET ajst-19323 67 12 features feature NOUN ajst-19323 67 13 selected select VERB ajst-19323 67 14 by by ADP ajst-19323 67 15 each each DET ajst-19323 67 16 label label NOUN ajst-19323 67 17 are be AUX ajst-19323 67 18 sparse sparse ADJ ajst-19323 67 19 and and CCONJ ajst-19323 67 20 representative representative NOUN ajst-19323 67 21 . . PUNCT ajst-19323 68 1 the the DET ajst-19323 68 2 objective objective ADJ ajst-19323 68 3 function function NOUN ajst-19323 68 4 is be AUX ajst-19323 68 5 as as SCONJ ajst-19323 68 6 follows follow VERB ajst-19323 68 7 : : PUNCT ajst-19323 68 8 min min NOUN ajst-19323 68 9 1 1 NUM ajst-19323 68 10 2 2 NUM ajst-19323 68 11 β€–π‘‹π‘Š β€–π‘‹π‘Š NOUN ajst-19323 68 12 π‘Œβ€– π‘Œβ€– PROPN ajst-19323 69 1 πœ†β€–π‘Šβ€– πœ†β€–π‘Šβ€– PROPN ajst-19323 69 2 1 1 NUM ajst-19323 69 3 where where SCONJ ajst-19323 69 4 x x PRON ajst-19323 69 5 is be AUX ajst-19323 69 6 the the DET ajst-19323 69 7 eigenmatrix eigenmatrix NOUN ajst-19323 69 8 of of ADP ajst-19323 69 9 the the DET ajst-19323 69 10 instance instance NOUN ajst-19323 69 11 , , PUNCT ajst-19323 69 12 y y PROPN ajst-19323 69 13 is be AUX ajst-19323 69 14 the the DET ajst-19323 69 15 label label NOUN ajst-19323 69 16 matrix matrix NOUN ajst-19323 69 17 containing contain VERB ajst-19323 69 18 the the DET ajst-19323 69 19 missing miss VERB ajst-19323 69 20 label label NOUN ajst-19323 69 21 , , PUNCT ajst-19323 69 22 and and CCONJ ajst-19323 69 23 Ξ» Ξ» PROPN ajst-19323 69 24 is be AUX ajst-19323 69 25 the the DET ajst-19323 69 26 parameter parameter NOUN ajst-19323 69 27 . . PUNCT ajst-19323 70 1 w w PROPN ajst-19323 70 2 is be AUX ajst-19323 70 3 the the DET ajst-19323 70 4 coefficient coefficient NOUN ajst-19323 70 5 matrix matrix NOUN ajst-19323 70 6 , , PUNCT ajst-19323 70 7 𝑀 𝑀 PART ajst-19323 70 8 represents represent VERB ajst-19323 70 9 the the DET ajst-19323 70 10 i i PROPN ajst-19323 70 11 row row NOUN ajst-19323 70 12 in in ADP ajst-19323 70 13 the the DET ajst-19323 70 14 matrix matrix NOUN ajst-19323 70 15 w w NOUN ajst-19323 70 16 , , PUNCT ajst-19323 70 17 the the DET ajst-19323 70 18 row row NOUN ajst-19323 70 19 vector vector NOUN ajst-19323 70 20 represents represent VERB ajst-19323 70 21 the the DET ajst-19323 70 22 feature feature NOUN ajst-19323 70 23 vector vector NOUN ajst-19323 70 24 , , PUNCT ajst-19323 70 25 and and CCONJ ajst-19323 70 26 the the DET ajst-19323 70 27 column column NOUN ajst-19323 70 28 vector vector NOUN ajst-19323 70 29 represents represent VERB ajst-19323 70 30 the the DET ajst-19323 70 31 label label NOUN ajst-19323 70 32 vector vector NOUN ajst-19323 70 33 . . PUNCT ajst-19323 71 1 if if SCONJ ajst-19323 71 2 the the DET ajst-19323 71 3 value value NOUN ajst-19323 71 4 of of ADP ajst-19323 71 5 a a DET ajst-19323 71 6 certain certain ADJ ajst-19323 71 7 place place NOUN ajst-19323 71 8 in in ADP ajst-19323 71 9 w w PROPN ajst-19323 71 10 is be AUX ajst-19323 71 11 0 0 NUM ajst-19323 71 12 , , PUNCT ajst-19323 71 13 it it PRON ajst-19323 71 14 means mean VERB ajst-19323 71 15 that that SCONJ ajst-19323 71 16 there there PRON ajst-19323 71 17 is be VERB ajst-19323 71 18 no no DET ajst-19323 71 19 correlation correlation NOUN ajst-19323 71 20 between between ADP ajst-19323 71 21 the the DET ajst-19323 71 22 feature feature NOUN ajst-19323 71 23 and and CCONJ ajst-19323 71 24 the the DET ajst-19323 71 25 label label NOUN ajst-19323 71 26 at at ADP ajst-19323 71 27 that that DET ajst-19323 71 28 place place NOUN ajst-19323 71 29 . . PUNCT ajst-19323 72 1 2.2 2.2 NUM ajst-19323 72 2 . . PUNCT ajst-19323 72 3 build build VERB ajst-19323 72 4 the the DET ajst-19323 72 5 instance instance NOUN ajst-19323 72 6 manifold manifold ADJ ajst-19323 72 7 and and CCONJ ajst-19323 72 8 feature feature NOUN ajst-19323 72 9 manifold manifold ADJ ajst-19323 72 10 models model NOUN ajst-19323 72 11 the the DET ajst-19323 72 12 core core ADJ ajst-19323 72 13 idea idea NOUN ajst-19323 72 14 of of ADP ajst-19323 72 15 manifold manifold ADJ ajst-19323 72 16 learning learning NOUN ajst-19323 72 17 is be AUX ajst-19323 72 18 that that SCONJ ajst-19323 72 19 data datum NOUN ajst-19323 72 20 in in ADP ajst-19323 72 21 highdimensional highdimensional ADJ ajst-19323 72 22 space space NOUN ajst-19323 72 23 is be AUX ajst-19323 72 24 actually actually ADV ajst-19323 72 25 distributed distribute VERB ajst-19323 72 26 on on ADP ajst-19323 72 27 a a DET ajst-19323 72 28 low low ADJ ajst-19323 72 29 dimensional dimensional ADJ ajst-19323 72 30 manifold manifold ADJ ajst-19323 72 31 structure structure NOUN ajst-19323 72 32 , , PUNCT ajst-19323 72 33 that that ADV ajst-19323 72 34 is is ADV ajst-19323 72 35 , , PUNCT ajst-19323 72 36 although although SCONJ ajst-19323 72 37 data datum NOUN ajst-19323 72 38 may may AUX ajst-19323 72 39 appear appear VERB ajst-19323 72 40 complex complex ADJ ajst-19323 72 41 and and CCONJ ajst-19323 72 42 difficult difficult ADJ ajst-19323 72 43 to to PART ajst-19323 72 44 analyze analyze VERB ajst-19323 72 45 in in ADP ajst-19323 72 46 the the DET ajst-19323 72 47 original original ADJ ajst-19323 72 48 highdimensional highdimensional ADJ ajst-19323 72 49 space space NOUN ajst-19323 72 50 , , PUNCT ajst-19323 72 51 their their PRON ajst-19323 72 52 essential essential ADJ ajst-19323 72 53 structure structure NOUN ajst-19323 72 54 can can AUX ajst-19323 72 55 often often ADV ajst-19323 72 56 be be AUX ajst-19323 72 57 characterized characterize VERB ajst-19323 72 58 by by ADP ajst-19323 72 59 a a DET ajst-19323 72 60 low low ADJ ajst-19323 72 61 dimensional dimensional ADJ ajst-19323 72 62 manifold manifold NOUN ajst-19323 72 63 . . PUNCT ajst-19323 73 1 the the DET ajst-19323 73 2 characteristic characteristic NOUN ajst-19323 73 3 makes make VERB ajst-19323 73 4 manifold manifold ADJ ajst-19323 73 5 learning learning NOUN ajst-19323 73 6 suitable suitable ADJ ajst-19323 73 7 for for ADP ajst-19323 73 8 handling handle VERB ajst-19323 73 9 nonlinear nonlinear ADJ ajst-19323 73 10 relationships relationship NOUN ajst-19323 73 11 between between ADP ajst-19323 73 12 data datum NOUN ajst-19323 73 13 . . PUNCT ajst-19323 74 1 the the DET ajst-19323 74 2 goal goal NOUN ajst-19323 74 3 of of ADP ajst-19323 74 4 manifold manifold ADJ ajst-19323 74 5 learning learning NOUN ajst-19323 74 6 is be AUX ajst-19323 74 7 to to PART ajst-19323 74 8 effectively effectively ADV ajst-19323 74 9 map map VERB ajst-19323 74 10 points point NOUN ajst-19323 74 11 on on ADP ajst-19323 74 12 manifold manifold ADJ ajst-19323 74 13 m m VERB ajst-19323 74 14 in in ADP ajst-19323 74 15 an an DET ajst-19323 74 16 ndimensional ndimensional ADJ ajst-19323 74 17 space space NOUN ajst-19323 74 18 to to ADP ajst-19323 74 19 a a DET ajst-19323 74 20 low low ADJ ajst-19323 74 21 dimensional dimensional ADJ ajst-19323 74 22 n n CCONJ ajst-19323 74 23 - - PUNCT ajst-19323 74 24 dimensional dimensional ADJ ajst-19323 74 25 space space NOUN ajst-19323 74 26 , , PUNCT ajst-19323 74 27 where where SCONJ ajst-19323 74 28 n n CCONJ ajst-19323 74 29 < < X ajst-19323 74 30 n. n. ADJ ajst-19323 74 31 popular popular ADJ ajst-19323 74 32 regularization regularization NOUN ajst-19323 74 33 refers refer VERB ajst-19323 74 34 to to ADP ajst-19323 74 35 adding add VERB ajst-19323 74 36 manifold manifold ADJ ajst-19323 74 37 related related ADJ ajst-19323 74 38 terms term NOUN ajst-19323 74 39 to to ADP ajst-19323 74 40 machine machine NOUN ajst-19323 74 41 learning learning NOUN ajst-19323 74 42 problems problem NOUN ajst-19323 74 43 , , PUNCT ajst-19323 74 44 with with ADP ajst-19323 74 45 the the DET ajst-19323 74 46 aim aim NOUN ajst-19323 74 47 of of ADP ajst-19323 74 48 constraining constrain VERB ajst-19323 74 49 the the DET ajst-19323 74 50 coefficient coefficient NOUN ajst-19323 74 51 matrix matrix NOUN ajst-19323 74 52 w w ADP ajst-19323 74 53 through through ADP ajst-19323 74 54 manifold manifold ADJ ajst-19323 74 55 regularization regularization NOUN ajst-19323 74 56 terms term NOUN ajst-19323 74 57 . . PUNCT ajst-19323 75 1 two two NUM ajst-19323 75 2 strongly strongly ADV ajst-19323 75 3 related relate VERB ajst-19323 75 4 instances instance NOUN ajst-19323 75 5 can can AUX ajst-19323 75 6 share share VERB ajst-19323 75 7 related related ADJ ajst-19323 75 8 subsets subset NOUN ajst-19323 75 9 of of ADP ajst-19323 75 10 each each DET ajst-19323 75 11 other other ADJ ajst-19323 75 12 [ [ X ajst-19323 75 13 15 15 NUM ajst-19323 75 14 ] ] PUNCT ajst-19323 75 15 . . PUNCT ajst-19323 76 1 if if SCONJ ajst-19323 76 2 there there PRON ajst-19323 76 3 are be VERB ajst-19323 76 4 two two NUM ajst-19323 76 5 strongly strongly ADV ajst-19323 76 6 correlated correlate VERB ajst-19323 76 7 instances instance NOUN ajst-19323 76 8 , , PUNCT ajst-19323 76 9 then then ADV ajst-19323 76 10 the the DET ajst-19323 76 11 corresponding corresponding ADJ ajst-19323 76 12 label label NOUN ajst-19323 76 13 sets set NOUN ajst-19323 76 14 or or CCONJ ajst-19323 76 15 feature feature NOUN ajst-19323 76 16 sets set NOUN ajst-19323 76 17 of of ADP ajst-19323 76 18 the the DET ajst-19323 76 19 two two NUM ajst-19323 76 20 instances instance NOUN ajst-19323 76 21 are be AUX ajst-19323 76 22 similar similar ADJ ajst-19323 76 23 , , PUNCT ajst-19323 76 24 and and CCONJ ajst-19323 76 25 the the DET ajst-19323 76 26 prediction prediction NOUN ajst-19323 76 27 labels label NOUN ajst-19323 76 28 obtained obtain VERB ajst-19323 76 29 by by ADP ajst-19323 76 30 using use VERB ajst-19323 76 31 the the DET ajst-19323 76 32 corresponding correspond VERB ajst-19323 76 33 w w NOUN ajst-19323 76 34 are be AUX ajst-19323 76 35 also also ADV ajst-19323 76 36 very very ADV ajst-19323 76 37 similar similar ADJ ajst-19323 76 38 . . PUNCT ajst-19323 77 1 the the DET ajst-19323 77 2 purpose purpose NOUN ajst-19323 77 3 of of ADP ajst-19323 77 4 laplacian laplacian ADJ ajst-19323 77 5 eigenmap eigenmap NOUN ajst-19323 77 6 in in ADP ajst-19323 77 7 manifold manifold ADJ ajst-19323 77 8 learning learning NOUN ajst-19323 77 9 is be AUX ajst-19323 77 10 to to PART ajst-19323 77 11 ensure ensure VERB ajst-19323 77 12 that that SCONJ ajst-19323 77 13 two two NUM ajst-19323 77 14 strongly strongly ADV ajst-19323 77 15 correlated correlate VERB ajst-19323 77 16 instances instance NOUN ajst-19323 77 17 are be AUX ajst-19323 77 18 as as ADV ajst-19323 77 19 close close ADJ ajst-19323 77 20 as as ADP ajst-19323 77 21 possible possible ADJ ajst-19323 77 22 after after ADP ajst-19323 77 23 dimensionality dimensionality NOUN ajst-19323 77 24 reduction reduction NOUN ajst-19323 77 25 , , PUNCT ajst-19323 77 26 and and CCONJ ajst-19323 77 27 to to PART ajst-19323 77 28 mine mine VERB ajst-19323 77 29 the the DET ajst-19323 77 30 correlation correlation NOUN ajst-19323 77 31 between between ADP ajst-19323 77 32 the the DET ajst-19323 77 33 instances instance NOUN ajst-19323 77 34 by by ADP ajst-19323 77 35 using use VERB ajst-19323 77 36 the the DET ajst-19323 77 37 manifold manifold ADJ ajst-19323 77 38 model model NOUN ajst-19323 77 39 between between ADP ajst-19323 77 40 the the DET ajst-19323 77 41 instances instance NOUN ajst-19323 77 42 . . PUNCT ajst-19323 78 1 add add VERB ajst-19323 78 2 the the DET ajst-19323 78 3 instance instance NOUN ajst-19323 78 4 correlation correlation NOUN ajst-19323 78 5 to to ADP ajst-19323 78 6 the the DET ajst-19323 78 7 equation equation NOUN ajst-19323 78 8 ( ( PUNCT ajst-19323 78 9 1 1 NUM ajst-19323 78 10 ) ) PUNCT ajst-19323 78 11 to to PART ajst-19323 78 12 get get VERB ajst-19323 78 13 the the DET ajst-19323 78 14 objective objective ADJ ajst-19323 78 15 function function NOUN ajst-19323 78 16 : : PUNCT ajst-19323 78 17 π‘šπ‘–π‘› π‘šπ‘–π‘› NOUN ajst-19323 78 18 1 1 NUM ajst-19323 78 19 2 2 NUM ajst-19323 78 20 β€–π‘‹π‘Š β€–π‘‹π‘Š NOUN ajst-19323 78 21 π‘Œβ€– π‘Œβ€– NOUN ajst-19323 79 1 𝛼 𝛼 PRON ajst-19323 79 2 2 2 NUM ajst-19323 79 3 π‘‘π‘Ÿ π‘‘π‘Ÿ NOUN ajst-19323 79 4 π‘Š π‘Š PROPN ajst-19323 79 5 𝑋 𝑋 NOUN ajst-19323 79 6 𝐿 𝐿 PROPN ajst-19323 79 7 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 79 8 πœ†β€–π‘Šβ€– πœ†β€–π‘Šβ€– PROPN ajst-19323 79 9 2 2 NUM ajst-19323 79 10 where where SCONJ ajst-19323 79 11 , , PUNCT ajst-19323 79 12 π›ΌοΌˆπ›Ό π›ΌοΌˆπ›Ό NOUN ajst-19323 79 13 0 0 NUM ajst-19323 79 14 οΌ‰ οΌ‰ PROPN ajst-19323 79 15 is be AUX ajst-19323 79 16 the the DET ajst-19323 79 17 parameter parameter NOUN ajst-19323 79 18 ; ; PUNCT ajst-19323 79 19 𝐿 𝐿 PROPN ajst-19323 79 20 𝐷 𝐷 PROPN ajst-19323 79 21 𝑆 𝑆 PROPN ajst-19323 79 22 , , PUNCT ajst-19323 79 23 𝐿 𝐿 PROPN ajst-19323 79 24 refers refer VERB ajst-19323 79 25 to to ADP ajst-19323 79 26 the the DET ajst-19323 79 27 laplacian laplacian ADJ ajst-19323 79 28 matrix matrix NOUN ajst-19323 79 29 , , PUNCT ajst-19323 79 30 𝐷 𝐷 PROPN ajst-19323 79 31 is be AUX ajst-19323 79 32 a a DET ajst-19323 79 33 diagonal diagonal ADJ ajst-19323 79 34 matrix matrix NOUN ajst-19323 79 35 , , PUNCT ajst-19323 79 36 𝑆 𝑆 PROPN ajst-19323 79 37 ∈ ∈ PROPN ajst-19323 79 38 𝑅 𝑅 PROPN ajst-19323 79 39 represents represent VERB ajst-19323 79 40 the the DET ajst-19323 79 41 instance instance NOUN ajst-19323 79 42 similarity similarity NOUN ajst-19323 79 43 matrix matrix NOUN ajst-19323 79 44 . . PUNCT ajst-19323 80 1 for for ADP ajst-19323 80 2 two two NUM ajst-19323 80 3 related related ADJ ajst-19323 80 4 features feature NOUN ajst-19323 80 5 of of ADP ajst-19323 80 6 a a DET ajst-19323 80 7 label label NOUN ajst-19323 80 8 , , PUNCT ajst-19323 80 9 the the DET ajst-19323 80 10 label label NOUN ajst-19323 80 11 has have VERB ajst-19323 80 12 a a DET ajst-19323 80 13 high high ADJ ajst-19323 80 14 probability probability NOUN ajst-19323 80 15 of of ADP ajst-19323 80 16 having have VERB ajst-19323 80 17 one one NUM ajst-19323 80 18 feature feature NOUN ajst-19323 80 19 and and CCONJ ajst-19323 80 20 the the DET ajst-19323 80 21 other other ADJ ajst-19323 80 22 feature feature NOUN ajst-19323 80 23 . . PUNCT ajst-19323 81 1 feature feature NOUN ajst-19323 81 2 correlation correlation NOUN ajst-19323 81 3 considers consider VERB ajst-19323 81 4 the the DET ajst-19323 81 5 interaction interaction NOUN ajst-19323 81 6 between between ADP ajst-19323 81 7 features feature NOUN ajst-19323 81 8 . . PUNCT ajst-19323 82 1 add add VERB ajst-19323 82 2 the the DET ajst-19323 82 3 feature feature NOUN ajst-19323 82 4 correlation correlation NOUN ajst-19323 82 5 to to ADP ajst-19323 82 6 the the DET ajst-19323 82 7 equation equation NOUN ajst-19323 82 8 ( ( PUNCT ajst-19323 82 9 2 2 NUM ajst-19323 82 10 ) ) PUNCT ajst-19323 82 11 to to PART ajst-19323 82 12 get get VERB ajst-19323 82 13 the the DET ajst-19323 82 14 final final ADJ ajst-19323 82 15 objective objective ADJ ajst-19323 82 16 function function NOUN ajst-19323 82 17 : : PUNCT ajst-19323 82 18 min min NOUN ajst-19323 82 19 1 1 NUM ajst-19323 82 20 2 2 NUM ajst-19323 82 21 β€–π‘‹π‘Š β€–π‘‹π‘Š NOUN ajst-19323 82 22 π‘Œβ€– π‘Œβ€– NOUN ajst-19323 83 1 𝛼 𝛼 PRON ajst-19323 83 2 2 2 NUM ajst-19323 83 3 π‘‘π‘Ÿ π‘‘π‘Ÿ NOUN ajst-19323 83 4 π‘Š π‘Š PROPN ajst-19323 83 5 𝑋 𝑋 NOUN ajst-19323 83 6 𝐿 𝐿 PROPN ajst-19323 83 7 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 83 8 𝛽 𝛽 NOUN ajst-19323 83 9 2 2 NUM ajst-19323 83 10 π‘‘π‘Ÿ π‘‘π‘Ÿ VERB ajst-19323 83 11 π‘Š π‘Š PROPN ajst-19323 83 12 𝐿 𝐿 NOUN ajst-19323 83 13 π‘Š π‘Š PROPN ajst-19323 83 14 πœ†β€–π‘Šβ€– πœ†β€–π‘Šβ€– PROPN ajst-19323 83 15 3 3 NUM ajst-19323 83 16 where where SCONJ ajst-19323 83 17 , , PUNCT ajst-19323 83 18 π›½οΌˆπ›½ π›½οΌˆπ›½ PROPN ajst-19323 83 19 0 0 NUM ajst-19323 83 20 οΌ‰ οΌ‰ X ajst-19323 83 21 is be AUX ajst-19323 83 22 a a DET ajst-19323 83 23 parameter parameter NOUN ajst-19323 83 24 ; ; PUNCT ajst-19323 83 25 𝐿 𝐿 PROPN ajst-19323 83 26 𝐷 𝐷 PROPN ajst-19323 83 27 𝑆 𝑆 PROPN ajst-19323 83 28 , , PUNCT ajst-19323 83 29 𝐿 𝐿 PROPN ajst-19323 83 30 refers refer VERB ajst-19323 83 31 to to ADP ajst-19323 83 32 the the DET ajst-19323 83 33 laplacian laplacian ADJ ajst-19323 83 34 matrix matrix NOUN ajst-19323 83 35 , , PUNCT ajst-19323 83 36 𝐷 𝐷 PROPN ajst-19323 83 37 is be AUX ajst-19323 83 38 a a DET ajst-19323 83 39 diagonal diagonal ADJ ajst-19323 83 40 matrix matrix NOUN ajst-19323 83 41 , , PUNCT ajst-19323 83 42 𝑆 𝑆 PROPN ajst-19323 83 43 ∈ ∈ PROPN ajst-19323 83 44 𝑅 𝑅 PROPN ajst-19323 83 45 represents represent VERB ajst-19323 83 46 the the DET ajst-19323 83 47 feature feature NOUN ajst-19323 83 48 similarity similarity NOUN ajst-19323 83 49 matrix matrix NOUN ajst-19323 83 50 . . PUNCT ajst-19323 84 1 in in ADP ajst-19323 84 2 equation equation NOUN ajst-19323 84 3 ( ( PUNCT ajst-19323 84 4 1 1 NUM ajst-19323 84 5 ) ) PUNCT ajst-19323 84 6 , , PUNCT ajst-19323 84 7 the the DET ajst-19323 84 8 𝑙 𝑙 PRON ajst-19323 84 9 norm norm NOUN ajst-19323 84 10 is be AUX ajst-19323 84 11 used use VERB ajst-19323 84 12 to to PART ajst-19323 84 13 regularize regularize VERB ajst-19323 84 14 the the DET ajst-19323 84 15 coefficient coefficient NOUN ajst-19323 84 16 matrix matrix NOUN ajst-19323 84 17 to to PART ajst-19323 84 18 make make VERB ajst-19323 84 19 it it PRON ajst-19323 84 20 sparse sparse VERB ajst-19323 84 21 and and CCONJ ajst-19323 84 22 exclude exclude VERB ajst-19323 84 23 some some DET ajst-19323 84 24 redundant redundant ADJ ajst-19323 84 25 features feature NOUN ajst-19323 84 26 in in ADP ajst-19323 84 27 the the DET ajst-19323 84 28 feature feature NOUN ajst-19323 84 29 space space NOUN ajst-19323 84 30 . . PUNCT ajst-19323 85 1 2.3 2.3 NUM ajst-19323 85 2 . . PUNCT ajst-19323 86 1 fill fill VERB ajst-19323 86 2 missing miss VERB ajst-19323 86 3 labels label NOUN ajst-19323 86 4 in in ADP ajst-19323 86 5 the the DET ajst-19323 86 6 case case NOUN ajst-19323 86 7 of of ADP ajst-19323 86 8 missing miss VERB ajst-19323 86 9 labels label NOUN ajst-19323 86 10 , , PUNCT ajst-19323 86 11 it it PRON ajst-19323 86 12 is be AUX ajst-19323 86 13 difficult difficult ADJ ajst-19323 86 14 to to PART ajst-19323 86 15 learn learn VERB ajst-19323 86 16 the the DET ajst-19323 86 17 real real ADJ ajst-19323 86 18 mapping mapping NOUN ajst-19323 86 19 relationship relationship NOUN ajst-19323 86 20 between between ADP ajst-19323 86 21 features feature NOUN ajst-19323 86 22 and and CCONJ ajst-19323 86 23 labels label NOUN ajst-19323 86 24 , , PUNCT ajst-19323 86 25 which which PRON ajst-19323 86 26 leads lead VERB ajst-19323 86 27 to to ADP ajst-19323 86 28 the the DET ajst-19323 86 29 unsatisfactory unsatisfactory ADJ ajst-19323 86 30 classification classification NOUN ajst-19323 86 31 effect effect NOUN ajst-19323 86 32 of of ADP ajst-19323 86 33 multi multi ADJ ajst-19323 86 34 - - ADJ ajst-19323 86 35 label label ADJ ajst-19323 86 36 tasks task NOUN ajst-19323 86 37 . . PUNCT ajst-19323 87 1 the the DET ajst-19323 87 2 missing miss VERB ajst-19323 87 3 label label NOUN ajst-19323 87 4 problem problem NOUN ajst-19323 87 5 is be AUX ajst-19323 87 6 dealt deal VERB ajst-19323 87 7 with with ADP ajst-19323 87 8 by by ADP ajst-19323 87 9 the the DET ajst-19323 87 10 method method NOUN ajst-19323 87 11 of of ADP ajst-19323 87 12 restoring restore VERB ajst-19323 87 13 it it PRON ajst-19323 87 14 . . PUNCT ajst-19323 88 1 the the DET ajst-19323 88 2 iterative iterative NOUN ajst-19323 88 3 process process NOUN ajst-19323 88 4 of of ADP ajst-19323 88 5 solving solve VERB ajst-19323 88 6 coefficient coefficient NOUN ajst-19323 88 7 366 366 NUM ajst-19323 88 8 matrix matrix NOUN ajst-19323 88 9 π‘Š π‘Š NOUN ajst-19323 88 10 and and CCONJ ajst-19323 88 11 the the DET ajst-19323 88 12 recovery recovery NOUN ajst-19323 88 13 process process NOUN ajst-19323 88 14 of of ADP ajst-19323 88 15 missing miss VERB ajst-19323 88 16 label label NOUN ajst-19323 88 17 are be AUX ajst-19323 88 18 carried carry VERB ajst-19323 88 19 out out ADP ajst-19323 88 20 simultaneously simultaneously ADV ajst-19323 88 21 . . PUNCT ajst-19323 89 1 according accord VERB ajst-19323 89 2 to to ADP ajst-19323 89 3 literature literature NOUN ajst-19323 89 4 [ [ X ajst-19323 89 5 16 16 NUM ajst-19323 89 6 ] ] PUNCT ajst-19323 89 7 , , PUNCT ajst-19323 89 8 the the DET ajst-19323 89 9 known know VERB ajst-19323 89 10 labels label NOUN ajst-19323 89 11 in in ADP ajst-19323 89 12 the the DET ajst-19323 89 13 original original ADJ ajst-19323 89 14 label label NOUN ajst-19323 89 15 matrix matrix NOUN ajst-19323 89 16 are be AUX ajst-19323 89 17 assigned assign VERB ajst-19323 89 18 to to ADP ajst-19323 89 19 a a DET ajst-19323 89 20 new new ADJ ajst-19323 89 21 matrix matrix NOUN ajst-19323 89 22 𝐴 𝐴 PROPN ajst-19323 89 23 , , PUNCT ajst-19323 89 24 then then ADV ajst-19323 89 25 𝐴 𝐴 PROPN ajst-19323 89 26 π‘Œ π‘Œ PROPN ajst-19323 89 27 ; ; PUNCT ajst-19323 89 28 and and CCONJ ajst-19323 89 29 where where SCONJ ajst-19323 89 30 the the DET ajst-19323 89 31 label label NOUN ajst-19323 89 32 is be AUX ajst-19323 89 33 missing miss VERB ajst-19323 89 34 , , PUNCT ajst-19323 89 35 𝐴 𝐴 PROPN ajst-19323 89 36 0 0 NUM ajst-19323 89 37 . . PUNCT ajst-19323 90 1 in in ADP ajst-19323 90 2 each each DET ajst-19323 90 3 iteration iteration NOUN ajst-19323 90 4 process process NOUN ajst-19323 90 5 of of ADP ajst-19323 90 6 solving solve VERB ajst-19323 90 7 coefficient coefficient NOUN ajst-19323 90 8 matrix matrix NOUN ajst-19323 90 9 π‘Š π‘Š PROPN ajst-19323 90 10 , , PUNCT ajst-19323 90 11 the the DET ajst-19323 90 12 missing miss VERB ajst-19323 90 13 label label NOUN ajst-19323 90 14 value value NOUN ajst-19323 90 15 in in ADP ajst-19323 90 16 𝐴 𝐴 PROPN ajst-19323 90 17 is be AUX ajst-19323 90 18 updated update VERB ajst-19323 90 19 according accord VERB ajst-19323 90 20 to to ADP ajst-19323 90 21 equation equation NOUN ajst-19323 90 22 ( ( PUNCT ajst-19323 90 23 4 4 NUM ajst-19323 90 24 ) ) PUNCT ajst-19323 90 25 , , PUNCT ajst-19323 90 26 where where SCONJ ajst-19323 90 27 𝐴 𝐴 PROPN ajst-19323 90 28 π‘‹π‘Š π‘‹π‘Š PROPN ajst-19323 90 29 , , PUNCT ajst-19323 90 30 and and CCONJ ajst-19323 90 31 the the DET ajst-19323 90 32 updated update VERB ajst-19323 90 33 𝐴 𝐴 PROPN ajst-19323 90 34 is be AUX ajst-19323 90 35 continued continue VERB ajst-19323 90 36 for for ADP ajst-19323 90 37 subsequent subsequent ADJ ajst-19323 90 38 iterations iteration NOUN ajst-19323 90 39 . . PUNCT ajst-19323 91 1 𝐴 𝐴 PROPN ajst-19323 91 2 1，𝐴 1，𝐴 PROPN ajst-19323 91 3 1 1 NUM ajst-19323 91 4 𝐴 𝐴 NOUN ajst-19323 91 5 ， ， PUNCT ajst-19323 91 6 1 1 NUM ajst-19323 91 7 𝐴 𝐴 NOUN ajst-19323 91 8 1 1 NUM ajst-19323 91 9 1，𝐴 1，𝐴 NUM ajst-19323 91 10 1 1 NUM ajst-19323 91 11 4 4 NUM ajst-19323 91 12 the the DET ajst-19323 91 13 continuous continuous ADJ ajst-19323 91 14 values value NOUN ajst-19323 91 15 in in ADP ajst-19323 91 16 𝐴 𝐴 PROPN ajst-19323 91 17 are be AUX ajst-19323 91 18 restored restore VERB ajst-19323 91 19 to to PART ajst-19323 91 20 discrete discrete VERB ajst-19323 91 21 values value NOUN ajst-19323 91 22 , , PUNCT ajst-19323 91 23 and and CCONJ ajst-19323 91 24 the the DET ajst-19323 91 25 final final ADJ ajst-19323 91 26 matrix matrix NOUN ajst-19323 91 27 𝐴 𝐴 NOUN ajst-19323 91 28 can can AUX ajst-19323 91 29 be be AUX ajst-19323 91 30 obtained obtain VERB ajst-19323 91 31 according accord VERB ajst-19323 91 32 to to ADP ajst-19323 91 33 equation equation NOUN ajst-19323 91 34 ( ( PUNCT ajst-19323 91 35 5 5 NUM ajst-19323 91 36 ) ) PUNCT ajst-19323 91 37 . . PUNCT ajst-19323 92 1 at at ADP ajst-19323 92 2 the the DET ajst-19323 92 3 same same ADJ ajst-19323 92 4 time time NOUN ajst-19323 92 5 , , PUNCT ajst-19323 92 6 let let VERB ajst-19323 92 7 π‘Œ π‘Œ PROPN ajst-19323 92 8 𝐴. 𝐴. PROPN ajst-19323 92 9 𝐴 𝐴 PROPN ajst-19323 92 10 1，𝐴 1，𝐴 VERB ajst-19323 92 11 0.1 0.1 NUM ajst-19323 92 12 1，𝐴 1，𝐴 NUM ajst-19323 92 13 0.1 0.1 NUM ajst-19323 92 14 5 5 NUM ajst-19323 92 15 2.4 2.4 NUM ajst-19323 92 16 . . PUNCT ajst-19323 93 1 model model NOUN ajst-19323 93 2 optimization optimization NOUN ajst-19323 93 3 it it PRON ajst-19323 93 4 can can AUX ajst-19323 93 5 be be AUX ajst-19323 93 6 seen see VERB ajst-19323 93 7 that that SCONJ ajst-19323 93 8 the the DET ajst-19323 93 9 objective objective ADJ ajst-19323 93 10 function function NOUN ajst-19323 93 11 belongs belong VERB ajst-19323 93 12 to to ADP ajst-19323 93 13 the the DET ajst-19323 93 14 nonsmooth nonsmooth ADJ ajst-19323 93 15 convex convex NOUN ajst-19323 93 16 optimization optimization NOUN ajst-19323 93 17 problem problem NOUN ajst-19323 93 18 , , PUNCT ajst-19323 93 19 so so CCONJ ajst-19323 93 20 the the DET ajst-19323 93 21 method method NOUN ajst-19323 93 22 of of ADP ajst-19323 93 23 accelerated accelerate VERB ajst-19323 93 24 proximal proximal ADJ ajst-19323 93 25 gradient gradient NOUN ajst-19323 93 26 can can AUX ajst-19323 93 27 be be AUX ajst-19323 93 28 used use VERB ajst-19323 93 29 to to PART ajst-19323 93 30 deal deal VERB ajst-19323 93 31 with with ADP ajst-19323 93 32 the the DET ajst-19323 93 33 optimization optimization NOUN ajst-19323 93 34 problem problem NOUN ajst-19323 93 35 . . PUNCT ajst-19323 94 1 equation equation NOUN ajst-19323 94 2 ( ( PUNCT ajst-19323 94 3 3 3 X ajst-19323 94 4 ) ) PUNCT ajst-19323 94 5 is be AUX ajst-19323 94 6 divided divide VERB ajst-19323 94 7 into into ADP ajst-19323 94 8 two two NUM ajst-19323 94 9 parts part NOUN ajst-19323 94 10 : : PUNCT ajst-19323 94 11 min min NOUN ajst-19323 94 12 𝐹 𝐹 PROPN ajst-19323 94 13 π‘Š π‘Š VERB ajst-19323 94 14 𝑓 𝑓 DET ajst-19323 94 15 π‘Š π‘Š PROPN ajst-19323 94 16 𝑔 𝑔 NOUN ajst-19323 94 17 π‘Š π‘Š PROPN ajst-19323 94 18 6 6 NUM ajst-19323 94 19 among among ADP ajst-19323 94 20 them they PRON ajst-19323 94 21 , , PUNCT ajst-19323 94 22 𝑓 𝑓 DET ajst-19323 94 23 π‘Š π‘Š NOUN ajst-19323 94 24 1 1 NUM ajst-19323 94 25 2 2 NUM ajst-19323 94 26 β€–π‘‹π‘Š β€–π‘‹π‘Š NOUN ajst-19323 94 27 π‘Œβ€– π‘Œβ€– NOUN ajst-19323 95 1 𝛼 𝛼 PRON ajst-19323 95 2 2 2 NUM ajst-19323 95 3 π‘‘π‘Ÿ π‘‘π‘Ÿ NOUN ajst-19323 95 4 π‘Š π‘Š PROPN ajst-19323 95 5 𝑋 𝑋 NOUN ajst-19323 95 6 𝐿 𝐿 PROPN ajst-19323 95 7 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 95 8 𝛽 𝛽 NOUN ajst-19323 95 9 2 2 NUM ajst-19323 95 10 π‘‘π‘Ÿ π‘‘π‘Ÿ VERB ajst-19323 95 11 π‘Šπ‘Š π‘Šπ‘Š PROPN ajst-19323 95 12 𝐿 𝐿 PROPN ajst-19323 95 13 7 7 NUM ajst-19323 95 14 𝑔 𝑔 PROPN ajst-19323 95 15 π‘Š π‘Š PROPN ajst-19323 95 16 πœ†β€–π‘Šβ€– πœ†β€–π‘Šβ€– PROPN ajst-19323 95 17 8 8 NUM ajst-19323 95 18 each each DET ajst-19323 95 19 element element NOUN ajst-19323 95 20 in in ADP ajst-19323 95 21 the the DET ajst-19323 95 22 coefficient coefficient NOUN ajst-19323 95 23 matrix matrix NOUN ajst-19323 95 24 π‘Š π‘Š NOUN ajst-19323 95 25 is be AUX ajst-19323 95 26 calculated calculate VERB ajst-19323 95 27 using use VERB ajst-19323 95 28 a a DET ajst-19323 95 29 soft soft ADJ ajst-19323 95 30 threshold threshold NOUN ajst-19323 95 31 , , PUNCT ajst-19323 95 32 which which PRON ajst-19323 95 33 is be AUX ajst-19323 95 34 : : PUNCT ajst-19323 95 35 π‘Š π‘Š VERB ajst-19323 95 36 π‘π‘Ÿπ‘œπ‘₯ π‘π‘Ÿπ‘œπ‘₯ NOUN ajst-19323 95 37 𝐺 𝐺 NOUN ajst-19323 95 38 9 9 NUM ajst-19323 95 39 the the DET ajst-19323 95 40 method method NOUN ajst-19323 95 41 proposed propose VERB ajst-19323 95 42 by by ADP ajst-19323 95 43 lin lin PROPN ajst-19323 95 44 et et PROPN ajst-19323 95 45 al[17 al[17 PROPN ajst-19323 95 46 ] ] X ajst-19323 95 47 is be AUX ajst-19323 95 48 adopted adopt VERB ajst-19323 95 49 to to PART ajst-19323 95 50 accelerate accelerate VERB ajst-19323 95 51 the the DET ajst-19323 95 52 solution solution NOUN ajst-19323 95 53 of of ADP ajst-19323 95 54 π‘Š π‘Š PROPN ajst-19323 95 55 , , PUNCT ajst-19323 95 56 which which PRON ajst-19323 95 57 can can AUX ajst-19323 95 58 be be AUX ajst-19323 95 59 set set VERB ajst-19323 95 60 π‘Š π‘Š PROPN ajst-19323 95 61 π‘Š π‘Š NOUN ajst-19323 95 62 𝑏 𝑏 DET ajst-19323 95 63 1 1 NUM ajst-19323 95 64 𝑏 𝑏 NOUN ajst-19323 95 65 π‘Š π‘Š NOUN ajst-19323 95 66 π‘Š π‘Š NOUN ajst-19323 95 67 10 10 NUM ajst-19323 95 68 significant significant ADJ ajst-19323 95 69 runtime runtime NOUN ajst-19323 95 70 savings saving NOUN ajst-19323 95 71 . . PUNCT ajst-19323 96 1 at at ADP ajst-19323 96 2 the the DET ajst-19323 96 3 same same ADJ ajst-19323 96 4 time time NOUN ajst-19323 96 5 , , PUNCT ajst-19323 96 6 when when SCONJ ajst-19323 96 7 solving solve VERB ajst-19323 96 8 the the DET ajst-19323 96 9 optimization optimization NOUN ajst-19323 96 10 problem problem NOUN ajst-19323 96 11 of of ADP ajst-19323 96 12 equation equation NOUN ajst-19323 96 13 ( ( PUNCT ajst-19323 96 14 7 7 NUM ajst-19323 96 15 ) ) PUNCT ajst-19323 96 16 , , PUNCT ajst-19323 96 17 lipschitz lipschitz VERB ajst-19323 96 18 continuity continuity NOUN ajst-19323 96 19 is be AUX ajst-19323 96 20 satisfied satisfied ADJ ajst-19323 96 21 : : PUNCT ajst-19323 96 22 β€–βˆ‡π‘“ β€–βˆ‡π‘“ PROPN ajst-19323 96 23 π‘Š π‘Š NOUN ajst-19323 96 24 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 96 25 π‘Š π‘Š PROPN ajst-19323 96 26 β€– β€– PROPN ajst-19323 96 27 𝐿 𝐿 PROPN ajst-19323 96 28 β€–π‘Š β€–π‘Š PROPN ajst-19323 96 29 π‘Š π‘Š PROPN ajst-19323 96 30 β€– β€– PROPN ajst-19323 96 31 , , PUNCT ajst-19323 96 32 𝐿 𝐿 PROPN ajst-19323 96 33 is be AUX ajst-19323 96 34 lipschitz lipschitz NOUN ajst-19323 96 35 constant constant ADJ ajst-19323 96 36 , , PUNCT ajst-19323 96 37 and and CCONJ ajst-19323 96 38 𝐿 𝐿 PROPN ajst-19323 96 39 can can AUX ajst-19323 96 40 be be AUX ajst-19323 96 41 obtained obtain VERB ajst-19323 96 42 by by ADP ajst-19323 96 43 proving prove VERB ajst-19323 96 44 lipschitz lipschitz NOUN ajst-19323 96 45 continuity continuity NOUN ajst-19323 96 46 . . PUNCT ajst-19323 97 1 the the DET ajst-19323 97 2 proof proof NOUN ajst-19323 97 3 process process NOUN ajst-19323 97 4 is be AUX ajst-19323 97 5 as as SCONJ ajst-19323 97 6 follows follow VERB ajst-19323 97 7 : : PUNCT ajst-19323 97 8 the the DET ajst-19323 97 9 gradient gradient NOUN ajst-19323 97 10 of of ADP ajst-19323 97 11 π‘Š π‘Š PROPN ajst-19323 97 12 in in ADP ajst-19323 97 13 equation equation NOUN ajst-19323 97 14 ( ( PUNCT ajst-19323 97 15 7 7 X ajst-19323 97 16 ) ) PUNCT ajst-19323 97 17 is be AUX ajst-19323 97 18 calculated calculate VERB ajst-19323 97 19 as as ADP ajst-19323 97 20 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 97 21 π‘Š π‘Š PROPN ajst-19323 97 22 𝑋 𝑋 NOUN ajst-19323 97 23 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 24 𝑋 𝑋 PROPN ajst-19323 97 25 π‘Œ π‘Œ PROPN ajst-19323 97 26 𝛼𝑋 𝛼𝑋 PROPN ajst-19323 97 27 𝐿 𝐿 PROPN ajst-19323 97 28 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 29 𝛽𝐿 𝛽𝐿 NOUN ajst-19323 97 30 π‘Š π‘Š PROPN ajst-19323 97 31 11 11 NUM ajst-19323 97 32 assuming assume VERB ajst-19323 97 33 that that SCONJ ajst-19323 97 34 there there PRON ajst-19323 97 35 are be VERB ajst-19323 97 36 parameters parameter NOUN ajst-19323 97 37 π‘Š π‘Š PROPN ajst-19323 97 38 and and CCONJ ajst-19323 97 39 π‘Š π‘Š PROPN ajst-19323 97 40 , , PUNCT ajst-19323 97 41 substituting substitute VERB ajst-19323 97 42 parameters parameter NOUN ajst-19323 97 43 π‘Š π‘Š PROPN ajst-19323 97 44 and and CCONJ ajst-19323 97 45 π‘Š π‘Š NOUN ajst-19323 97 46 into into ADP ajst-19323 97 47 equation equation NOUN ajst-19323 97 48 ( ( PUNCT ajst-19323 97 49 11 11 NUM ajst-19323 97 50 ) ) PUNCT ajst-19323 97 51 respectively respectively ADV ajst-19323 97 52 , , PUNCT ajst-19323 97 53 we we PRON ajst-19323 97 54 can can AUX ajst-19323 97 55 get get VERB ajst-19323 97 56 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 97 57 π‘Š π‘Š NOUN ajst-19323 97 58 𝑋 𝑋 NOUN ajst-19323 97 59 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 60 𝑋 𝑋 PROPN ajst-19323 97 61 π‘Œ π‘Œ PROPN ajst-19323 97 62 𝛼𝑋 𝛼𝑋 PROPN ajst-19323 97 63 𝐿 𝐿 PROPN ajst-19323 97 64 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 65 𝛽𝐿 𝛽𝐿 NOUN ajst-19323 97 66 π‘Š π‘Š PROPN ajst-19323 97 67 12 12 NUM ajst-19323 97 68 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 97 69 π‘Š π‘Š NOUN ajst-19323 97 70 𝑋 𝑋 NOUN ajst-19323 97 71 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 72 𝑋 𝑋 PROPN ajst-19323 97 73 π‘Œ π‘Œ PROPN ajst-19323 97 74 𝛼𝑋 𝛼𝑋 PROPN ajst-19323 97 75 𝐿 𝐿 PROPN ajst-19323 97 76 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 97 77 𝛽𝐿 𝛽𝐿 NOUN ajst-19323 97 78 π‘Š π‘Š PROPN ajst-19323 97 79 13 13 NUM ajst-19323 97 80 according accord VERB ajst-19323 97 81 to to ADP ajst-19323 97 82 equation equation NOUN ajst-19323 97 83 ( ( PUNCT ajst-19323 97 84 12 12 NUM ajst-19323 97 85 ) ) PUNCT ajst-19323 97 86 and and CCONJ ajst-19323 97 87 equation equation NOUN ajst-19323 97 88 ( ( PUNCT ajst-19323 97 89 13 13 NUM ajst-19323 97 90 ) ) PUNCT ajst-19323 97 91 , , PUNCT ajst-19323 97 92 and and CCONJ ajst-19323 97 93 βˆ†π‘Š βˆ†π‘Š ADP ajst-19323 97 94 π‘Š π‘Š PROPN ajst-19323 97 95 π‘Š π‘Š PROPN ajst-19323 97 96 , , PUNCT ajst-19323 97 97 can can AUX ajst-19323 97 98 be be AUX ajst-19323 97 99 obtained obtain VERB ajst-19323 97 100 β€–βˆ‡π‘“ β€–βˆ‡π‘“ PROPN ajst-19323 97 101 π‘Š π‘Š PROPN ajst-19323 97 102 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 97 103 π‘Š π‘Š NOUN ajst-19323 97 104 β€– β€– NOUN ajst-19323 97 105 𝑋 𝑋 NOUN ajst-19323 97 106 π‘‹βˆ†π‘Š π‘‹βˆ†π‘Š PROPN ajst-19323 97 107 𝛼𝑋 𝛼𝑋 NOUN ajst-19323 97 108 𝐿 𝐿 PROPN ajst-19323 97 109 π‘‹βˆ†π‘Š π‘‹βˆ†π‘Š NOUN ajst-19323 97 110 𝛽𝐿 𝛽𝐿 PROPN ajst-19323 97 111 βˆ†π‘Š βˆ†π‘Š ADP ajst-19323 97 112 2‖𝑋 2‖𝑋 NUM ajst-19323 97 113 π‘‹βˆ†π‘Šβ€– π‘‹βˆ†π‘Šβ€– ADV ajst-19323 97 114 2‖𝛼𝑋 2‖𝛼𝑋 NUM ajst-19323 97 115 𝐿 𝐿 PROPN ajst-19323 97 116 π‘‹βˆ†π‘Šβ€– π‘‹βˆ†π‘Šβ€– ADJ ajst-19323 97 117 2 2 NUM ajst-19323 97 118 𝛽𝐿 𝛽𝐿 PROPN ajst-19323 97 119 βˆ†π‘Š βˆ†π‘Š ADP ajst-19323 97 120 2‖𝑋 2‖𝑋 NUM ajst-19323 97 121 𝑋‖ 𝑋‖ ADJ ajst-19323 97 122 β€–βˆ†π‘Šβ€– β€–βˆ†π‘Šβ€– ADJ ajst-19323 97 123 2‖𝛼𝑋 2‖𝛼𝑋 NUM ajst-19323 97 124 𝐿 𝐿 PROPN ajst-19323 97 125 𝑋‖ 𝑋‖ ADJ ajst-19323 97 126 β€–βˆ†π‘Šβ€– β€–βˆ†π‘Šβ€– ADJ ajst-19323 97 127 2 2 NUM ajst-19323 97 128 𝛽𝐿 𝛽𝐿 NOUN ajst-19323 97 129 β€–βˆ†π‘Šβ€– β€–βˆ†π‘Šβ€– NOUN ajst-19323 97 130 2 2 NUM ajst-19323 97 131 ‖𝑋 ‖𝑋 NOUN ajst-19323 97 132 𝑋‖ 𝑋‖ ADJ ajst-19323 97 133 ‖𝛼𝑋 ‖𝛼𝑋 PROPN ajst-19323 97 134 𝐿 𝐿 PROPN ajst-19323 97 135 𝑋‖ 𝑋‖ ADJ ajst-19323 97 136 𝛽𝐿 𝛽𝐿 ADJ ajst-19323 97 137 β€–βˆ†π‘Šβ€– β€–βˆ†π‘Šβ€– NOUN ajst-19323 97 138 14 14 NUM ajst-19323 97 139 therefore therefore ADV ajst-19323 97 140 β€–βˆ‡π‘“ β€–βˆ‡π‘“ PROPN ajst-19323 97 141 π‘Š π‘Š PROPN ajst-19323 97 142 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 97 143 π‘Š π‘Š NOUN ajst-19323 97 144 β€– β€– PROPN ajst-19323 97 145 2 2 NUM ajst-19323 97 146 ‖𝑋 ‖𝑋 NOUN ajst-19323 97 147 𝑋‖ 𝑋‖ ADJ ajst-19323 97 148 ‖𝛼𝑋 ‖𝛼𝑋 PROPN ajst-19323 97 149 𝐿 𝐿 PROPN ajst-19323 97 150 𝑋‖ 𝑋‖ ADJ ajst-19323 97 151 𝛽𝐿 𝛽𝐿 ADJ ajst-19323 97 152 β€–βˆ†π‘Šβ€– β€–βˆ†π‘Šβ€– PROPN ajst-19323 97 153 15 15 NUM ajst-19323 97 154 therefore therefore ADV ajst-19323 97 155 𝐿 𝐿 PROPN ajst-19323 97 156 2 2 NUM ajst-19323 97 157 ‖𝑋 ‖𝑋 NOUN ajst-19323 97 158 𝑋‖ 𝑋‖ ADJ ajst-19323 98 1 ‖𝛼𝑋 ‖𝛼𝑋 PROPN ajst-19323 98 2 𝐿 𝐿 PROPN ajst-19323 98 3 𝑋‖ 𝑋‖ PROPN ajst-19323 98 4 𝛽𝐿 𝛽𝐿 PROPN ajst-19323 98 5 16 16 NUM ajst-19323 98 6 ml ml PROPN ajst-19323 98 7 - - PUNCT ajst-19323 98 8 lsf lsf NOUN ajst-19323 98 9 algorithm algorithm NOUN ajst-19323 98 10 as as SCONJ ajst-19323 98 11 shown show VERB ajst-19323 98 12 in in ADP ajst-19323 98 13 algorithm algorithm NOUN ajst-19323 98 14 1 1 NUM ajst-19323 98 15 . . PUNCT ajst-19323 98 16 algorithm algorithm NOUN ajst-19323 98 17 1 1 NUM ajst-19323 98 18 : : PUNCT ajst-19323 98 19 multi multi ADJ ajst-19323 98 20 - - ADJ ajst-19323 98 21 label label ADJ ajst-19323 98 22 feature feature NOUN ajst-19323 98 23 selection selection NOUN ajst-19323 98 24 based base VERB ajst-19323 98 25 on on ADP ajst-19323 98 26 labelspecific labelspecific ADJ ajst-19323 98 27 feature feature NOUN ajst-19323 98 28 and and CCONJ ajst-19323 98 29 manifold manifold ADJ ajst-19323 98 30 learning learning NOUN ajst-19323 98 31 input input NOUN ajst-19323 98 32 : : PUNCT ajst-19323 98 33 training training NOUN ajst-19323 98 34 matrix matrix NOUN ajst-19323 98 35 𝑋 𝑋 PROPN ajst-19323 98 36 ∈ ∈ PROPN ajst-19323 98 37 𝑅 𝑅 PROPN ajst-19323 98 38 , , PUNCT ajst-19323 98 39 label label NOUN ajst-19323 98 40 matrix matrix NOUN ajst-19323 98 41 π‘Œ π‘Œ PROPN ajst-19323 98 42 ∈ ∈ PROPN ajst-19323 98 43 𝑅 𝑅 PROPN ajst-19323 98 44 with with ADP ajst-19323 98 45 missing miss VERB ajst-19323 98 46 labels label NOUN ajst-19323 98 47 , , PUNCT ajst-19323 98 48 parameters parameter NOUN ajst-19323 98 49 πœ† πœ† ADP ajst-19323 98 50 , , PUNCT ajst-19323 98 51 𝛼 𝛼 PROPN ajst-19323 98 52 , , PUNCT ajst-19323 98 53 𝛽 𝛽 NOUN ajst-19323 98 54 ; ; PUNCT ajst-19323 98 55 output output NOUN ajst-19323 98 56 : : PUNCT ajst-19323 98 57 coefficient coefficient NOUN ajst-19323 98 58 matrix matrix NOUN ajst-19323 98 59 π‘Šβˆ— π‘Šβˆ— NUM ajst-19323 98 60 1οΌ‰initialization 1οΌ‰initialization NUM ajst-19323 98 61 : : PUNCT ajst-19323 98 62 π‘Š π‘Š NOUN ajst-19323 98 63 π‘Š π‘Š NOUN ajst-19323 98 64 𝑋 𝑋 NOUN ajst-19323 98 65 𝑋 𝑋 PROPN ajst-19323 98 66 𝛾𝐼 𝛾𝐼 PROPN ajst-19323 98 67 𝑋 𝑋 PROPN ajst-19323 98 68 π‘Œ π‘Œ PROPN ajst-19323 98 69 , , PUNCT ajst-19323 98 70 𝑏 𝑏 PROPN ajst-19323 98 71 𝑏 𝑏 PROPN ajst-19323 98 72 1 1 NUM ajst-19323 98 73 , , PUNCT ajst-19323 98 74 𝑑 𝑑 ADP ajst-19323 98 75 1 1 NUM ajst-19323 98 76 2οΌ‰repeat 2οΌ‰repeat NUM ajst-19323 98 77 : : PUNCT ajst-19323 98 78 3 3 NUM ajst-19323 98 79 οΌ‰ οΌ‰ PUNCT ajst-19323 98 80 calculate calculate VERB ajst-19323 98 81 the the DET ajst-19323 98 82 laplacian laplacian ADJ ajst-19323 98 83 matrix matrix NOUN ajst-19323 98 84 𝐿 𝐿 PROPN ajst-19323 98 85 , , PUNCT ajst-19323 98 86 𝐿 𝐿 PROPN ajst-19323 98 87 4 4 NUM ajst-19323 98 88 οΌ‰ οΌ‰ NOUN ajst-19323 98 89 calculate calculate NOUN ajst-19323 98 90 𝐿 𝐿 PROPN ajst-19323 98 91 according accord VERB ajst-19323 98 92 to to ADP ajst-19323 98 93 equation(16 equation(16 NOUN ajst-19323 98 94 ) ) PUNCT ajst-19323 99 1 5 5 NUM ajst-19323 99 2 οΌ‰ οΌ‰ X ajst-19323 99 3 π‘Š π‘Š VERB ajst-19323 99 4 π‘Š π‘Š NOUN ajst-19323 99 5 π‘Š π‘Š NOUN ajst-19323 99 6 π‘Š π‘Š NOUN ajst-19323 99 7 6 6 NUM ajst-19323 99 8 οΌ‰ οΌ‰ NOUN ajst-19323 99 9 𝐺 𝐺 NOUN ajst-19323 99 10 π‘Š π‘Š NOUN ajst-19323 99 11 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 99 12 π‘Š π‘Š PROPN ajst-19323 99 13 7 7 NUM ajst-19323 99 14 οΌ‰ οΌ‰ SYM ajst-19323 99 15 π‘Š π‘Š NOUN ajst-19323 99 16 π‘π‘Ÿπ‘œπ‘₯ π‘π‘Ÿπ‘œπ‘₯ NOUN ajst-19323 99 17 π‘Š π‘Š VERB ajst-19323 99 18 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 99 19 π‘Š π‘Š PROPN ajst-19323 99 20 8 8 NUM ajst-19323 99 21 οΌ‰ οΌ‰ NOUN ajst-19323 99 22 π‘Š π‘Š VERB ajst-19323 99 23 π‘Š π‘Š NOUN ajst-19323 99 24 9 9 NUM ajst-19323 99 25 οΌ‰ οΌ‰ NOUN ajst-19323 99 26 update update NOUN ajst-19323 99 27 π‘Œ π‘Œ PROPN ajst-19323 99 28 by by ADP ajst-19323 99 29 π‘Œ π‘Œ PROPN ajst-19323 99 30 π‘‹π‘Š π‘‹π‘Š NOUN ajst-19323 99 31 , , PUNCT ajst-19323 99 32 according accord VERB ajst-19323 99 33 to to ADP ajst-19323 99 34 equation equation NOUN ajst-19323 99 35 ( ( PUNCT ajst-19323 99 36 4 4 NUM ajst-19323 99 37 ) ) PUNCT ajst-19323 99 38 10 10 NUM ajst-19323 99 39 οΌ‰ οΌ‰ PUNCT ajst-19323 99 40 𝑑 𝑑 PROPN ajst-19323 99 41 𝑑 𝑑 VERB ajst-19323 99 42 1 1 NUM ajst-19323 99 43 11 11 NUM ajst-19323 99 44 οΌ‰ οΌ‰ SYM ajst-19323 100 1 𝑏 𝑏 DET ajst-19323 100 2 12οΌ‰until 12οΌ‰until NUM ajst-19323 100 3 convergence convergence NOUN ajst-19323 100 4 13 13 NUM ajst-19323 100 5 οΌ‰ οΌ‰ PUNCT ajst-19323 100 6 restore restore VERB ajst-19323 100 7 the the DET ajst-19323 100 8 label label NOUN ajst-19323 100 9 matrix matrix NOUN ajst-19323 100 10 π‘Œ π‘Œ PROPN ajst-19323 100 11 with with ADP ajst-19323 100 12 missing miss VERB ajst-19323 100 13 labels label NOUN ajst-19323 100 14 according accord VERB ajst-19323 100 15 to to ADP ajst-19323 100 16 equation equation NOUN ajst-19323 100 17 ( ( PUNCT ajst-19323 100 18 5 5 NUM ajst-19323 100 19 ) ) PUNCT ajst-19323 100 20 14οΌ‰π‘Šβˆ— 14οΌ‰π‘Šβˆ— NUM ajst-19323 100 21 π‘Š π‘Š PROPN ajst-19323 100 22 15οΌ‰compute 15οΌ‰compute NUM ajst-19323 100 23 ‖𝑀 ‖𝑀 NOUN ajst-19323 100 24 β€– β€– PROPN ajst-19323 100 25 , , PUNCT ajst-19323 100 26 𝑖 𝑖 SYM ajst-19323 100 27 ∈ ∈ PROPN ajst-19323 100 28 1 1 NUM ajst-19323 100 29 , , PUNCT ajst-19323 100 30 𝑑 𝑑 PROPN ajst-19323 100 31 16οΌ‰sort 16οΌ‰sort NUM ajst-19323 100 32 and and CCONJ ajst-19323 100 33 select select VERB ajst-19323 100 34 the the DET ajst-19323 100 35 first first ADJ ajst-19323 100 36 𝑙 𝑙 NOUN ajst-19323 100 37 features feature VERB ajst-19323 100 38 2.5 2.5 NUM ajst-19323 100 39 . . PUNCT ajst-19323 101 1 time time NOUN ajst-19323 101 2 complexity complexity NOUN ajst-19323 101 3 in in ADP ajst-19323 101 4 step step NOUN ajst-19323 101 5 1 1 NUM ajst-19323 101 6 , , PUNCT ajst-19323 101 7 π‘Š π‘Š PROPN ajst-19323 101 8 needs need VERB ajst-19323 101 9 to to PART ajst-19323 101 10 be be AUX ajst-19323 101 11 initialized initialize VERB ajst-19323 101 12 and and CCONJ ajst-19323 101 13 the the DET ajst-19323 101 14 time time NOUN ajst-19323 101 15 complexity complexity NOUN ajst-19323 101 16 is be AUX ajst-19323 101 17 𝑂 𝑂 PROPN ajst-19323 101 18 𝑛𝑑𝑙 𝑛𝑑𝑙 NOUN ajst-19323 101 19 𝑛𝑑 𝑛𝑑 ADP ajst-19323 101 20 𝑑 𝑑 NOUN ajst-19323 101 21 ; ; PUNCT ajst-19323 101 22 in in ADP ajst-19323 101 23 step step NOUN ajst-19323 101 24 4 4 NUM ajst-19323 101 25 , , PUNCT ajst-19323 101 26 we we PRON ajst-19323 101 27 need need VERB ajst-19323 101 28 to to PART ajst-19323 101 29 compute compute VERB ajst-19323 101 30 𝐿 𝐿 PROPN ajst-19323 101 31 , , PUNCT ajst-19323 101 32 and and CCONJ ajst-19323 101 33 the the DET ajst-19323 101 34 time time NOUN ajst-19323 101 35 complexity complexity NOUN ajst-19323 101 36 is be AUX ajst-19323 101 37 𝑂 𝑂 PROPN ajst-19323 101 38 𝑑 𝑑 NOUN ajst-19323 101 39 𝑙 𝑙 X ajst-19323 101 40 ; ; PUNCT ajst-19323 101 41 in in ADP ajst-19323 101 42 step step NOUN ajst-19323 101 43 6 6 NUM ajst-19323 101 44 , , PUNCT ajst-19323 101 45 we we PRON ajst-19323 101 46 need need VERB ajst-19323 101 47 to to PART ajst-19323 101 48 compute compute VERB ajst-19323 101 49 βˆ‡π‘“ βˆ‡π‘“ NOUN ajst-19323 101 50 π‘Š π‘Š PROPN ajst-19323 101 51 , , PUNCT ajst-19323 101 52 the the DET ajst-19323 101 53 time time NOUN ajst-19323 101 54 complexity complexity NOUN ajst-19323 101 55 is be AUX ajst-19323 101 56 𝑂 𝑂 PROPN ajst-19323 101 57 𝑑 𝑑 PROPN ajst-19323 101 58 𝑙 𝑙 X ajst-19323 101 59 𝑑𝑙 𝑑𝑙 NOUN ajst-19323 101 60 , , PUNCT ajst-19323 101 61 and and CCONJ ajst-19323 101 62 after after ADP ajst-19323 101 63 𝑑 𝑑 NOUN ajst-19323 101 64 iterations iteration NOUN ajst-19323 101 65 , , PUNCT ajst-19323 101 66 the the DET ajst-19323 101 67 time time NOUN ajst-19323 101 68 complexity complexity NOUN ajst-19323 101 69 is be AUX ajst-19323 101 70 𝑂 𝑂 PROPN ajst-19323 101 71 𝑑 𝑑 PROPN ajst-19323 101 72 𝑑 𝑑 PROPN ajst-19323 101 73 𝑙 𝑙 X ajst-19323 101 74 𝑑𝑙 𝑑𝑙 NOUN ajst-19323 101 75 . . PUNCT ajst-19323 102 1 as as ADP ajst-19323 102 2 a a DET ajst-19323 102 3 result result NOUN ajst-19323 102 4 , , PUNCT ajst-19323 102 5 the the DET ajst-19323 102 6 algorithm algorithm NOUN ajst-19323 102 7 's 's PART ajst-19323 102 8 time time NOUN ajst-19323 102 9 complexity complexity NOUN ajst-19323 102 10 is be AUX ajst-19323 102 11 𝑂 𝑂 PROPN ajst-19323 102 12 𝑛𝑑𝑙 𝑛𝑑𝑙 NOUN ajst-19323 102 13 𝑛𝑑 𝑛𝑑 ADP ajst-19323 102 14 𝑑 𝑑 NOUN ajst-19323 102 15 𝑂 𝑂 PROPN ajst-19323 102 16 𝑑 𝑑 PROPN ajst-19323 102 17 𝑙 𝑙 NOUN ajst-19323 102 18 𝑂 𝑂 PROPN ajst-19323 102 19 𝑑 𝑑 PROPN ajst-19323 102 20 𝑑 𝑑 PROPN ajst-19323 102 21 𝑙 𝑙 X ajst-19323 102 22 𝑑𝑙 𝑑𝑙 NOUN ajst-19323 102 23 . . PUNCT ajst-19323 103 1 367 367 NUM ajst-19323 103 2 3 3 NUM ajst-19323 103 3 . . PUNCT ajst-19323 103 4 experimental experimental ADJ ajst-19323 103 5 design design NOUN ajst-19323 103 6 and and CCONJ ajst-19323 103 7 results result VERB ajst-19323 103 8 3.1 3.1 NUM ajst-19323 103 9 . . PUNCT ajst-19323 104 1 data datum NOUN ajst-19323 104 2 set set VERB ajst-19323 104 3 and and CCONJ ajst-19323 104 4 experiment experiment NOUN ajst-19323 104 5 setup setup NOUN ajst-19323 104 6 in in ADP ajst-19323 104 7 order order NOUN ajst-19323 104 8 to to PART ajst-19323 104 9 verify verify VERB ajst-19323 104 10 the the DET ajst-19323 104 11 effectiveness effectiveness NOUN ajst-19323 104 12 of of ADP ajst-19323 104 13 ml ml NOUN ajst-19323 104 14 - - PUNCT ajst-19323 104 15 lsf lsf NOUN ajst-19323 104 16 , , PUNCT ajst-19323 104 17 mulan mulan PROPN ajst-19323 104 18 public public ADJ ajst-19323 104 19 data datum NOUN ajst-19323 104 20 sets set NOUN ajst-19323 104 21 from from ADP ajst-19323 104 22 multi multi ADJ ajst-19323 104 23 - - ADJ ajst-19323 104 24 label label ADJ ajst-19323 104 25 library library NOUN ajst-19323 104 26 ( ( PUNCT ajst-19323 104 27 http://mulan.sourceforge.net/datasets.html http://mulan.sourceforge.net/datasets.html NOUN ajst-19323 104 28 ) ) PUNCT ajst-19323 104 29 to to PART ajst-19323 104 30 choose choose VERB ajst-19323 104 31 five five NUM ajst-19323 104 32 classic classic ADJ ajst-19323 104 33 data datum NOUN ajst-19323 104 34 set set VERB ajst-19323 104 35 , , PUNCT ajst-19323 104 36 as as SCONJ ajst-19323 104 37 shown show VERB ajst-19323 104 38 in in ADP ajst-19323 104 39 table table NOUN ajst-19323 104 40 1 1 NUM ajst-19323 104 41 . . PUNCT ajst-19323 104 42 table table NOUN ajst-19323 104 43 1 1 NUM ajst-19323 104 44 . . PUNCT ajst-19323 105 1 multi multi ADJ ajst-19323 105 2 - - ADJ ajst-19323 105 3 label label ADJ ajst-19323 105 4 datasets dataset NOUN ajst-19323 105 5 data datum NOUN ajst-19323 105 6 set set VERB ajst-19323 105 7 domain domain NOUN ajst-19323 105 8 sample sample NOUN ajst-19323 105 9 feature feature NOUN ajst-19323 105 10 class class NOUN ajst-19323 105 11 arts art NOUN ajst-19323 105 12 text text NOUN ajst-19323 105 13 5000 5000 NUM ajst-19323 105 14 462 462 NUM ajst-19323 105 15 26 26 NUM ajst-19323 105 16 entertain entertain AUX ajst-19323 105 17 text text NOUN ajst-19323 105 18 5000 5000 NUM ajst-19323 105 19 640 640 NUM ajst-19323 105 20 21 21 NUM ajst-19323 105 21 emotions emotion NOUN ajst-19323 105 22 music music NOUN ajst-19323 105 23 593 593 NUM ajst-19323 105 24 72 72 NUM ajst-19323 105 25 6 6 NUM ajst-19323 105 26 reference reference NOUN ajst-19323 105 27 text text NOUN ajst-19323 105 28 5000 5000 NUM ajst-19323 105 29 793 793 NUM ajst-19323 105 30 33 33 NUM ajst-19323 105 31 scene scene NOUN ajst-19323 105 32 image image NOUN ajst-19323 105 33 2407 2407 NUM ajst-19323 105 34 294 294 NUM ajst-19323 105 35 6 6 NUM ajst-19323 105 36 for for ADP ajst-19323 105 37 the the DET ajst-19323 105 38 algorithm algorithm NOUN ajst-19323 105 39 ml ml PROPN ajst-19323 105 40 - - PUNCT ajst-19323 105 41 lsf lsf NOUN ajst-19323 105 42 in in ADP ajst-19323 105 43 this this DET ajst-19323 105 44 paper paper NOUN ajst-19323 105 45 , , PUNCT ajst-19323 105 46 there there PRON ajst-19323 105 47 are be VERB ajst-19323 105 48 three three NUM ajst-19323 105 49 parameters parameter NOUN ajst-19323 105 50 πœ† πœ† PRON ajst-19323 105 51 , , PUNCT ajst-19323 105 52 𝛼 𝛼 INTJ ajst-19323 105 53 , , PUNCT ajst-19323 105 54 𝛽 𝛽 NOUN ajst-19323 105 55 , , PUNCT ajst-19323 105 56 and and CCONJ ajst-19323 105 57 the the DET ajst-19323 105 58 parameter parameter NOUN ajst-19323 105 59 πœ† πœ† NOUN ajst-19323 105 60 is be AUX ajst-19323 105 61 set set VERB ajst-19323 105 62 to to ADP ajst-19323 105 63 10 10 NUM ajst-19323 105 64 ， ， PROPN ajst-19323 105 65 10 10 NUM ajst-19323 105 66 ， ， PROPN ajst-19323 105 67 … … PUNCT ajst-19323 105 68 ， ， ADJ ajst-19323 105 69 10 10 NUM ajst-19323 105 70 range range NOUN ajst-19323 105 71 ; ; PUNCT ajst-19323 105 72 parameter parameter NOUN ajst-19323 105 73 𝛼 𝛼 PROPN ajst-19323 105 74 2 2 NUM ajst-19323 105 75 ， ， PROPN ajst-19323 105 76 2 2 NUM ajst-19323 105 77 ， ， X ajst-19323 105 78 … … PUNCT ajst-19323 105 79 ， ， X ajst-19323 105 80 2 2 NUM ajst-19323 105 81 , , PUNCT ajst-19323 105 82 parameter parameter NOUN ajst-19323 105 83 𝛽 𝛽 NOUN ajst-19323 105 84 2 2 NUM ajst-19323 105 85 ， ， PROPN ajst-19323 105 86 2 2 NUM ajst-19323 105 87 ， ， X ajst-19323 105 88 … … PUNCT ajst-19323 105 89 ， ， PROPN ajst-19323 105 90 2 2 NUM ajst-19323 105 91 , , PUNCT ajst-19323 105 92 k k NOUN ajst-19323 105 93 - - PUNCT ajst-19323 105 94 nearest near ADJ ajst-19323 105 95 neighbor neighbor NOUN ajst-19323 105 96 probability probability NOUN ajst-19323 105 97 graph graph NOUN ajst-19323 105 98 model model NOUN ajst-19323 105 99 𝐾 𝐾 PROPN ajst-19323 105 100 10 10 NUM ajst-19323 105 101 for for ADP ajst-19323 105 102 instance instance NOUN ajst-19323 105 103 correlation correlation NOUN ajst-19323 105 104 and and CCONJ ajst-19323 105 105 feature feature NOUN ajst-19323 105 106 correlation correlation NOUN ajst-19323 105 107 . . PUNCT ajst-19323 106 1 set set VERB ajst-19323 106 2 the the DET ajst-19323 106 3 label label NOUN ajst-19323 106 4 missing miss VERB ajst-19323 106 5 rate rate NOUN ajst-19323 106 6 to to ADP ajst-19323 106 7 0 0 NUM ajst-19323 106 8 % % NOUN ajst-19323 106 9 , , PUNCT ajst-19323 106 10 10 10 NUM ajst-19323 106 11 % % NOUN ajst-19323 106 12 , , PUNCT ajst-19323 106 13 and and CCONJ ajst-19323 106 14 20 20 NUM ajst-19323 106 15 % % NOUN ajst-19323 106 16 . . PUNCT ajst-19323 107 1 as as ADP ajst-19323 107 2 with with ADP ajst-19323 107 3 all all DET ajst-19323 107 4 comparison comparison NOUN ajst-19323 107 5 algorithms algorithm NOUN ajst-19323 107 6 , , PUNCT ajst-19323 107 7 the the DET ajst-19323 107 8 classical classical ADJ ajst-19323 107 9 ml ml NOUN ajst-19323 107 10 - - PUNCT ajst-19323 107 11 knn knn PROPN ajst-19323 107 12 classifier classifier NOUN ajst-19323 107 13 is be AUX ajst-19323 107 14 used use VERB ajst-19323 107 15 , , PUNCT ajst-19323 107 16 where where SCONJ ajst-19323 107 17 the the DET ajst-19323 107 18 parameters parameter NOUN ajst-19323 107 19 num num ADJ ajst-19323 107 20 and and CCONJ ajst-19323 107 21 smooth smooth ADJ ajst-19323 107 22 are be AUX ajst-19323 107 23 set set VERB ajst-19323 107 24 to to ADP ajst-19323 107 25 10 10 NUM ajst-19323 107 26 and and CCONJ ajst-19323 107 27 1 1 NUM ajst-19323 107 28 , , PUNCT ajst-19323 107 29 respectively respectively ADV ajst-19323 107 30 . . PUNCT ajst-19323 108 1 3.2 3.2 NUM ajst-19323 108 2 . . PUNCT ajst-19323 109 1 experimental experimental ADJ ajst-19323 109 2 results result NOUN ajst-19323 109 3 and and CCONJ ajst-19323 109 4 analysis analysis NOUN ajst-19323 109 5 in in ADP ajst-19323 109 6 order order NOUN ajst-19323 109 7 to to PART ajst-19323 109 8 verify verify VERB ajst-19323 109 9 the the DET ajst-19323 109 10 performance performance NOUN ajst-19323 109 11 of of ADP ajst-19323 109 12 ml ml NOUN ajst-19323 109 13 - - PUNCT ajst-19323 109 14 lsf lsf NOUN ajst-19323 109 15 algorithm algorithm PROPN ajst-19323 109 16 , , PUNCT ajst-19323 109 17 pmu pmu NOUN ajst-19323 109 18 , , PUNCT ajst-19323 109 19 mfml mfml ADJ ajst-19323 109 20 , , PUNCT ajst-19323 109 21 mlmlfs mlmlfs NOUN ajst-19323 109 22 and and CCONJ ajst-19323 109 23 lsf lsf PROPN ajst-19323 109 24 - - PROPN ajst-19323 109 25 ci ci PROPN ajst-19323 109 26 algorithms algorithm NOUN ajst-19323 109 27 are be AUX ajst-19323 109 28 selected select VERB ajst-19323 109 29 to to PART ajst-19323 109 30 compare compare VERB ajst-19323 109 31 with with ADP ajst-19323 109 32 ml ml NOUN ajst-19323 109 33 - - PUNCT ajst-19323 109 34 lsf lsf NOUN ajst-19323 109 35 algorithm algorithm NOUN ajst-19323 109 36 . . PUNCT ajst-19323 110 1 the the DET ajst-19323 110 2 first first ADJ ajst-19323 110 3 three three NUM ajst-19323 110 4 algorithms algorithm NOUN ajst-19323 110 5 are be AUX ajst-19323 110 6 popular popular ADJ ajst-19323 110 7 multi multi ADJ ajst-19323 110 8 - - ADJ ajst-19323 110 9 label label ADJ ajst-19323 110 10 feature feature NOUN ajst-19323 110 11 selection selection NOUN ajst-19323 110 12 algorithms algorithm NOUN ajst-19323 110 13 and and CCONJ ajst-19323 110 14 the the DET ajst-19323 110 15 last last ADJ ajst-19323 110 16 one one NOUN ajst-19323 110 17 is be AUX ajst-19323 110 18 label label NOUN ajst-19323 110 19 - - PUNCT ajst-19323 110 20 specific specific ADJ ajst-19323 110 21 features feature NOUN ajst-19323 110 22 multilabel multilabel NOUN ajst-19323 110 23 learning learn VERB ajst-19323 110 24 algorithm algorithm NOUN ajst-19323 110 25 . . PUNCT ajst-19323 111 1 the the DET ajst-19323 111 2 best good ADJ ajst-19323 111 3 results result NOUN ajst-19323 111 4 are be AUX ajst-19323 111 5 highlighted highlight VERB ajst-19323 111 6 in in ADP ajst-19323 111 7 bold bold ADJ ajst-19323 111 8 in in ADP ajst-19323 111 9 the the DET ajst-19323 111 10 table table NOUN ajst-19323 111 11 . . PUNCT ajst-19323 112 1 the the DET ajst-19323 112 2 experimental experimental ADJ ajst-19323 112 3 results result NOUN ajst-19323 112 4 from from ADP ajst-19323 112 5 table table NOUN ajst-19323 112 6 2 2 NUM ajst-19323 112 7 to to PART ajst-19323 112 8 table table NOUN ajst-19323 112 9 5 5 NUM ajst-19323 112 10 were be AUX ajst-19323 112 11 analyzed analyze VERB ajst-19323 112 12 : : PUNCT ajst-19323 112 13 when when SCONJ ajst-19323 112 14 the the DET ajst-19323 112 15 label label NOUN ajst-19323 112 16 missing miss VERB ajst-19323 112 17 rate rate NOUN ajst-19323 112 18 is be AUX ajst-19323 112 19 10 10 NUM ajst-19323 112 20 % % NOUN ajst-19323 112 21 , , PUNCT ajst-19323 112 22 the the DET ajst-19323 112 23 performance performance NOUN ajst-19323 112 24 of of ADP ajst-19323 112 25 ml ml NOUN ajst-19323 112 26 - - PUNCT ajst-19323 112 27 lsf lsf NOUN ajst-19323 112 28 algorithm algorithm NOUN ajst-19323 112 29 is be AUX ajst-19323 112 30 significantly significantly ADV ajst-19323 112 31 improved improve VERB ajst-19323 112 32 compared compare VERB ajst-19323 112 33 with with ADP ajst-19323 112 34 the the DET ajst-19323 112 35 other other ADJ ajst-19323 112 36 four four NUM ajst-19323 112 37 algorithms algorithm NOUN ajst-19323 112 38 . . PUNCT ajst-19323 113 1 when when SCONJ ajst-19323 113 2 a a DET ajst-19323 113 3 label label NOUN ajst-19323 113 4 is be AUX ajst-19323 113 5 missing miss VERB ajst-19323 113 6 , , PUNCT ajst-19323 113 7 the the DET ajst-19323 113 8 method method NOUN ajst-19323 113 9 of of ADP ajst-19323 113 10 first first ADV ajst-19323 113 11 completing complete VERB ajst-19323 113 12 the the DET ajst-19323 113 13 missing miss VERB ajst-19323 113 14 label label NOUN ajst-19323 113 15 can can AUX ajst-19323 113 16 effectively effectively ADV ajst-19323 113 17 improve improve VERB ajst-19323 113 18 the the DET ajst-19323 113 19 performance performance NOUN ajst-19323 113 20 of of ADP ajst-19323 113 21 the the DET ajst-19323 113 22 algorithm algorithm NOUN ajst-19323 113 23 . . PUNCT ajst-19323 114 1 mfml mfml ADJ ajst-19323 114 2 algorithm algorithm PROPN ajst-19323 114 3 uses use VERB ajst-19323 114 4 feature feature NOUN ajst-19323 114 5 interaction interaction NOUN ajst-19323 114 6 to to PART ajst-19323 114 7 select select VERB ajst-19323 114 8 more more ADV ajst-19323 114 9 valuable valuable ADJ ajst-19323 114 10 features feature NOUN ajst-19323 114 11 that that PRON ajst-19323 114 12 may may AUX ajst-19323 114 13 be be AUX ajst-19323 114 14 ignored ignore VERB ajst-19323 114 15 due due ADJ ajst-19323 114 16 to to ADP ajst-19323 114 17 incomplete incomplete ADJ ajst-19323 114 18 label label NOUN ajst-19323 114 19 space space NOUN ajst-19323 114 20 . . PUNCT ajst-19323 115 1 in in ADP ajst-19323 115 2 this this DET ajst-19323 115 3 way way NOUN ajst-19323 115 4 , , PUNCT ajst-19323 115 5 the the DET ajst-19323 115 6 multi multi ADJ ajst-19323 115 7 - - ADJ ajst-19323 115 8 label label ADJ ajst-19323 115 9 feature feature NOUN ajst-19323 115 10 selection selection NOUN ajst-19323 115 11 problem problem NOUN ajst-19323 115 12 under under ADP ajst-19323 115 13 missing miss VERB ajst-19323 115 14 labels label NOUN ajst-19323 115 15 can can AUX ajst-19323 115 16 be be AUX ajst-19323 115 17 solved solve VERB ajst-19323 115 18 . . PUNCT ajst-19323 116 1 mlmlfs mlmlfs NOUN ajst-19323 116 2 algorithm algorithm PROPN ajst-19323 116 3 uses use VERB ajst-19323 116 4 linear linear PROPN ajst-19323 116 5 regression regression NOUN ajst-19323 116 6 model model NOUN ajst-19323 116 7 to to PART ajst-19323 116 8 recover recover VERB ajst-19323 116 9 missing missing ADJ ajst-19323 116 10 labels label NOUN ajst-19323 116 11 , , PUNCT ajst-19323 116 12 which which PRON ajst-19323 116 13 improves improve VERB ajst-19323 116 14 the the DET ajst-19323 116 15 accuracy accuracy NOUN ajst-19323 116 16 of of ADP ajst-19323 116 17 its its PRON ajst-19323 116 18 modeling modeling NOUN ajst-19323 116 19 . . PUNCT ajst-19323 117 1 the the DET ajst-19323 117 2 experimental experimental ADJ ajst-19323 117 3 results result NOUN ajst-19323 117 4 show show VERB ajst-19323 117 5 that that SCONJ ajst-19323 117 6 these these DET ajst-19323 117 7 two two NUM ajst-19323 117 8 algorithms algorithm NOUN ajst-19323 117 9 can can AUX ajst-19323 117 10 also also ADV ajst-19323 117 11 alleviate alleviate VERB ajst-19323 117 12 the the DET ajst-19323 117 13 performance performance NOUN ajst-19323 117 14 problems problem NOUN ajst-19323 117 15 caused cause VERB ajst-19323 117 16 by by ADP ajst-19323 117 17 missing miss VERB ajst-19323 117 18 labels label NOUN ajst-19323 117 19 . . PUNCT ajst-19323 118 1 however however ADV ajst-19323 118 2 , , PUNCT ajst-19323 118 3 the the DET ajst-19323 118 4 pmu pmu PROPN ajst-19323 118 5 algorithm algorithm NOUN ajst-19323 118 6 can can AUX ajst-19323 118 7 effectively effectively ADV ajst-19323 118 8 handle handle VERB ajst-19323 118 9 multi multi ADJ ajst-19323 118 10 - - ADJ ajst-19323 118 11 label label ADJ ajst-19323 118 12 feature feature NOUN ajst-19323 118 13 selection selection NOUN ajst-19323 118 14 when when SCONJ ajst-19323 118 15 the the DET ajst-19323 118 16 label label NOUN ajst-19323 118 17 space space NOUN ajst-19323 118 18 is be AUX ajst-19323 118 19 complete complete ADJ ajst-19323 118 20 . . PUNCT ajst-19323 119 1 however however ADV ajst-19323 119 2 , , PUNCT ajst-19323 119 3 when when SCONJ ajst-19323 119 4 there there PRON ajst-19323 119 5 are be VERB ajst-19323 119 6 missing miss VERB ajst-19323 119 7 labels label NOUN ajst-19323 119 8 , , PUNCT ajst-19323 119 9 the the DET ajst-19323 119 10 label label NOUN ajst-19323 119 11 space space NOUN ajst-19323 119 12 is be AUX ajst-19323 119 13 damaged damage VERB ajst-19323 119 14 to to ADP ajst-19323 119 15 a a DET ajst-19323 119 16 certain certain ADJ ajst-19323 119 17 extent extent NOUN ajst-19323 119 18 , , PUNCT ajst-19323 119 19 resulting result VERB ajst-19323 119 20 in in ADP ajst-19323 119 21 the the DET ajst-19323 119 22 algorithm algorithm NOUN ajst-19323 119 23 being be AUX ajst-19323 119 24 unable unable ADJ ajst-19323 119 25 to to PART ajst-19323 119 26 accurately accurately ADV ajst-19323 119 27 model model VERB ajst-19323 119 28 the the DET ajst-19323 119 29 relationship relationship NOUN ajst-19323 119 30 between between ADP ajst-19323 119 31 features feature NOUN ajst-19323 119 32 and and CCONJ ajst-19323 119 33 labels label NOUN ajst-19323 119 34 . . PUNCT ajst-19323 120 1 in in ADP ajst-19323 120 2 general general ADJ ajst-19323 120 3 , , PUNCT ajst-19323 120 4 the the DET ajst-19323 120 5 classification classification NOUN ajst-19323 120 6 performance performance NOUN ajst-19323 120 7 of of ADP ajst-19323 120 8 ml ml NOUN ajst-19323 120 9 - - PUNCT ajst-19323 120 10 lsf lsf NOUN ajst-19323 120 11 algorithm algorithm NOUN ajst-19323 120 12 is be AUX ajst-19323 120 13 better well ADJ ajst-19323 120 14 than than ADP ajst-19323 120 15 other other ADJ ajst-19323 120 16 algorithms algorithm NOUN ajst-19323 120 17 . . PUNCT ajst-19323 121 1 table table NOUN ajst-19323 121 2 2 2 NUM ajst-19323 121 3 . . X ajst-19323 121 4 comparison comparison NOUN ajst-19323 121 5 results result VERB ajst-19323 121 6 under under ADP ajst-19323 121 7 10 10 NUM ajst-19323 121 8 % % NOUN ajst-19323 121 9 missing miss VERB ajst-19323 121 10 labels label NOUN ajst-19323 121 11 data datum NOUN ajst-19323 121 12 set set VERB ajst-19323 121 13 pmu pmu PROPN ajst-19323 121 14 mfml mfml PROPN ajst-19323 121 15 mlmlfs mlmlfs PROPN ajst-19323 121 16 lsf lsf PROPN ajst-19323 121 17 - - PROPN ajst-19323 121 18 ci ci PROPN ajst-19323 121 19 ml ml PROPN ajst-19323 121 20 - - PUNCT ajst-19323 121 21 lsf lsf NOUN ajst-19323 121 22 arts arts PROPN ajst-19323 121 23 0.0625 0.0625 NUM ajst-19323 121 24 0.0590 0.0590 NUM ajst-19323 121 25 0.0589 0.0589 NUM ajst-19323 121 26 0.0610 0.0610 NUM ajst-19323 121 27 0.0588 0.0588 NUM ajst-19323 121 28 entertain entertain VERB ajst-19323 121 29 0.0631 0.0631 NUM ajst-19323 121 30 0.0587 0.0587 NUM ajst-19323 121 31 0.0580 0.0580 NUM ajst-19323 121 32 0.0617 0.0617 NUM ajst-19323 121 33 0.0576 0.0576 NUM ajst-19323 121 34 emotions emotion NOUN ajst-19323 121 35 0.2533 0.2533 NUM ajst-19323 122 1 0.2458 0.2458 NUM ajst-19323 122 2 0.2436 0.2436 NUM ajst-19323 122 3 0.2564 0.2564 NUM ajst-19323 122 4 0.2389 0.2389 NUM ajst-19323 122 5 reference reference NOUN ajst-19323 122 6 0.0312 0.0312 NUM ajst-19323 122 7 0.0280 0.0280 NUM ajst-19323 122 8 0.0283 0.0283 NUM ajst-19323 122 9 0.0305 0.0305 NUM ajst-19323 122 10 0.0278 0.0278 NUM ajst-19323 122 11 scene scene NOUN ajst-19323 122 12 0.1121 0.1121 NUM ajst-19323 122 13 0.1097 0.1097 NUM ajst-19323 122 14 0.1092 0.1092 NUM ajst-19323 122 15 0.1103 0.1103 NUM ajst-19323 122 16 0.1091 0.1091 NUM ajst-19323 122 17 average average NOUN ajst-19323 122 18 0.1044 0.1044 NUM ajst-19323 122 19 0.1002 0.1002 NUM ajst-19323 122 20 0.0996 0.0996 NUM ajst-19323 122 21 0.1040 0.1040 NUM ajst-19323 122 22 0.0984 0.0984 NUM ajst-19323 122 23 table table NOUN ajst-19323 122 24 3 3 NUM ajst-19323 122 25 . . X ajst-19323 122 26 comparison comparison NOUN ajst-19323 122 27 results result VERB ajst-19323 122 28 under under ADP ajst-19323 122 29 10 10 NUM ajst-19323 122 30 % % NOUN ajst-19323 122 31 missing miss VERB ajst-19323 122 32 labels label NOUN ajst-19323 122 33 data datum NOUN ajst-19323 122 34 set set VERB ajst-19323 122 35 pmu pmu PROPN ajst-19323 122 36 mfml mfml PROPN ajst-19323 122 37 mlmlfs mlmlfs PROPN ajst-19323 122 38 lsf lsf PROPN ajst-19323 122 39 - - PROPN ajst-19323 122 40 ci ci PROPN ajst-19323 122 41 ml ml PROPN ajst-19323 122 42 - - PUNCT ajst-19323 122 43 lsf lsf NOUN ajst-19323 122 44 arts art NOUN ajst-19323 122 45 0.4630 0.4630 NUM ajst-19323 122 46 0.4987 0.4987 NUM ajst-19323 122 47 0.5012 0.5012 NUM ajst-19323 122 48 0.5011 0.5011 NUM ajst-19323 122 49 0.5197 0.5197 NUM ajst-19323 122 50 entertain entertain NOUN ajst-19323 122 51 0.5512 0.5512 NUM ajst-19323 122 52 0.5678 0.5678 NUM ajst-19323 122 53 0.5519 0.5519 NUM ajst-19323 122 54 0.5612 0.5612 NUM ajst-19323 122 55 0.5803 0.5803 NUM ajst-19323 122 56 emotions emotion NOUN ajst-19323 122 57 0.7236 0.7236 NUM ajst-19323 122 58 0.6830 0.6830 NUM ajst-19323 122 59 0.7326 0.7326 NUM ajst-19323 122 60 0.7130 0.7130 NUM ajst-19323 122 61 0.7401 0.7401 NUM ajst-19323 122 62 reference reference NOUN ajst-19323 122 63 0.6153 0.6153 NUM ajst-19323 122 64 0.6055 0.6055 NUM ajst-19323 122 65 0.5918 0.5918 NUM ajst-19323 122 66 0.6011 0.6011 NUM ajst-19323 122 67 0.6183 0.6183 NUM ajst-19323 122 68 scene scene NOUN ajst-19323 123 1 0.7989 0.7989 NUM ajst-19323 123 2 0.6497 0.6497 NUM ajst-19323 123 3 0.6501 0.6501 NUM ajst-19323 123 4 0.7991 0.7991 NUM ajst-19323 123 5 0.8137 0.8137 NUM ajst-19323 123 6 average average NOUN ajst-19323 123 7 0.6304 0.6304 NUM ajst-19323 123 8 0.6009 0.6009 NUM ajst-19323 123 9 0.6055 0.6055 NUM ajst-19323 123 10 0.6351 0.6351 NUM ajst-19323 123 11 0.6544 0.6544 NUM ajst-19323 123 12 table table NOUN ajst-19323 123 13 4 4 NUM ajst-19323 123 14 . . PUNCT ajst-19323 123 15 comparison comparison NOUN ajst-19323 123 16 results result NOUN ajst-19323 123 17 under under ADP ajst-19323 123 18 10 10 NUM ajst-19323 123 19 % % NOUN ajst-19323 123 20 missing miss VERB ajst-19323 123 21 labels label NOUN ajst-19323 123 22 data datum NOUN ajst-19323 123 23 set set VERB ajst-19323 123 24 pmu pmu PROPN ajst-19323 123 25 mfml mfml PROPN ajst-19323 123 26 mlmlfs mlmlfs PROPN ajst-19323 123 27 lsf lsf PROPN ajst-19323 123 28 - - PROPN ajst-19323 123 29 ci ci PROPN ajst-19323 123 30 ml ml PROPN ajst-19323 123 31 - - PUNCT ajst-19323 123 32 lsf lsf NOUN ajst-19323 123 33 arts arts PROPN ajst-19323 123 34 5.5996 5.5996 NUM ajst-19323 123 35 5.4869 5.4869 NUM ajst-19323 123 36 5.5016 5.5016 NUM ajst-19323 123 37 5.4756 5.4756 NUM ajst-19323 123 38 5.3817 5.3817 NUM ajst-19323 123 39 entertain entertain VERB ajst-19323 123 40 3.4021 3.4021 NUM ajst-19323 123 41 3.2659 3.2659 NUM ajst-19323 123 42 3.2854 3.2854 NUM ajst-19323 123 43 3.3520 3.3520 NUM ajst-19323 123 44 3.1978 3.1978 NUM ajst-19323 123 45 emotions emotion NOUN ajst-19323 123 46 2.3016 2.3016 NUM ajst-19323 123 47 2.2356 2.2356 NUM ajst-19323 123 48 2.2217 2.2217 NUM ajst-19323 123 49 2.2231 2.2231 NUM ajst-19323 123 50 2.0356 2.0356 NUM ajst-19323 123 51 reference reference NOUN ajst-19323 123 52 3.5325 3.5325 NUM ajst-19323 123 53 3.5023 3.5023 NUM ajst-19323 123 54 3.4958 3.4958 NUM ajst-19323 123 55 3.5124 3.5124 NUM ajst-19323 123 56 3.4726 3.4726 NUM ajst-19323 123 57 scene scene NOUN ajst-19323 123 58 0.7326 0.7326 NUM ajst-19323 123 59 0.6989 0.6989 NUM ajst-19323 123 60 0.6807 0.6807 NUM ajst-19323 123 61 0.7012 0.7012 NUM ajst-19323 123 62 0.6695 0.6695 NUM ajst-19323 123 63 average average ADJ ajst-19323 123 64 3.1137 3.1137 NUM ajst-19323 123 65 3.0379 3.0379 NUM ajst-19323 123 66 3.0370 3.0370 NUM ajst-19323 123 67 3.0529 3.0529 NUM ajst-19323 123 68 2.9514 2.9514 NUM ajst-19323 123 69 368 368 NUM ajst-19323 123 70 table table NOUN ajst-19323 123 71 5 5 NUM ajst-19323 123 72 . . PUNCT ajst-19323 123 73 comparison comparison NOUN ajst-19323 123 74 results result VERB ajst-19323 123 75 under under ADP ajst-19323 123 76 10 10 NUM ajst-19323 123 77 % % NOUN ajst-19323 123 78 missing miss VERB ajst-19323 123 79 labels label NOUN ajst-19323 123 80 data datum NOUN ajst-19323 123 81 set set VERB ajst-19323 123 82 pmu pmu PROPN ajst-19323 123 83 mfml mfml PROPN ajst-19323 123 84 mlmlfs mlmlfs PROPN ajst-19323 123 85 lsf lsf PROPN ajst-19323 123 86 - - PROPN ajst-19323 123 87 ci ci PROPN ajst-19323 123 88 ml ml PROPN ajst-19323 123 89 - - PUNCT ajst-19323 123 90 lsf lsf NOUN ajst-19323 123 91 arts art NOUN ajst-19323 123 92 0.6658 0.6658 NUM ajst-19323 123 93 0.6271 0.6271 NUM ajst-19323 123 94 0.6411 0.6411 NUM ajst-19323 123 95 0.6315 0.6315 NUM ajst-19323 123 96 0.6154 0.6154 NUM ajst-19323 123 97 entertain entertain VERB ajst-19323 123 98 0.6119 0.6119 NUM ajst-19323 123 99 0.5903 0.5903 NUM ajst-19323 123 100 0.6123 0.6123 NUM ajst-19323 123 101 0.6124 0.6124 NUM ajst-19323 124 1 0.5189 0.5189 NUM ajst-19323 124 2 emotions emotion NOUN ajst-19323 124 3 0.3772 0.3772 NUM ajst-19323 124 4 0.4298 0.4298 NUM ajst-19323 124 5 0.3967 0.3967 NUM ajst-19323 124 6 0.4128 0.4128 NUM ajst-19323 124 7 0.3121 0.3121 NUM ajst-19323 124 8 reference reference NOUN ajst-19323 124 9 0.5120 0.5120 NUM ajst-19323 124 10 0.5109 0.5109 NUM ajst-19323 124 11 0.5116 0.5116 NUM ajst-19323 124 12 0.5056 0.5056 NUM ajst-19323 124 13 0.4713 0.4713 NUM ajst-19323 124 14 scene scene NOUN ajst-19323 124 15 0.3210 0.3210 NUM ajst-19323 124 16 0.5397 0.5397 NUM ajst-19323 124 17 0.3530 0.3530 NUM ajst-19323 124 18 0.3498 0.3498 NUM ajst-19323 124 19 0.3115 0.3115 NUM ajst-19323 124 20 average average ADJ ajst-19323 124 21 0.4976 0.4976 NUM ajst-19323 124 22 0.5396 0.5396 NUM ajst-19323 124 23 0.5029 0.5029 NUM ajst-19323 124 24 0.5024 0.5024 NUM ajst-19323 124 25 0.4458 0.4458 NUM ajst-19323 124 26 figure figure NOUN ajst-19323 124 27 1 1 NUM ajst-19323 124 28 . . PUNCT ajst-19323 124 29 results result NOUN ajst-19323 124 30 of of ADP ajst-19323 124 31 four four NUM ajst-19323 124 32 classification classification NOUN ajst-19323 124 33 indicators indicator NOUN ajst-19323 124 34 when when SCONJ ajst-19323 124 35 the the DET ajst-19323 124 36 label label NOUN ajst-19323 124 37 missing miss VERB ajst-19323 124 38 rate rate NOUN ajst-19323 124 39 is be AUX ajst-19323 124 40 10 10 NUM ajst-19323 124 41 % % NOUN ajst-19323 124 42 in in ADP ajst-19323 124 43 arts art NOUN ajst-19323 124 44 dataset dataset VERB ajst-19323 124 45 4 4 NUM ajst-19323 124 46 . . PUNCT ajst-19323 125 1 conclusions conclusion NOUN ajst-19323 125 2 the the DET ajst-19323 125 3 missing miss VERB ajst-19323 125 4 label label NOUN ajst-19323 125 5 problem problem NOUN ajst-19323 125 6 in in ADP ajst-19323 125 7 the the DET ajst-19323 125 8 multi multi ADJ ajst-19323 125 9 - - ADJ ajst-19323 125 10 label label ADJ ajst-19323 125 11 feature feature NOUN ajst-19323 125 12 selection selection NOUN ajst-19323 125 13 is be AUX ajst-19323 125 14 analyzed analyze VERB ajst-19323 125 15 , , PUNCT ajst-19323 125 16 and and CCONJ ajst-19323 125 17 a a DET ajst-19323 125 18 multi multi ADJ ajst-19323 125 19 - - ADJ ajst-19323 125 20 label label ADJ ajst-19323 125 21 feature feature NOUN ajst-19323 125 22 selection selection NOUN ajst-19323 125 23 algorithm algorithm NOUN ajst-19323 125 24 ml ml AUX ajst-19323 125 25 - - PUNCT ajst-19323 125 26 lsf lsf PROPN ajst-19323 125 27 is be AUX ajst-19323 125 28 proposed propose VERB ajst-19323 125 29 to to PART ajst-19323 125 30 deal deal VERB ajst-19323 125 31 with with ADP ajst-19323 125 32 the the DET ajst-19323 125 33 missing miss VERB ajst-19323 125 34 label label NOUN ajst-19323 125 35 problem problem NOUN ajst-19323 125 36 . . PUNCT ajst-19323 126 1 based base VERB ajst-19323 126 2 on on ADP ajst-19323 126 3 the the DET ajst-19323 126 4 linear linear PROPN ajst-19323 126 5 relationship relationship NOUN ajst-19323 126 6 between between ADP ajst-19323 126 7 labelspecific labelspecific ADJ ajst-19323 126 8 features feature NOUN ajst-19323 126 9 and and CCONJ ajst-19323 126 10 labels label NOUN ajst-19323 126 11 and and CCONJ ajst-19323 126 12 the the DET ajst-19323 126 13 nonlinear nonlinear ADJ ajst-19323 126 14 relationship relationship NOUN ajst-19323 126 15 between between ADP ajst-19323 126 16 data datum NOUN ajst-19323 126 17 , , PUNCT ajst-19323 126 18 we we PRON ajst-19323 126 19 learn learn VERB ajst-19323 126 20 the the DET ajst-19323 126 21 label label NOUN ajst-19323 126 22 - - PUNCT ajst-19323 126 23 specific specific ADJ ajst-19323 126 24 features feature NOUN ajst-19323 126 25 and and CCONJ ajst-19323 126 26 use use VERB ajst-19323 126 27 the the DET ajst-19323 126 28 linear linear ADJ ajst-19323 126 29 relationship relationship NOUN ajst-19323 126 30 between between ADP ajst-19323 126 31 label label NOUN ajst-19323 126 32 - - PUNCT ajst-19323 126 33 specific specific ADJ ajst-19323 126 34 features feature NOUN ajst-19323 126 35 and and CCONJ ajst-19323 126 36 labels label NOUN ajst-19323 126 37 to to PART ajst-19323 126 38 fill fill VERB ajst-19323 126 39 the the DET ajst-19323 126 40 missing miss VERB ajst-19323 126 41 labels label NOUN ajst-19323 126 42 . . PUNCT ajst-19323 127 1 the the DET ajst-19323 127 2 process process NOUN ajst-19323 127 3 of of ADP ajst-19323 127 4 feature feature NOUN ajst-19323 127 5 selection selection NOUN ajst-19323 127 6 and and CCONJ ajst-19323 127 7 label label NOUN ajst-19323 127 8 fill fill NOUN ajst-19323 127 9 is be AUX ajst-19323 127 10 iterative iterative NOUN ajst-19323 127 11 synchronously synchronously ADV ajst-19323 127 12 . . PUNCT ajst-19323 128 1 the the DET ajst-19323 128 2 final final ADJ ajst-19323 128 3 solution solution NOUN ajst-19323 128 4 is be AUX ajst-19323 128 5 a a DET ajst-19323 128 6 non non ADJ ajst-19323 128 7 - - ADJ ajst-19323 128 8 smooth smooth ADJ ajst-19323 128 9 , , PUNCT ajst-19323 128 10 convex convex ADJ ajst-19323 128 11 optimization optimization NOUN ajst-19323 128 12 problem problem NOUN ajst-19323 128 13 . . PUNCT ajst-19323 129 1 the the DET ajst-19323 129 2 accelerated accelerated ADJ ajst-19323 129 3 near near ADJ ajst-19323 129 4 - - PUNCT ajst-19323 129 5 end end NOUN ajst-19323 129 6 gradient gradient NOUN ajst-19323 129 7 method method NOUN ajst-19323 129 8 is be AUX ajst-19323 129 9 chosen choose VERB ajst-19323 129 10 to to PART ajst-19323 129 11 solve solve VERB ajst-19323 129 12 this this DET ajst-19323 129 13 problem problem NOUN ajst-19323 129 14 . . PUNCT ajst-19323 130 1 when when SCONJ ajst-19323 130 2 comparing compare VERB ajst-19323 130 3 ml ml PROPN ajst-19323 130 4 - - PUNCT ajst-19323 130 5 lsf lsf NOUN ajst-19323 130 6 with with ADP ajst-19323 130 7 other other ADJ ajst-19323 130 8 algorithms algorithm NOUN ajst-19323 130 9 , , PUNCT ajst-19323 130 10 the the DET ajst-19323 130 11 case case NOUN ajst-19323 130 12 of of ADP ajst-19323 130 13 missing miss VERB ajst-19323 130 14 label label NOUN ajst-19323 130 15 rate rate NOUN ajst-19323 130 16 of of ADP ajst-19323 130 17 10 10 NUM ajst-19323 130 18 % % NOUN ajst-19323 130 19 is be AUX ajst-19323 130 20 simulated simulate VERB ajst-19323 130 21 , , PUNCT ajst-19323 130 22 which which PRON ajst-19323 130 23 can can AUX ajst-19323 130 24 effectively effectively ADV ajst-19323 130 25 verify verify VERB ajst-19323 130 26 the the DET ajst-19323 130 27 effectiveness effectiveness NOUN ajst-19323 130 28 of of ADP ajst-19323 130 29 ml ml NOUN ajst-19323 130 30 - - PUNCT ajst-19323 130 31 lsf lsf NOUN ajst-19323 130 32 algorithm algorithm NOUN ajst-19323 130 33 in in ADP ajst-19323 130 34 processing process VERB ajst-19323 130 35 multi multi ADJ ajst-19323 130 36 - - ADJ ajst-19323 130 37 label label ADJ ajst-19323 130 38 data datum NOUN ajst-19323 130 39 with with ADP ajst-19323 130 40 missing miss VERB ajst-19323 130 41 label label NOUN ajst-19323 130 42 . . PUNCT ajst-19323 131 1 experimental experimental ADJ ajst-19323 131 2 results result NOUN ajst-19323 131 3 show show VERB ajst-19323 131 4 that that SCONJ ajst-19323 131 5 the the DET ajst-19323 131 6 proposed propose VERB ajst-19323 131 7 algorithm algorithm NOUN ajst-19323 131 8 has have VERB ajst-19323 131 9 good good ADJ ajst-19323 131 10 performance performance NOUN ajst-19323 131 11 and and CCONJ ajst-19323 131 12 is be AUX ajst-19323 131 13 an an DET ajst-19323 131 14 effective effective ADJ ajst-19323 131 15 multi multi ADJ ajst-19323 131 16 - - ADJ ajst-19323 131 17 label label ADJ ajst-19323 131 18 feature feature NOUN ajst-19323 131 19 selection selection NOUN ajst-19323 131 20 algorithm algorithm NOUN ajst-19323 131 21 . . PUNCT ajst-19323 132 1 references reference NOUN ajst-19323 132 2 [ [ X ajst-19323 132 3 1 1 NUM ajst-19323 132 4 ] ] PUNCT ajst-19323 132 5 lin lin PROPN ajst-19323 132 6 y y PROPN ajst-19323 132 7 j j PROPN ajst-19323 132 8 , , PUNCT ajst-19323 132 9 hu hu PROPN ajst-19323 133 1 q q PROPN ajst-19323 133 2 h h PROPN ajst-19323 133 3 , , PUNCT ajst-19323 133 4 liu liu PROPN ajst-19323 133 5 j j PROPN ajst-19323 133 6 h h PROPN ajst-19323 133 7 , , PUNCT ajst-19323 133 8 et et PROPN ajst-19323 133 9 al al PROPN ajst-19323 133 10 . . PUNCT ajst-19323 134 1 streaming stream VERB ajst-19323 134 2 feature feature NOUN ajst-19323 134 3 selection selection NOUN ajst-19323 134 4 for for ADP ajst-19323 134 5 multilabel multilabel NOUN ajst-19323 134 6 learning learn VERB ajst-19323 134 7 based base VERB ajst-19323 134 8 on on ADP ajst-19323 134 9 fuzzy fuzzy ADJ ajst-19323 134 10 mutual mutual ADJ ajst-19323 134 11 information information NOUN ajst-19323 135 1 [ [ X ajst-19323 135 2 j j X ajst-19323 135 3 ] ] X ajst-19323 135 4 . . PUNCT ajst-19323 136 1 ieee ieee NOUN ajst-19323 136 2 transactions transaction NOUN ajst-19323 136 3 on on ADP ajst-19323 136 4 fuzzy fuzzy ADJ ajst-19323 136 5 systems system NOUN ajst-19323 136 6 , , PUNCT ajst-19323 136 7 2017 2017 NUM ajst-19323 136 8 , , PUNCT ajst-19323 136 9 25(6 25(6 NUM ajst-19323 136 10 ): ): PUNCT ajst-19323 136 11 1491 1491 NUM ajst-19323 136 12 - - SYM ajst-19323 136 13 1507 1507 NUM ajst-19323 136 14 . . PUNCT ajst-19323 137 1 [ [ X ajst-19323 137 2 2 2 X ajst-19323 137 3 ] ] PUNCT ajst-19323 137 4 wang wang PROPN ajst-19323 137 5 c c PROPN ajst-19323 137 6 x x PROPN ajst-19323 137 7 , , PUNCT ajst-19323 137 8 lin lin PROPN ajst-19323 137 9 y y PROPN ajst-19323 137 10 j j PROPN ajst-19323 137 11 , , PUNCT ajst-19323 137 12 tang tang PROPN ajst-19323 137 13 l，et l，et PUNCT ajst-19323 137 14 al al PROPN ajst-19323 137 15 . . PUNCT ajst-19323 137 16 multi multi ADJ ajst-19323 137 17 - - ADJ ajst-19323 137 18 label label ADJ ajst-19323 137 19 feature feature NOUN ajst-19323 137 20 selection selection NOUN ajst-19323 137 21 based base VERB ajst-19323 137 22 on on ADP ajst-19323 137 23 information information NOUN ajst-19323 137 24 granulation[j granulation[j PROPN ajst-19323 137 25 ] ] PUNCT ajst-19323 137 26 . . PUNCT ajst-19323 137 27 pattern pattern NOUN ajst-19323 137 28 recognition recognition NOUN ajst-19323 137 29 and and CCONJ ajst-19323 137 30 artificial artificial ADJ ajst-19323 137 31 intelligence intelligence NOUN ajst-19323 137 32 , , PUNCT ajst-19323 137 33 2018 2018 NUM ajst-19323 137 34 , , PUNCT ajst-19323 137 35 31(2 31(2 NUM ajst-19323 137 36 ): ): PUNCT ajst-19323 137 37 123 123 NUM ajst-19323 137 38 - - SYM ajst-19323 137 39 131 131 NUM ajst-19323 137 40 . . PUNCT ajst-19323 138 1 [ [ X ajst-19323 138 2 3 3 X ajst-19323 138 3 ] ] X ajst-19323 138 4 liu liu PROPN ajst-19323 138 5 j j PROPN ajst-19323 138 6 h h PROPN ajst-19323 138 7 , , PUNCT ajst-19323 138 8 li li PROPN ajst-19323 138 9 y y PROPN ajst-19323 138 10 w w PROPN ajst-19323 138 11 , , PUNCT ajst-19323 138 12 weng weng PROPN ajst-19323 138 13 w w PROPN ajst-19323 138 14 , , PUNCT ajst-19323 138 15 et et PROPN ajst-19323 138 16 al al PROPN ajst-19323 138 17 . . PROPN ajst-19323 138 18 feature feature NOUN ajst-19323 138 19 selection selection NOUN ajst-19323 138 20 for for ADP ajst-19323 138 21 multilabel multilabel NOUN ajst-19323 138 22 learning learn VERB ajst-19323 138 23 with with ADP ajst-19323 138 24 streaming streaming NOUN ajst-19323 138 25 label label NOUN ajst-19323 138 26 [ [ X ajst-19323 138 27 j j X ajst-19323 138 28 ] ] X ajst-19323 138 29 . . PUNCT ajst-19323 139 1 neurocomputing neurocomputing NOUN ajst-19323 139 2 , , PUNCT ajst-19323 139 3 2020 2020 NUM ajst-19323 139 4 , , PUNCT ajst-19323 139 5 387 387 NUM ajst-19323 139 6 : : SYM ajst-19323 139 7 268 268 NUM ajst-19323 139 8 - - SYM ajst-19323 139 9 278 278 NUM ajst-19323 139 10 . . PUNCT ajst-19323 140 1 [ [ X ajst-19323 140 2 4 4 X ajst-19323 140 3 ] ] X ajst-19323 140 4 zhang zhang PROPN ajst-19323 140 5 m m PROPN ajst-19323 140 6 l l NOUN ajst-19323 140 7 , , PUNCT ajst-19323 140 8 zhou zhou PROPN ajst-19323 140 9 z z PROPN ajst-19323 140 10 h. h. PROPN ajst-19323 140 11 a a DET ajst-19323 140 12 review review NOUN ajst-19323 140 13 on on ADP ajst-19323 140 14 multi multi ADJ ajst-19323 140 15 - - ADJ ajst-19323 140 16 label label ADJ ajst-19323 140 17 learning learn VERB ajst-19323 140 18 algorithms algorithm NOUN ajst-19323 140 19 [ [ X ajst-19323 140 20 j j X ajst-19323 140 21 ] ] X ajst-19323 140 22 . . PUNCT ajst-19323 141 1 ieee ieee NOUN ajst-19323 141 2 transactions transaction NOUN ajst-19323 141 3 on on ADP ajst-19323 141 4 knowledge knowledge NOUN ajst-19323 141 5 and and CCONJ ajst-19323 141 6 data datum NOUN ajst-19323 141 7 engineering engineering NOUN ajst-19323 141 8 , , PUNCT ajst-19323 141 9 2014 2014 NUM ajst-19323 141 10 , , PUNCT ajst-19323 141 11 26(8 26(8 NUM ajst-19323 141 12 ): ): PUNCT ajst-19323 141 13 1819 1819 NUM ajst-19323 141 14 - - SYM ajst-19323 141 15 1837 1837 NUM ajst-19323 141 16 . . PUNCT ajst-19323 142 1 [ [ X ajst-19323 142 2 5 5 NUM ajst-19323 142 3 ] ] X ajst-19323 142 4 shaikh shaikh X ajst-19323 142 5 r r PROPN ajst-19323 142 6 , , PUNCT ajst-19323 142 7 rafi rafi PROPN ajst-19323 142 8 m m PROPN ajst-19323 142 9 , , PUNCT ajst-19323 142 10 mahoto mahoto NOUN ajst-19323 142 11 na na PROPN ajst-19323 142 12 , , PUNCT ajst-19323 142 13 sulaiman sulaiman PROPN ajst-19323 142 14 a a PRON ajst-19323 142 15 , , PUNCT ajst-19323 142 16 shaikh shaikh PROPN ajst-19323 142 17 a. a. NOUN ajst-19323 142 18 a a DET ajst-19323 142 19 filterbased filterbase VERB ajst-19323 142 20 feature feature NOUN ajst-19323 142 21 selection selection NOUN ajst-19323 142 22 approach approach NOUN ajst-19323 142 23 in in ADP ajst-19323 142 24 multilabel multilabel NOUN ajst-19323 142 25 classification classification NOUN ajst-19323 142 26 [ [ X ajst-19323 142 27 j j X ajst-19323 142 28 ] ] X ajst-19323 142 29 . . PUNCT ajst-19323 143 1 machine machine NOUN ajst-19323 143 2 learning learning NOUN ajst-19323 143 3 - - PUNCT ajst-19323 143 4 science science NOUN ajst-19323 143 5 and and CCONJ ajst-19323 143 6 technology technology NOUN ajst-19323 143 7 , , PUNCT ajst-19323 143 8 2023 2023 NUM ajst-19323 143 9 , , PUNCT ajst-19323 143 10 4(4 4(4 NUM ajst-19323 143 11 ) ) PUNCT ajst-19323 143 12 . . PUNCT ajst-19323 144 1 [ [ X ajst-19323 144 2 6 6 NUM ajst-19323 144 3 ] ] PUNCT ajst-19323 144 4 qian qian PROPN ajst-19323 144 5 wb wb PROPN ajst-19323 144 6 , , PUNCT ajst-19323 144 7 xu xu PROPN ajst-19323 144 8 fk fk PROPN ajst-19323 144 9 , , PUNCT ajst-19323 144 10 qian qian PROPN ajst-19323 144 11 j j PROPN ajst-19323 144 12 , , PUNCT ajst-19323 144 13 shu shu PROPN ajst-19323 144 14 wh wh PROPN ajst-19323 144 15 , , PUNCT ajst-19323 144 16 ding de VERB ajst-19323 144 17 wp wp PROPN ajst-19323 144 18 . . PUNCT ajst-19323 145 1 multi multi ADJ ajst-19323 145 2 - - ADJ ajst-19323 145 3 label label ADJ ajst-19323 145 4 feature feature NOUN ajst-19323 145 5 selection selection NOUN ajst-19323 145 6 based base VERB ajst-19323 145 7 on on ADP ajst-19323 145 8 rough rough ADJ ajst-19323 145 9 granular granular ADJ ajst-19323 145 10 - - PUNCT ajst-19323 145 11 ball ball NOUN ajst-19323 145 12 and and CCONJ ajst-19323 145 13 label label NOUN ajst-19323 145 14 distribution distribution NOUN ajst-19323 146 1 [ [ X ajst-19323 146 2 j j X ajst-19323 146 3 ] ] X ajst-19323 146 4 . . PUNCT ajst-19323 147 1 information information NOUN ajst-19323 147 2 sciences sciences PROPN ajst-19323 147 3 , , PUNCT ajst-19323 147 4 2023 2023 NUM ajst-19323 147 5 , , PUNCT ajst-19323 147 6 650 650 NUM ajst-19323 147 7 . . PUNCT ajst-19323 148 1 [ [ X ajst-19323 148 2 7 7 X ajst-19323 148 3 ] ] X ajst-19323 148 4 ma ma PROPN ajst-19323 148 5 xa xa PROPN ajst-19323 148 6 , , PUNCT ajst-19323 148 7 jiang jiang PROPN ajst-19323 148 8 wt wt PROPN ajst-19323 148 9 , , PUNCT ajst-19323 148 10 ling ling PROPN ajst-19323 148 11 y y PROPN ajst-19323 148 12 , , PUNCT ajst-19323 148 13 yang yang PROPN ajst-19323 148 14 bl bl PROPN ajst-19323 148 15 . . PUNCT ajst-19323 149 1 multi multi ADJ ajst-19323 149 2 - - ADJ ajst-19323 149 3 label label ADJ ajst-19323 149 4 feature feature NOUN ajst-19323 149 5 selection selection NOUN ajst-19323 149 6 via via ADP ajst-19323 149 7 maximum maximum ADJ ajst-19323 149 8 dynamic dynamic ADJ ajst-19323 149 9 correlation correlation NOUN ajst-19323 149 10 change change NOUN ajst-19323 149 11 and and CCONJ ajst-19323 149 12 minimum minimum NOUN ajst-19323 149 13 label label NOUN ajst-19323 149 14 redundancy redundancy NOUN ajst-19323 150 1 [ [ X ajst-19323 150 2 j j X ajst-19323 150 3 ] ] X ajst-19323 150 4 . . PUNCT ajst-19323 151 1 artificial artificial ADJ ajst-19323 151 2 intelligence intelligence NOUN ajst-19323 151 3 review review NOUN ajst-19323 151 4 , , PUNCT ajst-19323 151 5 2023 2023 NUM ajst-19323 151 6 . . PUNCT ajst-19323 152 1 [ [ X ajst-19323 152 2 8 8 NUM ajst-19323 152 3 ] ] X ajst-19323 152 4 li li PROPN ajst-19323 152 5 yuchen yuchen PROPN ajst-19323 152 6 , , PUNCT ajst-19323 152 7 wei wei PROPN ajst-19323 152 8 wei wei PROPN ajst-19323 152 9 , , PUNCT ajst-19323 152 10 bai bai PROPN ajst-19323 152 11 weiming weiming NOUN ajst-19323 152 12 , , PUNCT ajst-19323 152 13 et et PROPN ajst-19323 152 14 al al PROPN ajst-19323 152 15 . . PUNCT ajst-19323 152 16 multi multi ADJ ajst-19323 152 17 - - ADJ ajst-19323 152 18 label label ADJ ajst-19323 152 19 feature feature NOUN ajst-19323 152 20 selection selection NOUN ajst-19323 152 21 based base VERB ajst-19323 152 22 on on ADP ajst-19323 152 23 label label NOUN ajst-19323 152 24 co co NOUN ajst-19323 152 25 - - NOUN ajst-19323 152 26 occurrence occurrence ADJ ajst-19323 152 27 relationship relationship NOUN ajst-19323 153 1 [ [ X ajst-19323 153 2 j j X ajst-19323 153 3 ] ] X ajst-19323 153 4 . . PUNCT ajst-19323 154 1 computer computer NOUN ajst-19323 154 2 engineering engineering NOUN ajst-19323 154 3 and and CCONJ ajst-19323 154 4 science science NOUN ajst-19323 154 5 , , PUNCT ajst-19323 154 6 2021 2021 NUM ajst-19323 154 7 , , PUNCT ajst-19323 154 8 43(11 43(11 NUM ajst-19323 154 9 ): ): PUNCT ajst-19323 154 10 2049 2049 NUM ajst-19323 154 11 - - SYM ajst-19323 154 12 2055 2055 NUM ajst-19323 154 13 . . PUNCT ajst-19323 155 1 [ [ X ajst-19323 155 2 9 9 NUM ajst-19323 155 3 ] ] X ajst-19323 155 4 weng weng PROPN ajst-19323 155 5 wei wei PROPN ajst-19323 155 6 , , PUNCT ajst-19323 155 7 wei wei PROPN ajst-19323 155 8 bowen bowen PROPN ajst-19323 155 9 , , PUNCT ajst-19323 155 10 ke ke PROPN ajst-19323 155 11 wen wen PROPN ajst-19323 155 12 , , PUNCT ajst-19323 155 13 fan fan PROPN ajst-19323 155 14 yuling yuling PROPN ajst-19323 155 15 , , PUNCT ajst-19323 155 16 wang wang PROPN ajst-19323 155 17 jinbo jinbo PROPN ajst-19323 155 18 , , PUNCT ajst-19323 155 19 li li PROPN ajst-19323 155 20 yuwen yuwen PROPN ajst-19323 155 21 . . PUNCT ajst-19323 156 1 learning learn VERB ajst-19323 156 2 label label NOUN ajst-19323 156 3 - - PUNCT ajst-19323 156 4 specific specific ADJ ajst-19323 156 5 features feature NOUN ajst-19323 156 6 with with ADP ajst-19323 156 7 global global ADJ ajst-19323 156 8 and and CCONJ ajst-19323 156 9 local local ADJ ajst-19323 156 10 369 369 NUM ajst-19323 156 11 label label NOUN ajst-19323 156 12 correlation correlation NOUN ajst-19323 156 13 for for ADP ajst-19323 156 14 multi multi ADJ ajst-19323 156 15 - - ADJ ajst-19323 156 16 label label ADJ ajst-19323 156 17 classification[j classification[j NOUN ajst-19323 156 18 ] ] PUNCT ajst-19323 156 19 . . PUNCT ajst-19323 157 1 applied apply VERB ajst-19323 157 2 intelligence intelligence NOUN ajst-19323 157 3 , , PUNCT ajst-19323 157 4 2023 2023 NUM ajst-19323 157 5 , , PUNCT ajst-19323 157 6 53(3 53(3 NUM ajst-19323 157 7 ): ): PUNCT ajst-19323 157 8 3017 3017 NUM ajst-19323 157 9 - - SYM ajst-19323 157 10 3033 3033 NUM ajst-19323 157 11 . . PUNCT ajst-19323 158 1 [ [ X ajst-19323 158 2 10 10 NUM ajst-19323 158 3 ] ] X ajst-19323 158 4 li li PROPN ajst-19323 158 5 yonghao yonghao PROPN ajst-19323 158 6 , , PUNCT ajst-19323 158 7 hu hu PROPN ajst-19323 158 8 liang liang PROPN ajst-19323 158 9 , , PUNCT ajst-19323 158 10 gao gao PROPN ajst-19323 158 11 wanfu wanfu PROPN ajst-19323 158 12 . . PUNCT ajst-19323 159 1 label label NOUN ajst-19323 159 2 correlations correlation NOUN ajst-19323 159 3 variation variation NOUN ajst-19323 159 4 for for ADP ajst-19323 159 5 robust robust ADJ ajst-19323 159 6 multi multi ADJ ajst-19323 159 7 - - ADJ ajst-19323 159 8 label label ADJ ajst-19323 159 9 feature feature NOUN ajst-19323 159 10 selection[j selection[j NOUN ajst-19323 159 11 ] ] PUNCT ajst-19323 159 12 . . PUNCT ajst-19323 160 1 information information NOUN ajst-19323 160 2 sciences sciences PROPN ajst-19323 160 3 , , PUNCT ajst-19323 160 4 2022 2022 NUM ajst-19323 160 5 , , PUNCT ajst-19323 160 6 609 609 NUM ajst-19323 160 7 : : PUNCT ajst-19323 160 8 1075 1075 NUM ajst-19323 160 9 - - SYM ajst-19323 160 10 1097 1097 NUM ajst-19323 160 11 . . PUNCT ajst-19323 161 1 [ [ X ajst-19323 161 2 11 11 NUM ajst-19323 161 3 ] ] PUNCT ajst-19323 161 4 wang wang PROPN ajst-19323 161 5 c c PROPN ajst-19323 161 6 x x PROPN ajst-19323 161 7 , , PUNCT ajst-19323 161 8 lin lin PROPN ajst-19323 161 9 y y PROPN ajst-19323 161 10 j j PROPN ajst-19323 161 11 , , PUNCT ajst-19323 161 12 liu liu PROPN ajst-19323 161 13 j j PROPN ajst-19323 161 14 h. h. PROPN ajst-19323 161 15 feature feature PROPN ajst-19323 161 16 selection selection NOUN ajst-19323 161 17 for for ADP ajst-19323 161 18 multilabel multilabel NOUN ajst-19323 161 19 learning learn VERB ajst-19323 161 20 with with ADP ajst-19323 161 21 missing miss VERB ajst-19323 161 22 labels[j labels[j NOUN ajst-19323 161 23 ] ] PUNCT ajst-19323 161 24 . . PUNCT ajst-19323 161 25 applied apply VERB ajst-19323 161 26 intelligence intelligence NOUN ajst-19323 161 27 , , PUNCT ajst-19323 161 28 2019 2019 NUM ajst-19323 161 29 , , PUNCT ajst-19323 161 30 49 49 NUM ajst-19323 161 31 ( ( PUNCT ajst-19323 161 32 8) 8) NUM ajst-19323 161 33 : : PUNCT ajst-19323 161 34 3027 3027 NUM ajst-19323 161 35 - - SYM ajst-19323 161 36 3042 3042 NUM ajst-19323 161 37 . . PUNCT ajst-19323 162 1 [ [ X ajst-19323 162 2 12 12 NUM ajst-19323 162 3 ] ] X ajst-19323 162 4 zhu zhu NOUN ajst-19323 163 1 p p PROPN ajst-19323 163 2 f f PROPN ajst-19323 163 3 , , PUNCT ajst-19323 163 4 xu xu PROPN ajst-19323 164 1 q q X ajst-19323 164 2 , , PUNCT ajst-19323 164 3 hu hu PROPN ajst-19323 164 4 q q PROPN ajst-19323 164 5 h h PROPN ajst-19323 164 6 , , PUNCT ajst-19323 164 7 et et PROPN ajst-19323 164 8 al al PROPN ajst-19323 164 9 . . PUNCT ajst-19323 165 1 multi multi ADJ ajst-19323 165 2 - - ADJ ajst-19323 165 3 label label ADJ ajst-19323 165 4 feature feature NOUN ajst-19323 165 5 selection selection NOUN ajst-19323 165 6 with with ADP ajst-19323 165 7 missing miss VERB ajst-19323 165 8 labels label NOUN ajst-19323 166 1 [ [ X ajst-19323 166 2 j j X ajst-19323 166 3 ] ] X ajst-19323 166 4 . . PUNCT ajst-19323 167 1 pattern pattern NOUN ajst-19323 167 2 recognition recognition NOUN ajst-19323 167 3 , , PUNCT ajst-19323 167 4 2018 2018 NUM ajst-19323 167 5 , , PUNCT ajst-19323 167 6 74 74 NUM ajst-19323 167 7 : : SYM ajst-19323 167 8 488502 488502 NUM ajst-19323 167 9 . . PUNCT ajst-19323 168 1 [ [ X ajst-19323 168 2 13 13 NUM ajst-19323 168 3 ] ] X ajst-19323 168 4 zhi zhi PROPN ajst-19323 168 5 - - PUNCT ajst-19323 168 6 fen fen PROPN ajst-19323 168 7 he he PROPN ajst-19323 168 8 , , PUNCT ajst-19323 168 9 ming ming PROPN ajst-19323 168 10 yang yang PROPN ajst-19323 168 11 , , PUNCT ajst-19323 168 12 yang yang PROPN ajst-19323 168 13 gao gao PROPN ajst-19323 168 14 , , PUNCT ajst-19323 168 15 hui hui PROPN ajst-19323 168 16 - - PUNCT ajst-19323 168 17 dong dong PROPN ajst-19323 168 18 liu liu PROPN ajst-19323 168 19 , , PUNCT ajst-19323 168 20 yilong yilong PROPN ajst-19323 168 21 yin yin PROPN ajst-19323 168 22 . . PUNCT ajst-19323 169 1 joint joint ADJ ajst-19323 169 2 multi multi ADJ ajst-19323 169 3 - - ADJ ajst-19323 169 4 label label ADJ ajst-19323 169 5 classification classification NOUN ajst-19323 169 6 and and CCONJ ajst-19323 169 7 label label NOUN ajst-19323 169 8 correlations correlation NOUN ajst-19323 169 9 with with ADP ajst-19323 169 10 missing miss VERB ajst-19323 169 11 labels label NOUN ajst-19323 169 12 and and CCONJ ajst-19323 169 13 feature feature NOUN ajst-19323 169 14 selection[j selection[j NOUN ajst-19323 169 15 ] ] PUNCT ajst-19323 169 16 . . PUNCT ajst-19323 170 1 knowledge knowledge NOUN ajst-19323 170 2 - - PUNCT ajst-19323 170 3 based base VERB ajst-19323 170 4 systems system NOUN ajst-19323 170 5 , , PUNCT ajst-19323 170 6 2019 2019 NUM ajst-19323 170 7 , , PUNCT ajst-19323 170 8 163 163 NUM ajst-19323 170 9 : : SYM ajst-19323 170 10 145 145 NUM ajst-19323 170 11 - - SYM ajst-19323 170 12 158 158 NUM ajst-19323 170 13 . . PUNCT ajst-19323 171 1 [ [ X ajst-19323 171 2 14 14 NUM ajst-19323 171 3 ] ] X ajst-19323 171 4 zhao zhao PROPN ajst-19323 171 5 dw dw PROPN ajst-19323 171 6 , , PUNCT ajst-19323 171 7 tan tan PROPN ajst-19323 171 8 y y PROPN ajst-19323 171 9 , , PUNCT ajst-19323 171 10 sun sun PROPN ajst-19323 171 11 d d PROPN ajst-19323 171 12 , , PUNCT ajst-19323 171 13 gao gao PROPN ajst-19323 171 14 qw qw PROPN ajst-19323 171 15 , , PUNCT ajst-19323 171 16 lu lu PROPN ajst-19323 171 17 yx yx PROPN ajst-19323 171 18 , , PUNCT ajst-19323 171 19 zhu zhu PROPN ajst-19323 171 20 d. d. PROPN ajst-19323 171 21 multi multi PROPN ajst-19323 171 22 - - ADJ ajst-19323 171 23 label label ADJ ajst-19323 171 24 learning learning NOUN ajst-19323 171 25 of of ADP ajst-19323 171 26 missing miss VERB ajst-19323 171 27 labels label NOUN ajst-19323 171 28 using use VERB ajst-19323 171 29 label label NOUN ajst-19323 171 30 - - PUNCT ajst-19323 171 31 specific specific ADJ ajst-19323 171 32 features feature NOUN ajst-19323 171 33 : : PUNCT ajst-19323 171 34 an an DET ajst-19323 171 35 embedded embed VERB ajst-19323 171 36 packaging packaging NOUN ajst-19323 171 37 method method NOUN ajst-19323 171 38 [ [ X ajst-19323 171 39 j j X ajst-19323 171 40 ] ] X ajst-19323 171 41 . . PUNCT ajst-19323 172 1 applied apply VERB ajst-19323 172 2 intelligence intelligence NOUN ajst-19323 172 3 , , PUNCT ajst-19323 172 4 2023 2023 NUM ajst-19323 172 5 . . PUNCT ajst-19323 173 1 [ [ X ajst-19323 173 2 15 15 NUM ajst-19323 173 3 ] ] X ajst-19323 173 4 han han PROPN ajst-19323 173 5 h h PROPN ajst-19323 173 6 r r PROPN ajst-19323 173 7 , , PUNCT ajst-19323 173 8 huang huang PROPN ajst-19323 173 9 m m PROPN ajst-19323 173 10 x x PROPN ajst-19323 173 11 , , PUNCT ajst-19323 173 12 zhang zhang PROPN ajst-19323 173 13 y y PROPN ajst-19323 173 14 , , PUNCT ajst-19323 173 15 et et PROPN ajst-19323 173 16 al al PROPN ajst-19323 173 17 . . PUNCT ajst-19323 173 18 multi multi ADJ ajst-19323 173 19 - - ADJ ajst-19323 173 20 label label ADJ ajst-19323 173 21 learning learning NOUN ajst-19323 173 22 with with ADP ajst-19323 173 23 label label NOUN ajst-19323 173 24 specific specific ADJ ajst-19323 173 25 features feature NOUN ajst-19323 173 26 using use VERB ajst-19323 173 27 correlation correlation NOUN ajst-19323 173 28 information information NOUN ajst-19323 174 1 [ [ X ajst-19323 174 2 j j X ajst-19323 174 3 ] ] X ajst-19323 174 4 . . PUNCT ajst-19323 175 1 ieee ieee NOUN ajst-19323 175 2 access access NOUN ajst-19323 175 3 , , PUNCT ajst-19323 175 4 2019 2019 NUM ajst-19323 175 5 , , PUNCT ajst-19323 175 6 7 7 NUM ajst-19323 175 7 : : SYM ajst-19323 175 8 11474 11474 NUM ajst-19323 175 9 - - SYM ajst-19323 175 10 11484 11484 NUM ajst-19323 175 11 . . PUNCT ajst-19323 176 1 [ [ X ajst-19323 176 2 16 16 NUM ajst-19323 176 3 ] ] X ajst-19323 176 4 zhang zhang PROPN ajst-19323 176 5 zhihao zhihao PROPN ajst-19323 176 6 , , PUNCT ajst-19323 176 7 lin lin PROPN ajst-19323 176 8 yaojin yaojin PROPN ajst-19323 176 9 , , PUNCT ajst-19323 176 10 lu lu PROPN ajst-19323 176 11 shun shun PROPN ajst-19323 176 12 , , PUNCT ajst-19323 176 13 et et PROPN ajst-19323 176 14 al al PROPN ajst-19323 176 15 . . PUNCT ajst-19323 176 16 multi multi ADJ ajst-19323 176 17 - - ADJ ajst-19323 176 18 label label ADJ ajst-19323 176 19 feature feature NOUN ajst-19323 176 20 selection selection NOUN ajst-19323 176 21 based base VERB ajst-19323 176 22 on on ADP ajst-19323 176 23 label label NOUN ajst-19323 176 24 - - PUNCT ajst-19323 176 25 specific specific ADJ ajst-19323 176 26 features feature NOUN ajst-19323 176 27 under under ADP ajst-19323 176 28 missing miss VERB ajst-19323 176 29 labels label NOUN ajst-19323 177 1 [ [ X ajst-19323 177 2 j j X ajst-19323 177 3 ] ] X ajst-19323 177 4 . . PUNCT ajst-19323 178 1 computer computer NOUN ajst-19323 178 2 applications application NOUN ajst-19323 178 3 , , PUNCT ajst-19323 178 4 2021 2021 NUM ajst-19323 178 5 , , PUNCT ajst-19323 178 6 41(10 41(10 NUM ajst-19323 178 7 ): ): PUNCT ajst-19323 178 8 2849 2849 NUM ajst-19323 178 9 - - SYM ajst-19323 178 10 2857 2857 NUM ajst-19323 178 11 . . PUNCT ajst-19323 179 1 [ [ X ajst-19323 179 2 17 17 NUM ajst-19323 179 3 ] ] X ajst-19323 179 4 lin lin PROPN ajst-19323 179 5 z z PROPN ajst-19323 179 6 c c PROPN ajst-19323 179 7 , , PUNCT ajst-19323 179 8 ganesh ganesh NOUN ajst-19323 179 9 a a PROPN ajst-19323 179 10 , , PUNCT ajst-19323 179 11 wright wright PROPN ajst-19323 179 12 j j PROPN ajst-19323 179 13 , , PUNCT ajst-19323 179 14 et et PROPN ajst-19323 179 15 al al PROPN ajst-19323 179 16 . . PUNCT ajst-19323 180 1 fast fast ADJ ajst-19323 180 2 convex convex PROPN ajst-19323 180 3 optimization optimization NOUN ajst-19323 180 4 algorithms algorithm NOUN ajst-19323 180 5 for for ADP ajst-19323 180 6 exact exact ADJ ajst-19323 180 7 recovery recovery NOUN ajst-19323 180 8 of of ADP ajst-19323 180 9 a a DET ajst-19323 180 10 corrupted corrupted ADJ ajst-19323 180 11 low low ADJ ajst-19323 180 12 - - PUNCT ajst-19323 180 13 rank rank NOUN ajst-19323 180 14 matrix matrix NOUN ajst-19323 180 15 [ [ X ajst-19323 180 16 c/ c/ NOUN ajst-19323 180 17 ol ol PROPN ajst-19323 180 18 ] ] PUNCT ajst-19323 180 19 , , PUNCT ajst-19323 180 20 2020 2020 NUM ajst-19323 180 21 . . PUNCT