id sid tid token lemma pos bracis-33557 1 1 an an DET bracis-33557 1 2 instance instance NOUN bracis-33557 1 3 level level NOUN bracis-33557 1 4 analysis analysis NOUN bracis-33557 1 5 of of ADP bracis-33557 1 6     SPACE bracis-33557 1 7 classification classification NOUN bracis-33557 1 8 difficulty difficulty NOUN bracis-33557 1 9 for for ADP bracis-33557 1 10     SPACE bracis-33557 1 11 unlabeled unlabele VERB bracis-33557 1 12 data datum NOUN bracis-33557 1 13 | | ADP bracis-33557 1 14 springer springer NOUN bracis-33557 1 15 nature nature NOUN bracis-33557 1 16 link link PROPN bracis-33557 1 17 ( ( PUNCT bracis-33557 1 18 formerly formerly ADV bracis-33557 1 19 springerlink springerlink NOUN bracis-33557 1 20 ) ) PUNCT bracis-33557 1 21 skip skip VERB bracis-33557 1 22 to to ADP bracis-33557 1 23 main main ADJ bracis-33557 1 24 content content NOUN bracis-33557 1 25 advertisement advertisement NOUN bracis-33557 1 26 log log NOUN bracis-33557 1 27 in in ADP bracis-33557 1 28 menu menu NOUN bracis-33557 1 29 find find VERB bracis-33557 1 30 a a DET bracis-33557 1 31 journal journal NOUN bracis-33557 1 32 publish publish VERB bracis-33557 1 33 with with ADP bracis-33557 1 34 us we PRON bracis-33557 1 35 track track VERB bracis-33557 1 36 your your PRON bracis-33557 1 37 research research NOUN bracis-33557 1 38 search search NOUN bracis-33557 1 39 cart cart NOUN bracis-33557 1 40 home home NOUN bracis-33557 1 41 intelligent intelligent ADJ bracis-33557 1 42 systems system NOUN bracis-33557 1 43 conference conference NOUN bracis-33557 1 44 paper paper NOUN bracis-33557 1 45 an an DET bracis-33557 1 46 instance instance NOUN bracis-33557 1 47 level level NOUN bracis-33557 1 48 analysis analysis NOUN bracis-33557 1 49 of of ADP bracis-33557 1 50     SPACE bracis-33557 1 51 classification classification NOUN bracis-33557 1 52 difficulty difficulty NOUN bracis-33557 1 53 for for ADP bracis-33557 1 54     SPACE bracis-33557 1 55 unlabeled unlabele VERB bracis-33557 1 56 data data PROPN bracis-33557 1 57 conference conference PROPN bracis-33557 1 58 paper paper NOUN bracis-33557 1 59 first first ADV bracis-33557 1 60 online online ADV bracis-33557 1 61 : : PUNCT bracis-33557 1 62 30 30 NUM bracis-33557 1 63 january january NOUN bracis-33557 1 64 2025 2025 NUM bracis-33557 2 1 pp pp ADP bracis-33557 2 2 141–155 141–155 NUM bracis-33557 2 3 cite cite VERB bracis-33557 2 4 this this DET bracis-33557 2 5 conference conference NOUN bracis-33557 2 6 paper paper NOUN bracis-33557 2 7 access access NOUN bracis-33557 2 8 provided provide VERB bracis-33557 2 9 by by ADP bracis-33557 2 10 university university PROPN bracis-33557 2 11 of of ADP bracis-33557 2 12 notre notre PROPN bracis-33557 2 13 dame dame PROPN bracis-33557 2 14 hesburgh hesburgh PROPN bracis-33557 2 15 library library PROPN bracis-33557 2 16 download download PROPN bracis-33557 2 17 book book NOUN bracis-33557 2 18 pdf pdf PROPN bracis-33557 2 19 download download NOUN bracis-33557 2 20 book book NOUN bracis-33557 2 21 epub epub PROPN bracis-33557 2 22 intelligent intelligent ADJ bracis-33557 2 23 systems system NOUN bracis-33557 2 24 ( ( PUNCT bracis-33557 2 25 bracis bracis NOUN bracis-33557 2 26 2024 2024 NUM bracis-33557 2 27 ) ) PUNCT bracis-33557 2 28 an an DET bracis-33557 2 29 instance instance NOUN bracis-33557 2 30 level level NOUN bracis-33557 2 31 analysis analysis NOUN bracis-33557 2 32 of of ADP bracis-33557 2 33     SPACE bracis-33557 2 34 classification classification NOUN bracis-33557 2 35 difficulty difficulty NOUN bracis-33557 2 36 for for ADP bracis-33557 2 37     SPACE bracis-33557 2 38 unlabeled unlabele VERB bracis-33557 2 39 data datum NOUN bracis-33557 2 40 download download NOUN bracis-33557 2 41 book book NOUN bracis-33557 2 42 pdf pdf PROPN bracis-33557 2 43 download download NOUN bracis-33557 2 44 book book PROPN bracis-33557 2 45 epub epub PROPN bracis-33557 2 46 patricia patricia PROPN bracis-33557 2 47 s. s. PROPN bracis-33557 2 48 m. m. PROPN bracis-33557 2 49 ueda ueda PROPN bracis-33557 2 50     SPACE bracis-33557 2 51 orcid orcid PROPN bracis-33557 2 52 : : PUNCT bracis-33557 2 53 orcid.org/0009-0003-2200-59029 orcid.org/0009-0003-2200-59029 NOUN bracis-33557 2 54 , , PUNCT bracis-33557 2 55 adriano adriano PROPN bracis-33557 2 56 rivolli rivolli PROPN bracis-33557 2 57     SPACE bracis-33557 2 58 orcid orcid PROPN bracis-33557 2 59 : : PUNCT bracis-33557 2 60 orcid.org/0000-0001-6445-300710 orcid.org/0000-0001-6445-300710 PROPN bracis-33557 2 61 & & CCONJ bracis-33557 2 62 ana ana PROPN bracis-33557 2 63 carolina carolina PROPN bracis-33557 2 64 lorena lorena PROPN bracis-33557 2 65     SPACE bracis-33557 2 66 orcid orcid PROPN bracis-33557 2 67 : : PUNCT bracis-33557 2 68 orcid.org/0000-0002-6140-571x9 orcid.org/0000-0002-6140-571x9 ADP bracis-33557 2 69     SPACE bracis-33557 2 70 part part NOUN bracis-33557 2 71 of of ADP bracis-33557 2 72 the the DET bracis-33557 2 73 book book NOUN bracis-33557 2 74 series series NOUN bracis-33557 2 75 : : PUNCT bracis-33557 2 76 lecture lecture NOUN bracis-33557 2 77 notes note NOUN bracis-33557 2 78 in in ADP bracis-33557 2 79 computer computer NOUN bracis-33557 2 80 science science NOUN bracis-33557 2 81 ( ( PUNCT bracis-33557 2 82 ( ( PUNCT bracis-33557 2 83 lnai lnai ADJ bracis-33557 2 84 , , PUNCT bracis-33557 2 85 volume volume NOUN bracis-33557 2 86 15412 15412 NUM bracis-33557 2 87 ) ) PUNCT bracis-33557 2 88 ) ) PUNCT bracis-33557 2 89 included include VERB bracis-33557 2 90 in in ADP bracis-33557 2 91 the the DET bracis-33557 2 92 following follow VERB bracis-33557 2 93 conference conference NOUN bracis-33557 2 94 series series NOUN bracis-33557 2 95 : : PUNCT bracis-33557 2 96 brazilian brazilian ADJ bracis-33557 2 97 conference conference NOUN bracis-33557 2 98 on on ADP bracis-33557 2 99 intelligent intelligent ADJ bracis-33557 2 100 systems system NOUN bracis-33557 2 101 497 497 NUM bracis-33557 2 102 accesses access NOUN bracis-33557 2 103 abstract abstract ADJ bracis-33557 2 104 instance instance NOUN bracis-33557 2 105 hardness hardness NOUN bracis-33557 2 106 measures measure NOUN bracis-33557 2 107 allow allow VERB bracis-33557 2 108 us we PRON bracis-33557 2 109 to to PART bracis-33557 2 110 assess assess VERB bracis-33557 2 111 and and CCONJ bracis-33557 2 112 understand understand VERB bracis-33557 2 113 why why SCONJ bracis-33557 2 114 some some DET bracis-33557 2 115 observations observation NOUN bracis-33557 2 116 from from ADP bracis-33557 2 117 a a DET bracis-33557 2 118 dataset dataset NOUN bracis-33557 2 119 are be AUX bracis-33557 2 120 difficult difficult ADJ bracis-33557 2 121 to to PART bracis-33557 2 122 classify classify VERB bracis-33557 2 123 . . PUNCT bracis-33557 3 1 with with ADP bracis-33557 3 2 this this DET bracis-33557 3 3 information information NOUN bracis-33557 3 4 , , PUNCT bracis-33557 3 5 one one PRON bracis-33557 3 6 may may AUX bracis-33557 3 7 curate curate VERB bracis-33557 3 8 and and CCONJ bracis-33557 3 9 cleanse cleanse VERB bracis-33557 3 10 the the DET bracis-33557 3 11 training training NOUN bracis-33557 3 12 dataset dataset NOUN bracis-33557 3 13 for for ADP bracis-33557 3 14 improved improved ADJ bracis-33557 3 15 data datum NOUN bracis-33557 3 16 quality quality NOUN bracis-33557 3 17 . . PUNCT bracis-33557 4 1 however however ADV bracis-33557 4 2 , , PUNCT bracis-33557 4 3 these these DET bracis-33557 4 4 measures measure NOUN bracis-33557 4 5 require require VERB bracis-33557 4 6 data datum NOUN bracis-33557 4 7 to to PART bracis-33557 4 8 be be AUX bracis-33557 4 9 labeled label VERB bracis-33557 4 10 . . PUNCT bracis-33557 5 1 this this PRON bracis-33557 5 2 limits limit VERB bracis-33557 5 3 their their PRON bracis-33557 5 4 usage usage NOUN bracis-33557 5 5 in in ADP bracis-33557 5 6 the the DET bracis-33557 5 7 deployment deployment NOUN bracis-33557 5 8 stage stage NOUN bracis-33557 5 9 when when SCONJ bracis-33557 5 10 data datum NOUN bracis-33557 5 11 is be AUX bracis-33557 5 12 unlabeled unlabele VERB bracis-33557 5 13 . . PUNCT bracis-33557 6 1 this this DET bracis-33557 6 2 paper paper NOUN bracis-33557 6 3 investigates investigate VERB bracis-33557 6 4 whether whether SCONJ bracis-33557 6 5 it it PRON bracis-33557 6 6 is be AUX bracis-33557 6 7 possible possible ADJ bracis-33557 6 8 to to PART bracis-33557 6 9 identify identify VERB bracis-33557 6 10 observations observation NOUN bracis-33557 6 11 that that PRON bracis-33557 6 12 will will AUX bracis-33557 6 13 be be AUX bracis-33557 6 14 hard hard ADJ bracis-33557 6 15 to to PART bracis-33557 6 16 classify classify VERB bracis-33557 6 17 despite despite SCONJ bracis-33557 6 18 their their PRON bracis-33557 6 19 label label NOUN bracis-33557 6 20 . . PUNCT bracis-33557 7 1 for for ADP bracis-33557 7 2 such such ADJ bracis-33557 7 3 , , PUNCT bracis-33557 7 4 two two NUM bracis-33557 7 5 approaches approach NOUN bracis-33557 7 6 are be AUX bracis-33557 7 7 tested test VERB bracis-33557 7 8 . . PUNCT bracis-33557 8 1 the the DET bracis-33557 8 2 first first ADJ bracis-33557 8 3 adapts adapt VERB bracis-33557 8 4 known know VERB bracis-33557 8 5 instance instance NOUN bracis-33557 8 6 hardness hardness NOUN bracis-33557 8 7 measures measure NOUN bracis-33557 8 8 to to ADP bracis-33557 8 9 the the DET bracis-33557 8 10 unlabeled unlabeled ADJ bracis-33557 8 11 scenario scenario NOUN bracis-33557 8 12 . . PUNCT bracis-33557 9 1 the the DET bracis-33557 9 2 second second ADJ bracis-33557 9 3 learns learn VERB bracis-33557 9 4 regression regression NOUN bracis-33557 9 5 meta meta NOUN bracis-33557 9 6 - - PUNCT bracis-33557 9 7 models model NOUN bracis-33557 9 8 to to PART bracis-33557 9 9 estimate estimate VERB bracis-33557 9 10 the the DET bracis-33557 9 11 instance instance NOUN bracis-33557 9 12 hardness hardness NOUN bracis-33557 9 13 of of ADP bracis-33557 9 14 new new ADJ bracis-33557 9 15 data datum NOUN bracis-33557 9 16 observations observation NOUN bracis-33557 9 17 . . PUNCT bracis-33557 10 1 in in ADP bracis-33557 10 2 experiments experiment NOUN bracis-33557 10 3 , , PUNCT bracis-33557 10 4 both both DET bracis-33557 10 5 approaches approach NOUN bracis-33557 10 6 were be AUX bracis-33557 10 7 better well ADJ bracis-33557 10 8 at at ADP bracis-33557 10 9 identifying identify VERB bracis-33557 10 10 instances instance NOUN bracis-33557 10 11 lying lie VERB bracis-33557 10 12 in in ADP bracis-33557 10 13 borderline borderline ADJ bracis-33557 10 14 regions region NOUN bracis-33557 10 15 of of ADP bracis-33557 10 16 the the DET bracis-33557 10 17 dataset dataset NOUN bracis-33557 10 18 , , PUNCT bracis-33557 10 19 which which PRON bracis-33557 10 20 pose pose VERB bracis-33557 10 21 a a DET bracis-33557 10 22 greater great ADJ bracis-33557 10 23 difficulty difficulty NOUN bracis-33557 10 24 when when SCONJ bracis-33557 10 25 the the DET bracis-33557 10 26 label label NOUN bracis-33557 10 27 is be AUX bracis-33557 10 28 unknown unknown ADJ bracis-33557 10 29 . . PUNCT bracis-33557 11 1 access access NOUN bracis-33557 11 2 provided provide VERB bracis-33557 11 3 by by ADP bracis-33557 11 4 university university PROPN bracis-33557 11 5 of of ADP bracis-33557 11 6 notre notre PROPN bracis-33557 11 7 dame dame PROPN bracis-33557 11 8 hesburgh hesburgh PROPN bracis-33557 11 9 library library PROPN bracis-33557 11 10 . . PUNCT bracis-33557 12 1 download download PROPN bracis-33557 12 2 conference conference NOUN bracis-33557 12 3 paper paper NOUN bracis-33557 12 4 pdf pdf NOUN bracis-33557 12 5 similar similar ADJ bracis-33557 12 6 content content NOUN bracis-33557 12 7 being be AUX bracis-33557 12 8 viewed view VERB bracis-33557 12 9 by by ADP bracis-33557 12 10 others other NOUN bracis-33557 12 11 measuring measure VERB bracis-33557 12 12 difficulty difficulty NOUN bracis-33557 12 13 of of ADP bracis-33557 12 14     SPACE bracis-33557 12 15 learning learn VERB bracis-33557 12 16 using use VERB bracis-33557 12 17 ensemble ensemble ADJ bracis-33557 12 18 methods method NOUN bracis-33557 12 19 chapter chapter NOUN bracis-33557 12 20 © © PROPN bracis-33557 12 21 2022 2022 NUM bracis-33557 12 22 measuring measure VERB bracis-33557 12 23 instance instance NOUN bracis-33557 12 24 hardness hardness NOUN bracis-33557 12 25 using use VERB bracis-33557 12 26 data datum NOUN bracis-33557 12 27 complexity complexity NOUN bracis-33557 12 28 measures measure NOUN bracis-33557 12 29 chapter chapter NOUN bracis-33557 13 1 © © NOUN bracis-33557 13 2 2020 2020 NUM bracis-33557 13 3 active active ADJ bracis-33557 13 4 learning learning NOUN bracis-33557 13 5 using use VERB bracis-33557 13 6 difficult difficult ADJ bracis-33557 13 7 instances instance NOUN bracis-33557 13 8 chapter chapter NOUN bracis-33557 13 9 © © NOUN bracis-33557 13 10 2022 2022 NUM bracis-33557 13 11 1 1 NUM bracis-33557 13 12 introduction introduction NOUN bracis-33557 13 13 the the DET bracis-33557 13 14 machine machine NOUN bracis-33557 13 15 learning learning NOUN bracis-33557 13 16 ( ( PUNCT bracis-33557 13 17 ml ml NOUN bracis-33557 13 18 ) ) PUNCT bracis-33557 13 19 literature literature NOUN bracis-33557 13 20 extensively extensively ADV bracis-33557 13 21 provides provide VERB bracis-33557 13 22 algorithmic algorithmic ADJ bracis-33557 13 23 developments development NOUN bracis-33557 13 24 focused focus VERB bracis-33557 13 25 on on ADP bracis-33557 13 26 model model NOUN bracis-33557 13 27 hyperparameter hyperparameter NOUN bracis-33557 13 28 tuning tuning NOUN bracis-33557 13 29 and and CCONJ bracis-33557 13 30 related related ADJ bracis-33557 13 31 model model ADJ bracis-33557 13 32 - - ADJ bracis-33557 13 33 centric centric ADJ bracis-33557 13 34 tasks task NOUN bracis-33557 13 35 . . PUNCT bracis-33557 14 1 more more ADV bracis-33557 14 2 recently recently ADV bracis-33557 14 3 , , PUNCT bracis-33557 14 4 the the DET bracis-33557 14 5 community community NOUN bracis-33557 14 6 of of ADP bracis-33557 14 7 data datum NOUN bracis-33557 14 8 - - PUNCT bracis-33557 14 9 centric centric ADJ bracis-33557 14 10 artificial artificial ADJ bracis-33557 14 11 intelligence intelligence NOUN bracis-33557 14 12 ( ( PUNCT bracis-33557 14 13 ai ai NOUN bracis-33557 14 14 ) ) PUNCT bracis-33557 14 15 is be AUX bracis-33557 14 16 lighting light VERB bracis-33557 14 17 the the DET bracis-33557 14 18 focus focus NOUN bracis-33557 14 19 on on ADP bracis-33557 14 20 the the DET bracis-33557 14 21 effort effort NOUN bracis-33557 14 22 to to PART bracis-33557 14 23 understand understand VERB bracis-33557 14 24 more more ADJ bracis-33557 14 25 the the DET bracis-33557 14 26 data datum NOUN bracis-33557 14 27 and and CCONJ bracis-33557 14 28 its its PRON bracis-33557 14 29 quality quality NOUN bracis-33557 14 30 improvement improvement NOUN bracis-33557 14 31 than than ADP bracis-33557 14 32 on on ADP bracis-33557 14 33 developing develop VERB bracis-33557 14 34 more more ADV bracis-33557 14 35 complex complex ADJ bracis-33557 14 36 ml ml NOUN bracis-33557 14 37 models model NOUN bracis-33557 14 38 [ [ X bracis-33557 14 39 13 13 NUM bracis-33557 14 40 ] ] PUNCT bracis-33557 14 41 . . PUNCT bracis-33557 15 1 paving pave VERB bracis-33557 15 2 the the DET bracis-33557 15 3 way way NOUN bracis-33557 15 4 for for ADP bracis-33557 15 5 such such DET bracis-33557 15 6 a a DET bracis-33557 15 7 data data NOUN bracis-33557 15 8 - - PUNCT bracis-33557 15 9 centric centric ADJ bracis-33557 15 10 approach approach NOUN bracis-33557 15 11 is be AUX bracis-33557 15 12 a a DET bracis-33557 15 13 more more ADV bracis-33557 15 14 fine fine ADV bracis-33557 15 15 - - PUNCT bracis-33557 15 16 grained grain VERB bracis-33557 15 17 analysis analysis NOUN bracis-33557 15 18 of of ADP bracis-33557 15 19 the the DET bracis-33557 15 20 data datum NOUN bracis-33557 15 21 and and CCONJ bracis-33557 15 22 classification classification NOUN bracis-33557 15 23 performance performance NOUN bracis-33557 15 24 . . PUNCT bracis-33557 16 1 herewith herewith PROPN bracis-33557 16 2 , , PUNCT bracis-33557 16 3 aggregated aggregate VERB bracis-33557 16 4 measures measure NOUN bracis-33557 16 5 applied apply VERB bracis-33557 16 6 for for ADP bracis-33557 16 7 classification classification NOUN bracis-33557 16 8 problems problem NOUN bracis-33557 16 9 , , PUNCT bracis-33557 16 10 such such ADJ bracis-33557 16 11 as as ADP bracis-33557 16 12 accuracy accuracy NOUN bracis-33557 16 13 , , PUNCT bracis-33557 16 14 precision precision NOUN bracis-33557 16 15 , , PUNCT bracis-33557 16 16 or or CCONJ bracis-33557 16 17 similar similar ADJ bracis-33557 16 18 metrics metric NOUN bracis-33557 16 19 , , PUNCT bracis-33557 16 20 restrict restrict VERB bracis-33557 16 21 the the DET bracis-33557 16 22 understanding understanding NOUN bracis-33557 16 23 of of ADP bracis-33557 16 24 the the DET bracis-33557 16 25 particularities particularity NOUN bracis-33557 16 26 in in ADP bracis-33557 16 27 the the DET bracis-33557 16 28 data datum NOUN bracis-33557 16 29 the the DET bracis-33557 16 30 algorithms algorithm NOUN bracis-33557 16 31 are be AUX bracis-33557 16 32 modeling model VERB bracis-33557 16 33 . . PUNCT bracis-33557 17 1 those those DET bracis-33557 17 2 aggregated aggregate VERB bracis-33557 17 3 metrics metric NOUN bracis-33557 17 4 do do AUX bracis-33557 17 5 not not PART bracis-33557 17 6 provide provide VERB bracis-33557 17 7 information information NOUN bracis-33557 17 8 about about ADP bracis-33557 17 9 misclassification misclassification NOUN bracis-33557 17 10 at at ADP bracis-33557 17 11 the the DET bracis-33557 17 12 level level NOUN bracis-33557 17 13 of of ADP bracis-33557 17 14 an an DET bracis-33557 17 15 instance instance NOUN bracis-33557 17 16 or or CCONJ bracis-33557 17 17 why why SCONJ bracis-33557 17 18 they they PRON bracis-33557 17 19 are be AUX bracis-33557 17 20 misclassified misclassified ADJ bracis-33557 17 21 . . PUNCT bracis-33557 18 1 however however ADV bracis-33557 18 2 , , PUNCT bracis-33557 18 3 a a DET bracis-33557 18 4 more more ADV bracis-33557 18 5 reliable reliable ADJ bracis-33557 18 6 usage usage NOUN bracis-33557 18 7 of of ADP bracis-33557 18 8 ml ml ADP bracis-33557 18 9 algorithms algorithm NOUN bracis-33557 18 10 must must AUX bracis-33557 18 11 reveal reveal VERB bracis-33557 18 12 for for ADP bracis-33557 18 13 which which PRON bracis-33557 18 14 particular particular ADJ bracis-33557 18 15 instances instance VERB bracis-33557 18 16 a a DET bracis-33557 18 17 model model NOUN bracis-33557 18 18 struggles struggle VERB bracis-33557 18 19 to to PART bracis-33557 18 20 classify classify VERB bracis-33557 18 21 correctly correctly ADV bracis-33557 18 22 and and CCONJ bracis-33557 18 23 why why SCONJ bracis-33557 18 24 . . PUNCT bracis-33557 19 1 one one NUM bracis-33557 19 2 way way NOUN bracis-33557 19 3 to to PART bracis-33557 19 4 achieve achieve VERB bracis-33557 19 5 such such ADJ bracis-33557 19 6 understanding understanding NOUN bracis-33557 19 7 is be AUX bracis-33557 19 8 leveraging leverage VERB bracis-33557 19 9 knowledge knowledge NOUN bracis-33557 19 10 from from ADP bracis-33557 19 11 correlating correlate VERB bracis-33557 19 12 data datum NOUN bracis-33557 19 13 characteristics characteristic NOUN bracis-33557 19 14 extracted extract VERB bracis-33557 19 15 by by ADP bracis-33557 19 16 a a DET bracis-33557 19 17 set set NOUN bracis-33557 19 18 of of ADP bracis-33557 19 19 meta meta NOUN bracis-33557 19 20 - - PUNCT bracis-33557 19 21 features feature VERB bracis-33557 19 22 [ [ X bracis-33557 19 23 12 12 NUM bracis-33557 19 24 ] ] PUNCT bracis-33557 19 25 to to ADP bracis-33557 19 26 the the DET bracis-33557 19 27 predictive predictive ADJ bracis-33557 19 28 performance performance NOUN bracis-33557 19 29 of of ADP bracis-33557 19 30 multiple multiple ADJ bracis-33557 19 31 algorithms algorithm NOUN bracis-33557 19 32 , , PUNCT bracis-33557 19 33 in in ADP bracis-33557 19 34 a a DET bracis-33557 19 35 meta meta ADV bracis-33557 19 36 - - PUNCT bracis-33557 19 37 learning learn VERB bracis-33557 19 38 ( ( PUNCT bracis-33557 19 39 mtl mtl PROPN bracis-33557 19 40 ) ) PUNCT bracis-33557 19 41 approach approach NOUN bracis-33557 19 42 [ [ X bracis-33557 19 43 2 2 NUM bracis-33557 19 44 ] ] PUNCT bracis-33557 19 45 . . PUNCT bracis-33557 20 1 one one NUM bracis-33557 20 2 particular particular ADJ bracis-33557 20 3 set set NOUN bracis-33557 20 4 of of ADP bracis-33557 20 5 meta meta NOUN bracis-33557 20 6 - - PUNCT bracis-33557 20 7 features feature NOUN bracis-33557 20 8 is be AUX bracis-33557 20 9 the the DET bracis-33557 20 10 set set NOUN bracis-33557 20 11 of of ADP bracis-33557 20 12 data datum NOUN bracis-33557 20 13 complexity complexity NOUN bracis-33557 20 14 measures measure NOUN bracis-33557 20 15 , , PUNCT bracis-33557 20 16 previously previously ADV bracis-33557 20 17 proposed propose VERB bracis-33557 20 18 by by ADP bracis-33557 20 19 ho ho PROPN bracis-33557 20 20 and and CCONJ bracis-33557 20 21 basu basu PROPN bracis-33557 21 1 [ [ X bracis-33557 21 2 5 5 NUM bracis-33557 21 3 ] ] PUNCT bracis-33557 21 4 to to PART bracis-33557 21 5 explore explore VERB bracis-33557 21 6 the the DET bracis-33557 21 7 overall overall ADJ bracis-33557 21 8 complexity complexity NOUN bracis-33557 21 9 of of ADP bracis-33557 21 10 solving solve VERB bracis-33557 21 11 the the DET bracis-33557 21 12 classification classification NOUN bracis-33557 21 13 problem problem NOUN bracis-33557 21 14 given give VERB bracis-33557 21 15 the the DET bracis-33557 21 16 dataset dataset NOUN bracis-33557 21 17 available available ADJ bracis-33557 21 18 for for ADP bracis-33557 21 19 learning learn VERB bracis-33557 21 20 , , PUNCT bracis-33557 21 21 providing provide VERB bracis-33557 21 22 a a DET bracis-33557 21 23 global global ADJ bracis-33557 21 24 perspective perspective NOUN bracis-33557 21 25 of of ADP bracis-33557 21 26 the the DET bracis-33557 21 27 difficulty difficulty NOUN bracis-33557 21 28 of of ADP bracis-33557 21 29 the the DET bracis-33557 21 30 problem problem NOUN bracis-33557 21 31 [ [ X bracis-33557 21 32 1 1 NUM bracis-33557 21 33 ] ] PUNCT bracis-33557 21 34 . . PUNCT bracis-33557 22 1 since since SCONJ bracis-33557 22 2 these these DET bracis-33557 22 3 measures measure NOUN bracis-33557 22 4 can can AUX bracis-33557 22 5 fail fail VERB bracis-33557 22 6 to to PART bracis-33557 22 7 provide provide VERB bracis-33557 22 8 information information NOUN bracis-33557 22 9 at at ADP bracis-33557 22 10 the the DET bracis-33557 22 11 instance instance NOUN bracis-33557 22 12 - - PUNCT bracis-33557 22 13 level level NOUN bracis-33557 22 14 [ [ X bracis-33557 22 15 8 8 NUM bracis-33557 22 16 ] ] PUNCT bracis-33557 22 17 , , PUNCT bracis-33557 22 18 the the DET bracis-33557 22 19 instance instance NOUN bracis-33557 22 20 hardness hardness NOUN bracis-33557 22 21 measures measure NOUN bracis-33557 22 22 ( ( PUNCT bracis-33557 22 23 ihms ihm NOUN bracis-33557 22 24 ) ) PUNCT bracis-33557 22 25 were be AUX bracis-33557 22 26 introduced introduce VERB bracis-33557 22 27 by by ADP bracis-33557 22 28 smith smith PROPN bracis-33557 22 29 et et PROPN bracis-33557 22 30 al al PROPN bracis-33557 22 31 . . PUNCT bracis-33557 23 1 [ [ X bracis-33557 23 2 14 14 NUM bracis-33557 23 3 ] ] PUNCT bracis-33557 23 4 to to PART bracis-33557 23 5 characterize characterize VERB bracis-33557 23 6 the the DET bracis-33557 23 7 difficulty difficulty NOUN bracis-33557 23 8 level level NOUN bracis-33557 23 9 of of ADP bracis-33557 23 10 each each DET bracis-33557 23 11 individual individual ADJ bracis-33557 23 12 instance instance NOUN bracis-33557 23 13 of of ADP bracis-33557 23 14 a a DET bracis-33557 23 15 dataset dataset NOUN bracis-33557 23 16 , , PUNCT bracis-33557 23 17 giving give VERB bracis-33557 23 18 information information NOUN bracis-33557 23 19 on on ADP bracis-33557 23 20 which which PRON bracis-33557 23 21 particular particular ADJ bracis-33557 23 22 instances instance NOUN bracis-33557 23 23 are be AUX bracis-33557 23 24 misclassified misclassifie VERB bracis-33557 23 25 and and CCONJ bracis-33557 23 26 why why SCONJ bracis-33557 23 27 . . PUNCT bracis-33557 24 1 these these DET bracis-33557 24 2 developments development NOUN bracis-33557 24 3 attend attend VERB bracis-33557 24 4 to to ADP bracis-33557 24 5 a a DET bracis-33557 24 6 trending trend VERB bracis-33557 24 7 interest interest NOUN bracis-33557 24 8 in in ADP bracis-33557 24 9 responsible responsible ADJ bracis-33557 24 10 ai ai NOUN bracis-33557 24 11 that that PRON bracis-33557 24 12 has have AUX bracis-33557 24 13 emerged emerge VERB bracis-33557 24 14 in in ADP bracis-33557 24 15 recent recent ADJ bracis-33557 24 16 years year NOUN bracis-33557 24 17 , , PUNCT bracis-33557 24 18 making make VERB bracis-33557 24 19 researchers researcher NOUN bracis-33557 24 20 focus focus VERB bracis-33557 24 21 on on ADP bracis-33557 24 22 the the DET bracis-33557 24 23 reliability reliability NOUN bracis-33557 24 24 and and CCONJ bracis-33557 24 25 trustfulness trustfulness NOUN bracis-33557 24 26 of of ADP bracis-33557 24 27 the the DET bracis-33557 24 28 predictions prediction NOUN bracis-33557 24 29 obtained obtain VERB bracis-33557 24 30 by by ADP bracis-33557 24 31 ml ml NOUN bracis-33557 24 32 models model NOUN bracis-33557 24 33 . . PUNCT bracis-33557 25 1 nonetheless nonetheless ADV bracis-33557 25 2 , , PUNCT bracis-33557 25 3 the the DET bracis-33557 25 4 current current ADJ bracis-33557 25 5 ihms ihm NOUN bracis-33557 25 6 need need VERB bracis-33557 25 7 the the DET bracis-33557 25 8 instance instance NOUN bracis-33557 25 9 label label NOUN bracis-33557 25 10 to to PART bracis-33557 25 11 be be AUX bracis-33557 25 12 computed compute VERB bracis-33557 25 13 , , PUNCT bracis-33557 25 14 which which PRON bracis-33557 25 15 restricts restrict VERB bracis-33557 25 16 their their PRON bracis-33557 25 17 use use NOUN bracis-33557 25 18 for for ADP bracis-33557 25 19 analyzing analyze VERB bracis-33557 25 20 and and CCONJ bracis-33557 25 21 curating curate VERB bracis-33557 25 22 ml ml ADP bracis-33557 25 23 training training NOUN bracis-33557 25 24 datasets dataset NOUN bracis-33557 25 25 . . PUNCT bracis-33557 26 1 in in ADP bracis-33557 26 2 the the DET bracis-33557 26 3 use use NOUN bracis-33557 26 4 of of ADP bracis-33557 26 5 ml ml NOUN bracis-33557 26 6 in in ADP bracis-33557 26 7 production production NOUN bracis-33557 26 8 , , PUNCT bracis-33557 26 9 where where SCONJ bracis-33557 26 10 the the DET bracis-33557 26 11 class class NOUN bracis-33557 26 12 of of ADP bracis-33557 26 13 an an DET bracis-33557 26 14 instance instance NOUN bracis-33557 26 15 is be AUX bracis-33557 26 16 unknown unknown ADJ bracis-33557 26 17 , , PUNCT bracis-33557 26 18 adaptations adaptation NOUN bracis-33557 26 19 are be AUX bracis-33557 26 20 needed need VERB bracis-33557 26 21 . . PUNCT bracis-33557 27 1 this this DET bracis-33557 27 2 paper paper NOUN bracis-33557 27 3 proposes propose VERB bracis-33557 27 4 alternative alternative ADJ bracis-33557 27 5 instance instance NOUN bracis-33557 27 6 hardness hardness NOUN bracis-33557 27 7 measures measure NOUN bracis-33557 27 8 when when SCONJ bracis-33557 27 9 the the DET bracis-33557 27 10 instances instance NOUN bracis-33557 27 11 do do AUX bracis-33557 27 12 not not PART bracis-33557 27 13 have have VERB bracis-33557 27 14 a a DET bracis-33557 27 15 label label NOUN bracis-33557 27 16 . . PUNCT bracis-33557 28 1 the the DET bracis-33557 28 2 idea idea NOUN bracis-33557 28 3 is be AUX bracis-33557 28 4 to to PART bracis-33557 28 5 leverage leverage VERB bracis-33557 28 6 the the DET bracis-33557 28 7 knowledge knowledge NOUN bracis-33557 28 8 of of ADP bracis-33557 28 9 the the DET bracis-33557 28 10 hardness hardness NOUN bracis-33557 28 11 of of ADP bracis-33557 28 12 the the DET bracis-33557 28 13 training training NOUN bracis-33557 28 14 dataset dataset NOUN bracis-33557 28 15 , , PUNCT bracis-33557 28 16 which which PRON bracis-33557 28 17 is be AUX bracis-33557 28 18 labeled label VERB bracis-33557 28 19 , , PUNCT bracis-33557 28 20 to to PART bracis-33557 28 21 assess assess VERB bracis-33557 28 22 the the DET bracis-33557 28 23 hardness hardness ADJ bracis-33557 28 24 level level NOUN bracis-33557 28 25 of of ADP bracis-33557 28 26 new new ADJ bracis-33557 28 27 unlabeled unlabele VERB bracis-33557 28 28 instances instance NOUN bracis-33557 28 29 . . PUNCT bracis-33557 29 1 this this DET bracis-33557 29 2 knowledge knowledge NOUN bracis-33557 29 3 can can AUX bracis-33557 29 4 support support VERB bracis-33557 29 5 reject reject VERB bracis-33557 29 6 - - PUNCT bracis-33557 29 7 option option NOUN bracis-33557 29 8 strategies strategy NOUN bracis-33557 29 9 in in ADP bracis-33557 29 10 the the DET bracis-33557 29 11 future future NOUN bracis-33557 29 12 so so SCONJ bracis-33557 29 13 the the DET bracis-33557 29 14 ml ml PROPN bracis-33557 29 15 model model NOUN bracis-33557 29 16 might might AUX bracis-33557 29 17 opt opt VERB bracis-33557 29 18 for for ADP bracis-33557 29 19 abstaining abstain VERB bracis-33557 29 20 from from ADP bracis-33557 29 21 some some DET bracis-33557 29 22 predictions prediction NOUN bracis-33557 29 23 that that PRON bracis-33557 29 24 will will AUX bracis-33557 29 25 be be AUX bracis-33557 29 26 uncertain uncertain ADJ bracis-33557 29 27 [ [ X bracis-33557 29 28 4 4 NUM bracis-33557 29 29 ] ] PUNCT bracis-33557 29 30 . . PUNCT bracis-33557 30 1 firstly firstly ADV bracis-33557 30 2 , , PUNCT bracis-33557 30 3 a a DET bracis-33557 30 4 set set NOUN bracis-33557 30 5 of of ADP bracis-33557 30 6 ihms ihm NOUN bracis-33557 30 7 is be AUX bracis-33557 30 8 adapted adapt VERB bracis-33557 30 9 to to PART bracis-33557 30 10 disregard disregard VERB bracis-33557 30 11 the the DET bracis-33557 30 12 labels label NOUN bracis-33557 30 13 of of ADP bracis-33557 30 14 the the DET bracis-33557 30 15 new new ADJ bracis-33557 30 16 instances instance NOUN bracis-33557 30 17 in in ADP bracis-33557 30 18 their their PRON bracis-33557 30 19 computation computation NOUN bracis-33557 30 20 . . PUNCT bracis-33557 31 1 another another DET bracis-33557 31 2 strategy strategy NOUN bracis-33557 31 3 tested test VERB bracis-33557 31 4 was be AUX bracis-33557 31 5 generating generate VERB bracis-33557 31 6 regression regression NOUN bracis-33557 31 7 meta meta NOUN bracis-33557 31 8 - - PUNCT bracis-33557 31 9 models model NOUN bracis-33557 31 10 to to PART bracis-33557 31 11 estimate estimate VERB bracis-33557 31 12 the the DET bracis-33557 31 13 ihms ihm NOUN bracis-33557 31 14 for for ADP bracis-33557 31 15 new new ADJ bracis-33557 31 16 unlabeled unlabeled ADJ bracis-33557 31 17 observations observation NOUN bracis-33557 31 18 in in ADP bracis-33557 31 19 a a DET bracis-33557 31 20 meta meta ADJ bracis-33557 31 21 - - PUNCT bracis-33557 31 22 learning learn VERB bracis-33557 31 23 approach approach NOUN bracis-33557 31 24 at at ADP bracis-33557 31 25 the the DET bracis-33557 31 26 instance instance NOUN bracis-33557 31 27 level level NOUN bracis-33557 31 28 . . PUNCT bracis-33557 32 1 both both DET bracis-33557 32 2 approaches approach NOUN bracis-33557 32 3 are be AUX bracis-33557 32 4 compared compare VERB bracis-33557 32 5 experimentally experimentally ADV bracis-33557 32 6 using use VERB bracis-33557 32 7 one one NUM bracis-33557 32 8 synthetic synthetic ADJ bracis-33557 32 9 dataset dataset NOUN bracis-33557 32 10 and and CCONJ bracis-33557 32 11 four four NUM bracis-33557 32 12 datasets dataset NOUN bracis-33557 32 13 of of ADP bracis-33557 32 14 the the DET bracis-33557 32 15 health health NOUN bracis-33557 32 16 domain domain NOUN bracis-33557 32 17 , , PUNCT bracis-33557 32 18 known know VERB bracis-33557 32 19 for for ADP bracis-33557 32 20 presenting present VERB bracis-33557 32 21 hard hard ADJ bracis-33557 32 22 instances instance NOUN bracis-33557 32 23 . . PUNCT bracis-33557 33 1 instances instance NOUN bracis-33557 33 2 with with ADP bracis-33557 33 3 characteristics characteristic NOUN bracis-33557 33 4 making make VERB bracis-33557 33 5 them they PRON bracis-33557 33 6 lie lie VERB bracis-33557 33 7 in in ADP bracis-33557 33 8 overlapping overlap VERB bracis-33557 33 9 or or CCONJ bracis-33557 33 10 borderline borderline NOUN bracis-33557 33 11 regions region NOUN bracis-33557 33 12 of of ADP bracis-33557 33 13 the the DET bracis-33557 33 14 classes class NOUN bracis-33557 33 15 are be AUX bracis-33557 33 16 highlighted highlight VERB bracis-33557 33 17 as as ADP bracis-33557 33 18 hard hard ADJ bracis-33557 33 19 to to PART bracis-33557 33 20 classify classify VERB bracis-33557 33 21 by by ADP bracis-33557 33 22 both both DET bracis-33557 33 23 approaches approach NOUN bracis-33557 33 24 . . PUNCT bracis-33557 34 1 the the DET bracis-33557 34 2 adapted adapt VERB bracis-33557 34 3 measures measure NOUN bracis-33557 34 4 show show VERB bracis-33557 34 5 an an DET bracis-33557 34 6 increased increase VERB bracis-33557 34 7 correlation correlation NOUN bracis-33557 34 8 to to ADP bracis-33557 34 9 the the DET bracis-33557 34 10 original original ADJ bracis-33557 34 11 values value NOUN bracis-33557 34 12 of of ADP bracis-33557 34 13 the the DET bracis-33557 34 14 instance instance NOUN bracis-33557 34 15 hardness hardness NOUN bracis-33557 34 16 measures measure NOUN bracis-33557 34 17 and and CCONJ bracis-33557 34 18 prove prove VERB bracis-33557 34 19 to to PART bracis-33557 34 20 be be AUX bracis-33557 34 21 an an DET bracis-33557 34 22 adequate adequate ADJ bracis-33557 34 23 alternative alternative NOUN bracis-33557 34 24 to to PART bracis-33557 34 25 estimate estimate VERB bracis-33557 34 26 instance instance NOUN bracis-33557 34 27 hardness hardness ADV bracis-33557 34 28 in in ADP bracis-33557 34 29 the the DET bracis-33557 34 30 deployment deployment NOUN bracis-33557 34 31 stage stage NOUN bracis-33557 34 32 , , PUNCT bracis-33557 34 33 driving drive VERB bracis-33557 34 34 the the DET bracis-33557 34 35 solutions solution NOUN bracis-33557 34 36 to to ADP bracis-33557 34 37 a a DET bracis-33557 34 38 more more ADV bracis-33557 34 39 refined refined ADJ bracis-33557 34 40 level level NOUN bracis-33557 34 41 and and CCONJ bracis-33557 34 42 contributing contribute VERB bracis-33557 34 43 toward toward ADP bracis-33557 34 44 a a DET bracis-33557 34 45 more more ADV bracis-33557 34 46 trustful trustful ADJ bracis-33557 34 47 use use NOUN bracis-33557 34 48 of of ADP bracis-33557 34 49 ml ml NOUN bracis-33557 34 50 models model NOUN bracis-33557 34 51 . . PUNCT bracis-33557 35 1 the the DET bracis-33557 35 2 paper paper NOUN bracis-33557 35 3 is be AUX bracis-33557 35 4 organized organize VERB bracis-33557 35 5 as as SCONJ bracis-33557 35 6 follows follow VERB bracis-33557 35 7 : : PUNCT bracis-33557 35 8 sect sect NOUN bracis-33557 35 9 . . PUNCT bracis-33557 35 10     SPACE bracis-33557 36 1 2 2 NUM bracis-33557 36 2 details detail NOUN bracis-33557 36 3 the the DET bracis-33557 36 4 hardness hardness NOUN bracis-33557 36 5 measures measure NOUN bracis-33557 36 6 to to PART bracis-33557 36 7 apply apply VERB bracis-33557 36 8 to to ADP bracis-33557 36 9 unlabeled unlabeled ADJ bracis-33557 36 10 data datum NOUN bracis-33557 36 11 and and CCONJ bracis-33557 36 12 how how SCONJ bracis-33557 36 13 they they PRON bracis-33557 36 14 were be AUX bracis-33557 36 15 modified modify VERB bracis-33557 36 16 from from ADP bracis-33557 36 17 the the DET bracis-33557 36 18 original original ADJ bracis-33557 36 19 measures measure NOUN bracis-33557 36 20 . . PUNCT bracis-33557 37 1 section section NOUN bracis-33557 37 2     SPACE bracis-33557 37 3 3 3 NUM bracis-33557 37 4 presents present VERB bracis-33557 37 5 the the DET bracis-33557 37 6 materials material NOUN bracis-33557 37 7 and and CCONJ bracis-33557 37 8 methods method NOUN bracis-33557 37 9 used use VERB bracis-33557 37 10 in in ADP bracis-33557 37 11 experiments experiment NOUN bracis-33557 37 12 , , PUNCT bracis-33557 37 13 whose whose DET bracis-33557 37 14 results result NOUN bracis-33557 37 15 are be AUX bracis-33557 37 16 presented present VERB bracis-33557 37 17 in in ADP bracis-33557 37 18 sect sect NOUN bracis-33557 37 19 . . PUNCT bracis-33557 37 20     SPACE bracis-33557 38 1 4 4 X bracis-33557 38 2 . . PUNCT bracis-33557 39 1 finally finally ADV bracis-33557 39 2 , , PUNCT bracis-33557 39 3 sect sect NOUN bracis-33557 39 4 . . PUNCT bracis-33557 39 5     SPACE bracis-33557 39 6 5 5 NUM bracis-33557 39 7 presents present VERB bracis-33557 39 8 the the DET bracis-33557 39 9 conclusions conclusion NOUN bracis-33557 39 10 of of ADP bracis-33557 39 11 this this DET bracis-33557 39 12 work work NOUN bracis-33557 39 13 . . PUNCT bracis-33557 40 1 2 2 NUM bracis-33557 40 2 instance instance NOUN bracis-33557 40 3 hardness hardness NOUN bracis-33557 40 4 measures measure NOUN bracis-33557 40 5 the the DET bracis-33557 40 6 concept concept NOUN bracis-33557 40 7 of of ADP bracis-33557 40 8 instance instance NOUN bracis-33557 40 9 hardness hardness NOUN bracis-33557 40 10 was be AUX bracis-33557 40 11 introduced introduce VERB bracis-33557 40 12 in in ADP bracis-33557 40 13 the the DET bracis-33557 40 14 seminal seminal ADJ bracis-33557 40 15 work work NOUN bracis-33557 40 16 of of ADP bracis-33557 40 17 smith smith PROPN bracis-33557 40 18 et et PROPN bracis-33557 40 19 al al PROPN bracis-33557 40 20 . . PUNCT bracis-33557 41 1 [ [ X bracis-33557 41 2 14 14 NUM bracis-33557 41 3 ] ] PUNCT bracis-33557 41 4 as as ADP bracis-33557 41 5 an an DET bracis-33557 41 6 alternative alternative NOUN bracis-33557 41 7 for for ADP bracis-33557 41 8 a a DET bracis-33557 41 9 fine fine ADV bracis-33557 41 10 - - PUNCT bracis-33557 41 11 grained grain VERB bracis-33557 41 12 analysis analysis NOUN bracis-33557 41 13 of of ADP bracis-33557 41 14 classification classification NOUN bracis-33557 41 15 difficulty difficulty NOUN bracis-33557 41 16 . . PUNCT bracis-33557 42 1 they they PRON bracis-33557 42 2 define define VERB bracis-33557 42 3 an an DET bracis-33557 42 4 instance instance NOUN bracis-33557 42 5 as as ADP bracis-33557 42 6 hard hard ADJ bracis-33557 42 7 to to PART bracis-33557 42 8 classify classify VERB bracis-33557 42 9 if if SCONJ bracis-33557 42 10 it it PRON bracis-33557 42 11 gets get AUX bracis-33557 42 12 consistently consistently ADV bracis-33557 42 13 misclassified misclassifie VERB bracis-33557 42 14 by by ADP bracis-33557 42 15 a a DET bracis-33557 42 16 set set NOUN bracis-33557 42 17 of of ADP bracis-33557 42 18 classifiers classifier NOUN bracis-33557 42 19 of of ADP bracis-33557 42 20 different different ADJ bracis-33557 42 21 biases bias NOUN bracis-33557 42 22 . . PUNCT bracis-33557 43 1 they they PRON bracis-33557 43 2 also also ADV bracis-33557 43 3 define define VERB bracis-33557 43 4 a a DET bracis-33557 43 5 set set NOUN bracis-33557 43 6 of of ADP bracis-33557 43 7 measures measure NOUN bracis-33557 43 8 to to PART bracis-33557 43 9 explain explain VERB bracis-33557 43 10 possible possible ADJ bracis-33557 43 11 reasons reason NOUN bracis-33557 43 12 why why SCONJ bracis-33557 43 13 an an DET bracis-33557 43 14 instance instance NOUN bracis-33557 43 15 is be AUX bracis-33557 43 16 difficult difficult ADJ bracis-33557 43 17 to to PART bracis-33557 43 18 classify classify VERB bracis-33557 43 19 , , PUNCT bracis-33557 43 20 which which PRON bracis-33557 43 21 are be AUX bracis-33557 43 22 regarded regard VERB bracis-33557 43 23 as as ADP bracis-33557 43 24 instance instance NOUN bracis-33557 43 25 - - PUNCT bracis-33557 43 26 level level NOUN bracis-33557 43 27 meta meta NOUN bracis-33557 43 28 - - PUNCT bracis-33557 43 29 features feature NOUN bracis-33557 43 30 in in ADP bracis-33557 43 31 the the DET bracis-33557 43 32 literature literature NOUN bracis-33557 43 33 [ [ X bracis-33557 43 34 7 7 NUM bracis-33557 43 35 ] ] PUNCT bracis-33557 43 36 . . PUNCT bracis-33557 44 1 the the DET bracis-33557 44 2 base base NOUN bracis-33557 44 3 ihms ihm NOUN bracis-33557 44 4 adopted adopt VERB bracis-33557 44 5 in in ADP bracis-33557 44 6 this this DET bracis-33557 44 7 work work NOUN bracis-33557 44 8 are be AUX bracis-33557 44 9 presented present VERB bracis-33557 44 10 next next ADV bracis-33557 44 11 , , PUNCT bracis-33557 44 12 along along ADP bracis-33557 44 13 with with ADP bracis-33557 44 14 their their PRON bracis-33557 44 15 adaptations adaptation NOUN bracis-33557 44 16 , , PUNCT bracis-33557 44 17 which which PRON bracis-33557 44 18 are be AUX bracis-33557 44 19 indicated indicate VERB bracis-33557 44 20 by by ADP bracis-33557 44 21 the the DET bracis-33557 44 22 “ " PUNCT bracis-33557 44 23 \({\text \({\text PROPN bracis-33557 44 24 { { PUNCT bracis-33557 44 25 adj}}\ adj}}\ PROPN bracis-33557 44 26 ) ) PUNCT bracis-33557 44 27 ” " PUNCT bracis-33557 44 28 ( ( PUNCT bracis-33557 44 29 adjusted adjust VERB bracis-33557 44 30 ) ) PUNCT bracis-33557 44 31 extension extension NOUN bracis-33557 44 32 . . PUNCT bracis-33557 45 1 in in ADP bracis-33557 45 2 their their PRON bracis-33557 45 3 definition definition NOUN bracis-33557 45 4 , , PUNCT bracis-33557 45 5 let let VERB bracis-33557 45 6 \(\mathcal \(\mathcal ADJ bracis-33557 45 7 d\ d\ NOUN bracis-33557 45 8 ) ) PUNCT bracis-33557 45 9 be be VERB bracis-33557 45 10 a a DET bracis-33557 45 11 training training NOUN bracis-33557 45 12 dataset dataset VERB bracis-33557 45 13 with with ADP bracis-33557 45 14 n n DET bracis-33557 45 15 pairs pair NOUN bracis-33557 45 16 of of ADP bracis-33557 45 17 labeled label VERB bracis-33557 45 18 instances instance NOUN bracis-33557 45 19 \((\textbf{x}_i \((\textbf{x}_i NOUN bracis-33557 45 20 , , PUNCT bracis-33557 45 21 y_i)\ y_i)\ NUM bracis-33557 45 22 ) ) PUNCT bracis-33557 45 23 , , PUNCT bracis-33557 45 24 where where SCONJ bracis-33557 45 25 each each DET bracis-33557 45 26 \(\textbf{x}_i \(\textbf{x}_i PROPN bracis-33557 45 27 \in \in PROPN bracis-33557 45 28 \mathcal \mathcal PROPN bracis-33557 45 29 x\ x\ NOUN bracis-33557 45 30 ) ) PUNCT bracis-33557 45 31 is be AUX bracis-33557 45 32 described describe VERB bracis-33557 45 33 by by ADP bracis-33557 45 34 m m PROPN bracis-33557 45 35 input input NOUN bracis-33557 45 36 features feature NOUN bracis-33557 45 37 and and CCONJ bracis-33557 45 38 \(y_i \(y_i PUNCT bracis-33557 45 39 \in \in PROPN bracis-33557 45 40 \mathcal \mathcal PROPN bracis-33557 45 41 y\ y\ NOUN bracis-33557 45 42 ) ) PUNCT bracis-33557 45 43 is be AUX bracis-33557 45 44 the the DET bracis-33557 45 45 class class NOUN bracis-33557 45 46 of of ADP bracis-33557 45 47 the the DET bracis-33557 45 48 instance instance NOUN bracis-33557 45 49 in in ADP bracis-33557 45 50 the the DET bracis-33557 45 51 dataset dataset NOUN bracis-33557 45 52 . . PUNCT bracis-33557 46 1 the the DET bracis-33557 46 2 number number NOUN bracis-33557 46 3 of of ADP bracis-33557 46 4 classes class NOUN bracis-33557 46 5 is be AUX bracis-33557 46 6 denoted denote VERB bracis-33557 46 7 as as ADP bracis-33557 46 8 c. c. NOUN bracis-33557 46 9 and and CCONJ bracis-33557 46 10 let let VERB bracis-33557 46 11 \(\textbf{x}\ \(\textbf{x}\ X bracis-33557 46 12 ) ) PUNCT bracis-33557 46 13 be be AUX bracis-33557 46 14 a a DET bracis-33557 46 15 new new ADJ bracis-33557 46 16 instance instance NOUN bracis-33557 46 17 for for ADP bracis-33557 46 18 which which PRON bracis-33557 46 19 the the DET bracis-33557 46 20 label label NOUN bracis-33557 46 21 is be AUX bracis-33557 46 22 unknown unknown ADJ bracis-33557 46 23 . . PUNCT bracis-33557 47 1 to to PART bracis-33557 47 2 illustrate illustrate VERB bracis-33557 47 3 the the DET bracis-33557 47 4 concepts concept NOUN bracis-33557 47 5 , , PUNCT bracis-33557 47 6 consider consider VERB bracis-33557 47 7 the the DET bracis-33557 47 8 dataset dataset NOUN bracis-33557 47 9 in in ADP bracis-33557 47 10 fig fig NOUN bracis-33557 47 11 . . PUNCT bracis-33557 47 12     SPACE bracis-33557 48 1 1 1 NUM bracis-33557 48 2 containing contain VERB bracis-33557 48 3 two two NUM bracis-33557 48 4 classes class NOUN bracis-33557 48 5 , , PUNCT bracis-33557 48 6 red red ADJ bracis-33557 48 7 and and CCONJ bracis-33557 48 8 blue blue ADJ bracis-33557 48 9 . . PUNCT bracis-33557 49 1 two two NUM bracis-33557 49 2 instances instance NOUN bracis-33557 49 3 are be AUX bracis-33557 49 4 highlighted highlight VERB bracis-33557 49 5 : : PUNCT bracis-33557 49 6 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 49 7 ) ) PUNCT bracis-33557 49 8 and and CCONJ bracis-33557 49 9 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 49 10 ) ) PUNCT bracis-33557 49 11 . . PUNCT bracis-33557 50 1 the the DET bracis-33557 50 2 instance instance NOUN bracis-33557 50 3 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 50 4 ) ) PUNCT bracis-33557 50 5 is be AUX bracis-33557 50 6 in in ADP bracis-33557 50 7 a a DET bracis-33557 50 8 borderline borderline NOUN bracis-33557 50 9 area area NOUN bracis-33557 50 10 of of ADP bracis-33557 50 11 the the DET bracis-33557 50 12 classes class NOUN bracis-33557 50 13 and and CCONJ bracis-33557 50 14 might might AUX bracis-33557 50 15 be be AUX bracis-33557 50 16 difficult difficult ADJ bracis-33557 50 17 to to PART bracis-33557 50 18 classify classify VERB bracis-33557 50 19 despite despite SCONJ bracis-33557 50 20 its its PRON bracis-33557 50 21 class class NOUN bracis-33557 50 22 . . PUNCT bracis-33557 51 1 the the DET bracis-33557 51 2 instance instance NOUN bracis-33557 51 3 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 51 4 ) ) PUNCT bracis-33557 51 5 is be AUX bracis-33557 51 6 more more ADV bracis-33557 51 7 aligned aligned ADJ bracis-33557 51 8 to to ADP bracis-33557 51 9 the the DET bracis-33557 51 10 blue blue ADJ bracis-33557 51 11 class class NOUN bracis-33557 51 12 . . PUNCT bracis-33557 52 1 if if SCONJ bracis-33557 52 2 the the DET bracis-33557 52 3 label label NOUN bracis-33557 52 4 registered register VERB bracis-33557 52 5 for for ADP bracis-33557 52 6 it it PRON bracis-33557 52 7 in in ADP bracis-33557 52 8 the the DET bracis-33557 52 9 dataset dataset NOUN bracis-33557 52 10 is be AUX bracis-33557 52 11 blue blue ADJ bracis-33557 52 12 , , PUNCT bracis-33557 52 13 it it PRON bracis-33557 52 14 will will AUX bracis-33557 52 15 be be AUX bracis-33557 52 16 easily easily ADV bracis-33557 52 17 classified classify VERB bracis-33557 52 18 . . PUNCT bracis-33557 53 1 otherwise otherwise ADV bracis-33557 53 2 , , PUNCT bracis-33557 53 3 it it PRON bracis-33557 53 4 will will AUX bracis-33557 53 5 have have VERB bracis-33557 53 6 a a DET bracis-33557 53 7 hardness hardness ADJ bracis-33557 53 8 level level NOUN bracis-33557 53 9 higher high ADJ bracis-33557 53 10 than than ADP bracis-33557 53 11 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 53 12 ) ) PUNCT bracis-33557 53 13 . . PUNCT bracis-33557 54 1 standard standard ADJ bracis-33557 54 2 ihms ihm NOUN bracis-33557 54 3 need need VERB bracis-33557 54 4 to to PART bracis-33557 54 5 know know VERB bracis-33557 54 6 these these DET bracis-33557 54 7 labels label NOUN bracis-33557 54 8 , , PUNCT bracis-33557 54 9 so so SCONJ bracis-33557 54 10 that that SCONJ bracis-33557 54 11 both both DET bracis-33557 54 12 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 54 13 ) ) PUNCT bracis-33557 54 14 and and CCONJ bracis-33557 54 15 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 54 16 ) ) PUNCT bracis-33557 54 17 are be AUX bracis-33557 54 18 contained contain VERB bracis-33557 54 19 in in ADP bracis-33557 54 20 the the DET bracis-33557 54 21 labeled label VERB bracis-33557 54 22 dataset dataset VERB bracis-33557 54 23 \(\mathcal \(\mathcal ADJ bracis-33557 54 24 d\ d\ NOUN bracis-33557 54 25 ) ) PUNCT bracis-33557 54 26 . . PUNCT bracis-33557 55 1 this this DET bracis-33557 55 2 work work NOUN bracis-33557 55 3 introduces introduce VERB bracis-33557 55 4 adaptations adaptation NOUN bracis-33557 55 5 to to PART bracis-33557 55 6 estimate estimate VERB bracis-33557 55 7 the the DET bracis-33557 55 8 hardness hardness ADJ bracis-33557 55 9 level level NOUN bracis-33557 55 10 of of ADP bracis-33557 55 11 an an DET bracis-33557 55 12 instance instance NOUN bracis-33557 55 13 in in ADP bracis-33557 55 14 the the DET bracis-33557 55 15 absence absence NOUN bracis-33557 55 16 of of ADP bracis-33557 55 17 its its PRON bracis-33557 55 18 label label NOUN bracis-33557 55 19 , , PUNCT bracis-33557 55 20 meaning mean VERB bracis-33557 55 21 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 55 22 ) ) PUNCT bracis-33557 55 23 and and CCONJ bracis-33557 55 24 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 55 25 ) ) PUNCT bracis-33557 55 26 are be AUX bracis-33557 55 27 not not PART bracis-33557 55 28 in in ADP bracis-33557 55 29 the the DET bracis-33557 55 30 labeled label VERB bracis-33557 55 31 dataset dataset VERB bracis-33557 55 32 \(\mathcal \(\mathcal ADJ bracis-33557 55 33 d\ d\ NOUN bracis-33557 55 34 ) ) PUNCT bracis-33557 55 35 used use VERB bracis-33557 55 36 to to PART bracis-33557 55 37 estimate estimate VERB bracis-33557 55 38 the the DET bracis-33557 55 39 hardness hardness NOUN bracis-33557 55 40 levels level NOUN bracis-33557 55 41 . . PUNCT bracis-33557 56 1 please please INTJ bracis-33557 56 2 note note VERB bracis-33557 56 3 there there PRON bracis-33557 56 4 are be VERB bracis-33557 56 5 differences difference NOUN bracis-33557 56 6 between between ADP bracis-33557 56 7 the the DET bracis-33557 56 8 two two NUM bracis-33557 56 9 estimations estimation NOUN bracis-33557 56 10 . . PUNCT bracis-33557 57 1 based base VERB bracis-33557 57 2 on on ADP bracis-33557 57 3 the the DET bracis-33557 57 4 characteristics characteristic NOUN bracis-33557 57 5 of of ADP bracis-33557 57 6 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 57 7 ) ) PUNCT bracis-33557 57 8 , , PUNCT bracis-33557 57 9 it it PRON bracis-33557 57 10 will will AUX bracis-33557 57 11 probably probably ADV bracis-33557 57 12 be be AUX bracis-33557 57 13 easily easily ADV bracis-33557 57 14 classified classify VERB bracis-33557 57 15 as as ADP bracis-33557 57 16 blue blue ADJ bracis-33557 57 17 . . PUNCT bracis-33557 58 1 in in ADP bracis-33557 58 2 contrast contrast NOUN bracis-33557 58 3 , , PUNCT bracis-33557 58 4 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 58 5 ) ) PUNCT bracis-33557 58 6 will will AUX bracis-33557 58 7 probably probably ADV bracis-33557 58 8 be be AUX bracis-33557 58 9 considered consider VERB bracis-33557 58 10 hard hard ADJ bracis-33557 58 11 to to PART bracis-33557 58 12 classify classify VERB bracis-33557 58 13 in in ADP bracis-33557 58 14 both both DET bracis-33557 58 15 scenarios scenario NOUN bracis-33557 58 16 . . PUNCT bracis-33557 59 1 fig fig NOUN bracis-33557 59 2 . . PUNCT bracis-33557 60 1 1 1 NUM bracis-33557 60 2 . . X bracis-33557 60 3 example example NOUN bracis-33557 60 4 of of ADP bracis-33557 60 5 dataset dataset NOUN bracis-33557 60 6 with with ADP bracis-33557 60 7 highlighted highlight VERB bracis-33557 60 8 instances instance NOUN bracis-33557 60 9 : : PUNCT bracis-33557 60 10 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 60 11 ) ) PUNCT bracis-33557 60 12 is be AUX bracis-33557 60 13 in in ADP bracis-33557 60 14 a a DET bracis-33557 60 15 borderline borderline NOUN bracis-33557 60 16 region region NOUN bracis-33557 60 17 and and CCONJ bracis-33557 60 18 can can AUX bracis-33557 60 19 be be AUX bracis-33557 60 20 difficult difficult ADJ bracis-33557 60 21 to to PART bracis-33557 60 22 classify classify VERB bracis-33557 60 23 despite despite SCONJ bracis-33557 60 24 its its PRON bracis-33557 60 25 class class NOUN bracis-33557 60 26 ; ; PUNCT bracis-33557 60 27 \(\textbf{x}_2\ \(\textbf{x}_2\ X bracis-33557 60 28 ) ) PUNCT bracis-33557 60 29 might might AUX bracis-33557 60 30 be be AUX bracis-33557 60 31 easy easy ADJ bracis-33557 60 32 or or CCONJ bracis-33557 60 33 hard hard ADJ bracis-33557 60 34 to to PART bracis-33557 60 35 classify classify VERB bracis-33557 60 36 depending depend VERB bracis-33557 60 37 on on ADP bracis-33557 60 38 its its PRON bracis-33557 60 39 registered register VERB bracis-33557 60 40 label label NOUN bracis-33557 60 41 . . PUNCT bracis-33557 61 1 ( ( PUNCT bracis-33557 61 2 color color NOUN bracis-33557 61 3 figure figure NOUN bracis-33557 61 4 online online ADV bracis-33557 61 5 ) ) PUNCT bracis-33557 61 6 full full ADJ bracis-33557 61 7 size size NOUN bracis-33557 61 8 image image NOUN bracis-33557 61 9 2.1 2.1 NUM bracis-33557 61 10 neighborhood neighborhood NOUN bracis-33557 61 11 - - PUNCT bracis-33557 61 12 based base VERB bracis-33557 61 13 ihm ihm NOUN bracis-33557 61 14 the the DET bracis-33557 61 15 hardness hardness ADJ bracis-33557 61 16 level level NOUN bracis-33557 61 17 of of ADP bracis-33557 61 18 the the DET bracis-33557 61 19 instance instance NOUN bracis-33557 61 20 can can AUX bracis-33557 61 21 be be AUX bracis-33557 61 22 obtained obtain VERB bracis-33557 61 23 considering consider VERB bracis-33557 61 24 its its PRON bracis-33557 61 25 neighbourhood neighbourhood NOUN bracis-33557 61 26 in in ADP bracis-33557 61 27 the the DET bracis-33557 61 28 dataset dataset NOUN bracis-33557 61 29 . . PUNCT bracis-33557 62 1 in in ADP bracis-33557 62 2 the the DET bracis-33557 62 3 original original ADJ bracis-33557 62 4 ihms ihm NOUN bracis-33557 62 5 , , PUNCT bracis-33557 62 6 instances instance NOUN bracis-33557 62 7 surrounded surround VERB bracis-33557 62 8 by by ADP bracis-33557 62 9 elements element NOUN bracis-33557 62 10 sharing share VERB bracis-33557 62 11 the the DET bracis-33557 62 12 same same ADJ bracis-33557 62 13 label label NOUN bracis-33557 62 14 as as SCONJ bracis-33557 62 15 themselves themselves PRON bracis-33557 62 16 can can AUX bracis-33557 62 17 be be AUX bracis-33557 62 18 considered consider VERB bracis-33557 62 19 easier easy ADJ bracis-33557 62 20 to to PART bracis-33557 62 21 classify classify VERB bracis-33557 62 22 . . PUNCT bracis-33557 63 1 for for ADP bracis-33557 63 2 new new ADJ bracis-33557 63 3 data datum NOUN bracis-33557 63 4 without without ADP bracis-33557 63 5 labels label NOUN bracis-33557 63 6 , , PUNCT bracis-33557 63 7 our our PRON bracis-33557 63 8 approach approach NOUN bracis-33557 63 9 seeks seek VERB bracis-33557 63 10 the the DET bracis-33557 63 11 neighbourhood neighbourhood NOUN bracis-33557 63 12 of of ADP bracis-33557 63 13 the the DET bracis-33557 63 14 instance instance NOUN bracis-33557 63 15 in in ADP bracis-33557 63 16 the the DET bracis-33557 63 17 labeled label VERB bracis-33557 63 18 dataset dataset VERB bracis-33557 63 19 \(\mathcal \(\mathcal ADJ bracis-33557 63 20 d\ d\ NOUN bracis-33557 63 21 ) ) PUNCT bracis-33557 63 22 and and CCONJ bracis-33557 63 23 assigns assign VERB bracis-33557 63 24 a a DET bracis-33557 63 25 higher high ADJ bracis-33557 63 26 hardness hardness NOUN bracis-33557 63 27 level level NOUN bracis-33557 63 28 when when SCONJ bracis-33557 63 29 there there PRON bracis-33557 63 30 is be VERB bracis-33557 63 31 a a DET bracis-33557 63 32 mix mix NOUN bracis-33557 63 33 of of ADP bracis-33557 63 34 different different ADJ bracis-33557 63 35 classes class NOUN bracis-33557 63 36 in in ADP bracis-33557 63 37 this this DET bracis-33557 63 38 region region NOUN bracis-33557 63 39 . . PUNCT bracis-33557 64 1 k k ADJ bracis-33557 64 2 - - ADJ bracis-33557 64 3 disagreeing disagree VERB bracis-33557 64 4 neighbors neighbor NOUN bracis-33557 64 5 kdn kdn PROPN bracis-33557 64 6 : : PUNCT bracis-33557 64 7 the the DET bracis-33557 64 8 original original ADJ bracis-33557 64 9 kdn kdn NOUN bracis-33557 64 10 measure measure NOUN bracis-33557 64 11 computes compute VERB bracis-33557 64 12 the the DET bracis-33557 64 13 percentage percentage NOUN bracis-33557 64 14 of of ADP bracis-33557 64 15 the the DET bracis-33557 64 16 k k PROPN bracis-33557 64 17 nearest near ADJ bracis-33557 64 18 neighbors neighbor NOUN bracis-33557 64 19 of of ADP bracis-33557 64 20 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 64 21 ) ) PUNCT bracis-33557 64 22 in in ADP bracis-33557 64 23 the the DET bracis-33557 64 24 dataset dataset ADJ bracis-33557 64 25 \(\mathcal \(\mathcal ADJ bracis-33557 64 26 d\ d\ NOUN bracis-33557 64 27 ) ) PUNCT bracis-33557 64 28 that that PRON bracis-33557 64 29 have have VERB bracis-33557 64 30 a a DET bracis-33557 64 31 different different ADJ bracis-33557 64 32 label label NOUN bracis-33557 64 33 than than ADP bracis-33557 64 34 the the DET bracis-33557 64 35 refereed refereed ADJ bracis-33557 64 36 instance instance NOUN bracis-33557 64 37 : : PUNCT bracis-33557 64 38 $ $ SYM bracis-33557 64 39 $ $ SYM bracis-33557 64 40 \begin{aligned \begin{aligne VERB bracis-33557 64 41 } } PUNCT bracis-33557 64 42 \text \text ADP bracis-33557 64 43 { { PUNCT bracis-33557 64 44 kdn}(\textbf{x}_i kdn}(\textbf{x}_i ADJ bracis-33557 64 45 , , PUNCT bracis-33557 64 46 y_i y_i NOUN bracis-33557 64 47 ) ) PUNCT bracis-33557 64 48 = = PUNCT bracis-33557 65 1 \frac{\sharp \frac{\sharp NUM bracis-33557 65 2 \{\textbf{x}_j \{\textbf{x}_j PROPN bracis-33557 65 3 | | ADV bracis-33557 65 4 \textbf{x}_j \textbf{x}_j NOUN bracis-33557 65 5 \in \in PROPN bracis-33557 65 6 \text \text PROPN bracis-33557 65 7 { { PUNCT bracis-33557 65 8 knn}(\textbf{x}_i knn}(\textbf{x}_i NOUN bracis-33557 65 9 ) ) PUNCT bracis-33557 65 10 \wedge \wedge PROPN bracis-33557 65 11 y_j y_j SYM bracis-33557 65 12 \ne \ne PROPN bracis-33557 65 13 y_i\}}{k y_i\}}{k PROPN bracis-33557 65 14 } } PUNCT bracis-33557 65 15 , , PUNCT bracis-33557 65 16 \end{aligned}$$ \end{aligned}$$ X bracis-33557 65 17 ( ( PUNCT bracis-33557 65 18 1 1 NUM bracis-33557 65 19 ) ) PUNCT bracis-33557 65 20 where where SCONJ bracis-33557 65 21 \(\text \(\text NOUN bracis-33557 65 22 { { PUNCT bracis-33557 65 23 knn}(\textbf{x}_i)\ knn}(\textbf{x}_i)\ PROPN bracis-33557 65 24 ) ) PUNCT bracis-33557 65 25 represents represent VERB bracis-33557 65 26 the the DET bracis-33557 65 27 set set NOUN bracis-33557 65 28 of of ADP bracis-33557 65 29 k k NOUN bracis-33557 65 30 - - PUNCT bracis-33557 65 31 nearest near ADJ bracis-33557 65 32 neighbors neighbor NOUN bracis-33557 65 33 of of ADP bracis-33557 65 34 the the DET bracis-33557 65 35 instance instance NOUN bracis-33557 65 36 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 65 37 ) ) PUNCT bracis-33557 65 38 in in ADP bracis-33557 65 39 the the DET bracis-33557 65 40 dataset dataset NOUN bracis-33557 65 41 \(\mathcal \(\mathcal ADJ bracis-33557 65 42 d\ d\ NOUN bracis-33557 65 43 ) ) PUNCT bracis-33557 65 44 . . PUNCT bracis-33557 66 1 an an DET bracis-33557 66 2 instance instance NOUN bracis-33557 66 3 will will AUX bracis-33557 66 4 be be AUX bracis-33557 66 5 considered consider VERB bracis-33557 66 6 harder hard ADJ bracis-33557 66 7 to to PART bracis-33557 66 8 classify classify VERB bracis-33557 66 9 when when SCONJ bracis-33557 66 10 the the DET bracis-33557 66 11 value value NOUN bracis-33557 66 12 of of ADP bracis-33557 66 13 kdn\((\textbf{x}_i kdn\((\textbf{x}_i PROPN bracis-33557 66 14 , , PUNCT bracis-33557 66 15 y_i)\ y_i)\ NUM bracis-33557 66 16 ) ) PUNCT bracis-33557 66 17 is be AUX bracis-33557 66 18 higher high ADJ bracis-33557 66 19 . . PUNCT bracis-33557 67 1 values value NOUN bracis-33557 67 2 close close ADJ bracis-33557 67 3 to to PART bracis-33557 67 4 1 1 NUM bracis-33557 67 5 represent represent VERB bracis-33557 67 6 an an DET bracis-33557 67 7 instance instance NOUN bracis-33557 67 8 surrounded surround VERB bracis-33557 67 9 by by ADP bracis-33557 67 10 examples example NOUN bracis-33557 67 11 from from ADP bracis-33557 67 12 a a DET bracis-33557 67 13 different different ADJ bracis-33557 67 14 class class NOUN bracis-33557 67 15 of of ADP bracis-33557 67 16 itself itself PRON bracis-33557 67 17 . . PUNCT bracis-33557 68 1 this this PRON bracis-33557 68 2 would would AUX bracis-33557 68 3 be be AUX bracis-33557 68 4 \(\textbf{x}_2\ \(\textbf{x}_2\ ADJ bracis-33557 68 5 ) ) PUNCT bracis-33557 68 6 ’s ’s PART bracis-33557 68 7 case case NOUN bracis-33557 68 8 in in ADP bracis-33557 68 9 fig fig NOUN bracis-33557 68 10 . . PUNCT bracis-33557 68 11     SPACE bracis-33557 68 12 1 1 NUM bracis-33557 68 13 when when SCONJ bracis-33557 68 14 labeled label VERB bracis-33557 68 15 red red ADJ bracis-33557 68 16 in in ADP bracis-33557 68 17 \(\mathcal \(\mathcal ADJ bracis-33557 68 18 d\ d\ NOUN bracis-33557 68 19 ) ) PUNCT bracis-33557 68 20 . . PUNCT bracis-33557 69 1 intermediate intermediate ADJ bracis-33557 69 2 values value NOUN bracis-33557 69 3 of of ADP bracis-33557 69 4 kdn\((\textbf{x}_i kdn\((\textbf{x}_i PROPN bracis-33557 69 5 , , PUNCT bracis-33557 69 6 y_i)\ y_i)\ NUM bracis-33557 69 7 ) ) PUNCT bracis-33557 69 8 are be AUX bracis-33557 69 9 found find VERB bracis-33557 69 10 for for ADP bracis-33557 69 11 borderline borderline NOUN bracis-33557 69 12 instances instance NOUN bracis-33557 69 13 . . PUNCT bracis-33557 70 1 easier easy ADJ bracis-33557 70 2 instances instance NOUN bracis-33557 70 3 are be AUX bracis-33557 70 4 those those PRON bracis-33557 70 5 surrounded surround VERB bracis-33557 70 6 by by ADP bracis-33557 70 7 elements element NOUN bracis-33557 70 8 sharing share VERB bracis-33557 70 9 their their PRON bracis-33557 70 10 class class NOUN bracis-33557 70 11 label label NOUN bracis-33557 70 12 , , PUNCT bracis-33557 70 13 which which PRON bracis-33557 70 14 would would AUX bracis-33557 70 15 correspond correspond VERB bracis-33557 70 16 to to PART bracis-33557 70 17 \(\textbf{x}_2\ \(\textbf{x}_2\ VERB bracis-33557 70 18 ) ) PUNCT bracis-33557 70 19 when when SCONJ bracis-33557 70 20 it it PRON bracis-33557 70 21 has have VERB bracis-33557 70 22 a a DET bracis-33557 70 23 blue blue ADJ bracis-33557 70 24 label label NOUN bracis-33557 70 25 . . PUNCT bracis-33557 71 1 in in ADP bracis-33557 71 2 the the DET bracis-33557 71 3 absence absence NOUN bracis-33557 71 4 of of ADP bracis-33557 71 5 an an DET bracis-33557 71 6 instance instance NOUN bracis-33557 71 7 ’s ’s PART bracis-33557 71 8 label label NOUN bracis-33557 71 9 , , PUNCT bracis-33557 71 10 an an DET bracis-33557 71 11 alternative alternative ADJ bracis-33557 71 12 way way NOUN bracis-33557 71 13 to to PART bracis-33557 71 14 measure measure VERB bracis-33557 71 15 the the DET bracis-33557 71 16 mixture mixture NOUN bracis-33557 71 17 of of ADP bracis-33557 71 18 classes class NOUN bracis-33557 71 19 in in ADP bracis-33557 71 20 its its PRON bracis-33557 71 21 neighbourhood neighbourhood NOUN bracis-33557 71 22 is be AUX bracis-33557 71 23 to to PART bracis-33557 71 24 compute compute VERB bracis-33557 71 25 an an DET bracis-33557 71 26 entropy entropy NOUN bracis-33557 71 27 measure measure NOUN bracis-33557 71 28 . . PUNCT bracis-33557 72 1 specifically specifically ADV bracis-33557 72 2 , , PUNCT bracis-33557 72 3 the the DET bracis-33557 72 4 entropy entropy NOUN bracis-33557 72 5 is be AUX bracis-33557 72 6 computed compute VERB bracis-33557 72 7 based base VERB bracis-33557 72 8 on on ADP bracis-33557 72 9 the the DET bracis-33557 72 10 proportion proportion NOUN bracis-33557 72 11 of of ADP bracis-33557 72 12 the the DET bracis-33557 72 13 classes class NOUN bracis-33557 72 14 found find VERB bracis-33557 72 15 in in ADP bracis-33557 72 16 the the DET bracis-33557 72 17 instance instance NOUN bracis-33557 72 18 ’s ’s PART bracis-33557 72 19 neighbourhood neighbourhood NOUN bracis-33557 72 20 . . PUNCT bracis-33557 73 1 higher high ADJ bracis-33557 73 2 entropy entropy PROPN bracis-33557 73 3 values value NOUN bracis-33557 73 4 represent represent VERB bracis-33557 73 5 the the DET bracis-33557 73 6 new new ADJ bracis-33557 73 7 instance instance NOUN bracis-33557 73 8 is be AUX bracis-33557 73 9 in in ADP bracis-33557 73 10 regions region NOUN bracis-33557 73 11 from from ADP bracis-33557 73 12 \(\mathcal \(\mathcal ADJ bracis-33557 73 13 d\ d\ NOUN bracis-33557 73 14 ) ) PUNCT bracis-33557 73 15 near near ADP bracis-33557 73 16 elements element NOUN bracis-33557 73 17 from from ADP bracis-33557 73 18 different different ADJ bracis-33557 73 19 classes class NOUN bracis-33557 73 20 . . PUNCT bracis-33557 74 1 this this PRON bracis-33557 74 2 corresponds correspond VERB bracis-33557 74 3 to to ADP bracis-33557 74 4 the the DET bracis-33557 74 5 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 74 6 ) ) PUNCT bracis-33557 74 7 case case NOUN bracis-33557 74 8 in in ADP bracis-33557 74 9 fig fig NOUN bracis-33557 74 10 . . PUNCT bracis-33557 74 11     SPACE bracis-33557 75 1 1 1 X bracis-33557 75 2 . . PUNCT bracis-33557 76 1 in in ADP bracis-33557 76 2 contrast contrast NOUN bracis-33557 76 3 , , PUNCT bracis-33557 76 4 \(\textbf{x}_2\ \(\textbf{x}_2\ PROPN bracis-33557 76 5 ) ) PUNCT bracis-33557 76 6 will will AUX bracis-33557 76 7 be be AUX bracis-33557 76 8 regarded regard VERB bracis-33557 76 9 as as ADP bracis-33557 76 10 easy easy ADJ bracis-33557 76 11 to to PART bracis-33557 76 12 predict predict VERB bracis-33557 76 13 , , PUNCT bracis-33557 76 14 as as SCONJ bracis-33557 76 15 it it PRON bracis-33557 76 16 is be AUX bracis-33557 76 17 surrounded surround VERB bracis-33557 76 18 by by ADP bracis-33557 76 19 elements element NOUN bracis-33557 76 20 of of ADP bracis-33557 76 21 the the DET bracis-33557 76 22 blue blue ADJ bracis-33557 76 23 class class NOUN bracis-33557 76 24 . . PUNCT bracis-33557 77 1 $ $ SYM bracis-33557 77 2 $ $ SYM bracis-33557 77 3 \begin{aligned \begin{aligne VERB bracis-33557 77 4 } } PUNCT bracis-33557 77 5 \text \text ADP bracis-33557 77 6 { { PUNCT bracis-33557 77 7 kdn}_{\text kdn}_{\text X bracis-33557 77 8 { { PUNCT bracis-33557 77 9 adj}}(\textbf{x adj}}(\textbf{x VERB bracis-33557 77 10 } } PUNCT bracis-33557 77 11 ) ) PUNCT bracis-33557 78 1 = = PUNCT bracis-33557 78 2 -\sum -\sum NOUN bracis-33557 78 3 _ _ PUNCT bracis-33557 78 4 { { PUNCT bracis-33557 78 5 i=1}^{c i=1}^{c ADJ bracis-33557 78 6 } } PUNCT bracis-33557 78 7 p(y_j p(y_j NOUN bracis-33557 78 8 = = PUNCT bracis-33557 78 9 c_i c_i PRON bracis-33557 78 10 ) ) PUNCT bracis-33557 78 11 \log \log PROPN bracis-33557 78 12 p(y_j p(y_j PROPN bracis-33557 78 13 = = PUNCT bracis-33557 78 14 c_i c_i NUM bracis-33557 78 15 ) ) PUNCT bracis-33557 78 16 , , PUNCT bracis-33557 78 17 \text \text ADP bracis-33557 78 18 { { PUNCT bracis-33557 78 19 for for ADP bracis-33557 78 20 } } PUNCT bracis-33557 78 21 \textbf{x}_j \textbf{x}_j NOUN bracis-33557 78 22 \in \in PROPN bracis-33557 78 23 \text \text PROPN bracis-33557 78 24 { { PUNCT bracis-33557 78 25 knn}(\textbf{x knn}(\textbf{x NOUN bracis-33557 78 26 } } PUNCT bracis-33557 78 27 ) ) PUNCT bracis-33557 78 28 , , PUNCT bracis-33557 78 29 \end{aligned}$$ \end{aligned}$$ X bracis-33557 78 30 ( ( PUNCT bracis-33557 78 31 2 2 NUM bracis-33557 78 32 ) ) PUNCT bracis-33557 78 33 where where SCONJ bracis-33557 78 34 \(p(y_j \(p(y_j X bracis-33557 78 35 = = PUNCT bracis-33557 78 36 c_i)\ c_i)\ X bracis-33557 78 37 ) ) PUNCT bracis-33557 78 38 are be AUX bracis-33557 78 39 the the DET bracis-33557 78 40 proportions proportion NOUN bracis-33557 78 41 of of ADP bracis-33557 78 42 the the DET bracis-33557 78 43 classes class NOUN bracis-33557 78 44 of of ADP bracis-33557 78 45 the the DET bracis-33557 78 46 k k NOUN bracis-33557 78 47 - - PUNCT bracis-33557 78 48 nearest near ADJ bracis-33557 78 49 neighbours neighbour NOUN bracis-33557 78 50 of of ADP bracis-33557 78 51 \(\textbf{x}\ \(\textbf{x}\ NOUN bracis-33557 78 52 ) ) PUNCT bracis-33557 78 53 in in ADP bracis-33557 78 54 the the DET bracis-33557 78 55 dataset dataset NOUN bracis-33557 78 56 \(\mathcal \(\mathcal ADJ bracis-33557 78 57 d\ d\ NOUN bracis-33557 78 58 ) ) PUNCT bracis-33557 78 59 . . PUNCT bracis-33557 79 1 ratio ratio NOUN bracis-33557 79 2 of of ADP bracis-33557 79 3 the the DET bracis-33557 79 4 intra intra ADJ bracis-33557 79 5 - - ADJ bracis-33557 79 6 class class ADJ bracis-33557 79 7 and and CCONJ bracis-33557 79 8 extra extra ADJ bracis-33557 79 9 - - ADJ bracis-33557 79 10 class class ADJ bracis-33557 79 11 distances distance NOUN bracis-33557 79 12 . . PUNCT bracis-33557 80 1 n2\(_{\text n2\(_{\text X bracis-33557 80 2 { { PUNCT bracis-33557 80 3 ihm}}\ ihm}}\ PROPN bracis-33557 80 4 ): ): PUNCT bracis-33557 80 5 the the DET bracis-33557 80 6 original original ADJ bracis-33557 80 7 measure measure NOUN bracis-33557 80 8 takes take VERB bracis-33557 80 9 the the DET bracis-33557 80 10 complement complement NOUN bracis-33557 80 11 of of ADP bracis-33557 80 12 the the DET bracis-33557 80 13 ratio ratio NOUN bracis-33557 80 14 of of ADP bracis-33557 80 15 the the DET bracis-33557 80 16 distance distance NOUN bracis-33557 80 17 of of ADP bracis-33557 80 18 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-33557 80 19 ) ) PUNCT bracis-33557 80 20 to to ADP bracis-33557 80 21 the the DET bracis-33557 80 22 nearest near ADJ bracis-33557 80 23 example example NOUN bracis-33557 80 24 from from ADP bracis-33557 80 25 its its PRON bracis-33557 80 26 class class NOUN bracis-33557 80 27 in in ADP bracis-33557 80 28 \(\mathcal \(\mathcal ADJ bracis-33557 80 29 d\ d\ NOUN bracis-33557 80 30 ) ) PUNCT bracis-33557 80 31 to to ADP bracis-33557 80 32 the the DET bracis-33557 80 33 distance distance NOUN bracis-33557 80 34 it it PRON bracis-33557 80 35 has have VERB bracis-33557 80 36 to to ADP bracis-33557 80 37 the the DET bracis-33557 80 38 nearest near ADJ bracis-33557 80 39 instance instance NOUN bracis-33557 80 40 from from ADP bracis-33557 80 41 a a DET bracis-33557 80 42 different different ADJ bracis-33557 80 43 class class NOUN bracis-33557 80 44 ( ( PUNCT bracis-33557 80 45 nearest near ADJ bracis-33557 80 46 enemy enemy NOUN bracis-33557 80 47 ) ) PUNCT bracis-33557 80 48 in in ADP bracis-33557 80 49 \(\mathcal \(\mathcal ADJ bracis-33557 80 50 d\ d\ NOUN bracis-33557 80 51 ) ) PUNCT bracis-33557 80 52 with with ADP bracis-33557 80 53 a a DET bracis-33557 80 54 normalization normalization NOUN bracis-33557 80 55 as as SCONJ bracis-33557 80 56 presented present VERB bracis-33557 80 57 next next ADV bracis-33557 80 58 : : PUNCT bracis-33557 80 59 $ $ SYM bracis-33557 80 60 $ $ SYM bracis-33557 80 61 \begin{aligned \begin{aligne VERB bracis-33557 80 62 } } PUNCT bracis-33557 80 63 \text \text ADP bracis-33557 80 64 { { PUNCT bracis-33557 80 65 n2}_{\text n2}_{\text X bracis-33557 80 66 { { PUNCT bracis-33557 80 67 ihm}}(\textbf{x}_i ihm}}(\textbf{x}_i PROPN bracis-33557 80 68 , , PUNCT bracis-33557 80 69 y_i y_i NUM bracis-33557 80 70 ) ) PUNCT bracis-33557 80 71 = = NOUN bracis-33557 80 72 1 1 NUM bracis-33557 80 73 \frac{1}{\textrm{intrainter}(\textbf{x}_i)+1 \frac{1}{\textrm{intrainter}(\textbf{x}_i)+1 NUM bracis-33557 80 74 } } PUNCT bracis-33557 80 75 , , PUNCT bracis-33557 80 76 \end{aligned}$$ \end{aligned}$$ X bracis-33557 80 77 ( ( PUNCT bracis-33557 80 78 3 3 NUM bracis-33557 80 79 ) ) PUNCT bracis-33557 80 80 where where SCONJ bracis-33557 80 81 : : PUNCT bracis-33557 80 82 $ $ SYM bracis-33557 80 83 $ $ SYM bracis-33557 80 84 \begin{aligned \begin{aligne VERB bracis-33557 80 85 } } PUNCT bracis-33557 80 86 \textrm{intrainter}(\textbf{x}_i \textrm{intrainter}(\textbf{x}_i PROPN bracis-33557 80 87 , , PUNCT bracis-33557 80 88 y_i y_i NUM bracis-33557 80 89 ) ) PUNCT bracis-33557 80 90 = = SYM bracis-33557 80 91 \frac{d(\textbf{x}_i,\text \frac{d(\textbf{x}_i,\text PROPN bracis-33557 80 92 { { PUNCT bracis-33557 80 93 nn}(\textbf{x}_i\in nn}(\textbf{x}_i\in X bracis-33557 80 94 y_i y_i NOUN bracis-33557 80 95 ) ) PUNCT bracis-33557 80 96 ) ) PUNCT bracis-33557 80 97 } } PUNCT bracis-33557 80 98 { { PUNCT bracis-33557 80 99 d(\textbf{x}_i,\text d(\textbf{x}_i,\text PROPN bracis-33557 80 100 { { PUNCT bracis-33557 80 101 ne}(\textbf{x}_i ne}(\textbf{x}_i NOUN bracis-33557 80 102 ) ) PUNCT bracis-33557 80 103 ) ) PUNCT bracis-33557 80 104 } } PUNCT bracis-33557 80 105 , , PUNCT bracis-33557 80 106 \end{aligned}$$ \end{aligned}$$ X bracis-33557 80 107 ( ( PUNCT bracis-33557 80 108 4 4 NUM bracis-33557 80 109 ) ) PUNCT bracis-33557 80 110 where where SCONJ bracis-33557 80 111 d d NOUN bracis-33557 80 112 is be AUX bracis-33557 80 113 a a DET bracis-33557 80 114 distance distance NOUN bracis-33557 80 115 function function NOUN bracis-33557 80 116 , , PUNCT bracis-33557 80 117 nn\((\textbf{x}_i\in nn\((\textbf{x}_i\in NOUN bracis-33557 80 118 y_i)\ y_i)\ NUM bracis-33557 80 119 ) ) PUNCT bracis-33557 80 120 represents represent VERB bracis-33557 80 121 the the DET bracis-33557 80 122 nearest near ADJ bracis-33557 80 123 neighbor neighbor NOUN bracis-33557 80 124 of of ADP bracis-33557 80 125 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 80 126 ) ) PUNCT bracis-33557 80 127 from from ADP bracis-33557 80 128 its its PRON bracis-33557 80 129 class class NOUN bracis-33557 80 130 and and CCONJ bracis-33557 80 131 ne\((\textbf{x}_i)\ ne\((\textbf{x}_i)\ PROPN bracis-33557 80 132 ) ) PUNCT bracis-33557 80 133 is be AUX bracis-33557 80 134 the the DET bracis-33557 80 135 nearest near ADJ bracis-33557 80 136 enemy enemy NOUN bracis-33557 80 137 of of ADP bracis-33557 80 138 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 80 139 ) ) PUNCT bracis-33557 80 140 ( ( PUNCT bracis-33557 80 141 ne\((\textbf{x}_i ne\((\textbf{x}_i NOUN bracis-33557 80 142 ) ) PUNCT bracis-33557 80 143 = = PUNCT bracis-33557 81 1 \text \text NOUN bracis-33557 81 2 { { PUNCT bracis-33557 81 3 nn}(\textbf{x}_i nn}(\textbf{x}_i PART bracis-33557 81 4 \in \in PROPN bracis-33557 81 5 y_j y_j SYM bracis-33557 81 6 \ne \ne PROPN bracis-33557 81 7 y_i\ y_i\ NOUN bracis-33557 81 8 ) ) PUNCT bracis-33557 81 9 ) ) PUNCT bracis-33557 81 10 ) ) PUNCT bracis-33557 81 11 . . PUNCT bracis-33557 82 1 in in ADP bracis-33557 82 2 this this DET bracis-33557 82 3 formulation formulation NOUN bracis-33557 82 4 , , PUNCT bracis-33557 82 5 when when SCONJ bracis-33557 82 6 an an DET bracis-33557 82 7 instance instance NOUN bracis-33557 82 8 is be AUX bracis-33557 82 9 closer close ADJ bracis-33557 82 10 to to ADP bracis-33557 82 11 an an DET bracis-33557 82 12 example example NOUN bracis-33557 82 13 from from ADP bracis-33557 82 14 another another DET bracis-33557 82 15 class class NOUN bracis-33557 82 16 than than ADP bracis-33557 82 17 another another PRON bracis-33557 82 18 from from ADP bracis-33557 82 19 its its PRON bracis-33557 82 20 own own ADJ bracis-33557 82 21 class class NOUN bracis-33557 82 22 , , PUNCT bracis-33557 82 23 the the DET bracis-33557 82 24 n2\(_{\text n2\(_{\text X bracis-33557 82 25 { { PUNCT bracis-33557 82 26 ihm}}\ ihm}}\ PROPN bracis-33557 82 27 ) ) PUNCT bracis-33557 82 28 values value NOUN bracis-33557 82 29 will will AUX bracis-33557 82 30 be be AUX bracis-33557 82 31 larger large ADJ bracis-33557 82 32 , , PUNCT bracis-33557 82 33 indicating indicate VERB bracis-33557 82 34 that that SCONJ bracis-33557 82 35 this this DET bracis-33557 82 36 instance instance NOUN bracis-33557 82 37 is be AUX bracis-33557 82 38 harder hard ADJ bracis-33557 82 39 to to PART bracis-33557 82 40 classify classify VERB bracis-33557 82 41 . . PUNCT bracis-33557 83 1 this this PRON bracis-33557 83 2 would would AUX bracis-33557 83 3 correspond correspond VERB bracis-33557 83 4 to to ADP bracis-33557 83 5 the the DET bracis-33557 83 6 case case NOUN bracis-33557 83 7 where where SCONJ bracis-33557 83 8 \(\textbf{x}_2\ \(\textbf{x}_2\ ADP bracis-33557 83 9 ) ) PUNCT bracis-33557 83 10 in in ADP bracis-33557 83 11 fig fig NOUN bracis-33557 83 12 . . PUNCT bracis-33557 83 13     SPACE bracis-33557 84 1 1 1 NUM bracis-33557 84 2 has have VERB bracis-33557 84 3 the the DET bracis-33557 84 4 red red ADJ bracis-33557 84 5 label label NOUN bracis-33557 84 6 . . PUNCT bracis-33557 85 1 the the DET bracis-33557 85 2 alternative alternative ADJ bracis-33557 85 3 measure measure NOUN bracis-33557 85 4 for for ADP bracis-33557 85 5 unlabeled unlabeled ADJ bracis-33557 85 6 instances instance NOUN bracis-33557 85 7 can can AUX bracis-33557 85 8 be be AUX bracis-33557 85 9 obtained obtain VERB bracis-33557 85 10 by by ADP bracis-33557 85 11 taking take VERB bracis-33557 85 12 the the DET bracis-33557 85 13 ratio ratio NOUN bracis-33557 85 14 of of ADP bracis-33557 85 15 the the DET bracis-33557 85 16 minimum minimum ADJ bracis-33557 85 17 distance distance NOUN bracis-33557 85 18 from from ADP bracis-33557 85 19 \(\textbf{x}\ \(\textbf{x}\ NOUN bracis-33557 85 20 ) ) PUNCT bracis-33557 85 21 and and CCONJ bracis-33557 85 22 the the DET bracis-33557 85 23 closest close ADJ bracis-33557 85 24 element element NOUN bracis-33557 85 25 in in ADP bracis-33557 85 26 \(\mathcal \(\mathcal ADJ bracis-33557 85 27 d\ d\ NOUN bracis-33557 85 28 ) ) PUNCT bracis-33557 85 29 , , PUNCT bracis-33557 85 30 denoted denote VERB bracis-33557 85 31 as as ADP bracis-33557 85 32 \(\textbf{x}_j\ \(\textbf{x}_j\ NOUN bracis-33557 85 33 ) ) PUNCT bracis-33557 85 34 in in ADP bracis-33557 85 35 eq eq PROPN bracis-33557 85 36 . . PUNCT bracis-33557 85 37     SPACE bracis-33557 85 38 5 5 NUM bracis-33557 85 39 , , PUNCT bracis-33557 85 40 to to ADP bracis-33557 85 41 the the DET bracis-33557 85 42 distance distance NOUN bracis-33557 85 43 from from ADP bracis-33557 85 44 \(\textbf{x}\ \(\textbf{x}\ NOUN bracis-33557 85 45 ) ) PUNCT bracis-33557 85 46 and and CCONJ bracis-33557 85 47 the the DET bracis-33557 85 48 closest close ADJ bracis-33557 85 49 element element NOUN bracis-33557 85 50 from from ADP bracis-33557 85 51 another another DET bracis-33557 85 52 class class NOUN bracis-33557 85 53 in in ADP bracis-33557 85 54 \(\mathcal \(\mathcal ADJ bracis-33557 85 55 d\ d\ NOUN bracis-33557 85 56 ) ) PUNCT bracis-33557 85 57 , , PUNCT bracis-33557 85 58 that that ADV bracis-33557 85 59 is is ADV bracis-33557 85 60 , , PUNCT bracis-33557 85 61 a a DET bracis-33557 85 62 class class NOUN bracis-33557 85 63 different different ADJ bracis-33557 85 64 from from ADP bracis-33557 85 65 that that PRON bracis-33557 85 66 of of ADP bracis-33557 85 67 \(\textbf{x}_j\ \(\textbf{x}_j\ PROPN bracis-33557 85 68 ) ) PUNCT bracis-33557 85 69 . . PUNCT bracis-33557 86 1 this this DET bracis-33557 86 2 ratio ratio NOUN bracis-33557 86 3 will will AUX bracis-33557 86 4 assume assume VERB bracis-33557 86 5 value value NOUN bracis-33557 86 6 close close ADV bracis-33557 86 7 to to ADP bracis-33557 86 8 1 1 NUM bracis-33557 86 9 when when SCONJ bracis-33557 86 10 the the DET bracis-33557 86 11 instance instance NOUN bracis-33557 86 12 is be AUX bracis-33557 86 13 almost almost ADV bracis-33557 86 14 equally equally ADV bracis-33557 86 15 distant distant ADJ bracis-33557 86 16 from from ADP bracis-33557 86 17 different different ADJ bracis-33557 86 18 classes class NOUN bracis-33557 86 19 . . PUNCT bracis-33557 87 1 this this PRON bracis-33557 87 2 will will AUX bracis-33557 87 3 happen happen VERB bracis-33557 87 4 more more ADV bracis-33557 87 5 probably probably ADV bracis-33557 87 6 for for ADP bracis-33557 87 7 borderline borderline NOUN bracis-33557 87 8 instances instance NOUN bracis-33557 87 9 , , PUNCT bracis-33557 87 10 such such ADJ bracis-33557 87 11 as as ADP bracis-33557 87 12 \(\textbf{x}_1\ \(\textbf{x}_1\ NOUN bracis-33557 87 13 ) ) PUNCT bracis-33557 87 14 in in ADP bracis-33557 87 15 fig fig NOUN bracis-33557 87 16 . . PUNCT bracis-33557 87 17     SPACE bracis-33557 88 1 1 1 NUM bracis-33557 88 2 . . PUNCT bracis-33557 88 3 $ $ SYM bracis-33557 88 4 $ $ SYM bracis-33557 88 5 \begin{aligned \begin{aligne VERB bracis-33557 88 6 } } PUNCT bracis-33557 88 7 \text \text ADP bracis-33557 88 8 { { PUNCT bracis-33557 88 9 n2}_{\text n2}_{\text X bracis-33557 88 10 { { PUNCT bracis-33557 88 11 adj}}(\textbf{x adj}}(\textbf{x VERB bracis-33557 88 12 } } PUNCT bracis-33557 88 13 ) ) PUNCT bracis-33557 89 1 = = SYM bracis-33557 89 2 \frac{\min \frac{\min PROPN bracis-33557 89 3 ( ( PUNCT bracis-33557 89 4 d(\textbf{x},\textbf{x}_j))}{\min d(\textbf{x},\textbf{x}_j))}{\min X bracis-33557 89 5 ( ( PUNCT bracis-33557 89 6 d(\textbf{x},\textbf{x}_k)|y_k d(\textbf{x},\textbf{x}_k)|y_k PROPN bracis-33557 89 7 \ne \ne PROPN bracis-33557 89 8 y_j y_j NUM bracis-33557 89 9 ) ) PUNCT bracis-33557 89 10 } } PUNCT bracis-33557 89 11 \end{aligned}$$ \end{aligned}$$ X bracis-33557 89 12 ( ( PUNCT bracis-33557 89 13 5 5 NUM bracis-33557 89 14 ) ) PUNCT bracis-33557 89 15 2.2 2.2 NUM bracis-33557 89 16 class class NOUN bracis-33557 89 17 likelihood likelihood NOUN bracis-33557 89 18 ihm ihm NOUN bracis-33557 89 19 this this DET bracis-33557 89 20 type type NOUN bracis-33557 89 21 of of ADP bracis-33557 89 22 measure measure NOUN bracis-33557 89 23 captures capture NOUN bracis-33557 89 24 if if SCONJ bracis-33557 89 25 the the DET bracis-33557 89 26 instance instance NOUN bracis-33557 89 27 is be AUX bracis-33557 89 28 well well ADV bracis-33557 89 29 situated situated ADJ bracis-33557 89 30 in in ADP bracis-33557 89 31 its its PRON bracis-33557 89 32 class class NOUN bracis-33557 89 33 , , PUNCT bracis-33557 89 34 considering consider VERB bracis-33557 89 35 the the DET bracis-33557 89 36 general general ADJ bracis-33557 89 37 patterns pattern NOUN bracis-33557 89 38 of of ADP bracis-33557 89 39 this this DET bracis-33557 89 40 class class NOUN bracis-33557 89 41 . . PUNCT bracis-33557 90 1 the the DET bracis-33557 90 2 likelihood likelihood NOUN bracis-33557 90 3 can can AUX bracis-33557 90 4 be be AUX bracis-33557 90 5 estimated estimate VERB bracis-33557 90 6 for for ADP bracis-33557 90 7 that that PRON bracis-33557 90 8 , , PUNCT bracis-33557 90 9 considering consider VERB bracis-33557 90 10 the the DET bracis-33557 90 11 input input NOUN bracis-33557 90 12 features feature NOUN bracis-33557 90 13 are be AUX bracis-33557 90 14 independent independent ADJ bracis-33557 90 15 for for ADP bracis-33557 90 16 simplifying simplify VERB bracis-33557 90 17 the the DET bracis-33557 90 18 computations computation NOUN bracis-33557 90 19 . . PUNCT bracis-33557 91 1 class class NOUN bracis-33557 91 2 likelihood likelihood NOUN bracis-33557 91 3 difference difference NOUN bracis-33557 91 4 ] ] PUNCT bracis-33557 92 1 cld cld PROPN bracis-33557 92 2 : : PUNCT bracis-33557 92 3 the the DET bracis-33557 92 4 original original ADJ bracis-33557 92 5 measure measure NOUN bracis-33557 92 6 takes take VERB bracis-33557 92 7 the the DET bracis-33557 92 8 complement complement NOUN bracis-33557 92 9 of of ADP bracis-33557 92 10 the the DET bracis-33557 92 11 difference difference NOUN bracis-33557 92 12 between between ADP bracis-33557 92 13 the the DET bracis-33557 92 14 likelihood likelihood NOUN bracis-33557 92 15 that that SCONJ bracis-33557 92 16 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-33557 92 17 ) ) PUNCT bracis-33557 92 18 belongs belong VERB bracis-33557 92 19 to to ADP bracis-33557 92 20 its its PRON bracis-33557 92 21 class class NOUN bracis-33557 92 22 \(y_i\ \(y_i\ NOUN bracis-33557 92 23 ) ) PUNCT bracis-33557 92 24 and and CCONJ bracis-33557 92 25 the the DET bracis-33557 92 26 maximum maximum ADJ bracis-33557 92 27 likelihood likelihood NOUN bracis-33557 92 28 it it PRON bracis-33557 92 29 has have VERB bracis-33557 92 30 to to ADP bracis-33557 92 31 any any DET bracis-33557 92 32 other other ADJ bracis-33557 92 33 class class NOUN bracis-33557 92 34 . . PUNCT bracis-33557 93 1 this this DET bracis-33557 93 2 complement complement NOUN bracis-33557 93 3 is be AUX bracis-33557 93 4 taken take VERB bracis-33557 93 5 to to PART bracis-33557 93 6 standardize standardize VERB bracis-33557 93 7 the the DET bracis-33557 93 8 interpretation interpretation NOUN bracis-33557 93 9 of of ADP bracis-33557 93 10 the the DET bracis-33557 93 11 direction direction NOUN bracis-33557 93 12 of of ADP bracis-33557 93 13 hardness hardness NOUN bracis-33557 93 14 since since SCONJ bracis-33557 93 15 the the DET bracis-33557 93 16 confidence confidence NOUN bracis-33557 93 17 of of ADP bracis-33557 93 18 an an DET bracis-33557 93 19 instance instance NOUN bracis-33557 93 20 belongs belong VERB bracis-33557 93 21 to to ADP bracis-33557 93 22 its its PRON bracis-33557 93 23 class class NOUN bracis-33557 93 24 is be AUX bracis-33557 93 25 larger large ADJ bracis-33557 93 26 than than ADP bracis-33557 93 27 that that PRON bracis-33557 93 28 of of ADP bracis-33557 93 29 any any DET bracis-33557 93 30 other other ADJ bracis-33557 93 31 class class NOUN bracis-33557 94 1 [ [ X bracis-33557 94 2 9 9 NUM bracis-33557 94 3 ] ] SYM bracis-33557 94 4 : : PUNCT bracis-33557 94 5 $ $ SYM bracis-33557 94 6 $ $ SYM bracis-33557 94 7 \begin{aligned \begin{aligne VERB bracis-33557 94 8 } } PUNCT bracis-33557 94 9 \text \text ADP bracis-33557 94 10 { { PUNCT bracis-33557 94 11 cld}(\textbf{x}_i cld}(\textbf{x}_i PROPN bracis-33557 94 12 , , PUNCT bracis-33557 94 13 y_i y_i NUM bracis-33557 94 14 ) ) PUNCT bracis-33557 94 15 = = PRON bracis-33557 94 16 \frac{1 \frac{1 PROPN bracis-33557 94 17 -\left -\left PROPN bracis-33557 94 18 ( ( PUNCT bracis-33557 94 19 p(\textbf{x}_i|y_i)p(y_i p(\textbf{x}_i|y_i)p(y_i PROPN bracis-33557 94 20 ) ) PUNCT bracis-33557 94 21 { { PUNCT bracis-33557 94 22 \max \max PROPN bracis-33557 94 23 } } PUNCT bracis-33557 94 24 _ _ PUNCT bracis-33557 94 25 { { PUNCT bracis-33557 94 26 y_j y_j SYM bracis-33557 94 27 \ne \ne PROPN bracis-33557 94 28 y_i}[p(\textbf{x}_i y_i}[p(\textbf{x}_i NOUN bracis-33557 94 29 |y_j)p(y_j)]\right |y_j)p(y_j)]\right NUM bracis-33557 94 30 ) ) PUNCT bracis-33557 94 31 } } PUNCT bracis-33557 94 32 { { PUNCT bracis-33557 94 33 2 2 NUM bracis-33557 94 34 } } PUNCT bracis-33557 94 35 , , PUNCT bracis-33557 94 36 \end{aligned}$$ \end{aligned}$$ X bracis-33557 94 37 ( ( PUNCT bracis-33557 94 38 6 6 NUM bracis-33557 94 39 ) ) PUNCT bracis-33557 94 40 where where SCONJ bracis-33557 94 41 \(p(y_i)\ \(p(y_i)\ NOUN bracis-33557 94 42 ) ) PUNCT bracis-33557 94 43 is be AUX bracis-33557 94 44 the the DET bracis-33557 94 45 prior prior NOUN bracis-33557 94 46 of of ADP bracis-33557 94 47 class class NOUN bracis-33557 94 48 \(y_i\ \(y_i\ NOUN bracis-33557 94 49 ) ) PUNCT bracis-33557 94 50 , , PUNCT bracis-33557 94 51 set set VERB bracis-33557 94 52 as as ADP bracis-33557 94 53 \(\frac{1}{c}\ \(\frac{1}{c}\ ADV bracis-33557 94 54 ) ) PUNCT bracis-33557 94 55 for for ADP bracis-33557 94 56 all all DET bracis-33557 94 57 data datum NOUN bracis-33557 94 58 instances instance NOUN bracis-33557 94 59 . . PUNCT bracis-33557 95 1 \(p(\textbf{x}_i|y_i)\ \(p(\textbf{x}_i|y_i)\ PROPN bracis-33557 95 2 ) ) PUNCT bracis-33557 95 3 represents represent VERB bracis-33557 95 4 the the DET bracis-33557 95 5 likelihood likelihood NOUN bracis-33557 95 6 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 95 7 ) ) PUNCT bracis-33557 95 8 belongs belong VERB bracis-33557 95 9 to to ADP bracis-33557 95 10 class class NOUN bracis-33557 95 11 \(y_i\ \(y_i\ NOUN bracis-33557 95 12 ) ) PUNCT bracis-33557 95 13 and and CCONJ bracis-33557 95 14 it it PRON bracis-33557 95 15 can can AUX bracis-33557 95 16 be be AUX bracis-33557 95 17 estimated estimate VERB bracis-33557 95 18 considering consider VERB bracis-33557 95 19 the the DET bracis-33557 95 20 input input NOUN bracis-33557 95 21 features feature VERB bracis-33557 95 22 independent independent ADJ bracis-33557 95 23 of of ADP bracis-33557 95 24 each each DET bracis-33557 95 25 other other ADJ bracis-33557 95 26 , , PUNCT bracis-33557 95 27 as as ADP bracis-33557 95 28 in in ADP bracis-33557 95 29 naïve naïve ADJ bracis-33557 95 30 bayes bayes NOUN bracis-33557 95 31 classification classification NOUN bracis-33557 95 32 . . PUNCT bracis-33557 96 1 for for ADP bracis-33557 96 2 example example NOUN bracis-33557 96 3 , , PUNCT bracis-33557 96 4 if if SCONJ bracis-33557 96 5 \(\textbf{x}_2\ \(\textbf{x}_2\ ADP bracis-33557 96 6 ) ) PUNCT bracis-33557 96 7 in in ADP bracis-33557 96 8 fig fig NOUN bracis-33557 96 9 . . PUNCT bracis-33557 96 10     SPACE bracis-33557 96 11 1 1 NUM bracis-33557 96 12 is be AUX bracis-33557 96 13 labeled label VERB bracis-33557 96 14 as as ADP bracis-33557 96 15 blue blue ADJ bracis-33557 96 16 in in ADP bracis-33557 96 17 \(\mathcal \(\mathcal ADJ bracis-33557 96 18 d\ d\ PROPN bracis-33557 96 19 ) ) PUNCT bracis-33557 96 20 , , PUNCT bracis-33557 96 21 it it PRON bracis-33557 96 22 will will AUX bracis-33557 96 23 be be AUX bracis-33557 96 24 easy easy ADJ bracis-33557 96 25 according accord VERB bracis-33557 96 26 to to ADP bracis-33557 96 27 this this DET bracis-33557 96 28 measure measure NOUN bracis-33557 96 29 , , PUNCT bracis-33557 96 30 as as SCONJ bracis-33557 96 31 its its PRON bracis-33557 96 32 likelihood likelihood NOUN bracis-33557 96 33 to to ADP bracis-33557 96 34 the the DET bracis-33557 96 35 blue blue ADJ bracis-33557 96 36 class class NOUN bracis-33557 96 37 will will AUX bracis-33557 96 38 be be AUX bracis-33557 96 39 higher high ADJ bracis-33557 96 40 than than ADP bracis-33557 96 41 to to ADP bracis-33557 96 42 the the DET bracis-33557 96 43 red red ADJ bracis-33557 96 44 class class NOUN bracis-33557 96 45 . . PUNCT bracis-33557 97 1 when when SCONJ bracis-33557 97 2 the the DET bracis-33557 97 3 class class NOUN bracis-33557 97 4 of of ADP bracis-33557 97 5 an an DET bracis-33557 97 6 instance instance NOUN bracis-33557 97 7 can can AUX bracis-33557 97 8 not not PART bracis-33557 97 9 be be AUX bracis-33557 97 10 defined define VERB bracis-33557 97 11 in in ADP bracis-33557 97 12 advance advance NOUN bracis-33557 97 13 , , PUNCT bracis-33557 97 14 the the DET bracis-33557 97 15 hardness hardness NOUN bracis-33557 97 16 measure measure NOUN bracis-33557 97 17 can can AUX bracis-33557 97 18 be be AUX bracis-33557 97 19 estimated estimate VERB bracis-33557 97 20 by by ADP bracis-33557 97 21 the the DET bracis-33557 97 22 difference difference NOUN bracis-33557 97 23 between between ADP bracis-33557 97 24 the the DET bracis-33557 97 25 two two NUM bracis-33557 97 26 higher high ADJ bracis-33557 97 27 likelihoods likelihood NOUN bracis-33557 97 28 of of ADP bracis-33557 97 29 all all DET bracis-33557 97 30 possible possible ADJ bracis-33557 97 31 classes class NOUN bracis-33557 97 32 in in ADP bracis-33557 97 33 the the DET bracis-33557 97 34 dataset dataset NOUN bracis-33557 97 35 . . PUNCT bracis-33557 98 1 like like INTJ bracis-33557 98 2 in in ADP bracis-33557 98 3 the the DET bracis-33557 98 4 original original ADJ bracis-33557 98 5 measure measure NOUN bracis-33557 98 6 , , PUNCT bracis-33557 98 7 the the DET bracis-33557 98 8 complement complement NOUN bracis-33557 98 9 of of ADP bracis-33557 98 10 the the DET bracis-33557 98 11 difference difference NOUN bracis-33557 98 12 is be AUX bracis-33557 98 13 taken take VERB bracis-33557 98 14 to to PART bracis-33557 98 15 keep keep VERB bracis-33557 98 16 the the DET bracis-33557 98 17 interpretation interpretation NOUN bracis-33557 98 18 that that SCONJ bracis-33557 98 19 higher high ADJ bracis-33557 98 20 values value NOUN bracis-33557 98 21 are be AUX bracis-33557 98 22 found find VERB bracis-33557 98 23 for for ADP bracis-33557 98 24 instances instance NOUN bracis-33557 98 25 harder hard ADV bracis-33557 98 26 to to PART bracis-33557 98 27 classify classify VERB bracis-33557 98 28 . . PUNCT bracis-33557 99 1 the the DET bracis-33557 99 2 values value NOUN bracis-33557 99 3 of of ADP bracis-33557 99 4 this this DET bracis-33557 99 5 measure measure NOUN bracis-33557 99 6 will will AUX bracis-33557 99 7 tend tend VERB bracis-33557 99 8 to to PART bracis-33557 99 9 be be AUX bracis-33557 99 10 higher high ADJ bracis-33557 99 11 for for ADP bracis-33557 99 12 borderline borderline NOUN bracis-33557 99 13 instances instance NOUN bracis-33557 99 14 since since SCONJ bracis-33557 99 15 their their PRON bracis-33557 99 16 likelihood likelihood NOUN bracis-33557 99 17 of of ADP bracis-33557 99 18 being be AUX bracis-33557 99 19 in in ADP bracis-33557 99 20 different different ADJ bracis-33557 99 21 classes class NOUN bracis-33557 99 22 will will AUX bracis-33557 99 23 be be AUX bracis-33557 99 24 similar similar ADJ bracis-33557 99 25 . . PUNCT bracis-33557 100 1 $ $ SYM bracis-33557 100 2 $ $ SYM bracis-33557 100 3 \begin{aligned \begin{aligne VERB bracis-33557 100 4 } } PUNCT bracis-33557 100 5 \text \text ADJ bracis-33557 100 6 { { PUNCT bracis-33557 100 7 cld}_{\text cld}_{\text NOUN bracis-33557 100 8 { { PUNCT bracis-33557 100 9 adj}}(\textbf{x adj}}(\textbf{x VERB bracis-33557 100 10 } } PUNCT bracis-33557 100 11 ) ) PUNCT bracis-33557 101 1 = = PRON bracis-33557 102 1 \frac{1 \frac{1 PROPN bracis-33557 102 2 -\left -\left PROPN bracis-33557 102 3 ( ( PUNCT bracis-33557 102 4 { { PUNCT bracis-33557 102 5 \max \max PROPN bracis-33557 102 6 } } PUNCT bracis-33557 102 7 _ _ PUNCT bracis-33557 102 8 { { PUNCT bracis-33557 102 9 y_i}[p(\textbf{x}|y_i)p(y_i y_i}[p(\textbf{x}|y_i)p(y_i NOUN bracis-33557 102 10 ) ) PUNCT bracis-33557 102 11 ] ] PUNCT bracis-33557 102 12 { { PUNCT bracis-33557 102 13 \max \max PROPN bracis-33557 102 14 } } PUNCT bracis-33557 102 15 _ _ PUNCT bracis-33557 102 16 { { PUNCT bracis-33557 102 17 y_j y_j SYM bracis-33557 102 18 \ne \ne PROPN bracis-33557 102 19 y_i}[p(\textbf{x}|y_j)p(y_j)]\right y_i}[p(\textbf{x}|y_j)p(y_j)]\right NUM bracis-33557 102 20 ) ) PUNCT bracis-33557 102 21 } } PUNCT bracis-33557 102 22 { { PUNCT bracis-33557 102 23 2 2 NUM bracis-33557 102 24 } } PUNCT bracis-33557 102 25 . . PUNCT bracis-33557 103 1 \end{aligned}$$ \end{aligned}$$ PROPN bracis-33557 103 2 ( ( PUNCT bracis-33557 103 3 7 7 NUM bracis-33557 103 4 ) ) PUNCT bracis-33557 103 5 2.3 2.3 NUM bracis-33557 103 6 tree tree NOUN bracis-33557 103 7 - - PUNCT bracis-33557 103 8 based base VERB bracis-33557 103 9 ihm ihm ADJ bracis-33557 103 10 decision decision NOUN bracis-33557 103 11 trees tree NOUN bracis-33557 103 12 ( ( PUNCT bracis-33557 103 13 dts dts NOUN bracis-33557 103 14 ) ) PUNCT bracis-33557 103 15 can can AUX bracis-33557 103 16 be be AUX bracis-33557 103 17 used use VERB bracis-33557 103 18 to to PART bracis-33557 103 19 estimate estimate VERB bracis-33557 103 20 the the DET bracis-33557 103 21 hardness hardness ADJ bracis-33557 103 22 level level NOUN bracis-33557 103 23 of of ADP bracis-33557 103 24 an an DET bracis-33557 103 25 instance instance NOUN bracis-33557 103 26 based base VERB bracis-33557 103 27 on on ADP bracis-33557 103 28 the the DET bracis-33557 103 29 number number NOUN bracis-33557 103 30 of of ADP bracis-33557 103 31 splits split NOUN bracis-33557 103 32 necessary necessary ADJ bracis-33557 103 33 to to PART bracis-33557 103 34 classify classify VERB bracis-33557 103 35 it it PRON bracis-33557 103 36 . . PUNCT bracis-33557 104 1 if if SCONJ bracis-33557 104 2 many many ADJ bracis-33557 104 3 splits split NOUN bracis-33557 104 4 are be AUX bracis-33557 104 5 required require VERB bracis-33557 104 6 , , PUNCT bracis-33557 104 7 the the DET bracis-33557 104 8 instance instance NOUN bracis-33557 104 9 ’s ’s PART bracis-33557 104 10 classification classification NOUN bracis-33557 104 11 will will AUX bracis-33557 104 12 be be AUX bracis-33557 104 13 harder hard ADJ bracis-33557 104 14 . . PUNCT bracis-33557 105 1 the the DET bracis-33557 105 2 dt dt NOUN bracis-33557 105 3 is be AUX bracis-33557 105 4 built build VERB bracis-33557 105 5 based base VERB bracis-33557 105 6 on on ADP bracis-33557 105 7 the the DET bracis-33557 105 8 labeled label VERB bracis-33557 105 9 dataset dataset VERB bracis-33557 105 10 \(\mathcal \(\mathcal ADJ bracis-33557 105 11 d\ d\ NOUN bracis-33557 105 12 ) ) PUNCT bracis-33557 105 13 . . PUNCT bracis-33557 106 1 unlabeled unlabele VERB bracis-33557 106 2 instances instance NOUN bracis-33557 106 3 are be AUX bracis-33557 106 4 input input NOUN bracis-33557 106 5 to to ADP bracis-33557 106 6 the the DET bracis-33557 106 7 built build VERB bracis-33557 106 8 dt dt PROPN bracis-33557 106 9 , , PUNCT bracis-33557 106 10 and and CCONJ bracis-33557 106 11 the the DET bracis-33557 106 12 measure measure NOUN bracis-33557 106 13 can can AUX bracis-33557 106 14 be be AUX bracis-33557 106 15 computed compute VERB bracis-33557 106 16 based base VERB bracis-33557 106 17 on on ADP bracis-33557 106 18 where where SCONJ bracis-33557 106 19 it it PRON bracis-33557 106 20 is be AUX bracis-33557 106 21 classified classified ADJ bracis-33557 106 22 . . PUNCT bracis-33557 107 1 disjunct disjunct ADJ bracis-33557 107 2 class class NOUN bracis-33557 107 3 percentage percentage NOUN bracis-33557 107 4 dcp dcp NOUN bracis-33557 107 5 : : PUNCT bracis-33557 107 6 from from ADP bracis-33557 107 7 a a DET bracis-33557 107 8 pruned prune VERB bracis-33557 107 9 decision decision NOUN bracis-33557 107 10 tree tree NOUN bracis-33557 107 11 ( ( PUNCT bracis-33557 107 12 dt dt NOUN bracis-33557 107 13 ) ) PUNCT bracis-33557 107 14 using use VERB bracis-33557 107 15 \(\mathcal \(\mathcal ADJ bracis-33557 107 16 d\ d\ NOUN bracis-33557 107 17 ) ) PUNCT bracis-33557 107 18 , , PUNCT bracis-33557 107 19 the the DET bracis-33557 107 20 leaf leaf NOUN bracis-33557 107 21 node node NOUN bracis-33557 107 22 where where SCONJ bracis-33557 107 23 the the DET bracis-33557 107 24 instance instance NOUN bracis-33557 107 25 is be AUX bracis-33557 107 26 classified classified ADJ bracis-33557 107 27 is be AUX bracis-33557 107 28 considered consider VERB bracis-33557 107 29 the the DET bracis-33557 107 30 disjunct disjunct NOUN bracis-33557 107 31 of of ADP bracis-33557 107 32 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 107 33 ) ) PUNCT bracis-33557 107 34 . . PUNCT bracis-33557 108 1 the the DET bracis-33557 108 2 complement complement NOUN bracis-33557 108 3 of of ADP bracis-33557 108 4 the the DET bracis-33557 108 5 percentage percentage NOUN bracis-33557 108 6 of of ADP bracis-33557 108 7 instances instance NOUN bracis-33557 108 8 in in ADP bracis-33557 108 9 this this DET bracis-33557 108 10 disjunct disjunct NOUN bracis-33557 108 11 that that PRON bracis-33557 108 12 shares share VERB bracis-33557 108 13 the the DET bracis-33557 108 14 same same ADJ bracis-33557 108 15 label label NOUN bracis-33557 108 16 as as ADP bracis-33557 108 17 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-33557 108 18 ) ) PUNCT bracis-33557 108 19 gives give VERB bracis-33557 108 20 the the DET bracis-33557 108 21 original original ADJ bracis-33557 108 22 dcp dcp NOUN bracis-33557 108 23 measure measure NOUN bracis-33557 108 24 : : PUNCT bracis-33557 109 1 $ $ SYM bracis-33557 109 2 $ $ SYM bracis-33557 109 3 \begin{aligned \begin{aligne VERB bracis-33557 109 4 } } PUNCT bracis-33557 109 5 \text \text ADP bracis-33557 109 6 { { PUNCT bracis-33557 109 7 dcp}(\textbf{x}_i dcp}(\textbf{x}_i NOUN bracis-33557 109 8 , , PUNCT bracis-33557 109 9 y_i y_i NUM bracis-33557 109 10 ) ) PUNCT bracis-33557 109 11 = = NOUN bracis-33557 109 12 1\frac{\sharp 1\frac{\sharp NUM bracis-33557 109 13 \{\textbf{x}_j \{\textbf{x}_j NOUN bracis-33557 109 14 | | ADV bracis-33557 109 15 \textbf{x}_j \textbf{x}_j NOUN bracis-33557 109 16 \in \in PROPN bracis-33557 109 17 \text \text PROPN bracis-33557 109 18 { { PUNCT bracis-33557 109 19 disjunct}(\textbf{x}_i disjunct}(\textbf{x}_i PROPN bracis-33557 109 20 ) ) PUNCT bracis-33557 109 21 \wedge \wedge PROPN bracis-33557 109 22 y_j y_j PUNCT bracis-33557 109 23 = = SYM bracis-33557 109 24 y_i\}}{\sharp y_i\}}{\sharp NUM bracis-33557 109 25 \{\textbf{x}_j|\textbf{x}_j \{\textbf{x}_j|\textbf{x}_j PRON bracis-33557 109 26 \in \in PROPN bracis-33557 109 27 \text \text PROPN bracis-33557 109 28 { { PUNCT bracis-33557 109 29 disjunct}(\textbf{x}_i)\ disjunct}(\textbf{x}_i)\ PROPN bracis-33557 109 30 } } PUNCT bracis-33557 109 31 } } PUNCT bracis-33557 109 32 , , PUNCT bracis-33557 109 33 \end{aligned}$$ \end{aligned}$$ X bracis-33557 109 34 ( ( PUNCT bracis-33557 109 35 8) 8) NUM bracis-33557 109 36 where where SCONJ bracis-33557 109 37 disjunct\((\textbf{x}_i)\ disjunct\((\textbf{x}_i)\ NOUN bracis-33557 109 38 ) ) PUNCT bracis-33557 109 39 represents represent VERB bracis-33557 109 40 the the DET bracis-33557 109 41 instances instance NOUN bracis-33557 109 42 contained contain VERB bracis-33557 109 43 in in ADP bracis-33557 109 44 the the DET bracis-33557 109 45 disjunct disjunct ADJ bracis-33557 109 46 ( ( PUNCT bracis-33557 109 47 leaf leaf NOUN bracis-33557 109 48 node node NOUN bracis-33557 109 49 ) ) PUNCT bracis-33557 109 50 where where SCONJ bracis-33557 109 51 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-33557 109 52 ) ) PUNCT bracis-33557 109 53 is be AUX bracis-33557 109 54 placed place VERB bracis-33557 109 55 . . PUNCT bracis-33557 110 1 for for ADP bracis-33557 110 2 easy easy ADJ bracis-33557 110 3 instances instance NOUN bracis-33557 110 4 , , PUNCT bracis-33557 110 5 according accord VERB bracis-33557 110 6 to to ADP bracis-33557 110 7 this this DET bracis-33557 110 8 measure measure NOUN bracis-33557 110 9 , , PUNCT bracis-33557 110 10 larger large ADJ bracis-33557 110 11 percentages percentage NOUN bracis-33557 110 12 of of ADP bracis-33557 110 13 examples example NOUN bracis-33557 110 14 sharing share VERB bracis-33557 110 15 the the DET bracis-33557 110 16 same same ADJ bracis-33557 110 17 label label NOUN bracis-33557 110 18 as as SCONJ bracis-33557 110 19 the the DET bracis-33557 110 20 instance instance NOUN bracis-33557 110 21 will will AUX bracis-33557 110 22 be be AUX bracis-33557 110 23 found find VERB bracis-33557 110 24 in in ADP bracis-33557 110 25 their their PRON bracis-33557 110 26 disjunct disjunct NOUN bracis-33557 110 27 . . PUNCT bracis-33557 111 1 for for ADP bracis-33557 111 2 example example NOUN bracis-33557 111 3 , , PUNCT bracis-33557 111 4 if if SCONJ bracis-33557 111 5 \(\textbf{x}_2\ \(\textbf{x}_2\ ADP bracis-33557 111 6 ) ) PUNCT bracis-33557 111 7 in in ADP bracis-33557 111 8 fig fig NOUN bracis-33557 111 9 . . PUNCT bracis-33557 111 10     SPACE bracis-33557 111 11 1 1 NUM bracis-33557 111 12 has have VERB bracis-33557 111 13 the the DET bracis-33557 111 14 red red ADJ bracis-33557 111 15 label label NOUN bracis-33557 111 16 in in ADP bracis-33557 111 17 \(\mathcal \(\mathcal ADJ bracis-33557 111 18 d\ d\ NOUN bracis-33557 111 19 ) ) PUNCT bracis-33557 111 20 , , PUNCT bracis-33557 111 21 it it PRON bracis-33557 111 22 will will AUX bracis-33557 111 23 probably probably ADV bracis-33557 111 24 be be AUX bracis-33557 111 25 placed place VERB bracis-33557 111 26 in in ADP bracis-33557 111 27 a a DET bracis-33557 111 28 leaf leaf NOUN bracis-33557 111 29 node node NOUN bracis-33557 111 30 containing contain VERB bracis-33557 111 31 many many ADJ bracis-33557 111 32 elements element NOUN bracis-33557 111 33 of of ADP bracis-33557 111 34 the the DET bracis-33557 111 35 blue blue ADJ bracis-33557 111 36 class class NOUN bracis-33557 111 37 , , PUNCT bracis-33557 111 38 making make VERB bracis-33557 111 39 it it PRON bracis-33557 111 40 harder hard ADV bracis-33557 111 41 to to PART bracis-33557 111 42 classify classify VERB bracis-33557 111 43 according accord VERB bracis-33557 111 44 to to ADP bracis-33557 111 45 the the DET bracis-33557 111 46 interpretation interpretation NOUN bracis-33557 111 47 of of ADP bracis-33557 111 48 this this DET bracis-33557 111 49 measure measure NOUN bracis-33557 111 50 . . PUNCT bracis-33557 112 1 in in ADP bracis-33557 112 2 scenarios scenario NOUN bracis-33557 112 3 where where SCONJ bracis-33557 112 4 the the DET bracis-33557 112 5 instance instance NOUN bracis-33557 112 6 ’s ’s PART bracis-33557 112 7 class class NOUN bracis-33557 112 8 is be AUX bracis-33557 112 9 unknown unknown ADJ bracis-33557 112 10 , , PUNCT bracis-33557 112 11 we we PRON bracis-33557 112 12 take take VERB bracis-33557 112 13 the the DET bracis-33557 112 14 entropy entropy NOUN bracis-33557 112 15 of of ADP bracis-33557 112 16 the the DET bracis-33557 112 17 disjunct disjunct NOUN bracis-33557 112 18 where where SCONJ bracis-33557 112 19 the the DET bracis-33557 112 20 instance instance NOUN bracis-33557 112 21 is be AUX bracis-33557 112 22 placed place VERB bracis-33557 112 23 as as ADP bracis-33557 112 24 a a DET bracis-33557 112 25 hardness hardness ADJ bracis-33557 112 26 measure measure NOUN bracis-33557 112 27 , , PUNCT bracis-33557 112 28 similarly similarly ADV bracis-33557 112 29 to to ADP bracis-33557 112 30 what what PRON bracis-33557 112 31 has have AUX bracis-33557 112 32 been be AUX bracis-33557 112 33 done do VERB bracis-33557 112 34 for for ADP bracis-33557 112 35 kdn kdn NOUN bracis-33557 112 36 . . PUNCT bracis-33557 113 1 $ $ SYM bracis-33557 113 2 $ $ SYM bracis-33557 113 3 \begin{aligned \begin{aligne VERB bracis-33557 113 4 } } PUNCT bracis-33557 113 5 \text \text ADP bracis-33557 113 6 { { PUNCT bracis-33557 113 7 dcp}_{\text dcp}_{\text NOUN bracis-33557 113 8 { { PUNCT bracis-33557 113 9 adj}}(\textbf{x adj}}(\textbf{x VERB bracis-33557 113 10 } } PUNCT bracis-33557 113 11 ) ) PUNCT bracis-33557 113 12 = = SYM bracis-33557 113 13 \sum \sum NOUN bracis-33557 113 14 _ _ PUNCT bracis-33557 113 15 { { PUNCT bracis-33557 113 16 i=1}^{c i=1}^{c ADJ bracis-33557 113 17 } } PUNCT bracis-33557 113 18 p(y_j p(y_j NOUN bracis-33557 113 19 = = PUNCT bracis-33557 113 20 c_i c_i PRON bracis-33557 113 21 ) ) PUNCT bracis-33557 113 22 \log \log PROPN bracis-33557 113 23 p(y_j p(y_j PROPN bracis-33557 113 24 = = PUNCT bracis-33557 113 25 c_i c_i NUM bracis-33557 113 26 ) ) PUNCT bracis-33557 113 27 , , PUNCT bracis-33557 113 28 \text \text ADP bracis-33557 113 29 { { PUNCT bracis-33557 113 30 for for ADP bracis-33557 113 31 } } PUNCT bracis-33557 113 32 \textbf{x}_j \textbf{x}_j NOUN bracis-33557 113 33 \in \in PROPN bracis-33557 113 34 \text \text PROPN bracis-33557 113 35 { { PUNCT bracis-33557 113 36 disjunct}(\textbf{x disjunct}(\textbf{x NOUN bracis-33557 113 37 } } PUNCT bracis-33557 113 38 ) ) PUNCT bracis-33557 113 39 , , PUNCT bracis-33557 113 40 \end{aligned}$$ \end{aligned}$$ X bracis-33557 113 41 ( ( PUNCT bracis-33557 113 42 9 9 NUM bracis-33557 113 43 ) ) PUNCT bracis-33557 113 44 where where SCONJ bracis-33557 113 45 the the DET bracis-33557 113 46 proportions proportion NOUN bracis-33557 113 47 of of ADP bracis-33557 113 48 the the DET bracis-33557 113 49 classes class NOUN bracis-33557 113 50 are be AUX bracis-33557 113 51 taken take VERB bracis-33557 113 52 based base VERB bracis-33557 113 53 on on ADP bracis-33557 113 54 the the DET bracis-33557 113 55 disjunct disjunct ADJ bracis-33557 113 56 where where SCONJ bracis-33557 113 57 \(\textbf{x}\ \(\textbf{x}\ NOUN bracis-33557 113 58 ) ) PUNCT bracis-33557 113 59 is be AUX bracis-33557 113 60 placed place VERB bracis-33557 113 61 in in ADP bracis-33557 113 62 the the DET bracis-33557 113 63 dt dt NOUN bracis-33557 113 64 built build VERB bracis-33557 113 65 using use VERB bracis-33557 113 66 the the DET bracis-33557 113 67 dataset dataset NOUN bracis-33557 113 68 \(\mathcal \(\mathcal ADJ bracis-33557 113 69 { { PUNCT bracis-33557 113 70 d}\ d}\ PROPN bracis-33557 113 71 ) ) PUNCT bracis-33557 113 72 . . PUNCT bracis-33557 114 1 tree tree NOUN bracis-33557 114 2 depth depth NOUN bracis-33557 114 3 td td NOUN bracis-33557 114 4 : : PUNCT bracis-33557 114 5 the the DET bracis-33557 114 6 original original ADJ bracis-33557 114 7 measure measure NOUN bracis-33557 114 8 gives give VERB bracis-33557 114 9 the the DET bracis-33557 114 10 depth depth NOUN bracis-33557 114 11 of of ADP bracis-33557 114 12 the the DET bracis-33557 114 13 leaf leaf NOUN bracis-33557 114 14 node node NOUN bracis-33557 114 15 that that PRON bracis-33557 114 16 classifies classify VERB bracis-33557 114 17 \(\textbf{x}_i\ \(\textbf{x}_i\ NOUN bracis-33557 114 18 ) ) PUNCT bracis-33557 114 19 in in ADP bracis-33557 114 20 a a DET bracis-33557 114 21 dt dt NOUN bracis-33557 114 22 built build VERB bracis-33557 114 23 using use VERB bracis-33557 114 24 all all PRON bracis-33557 114 25 labeled label VERB bracis-33557 114 26 dataset dataset ADJ bracis-33557 114 27 \(\mathcal \(\mathcal ADJ bracis-33557 114 28 d\ d\ NOUN bracis-33557 114 29 ) ) PUNCT bracis-33557 114 30 , , PUNCT bracis-33557 114 31 normalized normalize VERB bracis-33557 114 32 by by ADP bracis-33557 114 33 the the DET bracis-33557 114 34 maximum maximum ADJ bracis-33557 114 35 depth depth NOUN bracis-33557 114 36 of of ADP bracis-33557 114 37 the the DET bracis-33557 114 38 tree tree NOUN bracis-33557 114 39 : : PUNCT bracis-33557 114 40 $ $ SYM bracis-33557 114 41 $ $ SYM bracis-33557 114 42 \begin{aligned \begin{aligne VERB bracis-33557 114 43 } } PUNCT bracis-33557 114 44 \text \text ADV bracis-33557 114 45 { { PUNCT bracis-33557 114 46 td}(\textbf{x}_i td}(\textbf{x}_i PROPN bracis-33557 114 47 , , PUNCT bracis-33557 114 48 y_i y_i NUM bracis-33557 114 49 ) ) PUNCT bracis-33557 114 50 = = NOUN bracis-33557 114 51 \frac{\text \frac{\text X bracis-33557 114 52 { { PUNCT bracis-33557 114 53 depth}_{\text depth}_{\text PROPN bracis-33557 114 54 { { PUNCT bracis-33557 114 55 dt}}(\textbf{x}_i)}{\max dt}}(\textbf{x}_i)}{\max X bracis-33557 114 56 ( ( PUNCT bracis-33557 114 57 \text \text X bracis-33557 114 58 { { PUNCT bracis-33557 114 59 depth}_{\text depth}_{\text PROPN bracis-33557 114 60 { { PUNCT bracis-33557 114 61 dt}}(\textbf{x}_j dt}}(\textbf{x}_j PROPN bracis-33557 114 62 \in \in PROPN bracis-33557 114 63 \mathcal \mathcal PROPN bracis-33557 114 64 d d NOUN bracis-33557 114 65 ) ) PUNCT bracis-33557 114 66 ) ) PUNCT bracis-33557 114 67 } } PUNCT bracis-33557 114 68 , , PUNCT bracis-33557 114 69 \end{aligned}$$ \end{aligned}$$ X bracis-33557 114 70 ( ( PUNCT bracis-33557 114 71 10 10 NUM bracis-33557 114 72 ) ) PUNCT bracis-33557 114 73 where where SCONJ bracis-33557 114 74 \(\text \(\text NOUN bracis-33557 114 75 { { PUNCT bracis-33557 114 76 depth}_{\text depth}_{\text PROPN bracis-33557 114 77 { { PUNCT bracis-33557 114 78 dt}}(\textbf{x}_i)\ dt}}(\textbf{x}_i)\ PROPN bracis-33557 114 79 ) ) PUNCT bracis-33557 114 80 gives give VERB bracis-33557 114 81 the the DET bracis-33557 114 82 depth depth NOUN bracis-33557 114 83 where where SCONJ bracis-33557 114 84 the the DET bracis-33557 114 85 instance instance NOUN bracis-33557 114 86 \(\textbf{x}_i\ \(\textbf{x}_i\ PROPN bracis-33557 114 87 ) ) PUNCT bracis-33557 114 88 is be AUX bracis-33557 114 89 placed place VERB bracis-33557 114 90 in in ADP bracis-33557 114 91 the the DET bracis-33557 114 92 dt dt PROPN bracis-33557 114 93 . . PUNCT bracis-33557 115 1 instances instance NOUN bracis-33557 115 2 harder hard ADV bracis-33557 115 3 to to PART bracis-33557 115 4 classify classify VERB bracis-33557 115 5 tend tend VERB bracis-33557 115 6 to to PART bracis-33557 115 7 be be AUX bracis-33557 115 8 placed place VERB bracis-33557 115 9 at at ADP bracis-33557 115 10 deeper deep ADJ bracis-33557 115 11 levels level NOUN bracis-33557 115 12 of of ADP bracis-33557 115 13 the the DET bracis-33557 115 14 tree tree NOUN bracis-33557 115 15 , , PUNCT bracis-33557 115 16 making make VERB bracis-33557 115 17 td td NOUN bracis-33557 115 18 higher higher ADV bracis-33557 115 19 . . PUNCT bracis-33557 116 1 there there PRON bracis-33557 116 2 are be VERB bracis-33557 116 3 two two NUM bracis-33557 116 4 versions version NOUN bracis-33557 116 5 of of ADP bracis-33557 116 6 this this DET bracis-33557 116 7 measure measure NOUN bracis-33557 116 8 . . PUNCT bracis-33557 117 1 one one NUM bracis-33557 117 2 derives derive VERB bracis-33557 117 3 from from ADP bracis-33557 117 4 a a DET bracis-33557 117 5 pruned prune VERB bracis-33557 117 6 tree tree NOUN bracis-33557 117 7 ( ( PUNCT bracis-33557 117 8 \(\text \(\text NOUN bracis-33557 117 9 { { PUNCT bracis-33557 117 10 td}_{\text td}_{\text NOUN bracis-33557 117 11 { { PUNCT bracis-33557 117 12 p}}\ p}}\ NOUN bracis-33557 117 13 ) ) PUNCT bracis-33557 117 14 ) ) PUNCT bracis-33557 117 15 and and CCONJ bracis-33557 117 16 the the DET bracis-33557 117 17 other other ADJ bracis-33557 117 18 from from ADP bracis-33557 117 19 an an DET bracis-33557 117 20 unpruned unpruned ADJ bracis-33557 117 21 tree tree NOUN bracis-33557 117 22 ( ( PUNCT bracis-33557 117 23 \(\text \(\text NOUN bracis-33557 117 24 { { PUNCT bracis-33557 117 25 td}_{\text td}_{\text NOUN bracis-33557 117 26 { { PUNCT bracis-33557 117 27 u}}\ u}}\ NOUN bracis-33557 117 28 ) ) PUNCT bracis-33557 117 29 ) ) PUNCT bracis-33557 117 30 . . PUNCT bracis-33557 118 1 for for ADP bracis-33557 118 2 unlabeled unlabeled ADJ bracis-33557 118 3 instances instance NOUN bracis-33557 118 4 , , PUNCT bracis-33557 118 5 the the DET bracis-33557 118 6 procedure procedure NOUN bracis-33557 118 7 for for ADP bracis-33557 118 8 hardness hardness NOUN bracis-33557 118 9 estimation estimation NOUN bracis-33557 118 10 is be AUX bracis-33557 118 11 the the DET bracis-33557 118 12 same same ADJ bracis-33557 118 13 as as ADP bracis-33557 118 14 in in ADP bracis-33557 118 15 dcp dcp NOUN bracis-33557 118 16 , , PUNCT bracis-33557 118 17 where where SCONJ bracis-33557 118 18 the the DET bracis-33557 118 19 dt dt NOUN bracis-33557 118 20 is be AUX bracis-33557 118 21 built build VERB bracis-33557 118 22 from from ADP bracis-33557 118 23 the the DET bracis-33557 118 24 labeled label VERB bracis-33557 118 25 set set VERB bracis-33557 118 26 \(\mathcal \(\mathcal ADJ bracis-33557 118 27 d\ d\ NOUN bracis-33557 118 28 ) ) PUNCT bracis-33557 118 29 , , PUNCT bracis-33557 118 30 and and CCONJ bracis-33557 118 31 next next ADV bracis-33557 118 32 , , PUNCT bracis-33557 118 33 the the DET bracis-33557 118 34 unlabeled unlabeled ADJ bracis-33557 118 35 instance instance NOUN bracis-33557 118 36 is be AUX bracis-33557 118 37 submitted submit VERB bracis-33557 118 38 to to ADP bracis-33557 118 39 the the DET bracis-33557 118 40 built build VERB bracis-33557 118 41 dt dt PROPN bracis-33557 118 42 . . PUNCT bracis-33557 119 1 the the DET bracis-33557 119 2 depth depth NOUN bracis-33557 119 3 of of ADP bracis-33557 119 4 the the DET bracis-33557 119 5 leaf leaf NOUN bracis-33557 119 6 node node NOUN bracis-33557 119 7 where where SCONJ bracis-33557 119 8 this this DET bracis-33557 119 9 instance instance NOUN bracis-33557 119 10 is be AUX bracis-33557 119 11 classified classify VERB bracis-33557 119 12 by by ADP bracis-33557 119 13 the the DET bracis-33557 119 14 dt dt NOUN bracis-33557 119 15 is be AUX bracis-33557 119 16 taken take VERB bracis-33557 119 17 and and CCONJ bracis-33557 119 18 used use VERB bracis-33557 119 19 in in ADP bracis-33557 119 20 the the DET bracis-33557 119 21 equation equation NOUN bracis-33557 119 22 : : PUNCT bracis-33557 119 23 $ $ SYM bracis-33557 119 24 $ $ SYM bracis-33557 119 25 \begin{aligned \begin{aligne VERB bracis-33557 119 26 } } PUNCT bracis-33557 119 27 \text \text ADP bracis-33557 119 28 { { PUNCT bracis-33557 119 29 td}_{\text td}_{\text NOUN bracis-33557 119 30 { { PUNCT bracis-33557 119 31 adj}}(\textbf{x adj}}(\textbf{x VERB bracis-33557 119 32 } } PUNCT bracis-33557 119 33 ) ) PUNCT bracis-33557 120 1 = = PUNCT bracis-33557 120 2 \frac{\text \frac{\text X bracis-33557 120 3 { { PUNCT bracis-33557 120 4 depth}_{\text depth}_{\text PROPN bracis-33557 120 5 { { PUNCT bracis-33557 120 6 dt}}(\textbf{x})}{\max dt}}(\textbf{x})}{\max PROPN bracis-33557 120 7 ( ( PUNCT bracis-33557 120 8 \text \text X bracis-33557 120 9 { { PUNCT bracis-33557 120 10 depth}_{\text depth}_{\text PROPN bracis-33557 120 11 { { PUNCT bracis-33557 120 12 dt}}(\textbf{x}_j dt}}(\textbf{x}_j PROPN bracis-33557 120 13 \in \in PROPN bracis-33557 120 14 \mathcal \mathcal PROPN bracis-33557 120 15 d d NOUN bracis-33557 120 16 ) ) PUNCT bracis-33557 120 17 ) ) PUNCT bracis-33557 120 18 } } PUNCT bracis-33557 120 19 , , PUNCT bracis-33557 120 20 \end{aligned}$$ \end{aligned}$$ X bracis-33557 120 21 ( ( PUNCT bracis-33557 120 22 11 11 NUM bracis-33557 120 23 ) ) PUNCT bracis-33557 120 24 2.4 2.4 NUM bracis-33557 120 25 using use VERB bracis-33557 120 26 meta meta NOUN bracis-33557 120 27 - - PUNCT bracis-33557 120 28 models model NOUN bracis-33557 120 29 to to PART bracis-33557 120 30     SPACE bracis-33557 120 31 estimate estimate VERB bracis-33557 120 32 ihm ihm PROPN bracis-33557 120 33 meta meta VERB bracis-33557 120 34 - - PUNCT bracis-33557 120 35 learning learning NOUN bracis-33557 120 36 is be AUX bracis-33557 120 37 a a DET bracis-33557 120 38 traditional traditional ADJ bracis-33557 120 39 ml ml NOUN bracis-33557 120 40 task task NOUN bracis-33557 120 41 that that PRON bracis-33557 120 42 uses use VERB bracis-33557 120 43 data datum NOUN bracis-33557 120 44 related relate VERB bracis-33557 120 45 to to ADP bracis-33557 120 46 ml ml ADP bracis-33557 120 47 itself itself PRON bracis-33557 120 48 [ [ X bracis-33557 120 49 2 2 NUM bracis-33557 120 50 ] ] PUNCT bracis-33557 120 51 . . PUNCT bracis-33557 121 1 here here ADV bracis-33557 121 2 , , PUNCT bracis-33557 121 3 mtl mtl PROPN bracis-33557 121 4 is be AUX bracis-33557 121 5 designed design VERB bracis-33557 121 6 to to PART bracis-33557 121 7 predict predict VERB bracis-33557 121 8 ihm ihm ADJ bracis-33557 121 9 values value NOUN bracis-33557 121 10 without without ADP bracis-33557 121 11 considering consider VERB bracis-33557 121 12 their their PRON bracis-33557 121 13 labels label NOUN bracis-33557 121 14 . . PUNCT bracis-33557 122 1 this this PRON bracis-33557 122 2 is be AUX bracis-33557 122 3 done do VERB bracis-33557 122 4 using use VERB bracis-33557 122 5 the the DET bracis-33557 122 6 original original ADJ bracis-33557 122 7 input input NOUN bracis-33557 122 8 features feature NOUN bracis-33557 122 9 from from ADP bracis-33557 122 10 the the DET bracis-33557 122 11 dataset dataset ADJ bracis-33557 122 12 \(\mathcal \(\mathcal ADJ bracis-33557 122 13 d\ d\ NOUN bracis-33557 122 14 ) ) PUNCT bracis-33557 122 15 to to PART bracis-33557 122 16 learn learn VERB bracis-33557 122 17 the the DET bracis-33557 122 18 expected expect VERB bracis-33557 122 19 ihm ihm ADJ bracis-33557 122 20 values value NOUN bracis-33557 122 21 in in ADP bracis-33557 122 22 a a DET bracis-33557 122 23 regression regression NOUN bracis-33557 122 24 task task NOUN bracis-33557 122 25 . . PUNCT bracis-33557 123 1 therefore therefore ADV bracis-33557 123 2 , , PUNCT bracis-33557 123 3 in in ADP bracis-33557 123 4 this this DET bracis-33557 123 5 approach approach NOUN bracis-33557 123 6 regression regression VERB bracis-33557 123 7 meta meta NOUN bracis-33557 123 8 - - PUNCT bracis-33557 123 9 models model NOUN bracis-33557 123 10 are be AUX bracis-33557 123 11 induced induce VERB bracis-33557 123 12 to to PART bracis-33557 123 13 estimate estimate VERB bracis-33557 123 14 the the DET bracis-33557 123 15 ihm ihm ADJ bracis-33557 123 16 values value NOUN bracis-33557 123 17 of of ADP bracis-33557 123 18 new new ADJ bracis-33557 123 19 instances instance NOUN bracis-33557 123 20 . . PUNCT bracis-33557 124 1 their their PRON bracis-33557 124 2 training training NOUN bracis-33557 124 3 datasets dataset NOUN bracis-33557 124 4 comprise comprise VERB bracis-33557 124 5 the the DET bracis-33557 124 6 original original ADJ bracis-33557 124 7 input input NOUN bracis-33557 124 8 features feature NOUN bracis-33557 124 9 of of ADP bracis-33557 124 10 \(\mathcal \(\mathcal ADJ bracis-33557 124 11 d\ d\ NOUN bracis-33557 124 12 ) ) PUNCT bracis-33557 124 13 and and CCONJ bracis-33557 124 14 a a DET bracis-33557 124 15 label label NOUN bracis-33557 124 16 corresponding correspond VERB bracis-33557 124 17 to to ADP bracis-33557 124 18 an an DET bracis-33557 124 19 ihm ihm NOUN bracis-33557 124 20 estimated estimate VERB bracis-33557 124 21 from from ADP bracis-33557 124 22 \(\mathcal \(\mathcal ADJ bracis-33557 124 23 d\ d\ NOUN bracis-33557 124 24 ) ) PUNCT bracis-33557 124 25 in in ADP bracis-33557 124 26 its its PRON bracis-33557 124 27 original original ADJ bracis-33557 124 28 formulation formulation NOUN bracis-33557 124 29 . . PUNCT bracis-33557 125 1 there there PRON bracis-33557 125 2 is be VERB bracis-33557 125 3 one one NUM bracis-33557 125 4 regression regression NOUN bracis-33557 125 5 model model NOUN bracis-33557 125 6 per per ADP bracis-33557 125 7 ihm ihm PROPN bracis-33557 125 8 . . PUNCT bracis-33557 126 1 the the DET bracis-33557 126 2 estimation estimation NOUN bracis-33557 126 3 of of ADP bracis-33557 126 4 the the DET bracis-33557 126 5 ihm ihm ADJ bracis-33557 126 6 values value NOUN bracis-33557 126 7 for for ADP bracis-33557 126 8 unlabeled unlabeled ADJ bracis-33557 126 9 data datum NOUN bracis-33557 126 10 with with ADP bracis-33557 126 11 this this DET bracis-33557 126 12 meta meta ADJ bracis-33557 126 13 - - PUNCT bracis-33557 126 14 learning learn VERB bracis-33557 126 15 approach approach NOUN bracis-33557 126 16 is be AUX bracis-33557 126 17 compared compare VERB bracis-33557 126 18 to to ADP bracis-33557 126 19 the the DET bracis-33557 126 20 usage usage NOUN bracis-33557 126 21 of of ADP bracis-33557 126 22 the the DET bracis-33557 126 23 adjusted adjust VERB bracis-33557 126 24 ihm ihm ADJ bracis-33557 126 25 values value NOUN bracis-33557 126 26 . . PUNCT bracis-33557 127 1 3 3 NUM bracis-33557 127 2 materials material NOUN bracis-33557 127 3 and and CCONJ bracis-33557 127 4     SPACE bracis-33557 127 5 methods method NOUN bracis-33557 127 6 in in ADP bracis-33557 127 7 this this DET bracis-33557 127 8 section section NOUN bracis-33557 127 9 , , PUNCT bracis-33557 127 10 we we PRON bracis-33557 127 11 describe describe VERB bracis-33557 127 12 the the DET bracis-33557 127 13 materials material NOUN bracis-33557 127 14 and and CCONJ bracis-33557 127 15 methods method NOUN bracis-33557 127 16 used use VERB bracis-33557 127 17 in in ADP bracis-33557 127 18 experiments experiment NOUN bracis-33557 127 19 performed perform VERB bracis-33557 127 20 to to PART bracis-33557 127 21 analyze analyze VERB bracis-33557 127 22 the the DET bracis-33557 127 23 behaviour behaviour NOUN bracis-33557 127 24 of of ADP bracis-33557 127 25 ihm ihm NOUN bracis-33557 127 26 for for ADP bracis-33557 127 27 unlabeled unlabeled ADJ bracis-33557 127 28 data datum NOUN bracis-33557 127 29 in in ADP bracis-33557 127 30 classification classification NOUN bracis-33557 127 31 problems problem NOUN bracis-33557 127 32 . . PUNCT bracis-33557 128 1 3.1 3.1 NUM bracis-33557 128 2 datasets dataset NOUN bracis-33557 128 3 five five NUM bracis-33557 128 4 datasets dataset NOUN bracis-33557 128 5 are be AUX bracis-33557 128 6 employed employ VERB bracis-33557 128 7 in in ADP bracis-33557 128 8 the the DET bracis-33557 128 9 experiments experiment NOUN bracis-33557 128 10 . . PUNCT bracis-33557 129 1 the the DET bracis-33557 129 2 first first ADJ bracis-33557 129 3 dataset dataset NOUN bracis-33557 129 4 was be AUX bracis-33557 129 5 created create VERB bracis-33557 129 6 synthetically synthetically ADV bracis-33557 129 7 , , PUNCT bracis-33557 129 8 containing contain VERB bracis-33557 129 9 three three NUM bracis-33557 129 10 classes class NOUN bracis-33557 129 11 with with ADP bracis-33557 129 12 some some DET bracis-33557 129 13 overlap overlap NOUN bracis-33557 129 14 . . PUNCT bracis-33557 130 1 the the DET bracis-33557 130 2 other other ADJ bracis-33557 130 3 four four NUM bracis-33557 130 4 datasets dataset NOUN bracis-33557 130 5 are be AUX bracis-33557 130 6 from from ADP bracis-33557 130 7 the the DET bracis-33557 130 8 health health NOUN bracis-33557 130 9 domain domain NOUN bracis-33557 130 10 , , PUNCT bracis-33557 130 11 for for ADP bracis-33557 130 12 which which PRON bracis-33557 130 13 some some DET bracis-33557 130 14 instances instance NOUN bracis-33557 130 15 are be AUX bracis-33557 130 16 hard hard ADJ bracis-33557 130 17 to to PART bracis-33557 130 18 classify classify VERB bracis-33557 130 19 due due ADP bracis-33557 130 20 to to ADP bracis-33557 130 21 the the DET bracis-33557 130 22 overlap overlap NOUN bracis-33557 130 23 of of ADP bracis-33557 130 24 attribute attribute NOUN bracis-33557 130 25 values value NOUN bracis-33557 130 26 for for ADP bracis-33557 130 27 different different ADJ bracis-33557 130 28 classes class NOUN bracis-33557 130 29 or or CCONJ bracis-33557 130 30 inconsistencies inconsistency NOUN bracis-33557 130 31 . . PUNCT bracis-33557 131 1 two two NUM bracis-33557 131 2 of of ADP bracis-33557 131 3 them they PRON bracis-33557 131 4 are be AUX bracis-33557 131 5 from from ADP bracis-33557 131 6 the the DET bracis-33557 131 7 uci uci PROPN bracis-33557 131 8 public public ADJ bracis-33557 131 9 repository repository NOUN bracis-33557 131 10 [ [ X bracis-33557 131 11 6 6 NUM bracis-33557 131 12 ] ] PUNCT bracis-33557 131 13 and and CCONJ bracis-33557 131 14 have have AUX bracis-33557 131 15 been be AUX bracis-33557 131 16 employed employ VERB bracis-33557 131 17 in in ADP bracis-33557 131 18 previous previous ADJ bracis-33557 131 19 related related ADJ bracis-33557 131 20 work work NOUN bracis-33557 131 21 [ [ X bracis-33557 131 22 8 8 NUM bracis-33557 131 23 , , PUNCT bracis-33557 131 24 11 11 NUM bracis-33557 131 25 ] ] PUNCT bracis-33557 131 26 . . PUNCT bracis-33557 132 1 the the DET bracis-33557 132 2 last last ADJ bracis-33557 132 3 two two NUM bracis-33557 132 4 are be AUX bracis-33557 132 5 related relate VERB bracis-33557 132 6 to to ADP bracis-33557 132 7 severe severe ADJ bracis-33557 132 8 covid-19 covid-19 PROPN bracis-33557 132 9 cases case NOUN bracis-33557 132 10 in in ADP bracis-33557 132 11 two two NUM bracis-33557 132 12 large large ADJ bracis-33557 132 13 hospitals hospital NOUN bracis-33557 132 14 from from ADP bracis-33557 132 15 the the DET bracis-33557 132 16 são são PROPN bracis-33557 132 17 paulo paulo PROPN bracis-33557 132 18 metropolitan metropolitan PROPN bracis-33557 132 19 area area PROPN bracis-33557 132 20 [ [ X bracis-33557 132 21 15 15 NUM bracis-33557 132 22 ] ] PUNCT bracis-33557 132 23 . . PUNCT bracis-33557 133 1 the the DET bracis-33557 133 2 main main ADJ bracis-33557 133 3 characteristics characteristic NOUN bracis-33557 133 4 of of ADP bracis-33557 133 5 the the DET bracis-33557 133 6 five five NUM bracis-33557 133 7 datasets dataset NOUN bracis-33557 133 8 are be AUX bracis-33557 133 9 presented present VERB bracis-33557 133 10 in in ADP bracis-33557 133 11 table table NOUN bracis-33557 133 12     SPACE bracis-33557 133 13 1 1 NUM bracis-33557 133 14 , , PUNCT bracis-33557 133 15 including include VERB bracis-33557 133 16 the the DET bracis-33557 133 17 number number NOUN bracis-33557 133 18 of of ADP bracis-33557 133 19 instances instance NOUN bracis-33557 133 20 , , PUNCT bracis-33557 133 21 classes class NOUN bracis-33557 133 22 and and CCONJ bracis-33557 133 23 input input NOUN bracis-33557 133 24 features feature NOUN bracis-33557 133 25 . . PUNCT bracis-33557 134 1 table table NOUN bracis-33557 134 2 1 1 NUM bracis-33557 134 3 . . PUNCT bracis-33557 135 1 summary summary NOUN bracis-33557 135 2 of of ADP bracis-33557 135 3 the the DET bracis-33557 135 4 datasets dataset NOUN bracis-33557 135 5 used use VERB bracis-33557 135 6 in in ADP bracis-33557 135 7 the the DET bracis-33557 135 8 study.full study.full NOUN bracis-33557 135 9 size size NOUN bracis-33557 135 10 table table NOUN bracis-33557 135 11 the the DET bracis-33557 135 12 dataset dataset NOUN bracis-33557 135 13 blobs blob NOUN bracis-33557 135 14 was be AUX bracis-33557 135 15 generated generate VERB bracis-33557 135 16 synthetically synthetically ADV bracis-33557 135 17 using use VERB bracis-33557 135 18 the the DET bracis-33557 135 19 make_blobs make_blob NOUN bracis-33557 135 20 package package NOUN bracis-33557 135 21 from from ADP bracis-33557 135 22 the the DET bracis-33557 135 23 scikit scikit NOUN bracis-33557 135 24 - - PUNCT bracis-33557 135 25 learn learn VERB bracis-33557 135 26 library library NOUN bracis-33557 136 1 [ [ X bracis-33557 136 2 10 10 NUM bracis-33557 136 3 ] ] PUNCT bracis-33557 136 4 , , PUNCT bracis-33557 136 5 which which PRON bracis-33557 136 6 can can AUX bracis-33557 136 7 generate generate VERB bracis-33557 136 8 isotropic isotropic ADJ bracis-33557 136 9 gaussian gaussian ADJ bracis-33557 136 10 blobs blob NOUN bracis-33557 136 11 in in ADP bracis-33557 136 12 space space NOUN bracis-33557 136 13 . . PUNCT bracis-33557 137 1 the the DET bracis-33557 137 2 standard standard ADJ bracis-33557 137 3 deviation deviation NOUN bracis-33557 137 4 between between ADP bracis-33557 137 5 the the DET bracis-33557 137 6 centers center NOUN bracis-33557 137 7 of of ADP bracis-33557 137 8 the the DET bracis-33557 137 9 classes class NOUN bracis-33557 137 10 was be AUX bracis-33557 137 11 set set VERB bracis-33557 137 12 as as ADP bracis-33557 137 13 2 2 NUM bracis-33557 137 14 to to PART bracis-33557 137 15 create create VERB bracis-33557 137 16 some some DET bracis-33557 137 17 overlap overlap NOUN bracis-33557 137 18 between between ADP bracis-33557 137 19 the the DET bracis-33557 137 20 input input NOUN bracis-33557 137 21 features feature NOUN bracis-33557 137 22 and and CCONJ bracis-33557 137 23 regions region NOUN bracis-33557 137 24 where where SCONJ bracis-33557 137 25 the the DET bracis-33557 137 26 difficulty difficulty NOUN bracis-33557 137 27 in in ADP bracis-33557 137 28 classifying classify VERB bracis-33557 137 29 the the DET bracis-33557 137 30 instances instance NOUN bracis-33557 137 31 is be AUX bracis-33557 137 32 harder hard ADJ bracis-33557 137 33 than than ADP bracis-33557 137 34 others other NOUN bracis-33557 137 35 . . PUNCT bracis-33557 138 1 figure figure VERB bracis-33557 138 2     SPACE bracis-33557 138 3 4 4 NUM bracis-33557 138 4 presents present VERB bracis-33557 138 5 this this DET bracis-33557 138 6 dataset dataset NOUN bracis-33557 138 7 , , PUNCT bracis-33557 138 8 where where SCONJ bracis-33557 138 9 it it PRON bracis-33557 138 10 is be AUX bracis-33557 138 11 possible possible ADJ bracis-33557 138 12 to to PART bracis-33557 138 13 notice notice VERB bracis-33557 138 14 some some DET bracis-33557 138 15 overlap overlap NOUN bracis-33557 138 16 in in ADP bracis-33557 138 17 the the DET bracis-33557 138 18 borderline borderline NOUN bracis-33557 138 19 regions region NOUN bracis-33557 138 20 of of ADP bracis-33557 138 21 the the DET bracis-33557 138 22 classes class NOUN bracis-33557 138 23 . . PUNCT bracis-33557 139 1 fig fig NOUN bracis-33557 139 2 . . PUNCT bracis-33557 140 1 2 2 X bracis-33557 140 2 . . X bracis-33557 140 3 illustration illustration NOUN bracis-33557 140 4 of of ADP bracis-33557 140 5 the the DET bracis-33557 140 6 blobs blob NOUN bracis-33557 140 7 dataset dataset VERB bracis-33557 140 8 . . PUNCT bracis-33557 141 1 full full ADJ bracis-33557 141 2 size size NOUN bracis-33557 141 3 image image NOUN bracis-33557 141 4 the the DET bracis-33557 141 5 diabetes diabetes NOUN bracis-33557 141 6 dataset dataset VERB bracis-33557 141 7 is be AUX bracis-33557 141 8 related relate VERB bracis-33557 141 9 to to ADP bracis-33557 141 10 the the DET bracis-33557 141 11 incidence incidence NOUN bracis-33557 141 12 of of ADP bracis-33557 141 13 diabetes diabetes NOUN bracis-33557 141 14 in in ADP bracis-33557 141 15 female female ADJ bracis-33557 141 16 patients patient NOUN bracis-33557 141 17 of of ADP bracis-33557 141 18 pima pima PROPN bracis-33557 141 19 indian indian PROPN bracis-33557 141 20 heritage heritage NOUN bracis-33557 141 21 who who PRON bracis-33557 141 22 are be AUX bracis-33557 141 23 at at ADV bracis-33557 141 24 least least ADV bracis-33557 141 25 21 21 NUM bracis-33557 141 26 years year NOUN bracis-33557 141 27 old old ADJ bracis-33557 141 28 . . PUNCT bracis-33557 142 1 the the DET bracis-33557 142 2 objective objective NOUN bracis-33557 142 3 is be AUX bracis-33557 142 4 to to PART bracis-33557 142 5 identify identify VERB bracis-33557 142 6 the the DET bracis-33557 142 7 presence presence NOUN bracis-33557 142 8 of of ADP bracis-33557 142 9 diabetes diabetes NOUN bracis-33557 142 10 . . PUNCT bracis-33557 143 1 the the DET bracis-33557 143 2 predictive predictive ADJ bracis-33557 143 3 variables variable NOUN bracis-33557 143 4 record record NOUN bracis-33557 143 5 blood blood NOUN bracis-33557 143 6 indices index NOUN bracis-33557 143 7 and and CCONJ bracis-33557 143 8 patient patient ADJ bracis-33557 143 9 characteristics characteristic NOUN bracis-33557 143 10 , , PUNCT bracis-33557 143 11 such such ADJ bracis-33557 143 12 as as ADP bracis-33557 143 13 number number NOUN bracis-33557 143 14 of of ADP bracis-33557 143 15 pregnancies pregnancy NOUN bracis-33557 143 16 and and CCONJ bracis-33557 143 17 age age NOUN bracis-33557 143 18 [ [ X bracis-33557 143 19 6 6 NUM bracis-33557 143 20 ] ] PUNCT bracis-33557 143 21 . . PUNCT bracis-33557 144 1 the the DET bracis-33557 144 2 heart heart NOUN bracis-33557 144 3 dataset dataset NOUN bracis-33557 144 4 registers register NOUN bracis-33557 144 5 heart heart NOUN bracis-33557 144 6 disease disease NOUN bracis-33557 144 7 in in ADP bracis-33557 144 8 patients patient NOUN bracis-33557 144 9 and and CCONJ bracis-33557 144 10 has have VERB bracis-33557 144 11 features feature NOUN bracis-33557 144 12 collected collect VERB bracis-33557 144 13 during during ADP bracis-33557 144 14 the the DET bracis-33557 144 15 exercise exercise NOUN bracis-33557 144 16 test test NOUN bracis-33557 144 17 , , PUNCT bracis-33557 144 18 others other NOUN bracis-33557 144 19 reflecting reflect VERB bracis-33557 144 20 blood blood NOUN bracis-33557 144 21 indices index NOUN bracis-33557 144 22 and and CCONJ bracis-33557 144 23 personal personal ADJ bracis-33557 144 24 characteristics characteristic NOUN bracis-33557 144 25 of of ADP bracis-33557 144 26 the the DET bracis-33557 144 27 patients patient NOUN bracis-33557 144 28 , , PUNCT bracis-33557 144 29 such such ADJ bracis-33557 144 30 as as ADP bracis-33557 144 31 age age NOUN bracis-33557 144 32 and and CCONJ bracis-33557 144 33 gender gender NOUN bracis-33557 145 1 [ [ X bracis-33557 145 2 6 6 NUM bracis-33557 145 3 ] ] PUNCT bracis-33557 145 4 . . PUNCT bracis-33557 146 1 the the DET bracis-33557 146 2 last last ADJ bracis-33557 146 3 two two NUM bracis-33557 146 4 datasets dataset NOUN bracis-33557 146 5 , , PUNCT bracis-33557 146 6 named name VERB bracis-33557 146 7 hospital hospital NOUN bracis-33557 146 8 , , PUNCT bracis-33557 146 9 were be AUX bracis-33557 146 10 extracted extract VERB bracis-33557 146 11 from from ADP bracis-33557 146 12 the the DET bracis-33557 146 13 raw raw ADJ bracis-33557 146 14 public public ADJ bracis-33557 146 15 database database NOUN bracis-33557 146 16 provided provide VERB bracis-33557 146 17 by by ADP bracis-33557 146 18 fapesp fapesp PROPN bracis-33557 146 19 covid covid PROPN bracis-33557 146 20 data data PROPN bracis-33557 146 21 sharing sharing NOUN bracis-33557 146 22 initiative initiative NOUN bracis-33557 146 23 [ [ X bracis-33557 146 24 3 3 NUM bracis-33557 146 25 ] ] PUNCT bracis-33557 146 26 . . PUNCT bracis-33557 147 1 the the DET bracis-33557 147 2 binary binary ADJ bracis-33557 147 3 response response NOUN bracis-33557 147 4 categorized categorize VERB bracis-33557 147 5 patients patient NOUN bracis-33557 147 6 as as ADP bracis-33557 147 7 severe severe ADJ bracis-33557 147 8 when when SCONJ bracis-33557 147 9 hospital hospital NOUN bracis-33557 147 10 stay stay NOUN bracis-33557 147 11 was be AUX bracis-33557 147 12 greater great ADJ bracis-33557 147 13 than than ADP bracis-33557 147 14 or or CCONJ bracis-33557 147 15 equal equal ADJ bracis-33557 147 16 to to ADP bracis-33557 147 17 14 14 NUM bracis-33557 147 18 days day NOUN bracis-33557 147 19 or or CCONJ bracis-33557 147 20 patients patient NOUN bracis-33557 147 21 who who PRON bracis-33557 147 22 progressed progress VERB bracis-33557 147 23 to to ADP bracis-33557 147 24 death death NOUN bracis-33557 147 25 . . PUNCT bracis-33557 148 1 the the DET bracis-33557 148 2 features feature NOUN bracis-33557 148 3 collected collect VERB bracis-33557 148 4 in in ADP bracis-33557 148 5 those those DET bracis-33557 148 6 datasets dataset NOUN bracis-33557 148 7 were be AUX bracis-33557 148 8 related relate VERB bracis-33557 148 9 to to ADP bracis-33557 148 10 blood blood NOUN bracis-33557 148 11 indices index NOUN bracis-33557 148 12 , , PUNCT bracis-33557 148 13 age age NOUN bracis-33557 148 14 and and CCONJ bracis-33557 148 15 gender gender NOUN bracis-33557 148 16 [ [ X bracis-33557 148 17 15 15 NUM bracis-33557 148 18 ] ] SYM bracis-33557 148 19 . . PUNCT bracis-33557 149 1 3.2 3.2 NUM bracis-33557 149 2 methodology methodology NOUN bracis-33557 149 3 the the DET bracis-33557 149 4 adjusted adjust VERB bracis-33557 149 5 ihm ihm NOUN bracis-33557 149 6 measures measure NOUN bracis-33557 149 7 proposed propose VERB bracis-33557 149 8 in in ADP bracis-33557 149 9 this this DET bracis-33557 149 10 paper paper NOUN bracis-33557 149 11 were be AUX bracis-33557 149 12 applied apply VERB bracis-33557 149 13 to to ADP bracis-33557 149 14 the the DET bracis-33557 149 15 datasets dataset NOUN bracis-33557 149 16 , , PUNCT bracis-33557 149 17 considering consider VERB bracis-33557 149 18 each each DET bracis-33557 149 19 instance instance NOUN bracis-33557 149 20 unlabeled unlabele VERB bracis-33557 149 21 once once ADV bracis-33557 149 22 at at ADP bracis-33557 149 23 a a DET bracis-33557 149 24 time time NOUN bracis-33557 149 25 , , PUNCT bracis-33557 149 26 and and CCONJ bracis-33557 149 27 the the DET bracis-33557 149 28 remaining remain VERB bracis-33557 149 29 instances instance NOUN bracis-33557 149 30 labeled label VERB bracis-33557 149 31 , , PUNCT bracis-33557 149 32 resembling resemble VERB bracis-33557 149 33 a a DET bracis-33557 149 34 leave leave VERB bracis-33557 149 35 - - PUNCT bracis-33557 149 36 one one NUM bracis-33557 149 37 - - PUNCT bracis-33557 149 38 out out NOUN bracis-33557 149 39 ( ( PUNCT bracis-33557 149 40 loo loo NOUN bracis-33557 149 41 ) ) PUNCT bracis-33557 149 42 cross cross ADJ bracis-33557 149 43 - - ADJ bracis-33557 149 44 validation validation ADJ bracis-33557 149 45 scheme scheme NOUN bracis-33557 149 46 . . PUNCT bracis-33557 150 1 the the DET bracis-33557 150 2 same same ADJ bracis-33557 150 3 procedure procedure NOUN bracis-33557 150 4 is be AUX bracis-33557 150 5 used use VERB bracis-33557 150 6 to to PART bracis-33557 150 7 generate generate VERB bracis-33557 150 8 the the DET bracis-33557 150 9 meta meta NOUN bracis-33557 150 10 - - PUNCT bracis-33557 150 11 models model NOUN bracis-33557 150 12 to to PART bracis-33557 150 13 predict predict VERB bracis-33557 150 14 the the DET bracis-33557 150 15 ihm ihm ADJ bracis-33557 150 16 values value NOUN bracis-33557 150 17 , , PUNCT bracis-33557 150 18 where where SCONJ bracis-33557 150 19 one one NUM bracis-33557 150 20 instance instance NOUN bracis-33557 150 21 is be AUX bracis-33557 150 22 left leave VERB bracis-33557 150 23 out out ADP bracis-33557 150 24 as as ADP bracis-33557 150 25 unlabeled unlabele VERB bracis-33557 150 26 at at ADP bracis-33557 150 27 a a DET bracis-33557 150 28 time time NOUN bracis-33557 150 29 . . PUNCT bracis-33557 151 1 the the DET bracis-33557 151 2 ihm ihm NOUN bracis-33557 151 3 of of ADP bracis-33557 151 4 the the DET bracis-33557 151 5 other other ADJ bracis-33557 151 6 instances instance NOUN bracis-33557 151 7 is be AUX bracis-33557 151 8 calculated calculate VERB bracis-33557 151 9 using use VERB bracis-33557 151 10 their their PRON bracis-33557 151 11 original original ADJ bracis-33557 151 12 formulations formulation NOUN bracis-33557 151 13 . . PUNCT bracis-33557 152 1 next next ADV bracis-33557 152 2 , , PUNCT bracis-33557 152 3 a a DET bracis-33557 152 4 meta meta ADJ bracis-33557 152 5 - - PUNCT bracis-33557 152 6 dataset dataset NOUN bracis-33557 152 7 is be AUX bracis-33557 152 8 built build VERB bracis-33557 152 9 , , PUNCT bracis-33557 152 10 mapping map VERB bracis-33557 152 11 the the DET bracis-33557 152 12 original original ADJ bracis-33557 152 13 features feature NOUN bracis-33557 152 14 of of ADP bracis-33557 152 15 the the DET bracis-33557 152 16 instances instance NOUN bracis-33557 152 17 to to ADP bracis-33557 152 18 the the DET bracis-33557 152 19 computed computed ADJ bracis-33557 152 20 ihm ihm ADJ bracis-33557 152 21 values value NOUN bracis-33557 152 22 . . PUNCT bracis-33557 153 1 regression regression VERB bracis-33557 153 2 meta meta NOUN bracis-33557 153 3 - - PUNCT bracis-33557 153 4 models model NOUN bracis-33557 153 5 are be AUX bracis-33557 153 6 induced induce VERB bracis-33557 153 7 to to PART bracis-33557 153 8 learn learn VERB bracis-33557 153 9 this this DET bracis-33557 153 10 relationship relationship NOUN bracis-33557 153 11 and and CCONJ bracis-33557 153 12 predict predict VERB bracis-33557 153 13 the the DET bracis-33557 153 14 expected expected ADJ bracis-33557 153 15 ihm ihm ADJ bracis-33557 153 16 value value NOUN bracis-33557 153 17 of of ADP bracis-33557 153 18 the the DET bracis-33557 153 19 left leave VERB bracis-33557 153 20 - - PUNCT bracis-33557 153 21 out out ADP bracis-33557 153 22 instance instance NOUN bracis-33557 153 23 . . PUNCT bracis-33557 154 1 one one NUM bracis-33557 154 2 meta meta ADJ bracis-33557 154 3 - - PUNCT bracis-33557 154 4 model model NOUN bracis-33557 154 5 is be AUX bracis-33557 154 6 induced induce VERB bracis-33557 154 7 per per ADP bracis-33557 154 8 ihm ihm ADJ bracis-33557 154 9 measure measure NOUN bracis-33557 154 10 considered consider VERB bracis-33557 154 11 . . PUNCT bracis-33557 155 1 we we PRON bracis-33557 155 2 used use VERB bracis-33557 155 3 the the DET bracis-33557 155 4 random random ADJ bracis-33557 155 5 forest forest NOUN bracis-33557 155 6 regressor regressor NOUN bracis-33557 155 7 ( ( PUNCT bracis-33557 155 8 rf rf NOUN bracis-33557 155 9 ) ) PUNCT bracis-33557 155 10 available available ADJ bracis-33557 155 11 in in ADP bracis-33557 155 12 the the DET bracis-33557 155 13 scikit scikit NOUN bracis-33557 155 14 - - PUNCT bracis-33557 155 15 learn learn VERB bracis-33557 155 16 library library NOUN bracis-33557 155 17 [ [ X bracis-33557 155 18 10 10 NUM bracis-33557 155 19 ] ] PUNCT bracis-33557 155 20 with with ADP bracis-33557 155 21 default default NOUN bracis-33557 155 22 hyperparameters hyperparameter NOUN bracis-33557 155 23 ’ ' PUNCT bracis-33557 155 24 values value NOUN bracis-33557 155 25 to to PART bracis-33557 155 26 generate generate VERB bracis-33557 155 27 these these DET bracis-33557 155 28 meta meta ADJ bracis-33557 155 29 - - PUNCT bracis-33557 155 30 models model NOUN bracis-33557 155 31 . . PUNCT bracis-33557 156 1 we we PRON bracis-33557 156 2 also also ADV bracis-33557 156 3 computed compute VERB bracis-33557 156 4 the the DET bracis-33557 156 5 original original ADJ bracis-33557 156 6 ihms ihm NOUN bracis-33557 156 7 for for ADP bracis-33557 156 8 the the DET bracis-33557 156 9 entire entire ADJ bracis-33557 156 10 datasets dataset NOUN bracis-33557 156 11 , , PUNCT bracis-33557 156 12 which which PRON bracis-33557 156 13 regard regard VERB bracis-33557 156 14 the the DET bracis-33557 156 15 labels label NOUN bracis-33557 156 16 of of ADP bracis-33557 156 17 all all DET bracis-33557 156 18 instances instance NOUN bracis-33557 156 19 . . PUNCT bracis-33557 157 1 next next ADV bracis-33557 157 2 , , PUNCT bracis-33557 157 3 we we PRON bracis-33557 157 4 compare compare VERB bracis-33557 157 5 the the DET bracis-33557 157 6 association association NOUN bracis-33557 157 7 of of ADP bracis-33557 157 8 the the DET bracis-33557 157 9 ihm ihm ADJ bracis-33557 157 10 values value NOUN bracis-33557 157 11 of of ADP bracis-33557 157 12 the the DET bracis-33557 157 13 original original ADJ bracis-33557 157 14 measures measure NOUN bracis-33557 157 15 to to ADP bracis-33557 157 16 those those PRON bracis-33557 157 17 of of ADP bracis-33557 157 18 the the DET bracis-33557 157 19 estimated estimate VERB bracis-33557 157 20 measures measure NOUN bracis-33557 157 21 , , PUNCT bracis-33557 157 22 where where SCONJ bracis-33557 157 23 the the DET bracis-33557 157 24 estimation estimation NOUN bracis-33557 157 25 is be AUX bracis-33557 157 26 taken take VERB bracis-33557 157 27 by by ADP bracis-33557 157 28 the the DET bracis-33557 157 29 adjusted adjust VERB bracis-33557 157 30 measures measure NOUN bracis-33557 157 31 or or CCONJ bracis-33557 157 32 the the DET bracis-33557 157 33 induced induce VERB bracis-33557 157 34 meta meta ADJ bracis-33557 157 35 - - PUNCT bracis-33557 157 36 models model NOUN bracis-33557 157 37 . . PUNCT bracis-33557 158 1 spearman spearman NOUN bracis-33557 158 2 ’s ’s PART bracis-33557 158 3 correlation correlation NOUN bracis-33557 158 4 provides provide VERB bracis-33557 158 5 a a DET bracis-33557 158 6 non non ADJ bracis-33557 158 7 - - ADJ bracis-33557 158 8 parametric parametric ADJ bracis-33557 158 9 estimation estimation NOUN bracis-33557 158 10 of of ADP bracis-33557 158 11 the the DET bracis-33557 158 12 association association NOUN bracis-33557 158 13 ( ( PUNCT bracis-33557 158 14 monotonic monotonic ADJ bracis-33557 158 15 relationship relationship NOUN bracis-33557 158 16 ) ) PUNCT bracis-33557 158 17 of of ADP bracis-33557 158 18 the the DET bracis-33557 158 19 modified modify VERB bracis-33557 158 20 measures measure NOUN bracis-33557 158 21 with with ADP bracis-33557 158 22 the the DET bracis-33557 158 23 original original ADJ bracis-33557 158 24 measure measure NOUN bracis-33557 158 25 . . PUNCT bracis-33557 159 1 this this DET bracis-33557 159 2 correlation correlation NOUN bracis-33557 159 3 captures capture VERB bracis-33557 159 4 if if SCONJ bracis-33557 159 5 the the DET bracis-33557 159 6 direction direction NOUN bracis-33557 159 7 of of ADP bracis-33557 159 8 the the DET bracis-33557 159 9 adjusted adjust VERB bracis-33557 159 10 / / SYM bracis-33557 159 11 estimated estimate VERB bracis-33557 159 12 ihm ihm NOUN bracis-33557 159 13 is be AUX bracis-33557 159 14 the the DET bracis-33557 159 15 same same ADJ bracis-33557 159 16 as as ADP bracis-33557 159 17 the the DET bracis-33557 159 18 value value NOUN bracis-33557 159 19 obtained obtain VERB bracis-33557 159 20 from from ADP bracis-33557 159 21 the the DET bracis-33557 159 22 original original ADJ bracis-33557 159 23 ihm ihm NOUN bracis-33557 159 24 . . PUNCT bracis-33557 160 1 higher high ADJ bracis-33557 160 2 values value NOUN bracis-33557 160 3 of of ADP bracis-33557 160 4 the the DET bracis-33557 160 5 spearman spearman NOUN bracis-33557 160 6 ’s ’s PART bracis-33557 160 7 correlation correlation NOUN bracis-33557 160 8 indicate indicate VERB bracis-33557 160 9 more more ADJ bracis-33557 160 10 association association NOUN bracis-33557 160 11 between between ADP bracis-33557 160 12 the the DET bracis-33557 160 13 estimated estimate VERB bracis-33557 160 14 and and CCONJ bracis-33557 160 15 original original ADJ bracis-33557 160 16 ihm ihm NOUN bracis-33557 160 17 . . PUNCT bracis-33557 161 1 we we PRON bracis-33557 161 2 expect expect VERB bracis-33557 161 3 medium medium ADJ bracis-33557 161 4 to to ADP bracis-33557 161 5 high high ADJ bracis-33557 161 6 correlations correlation NOUN bracis-33557 161 7 , , PUNCT bracis-33557 161 8 although although SCONJ bracis-33557 161 9 there there PRON bracis-33557 161 10 can can AUX bracis-33557 161 11 be be AUX bracis-33557 161 12 deviations deviation NOUN bracis-33557 161 13 of of ADP bracis-33557 161 14 values value NOUN bracis-33557 161 15 , , PUNCT bracis-33557 161 16 since since SCONJ bracis-33557 161 17 they they PRON bracis-33557 161 18 do do AUX bracis-33557 161 19 not not PART bracis-33557 161 20 strive strive VERB bracis-33557 161 21 to to PART bracis-33557 161 22 deliver deliver VERB bracis-33557 161 23 identical identical ADJ bracis-33557 161 24 ihm ihm ADJ bracis-33557 161 25 values value NOUN bracis-33557 161 26 . . PUNCT bracis-33557 162 1 indeed indeed ADV bracis-33557 162 2 , , PUNCT bracis-33557 162 3 instances instance NOUN bracis-33557 162 4 with with ADP bracis-33557 162 5 noisy noisy ADJ bracis-33557 162 6 labels label NOUN bracis-33557 162 7 in in ADP bracis-33557 162 8 the the DET bracis-33557 162 9 training training NOUN bracis-33557 162 10 datasets dataset NOUN bracis-33557 162 11 have have VERB bracis-33557 162 12 characteristics characteristic NOUN bracis-33557 162 13 that that PRON bracis-33557 162 14 make make VERB bracis-33557 162 15 them they PRON bracis-33557 162 16 aligned align VERB bracis-33557 162 17 to to ADP bracis-33557 162 18 another another DET bracis-33557 162 19 class class NOUN bracis-33557 162 20 and and CCONJ bracis-33557 162 21 are be AUX bracis-33557 162 22 expected expect VERB bracis-33557 162 23 to to PART bracis-33557 162 24 show show VERB bracis-33557 162 25 a a DET bracis-33557 162 26 lower low ADJ bracis-33557 162 27 correlation correlation NOUN bracis-33557 162 28 to to ADP bracis-33557 162 29 the the DET bracis-33557 162 30 original original ADJ bracis-33557 162 31 ihm ihm ADJ bracis-33557 162 32 values value NOUN bracis-33557 162 33 . . PUNCT bracis-33557 163 1 but but CCONJ bracis-33557 163 2 for for ADP bracis-33557 163 3 most most ADJ bracis-33557 163 4 cases case NOUN bracis-33557 163 5 , , PUNCT bracis-33557 163 6 we we PRON bracis-33557 163 7 expect expect VERB bracis-33557 163 8 the the DET bracis-33557 163 9 hardness hardness ADJ bracis-33557 163 10 directions direction NOUN bracis-33557 163 11 to to PART bracis-33557 163 12 be be AUX bracis-33557 163 13 maintained maintain VERB bracis-33557 163 14 . . PUNCT bracis-33557 164 1 all all DET bracis-33557 164 2 codes code NOUN bracis-33557 164 3 and and CCONJ bracis-33557 164 4 analyses analysis NOUN bracis-33557 164 5 are be AUX bracis-33557 164 6 implemented implement VERB bracis-33557 164 7 in in ADP bracis-33557 164 8 python python PROPN bracis-33557 164 9 . . PUNCT bracis-33557 165 1 the the DET bracis-33557 165 2 original original ADJ bracis-33557 165 3 ihms ihm NOUN bracis-33557 165 4 are be AUX bracis-33557 165 5 computed compute VERB bracis-33557 165 6 using use VERB bracis-33557 165 7 the the DET bracis-33557 165 8 pyhard pyhard ADJ bracis-33557 165 9 package package NOUN bracis-33557 166 1 [ [ X bracis-33557 166 2 7 7 NUM bracis-33557 166 3 , , PUNCT bracis-33557 166 4 9 9 NUM bracis-33557 166 5 ] ] PUNCT bracis-33557 166 6 . . PUNCT bracis-33557 167 1 codes code NOUN bracis-33557 167 2 of of ADP bracis-33557 167 3 the the DET bracis-33557 167 4 adjusted adjust VERB bracis-33557 167 5 measures measure NOUN bracis-33557 167 6 are be AUX bracis-33557 167 7 in in ADP bracis-33557 167 8 a a DET bracis-33557 167 9 public public ADJ bracis-33557 167 10 repository repository NOUN bracis-33557 167 11 https://anonymous.4open.science/r/adj-ihm-bf75 https://anonymous.4open.science/r/adj-ihm-bf75 NOUN bracis-33557 167 12 . . PUNCT bracis-33557 168 1 the the DET bracis-33557 168 2 k k PROPN bracis-33557 168 3 value value NOUN bracis-33557 168 4 in in ADP bracis-33557 168 5 kdn kdn NOUN bracis-33557 168 6 was be AUX bracis-33557 168 7 set set VERB bracis-33557 168 8 as as ADP bracis-33557 168 9 10 10 NUM bracis-33557 168 10 , , PUNCT bracis-33557 168 11 default default NOUN bracis-33557 168 12 value value NOUN bracis-33557 168 13 in in ADP bracis-33557 168 14 the the DET bracis-33557 168 15 pyhard pyhard NOUN bracis-33557 168 16 package package NOUN bracis-33557 168 17 . . PUNCT bracis-33557 169 1 4 4 NUM bracis-33557 169 2 results result VERB bracis-33557 169 3 the the DET bracis-33557 169 4 results result NOUN bracis-33557 169 5 of of ADP bracis-33557 169 6 the the DET bracis-33557 169 7 experiments experiment NOUN bracis-33557 169 8 performed perform VERB bracis-33557 169 9 are be AUX bracis-33557 169 10 presented present VERB bracis-33557 169 11 and and CCONJ bracis-33557 169 12 discussed discuss VERB bracis-33557 169 13 next next ADV bracis-33557 169 14 . . PUNCT bracis-33557 170 1 4.1 4.1 NUM bracis-33557 170 2 meta meta X bracis-33557 170 3 - - PUNCT bracis-33557 170 4 models model NOUN bracis-33557 170 5 first first ADV bracis-33557 170 6 , , PUNCT bracis-33557 170 7 we we PRON bracis-33557 170 8 present present VERB bracis-33557 170 9 the the DET bracis-33557 170 10 performance performance NOUN bracis-33557 170 11 of of ADP bracis-33557 170 12 the the DET bracis-33557 170 13 meta meta NOUN bracis-33557 170 14 - - PUNCT bracis-33557 170 15 models model NOUN bracis-33557 170 16 in in ADP bracis-33557 170 17 the the DET bracis-33557 170 18 regression regression NOUN bracis-33557 170 19 task task NOUN bracis-33557 170 20 . . PUNCT bracis-33557 171 1 table table NOUN bracis-33557 171 2     SPACE bracis-33557 171 3 2 2 NUM bracis-33557 171 4 presents present NOUN bracis-33557 171 5 the the DET bracis-33557 171 6 mean mean NOUN bracis-33557 171 7 squared square VERB bracis-33557 171 8 error error NOUN bracis-33557 171 9 ( ( PUNCT bracis-33557 171 10 mse mse NOUN bracis-33557 171 11 ) ) PUNCT bracis-33557 171 12 obtained obtain VERB bracis-33557 171 13 in in ADP bracis-33557 171 14 predicting predict VERB bracis-33557 171 15 the the DET bracis-33557 171 16 ihms ihm NOUN bracis-33557 171 17 using use VERB bracis-33557 171 18 the the DET bracis-33557 171 19 regression regression NOUN bracis-33557 171 20 meta meta ADJ bracis-33557 171 21 - - PUNCT bracis-33557 171 22 models model NOUN bracis-33557 171 23 . . PUNCT bracis-33557 172 1 lower low ADJ bracis-33557 172 2 values value NOUN bracis-33557 172 3 are be AUX bracis-33557 172 4 indicative indicative ADJ bracis-33557 172 5 of of ADP bracis-33557 172 6 better well ADJ bracis-33557 172 7 performances performance NOUN bracis-33557 172 8 in in ADP bracis-33557 172 9 predicting predict VERB bracis-33557 172 10 the the DET bracis-33557 172 11 original original ADJ bracis-33557 172 12 ihm ihm ADJ bracis-33557 172 13 values value NOUN bracis-33557 172 14 . . PUNCT bracis-33557 173 1 table table NOUN bracis-33557 173 2 2 2 NUM bracis-33557 173 3 . . X bracis-33557 173 4 mse mse PROPN bracis-33557 173 5 obtained obtain VERB bracis-33557 173 6 for for ADP bracis-33557 173 7 the the DET bracis-33557 173 8 rf rf ADJ bracis-33557 173 9 algorithm algorithm NOUN bracis-33557 173 10 concerning concern VERB bracis-33557 173 11 predicting predict VERB bracis-33557 173 12 the the DET bracis-33557 173 13 ihms ihm NOUN bracis-33557 173 14 to to ADP bracis-33557 173 15 different different ADJ bracis-33557 173 16 datasets.full datasets.full NUM bracis-33557 173 17 size size NOUN bracis-33557 173 18 table table NOUN bracis-33557 173 19 for for ADP bracis-33557 173 20 some some DET bracis-33557 173 21 measures measure NOUN bracis-33557 173 22 , , PUNCT bracis-33557 173 23 the the DET bracis-33557 173 24 mses ms NOUN bracis-33557 173 25 are be AUX bracis-33557 173 26 lower low ADJ bracis-33557 173 27 , , PUNCT bracis-33557 173 28 demonstrating demonstrate VERB bracis-33557 173 29 a a DET bracis-33557 173 30 better well ADJ bracis-33557 173 31 approximation approximation NOUN bracis-33557 173 32 of of ADP bracis-33557 173 33 the the DET bracis-33557 173 34 original original ADJ bracis-33557 173 35 ihm ihm ADJ bracis-33557 173 36 values value NOUN bracis-33557 173 37 . . PUNCT bracis-33557 174 1 this this PRON bracis-33557 174 2 happens happen VERB bracis-33557 174 3 mostly mostly ADV bracis-33557 174 4 for for ADP bracis-33557 174 5 tree tree NOUN bracis-33557 174 6 - - PUNCT bracis-33557 174 7 depth depth NOUN bracis-33557 174 8 measures measure NOUN bracis-33557 174 9 . . PUNCT bracis-33557 175 1 for for ADP bracis-33557 175 2 others other NOUN bracis-33557 175 3 , , PUNCT bracis-33557 175 4 the the DET bracis-33557 175 5 approximations approximation NOUN bracis-33557 175 6 are be AUX bracis-33557 175 7 not not PART bracis-33557 175 8 as as ADV bracis-33557 175 9 good good ADJ bracis-33557 175 10 ( ( PUNCT bracis-33557 175 11 e.g. e.g. ADV bracis-33557 175 12 for for ADP bracis-33557 175 13 kdn kdn NOUN bracis-33557 175 14 and and CCONJ bracis-33557 175 15 dcp dcp NOUN bracis-33557 175 16 ) ) PUNCT bracis-33557 175 17 . . PUNCT bracis-33557 176 1 one one NUM bracis-33557 176 2 possible possible ADJ bracis-33557 176 3 explanation explanation NOUN bracis-33557 176 4 is be AUX bracis-33557 176 5 that that SCONJ bracis-33557 176 6 the the DET bracis-33557 176 7 tree tree NOUN bracis-33557 176 8 depth depth NOUN bracis-33557 176 9 measures measure NOUN bracis-33557 176 10 do do AUX bracis-33557 176 11 not not PART bracis-33557 176 12 depend depend VERB bracis-33557 176 13 as as ADV bracis-33557 176 14 much much ADJ bracis-33557 176 15 on on ADP bracis-33557 176 16 the the DET bracis-33557 176 17 labels label NOUN bracis-33557 176 18 of of ADP bracis-33557 176 19 the the DET bracis-33557 176 20 instances instance NOUN bracis-33557 176 21 as as ADP bracis-33557 176 22 the the DET bracis-33557 176 23 others other NOUN bracis-33557 176 24 . . PUNCT bracis-33557 177 1 the the DET bracis-33557 177 2 only only ADJ bracis-33557 177 3 difference difference NOUN bracis-33557 177 4 between between ADP bracis-33557 177 5 the the DET bracis-33557 177 6 original original ADJ bracis-33557 177 7 tree tree NOUN bracis-33557 177 8 depth depth NOUN bracis-33557 177 9 measures measure NOUN bracis-33557 177 10 and and CCONJ bracis-33557 177 11 their their PRON bracis-33557 177 12 estimated estimate VERB bracis-33557 177 13 counterparts counterpart NOUN bracis-33557 177 14 is be AUX bracis-33557 177 15 excluding exclude VERB bracis-33557 177 16 one one NUM bracis-33557 177 17 instance instance NOUN bracis-33557 177 18 from from ADP bracis-33557 177 19 the the DET bracis-33557 177 20 decision decision NOUN bracis-33557 177 21 tree tree NOUN bracis-33557 177 22 induction induction NOUN bracis-33557 177 23 , , PUNCT bracis-33557 177 24 which which PRON bracis-33557 177 25 affects affect VERB bracis-33557 177 26 less less ADJ bracis-33557 177 27 the the DET bracis-33557 177 28 results result NOUN bracis-33557 177 29 . . PUNCT bracis-33557 178 1 for for ADP bracis-33557 178 2 other other ADJ bracis-33557 178 3 measures measure NOUN bracis-33557 178 4 , , PUNCT bracis-33557 178 5 if if SCONJ bracis-33557 178 6 an an DET bracis-33557 178 7 instance instance NOUN bracis-33557 178 8 is be AUX bracis-33557 178 9 incorrectly incorrectly ADV bracis-33557 178 10 labeled label VERB bracis-33557 178 11 , , PUNCT bracis-33557 178 12 the the DET bracis-33557 178 13 original original ADJ bracis-33557 178 14 measures measure NOUN bracis-33557 178 15 will will AUX bracis-33557 178 16 point point VERB bracis-33557 178 17 them they PRON bracis-33557 178 18 as as ADP bracis-33557 178 19 very very ADV bracis-33557 178 20 hard hard ADJ bracis-33557 178 21 to to PART bracis-33557 178 22 classify classify VERB bracis-33557 178 23 . . PUNCT bracis-33557 179 1 but but CCONJ bracis-33557 179 2 this this DET bracis-33557 179 3 instance instance NOUN bracis-33557 179 4 might might AUX bracis-33557 179 5 be be AUX bracis-33557 179 6 easily easily ADV bracis-33557 179 7 classified classify VERB bracis-33557 179 8 into into ADP bracis-33557 179 9 another another DET bracis-33557 179 10 class class NOUN bracis-33557 179 11 , , PUNCT bracis-33557 179 12 making make VERB bracis-33557 179 13 it it PRON bracis-33557 179 14 easy easy ADJ bracis-33557 179 15 without without ADP bracis-33557 179 16 the the DET bracis-33557 179 17 label label NOUN bracis-33557 179 18 information information NOUN bracis-33557 179 19 . . PUNCT bracis-33557 180 1 4.2 4.2 NUM bracis-33557 180 2 correlation correlation NOUN bracis-33557 180 3 analysis analysis NOUN bracis-33557 180 4 table table NOUN bracis-33557 180 5     SPACE bracis-33557 180 6 3 3 NUM bracis-33557 180 7 shows show VERB bracis-33557 180 8 spearman spearman NOUN bracis-33557 180 9 ’s ’s PART bracis-33557 180 10 correlation correlation NOUN bracis-33557 180 11 coefficient coefficient NOUN bracis-33557 180 12 between between ADP bracis-33557 180 13 the the DET bracis-33557 180 14 original original ADJ bracis-33557 180 15 ihms ihm NOUN bracis-33557 180 16 and and CCONJ bracis-33557 180 17 the the DET bracis-33557 180 18 measures measure NOUN bracis-33557 180 19 obtained obtain VERB bracis-33557 180 20 using use VERB bracis-33557 180 21 the the DET bracis-33557 180 22 meta meta ADJ bracis-33557 180 23 - - PUNCT bracis-33557 180 24 learning learn VERB bracis-33557 180 25 approach approach NOUN bracis-33557 180 26 . . PUNCT bracis-33557 181 1 values value NOUN bracis-33557 181 2 higher high ADJ bracis-33557 181 3 than than ADP bracis-33557 181 4 0.5 0.5 NUM bracis-33557 181 5 are be AUX bracis-33557 181 6 highlighted highlight VERB bracis-33557 181 7 in in ADP bracis-33557 181 8 bold bold ADJ bracis-33557 181 9 . . PUNCT bracis-33557 182 1 the the DET bracis-33557 182 2 values value NOUN bracis-33557 182 3 of of ADP bracis-33557 182 4 the the DET bracis-33557 182 5 estimated estimate VERB bracis-33557 182 6 tree tree NOUN bracis-33557 182 7 depth depth NOUN bracis-33557 182 8 measures measure NOUN bracis-33557 182 9 are be AUX bracis-33557 182 10 the the DET bracis-33557 182 11 highest high ADJ bracis-33557 182 12 , , PUNCT bracis-33557 182 13 especially especially ADV bracis-33557 182 14 for for ADP bracis-33557 182 15 the the DET bracis-33557 182 16 pruned prune VERB bracis-33557 182 17 version version NOUN bracis-33557 182 18 of of ADP bracis-33557 182 19 the the DET bracis-33557 182 20 measure measure NOUN bracis-33557 182 21 ( ( PUNCT bracis-33557 182 22 td\(_\text td\(_\text X bracis-33557 182 23 { { PUNCT bracis-33557 182 24 u}\ u}\ PROPN bracis-33557 182 25 ) ) PUNCT bracis-33557 182 26 ) ) PUNCT bracis-33557 182 27 . . PUNCT bracis-33557 183 1 this this PRON bracis-33557 183 2 happens happen VERB bracis-33557 183 3 because because SCONJ bracis-33557 183 4 , , PUNCT bracis-33557 183 5 in in ADP bracis-33557 183 6 the the DET bracis-33557 183 7 pruned prune VERB bracis-33557 183 8 version version NOUN bracis-33557 183 9 of of ADP bracis-33557 183 10 the the DET bracis-33557 183 11 tree tree NOUN bracis-33557 183 12 , , PUNCT bracis-33557 183 13 noisy noisy ADJ bracis-33557 183 14 and and CCONJ bracis-33557 183 15 outlier outlier NOUN bracis-33557 183 16 instances instance NOUN bracis-33557 183 17 tend tend VERB bracis-33557 183 18 to to PART bracis-33557 183 19 be be AUX bracis-33557 183 20 placed place VERB bracis-33557 183 21 in in ADP bracis-33557 183 22 nodes node NOUN bracis-33557 183 23 which which PRON bracis-33557 183 24 have have AUX bracis-33557 183 25 undergone undergo VERB bracis-33557 183 26 pruning pruning NOUN bracis-33557 183 27 . . PUNCT bracis-33557 184 1 therefore therefore ADV bracis-33557 184 2 , , PUNCT bracis-33557 184 3 the the DET bracis-33557 184 4 label label NOUN bracis-33557 184 5 of of ADP bracis-33557 184 6 the the DET bracis-33557 184 7 particular particular ADJ bracis-33557 184 8 instance instance NOUN bracis-33557 184 9 seems seem VERB bracis-33557 184 10 to to PART bracis-33557 184 11 matter matter VERB bracis-33557 184 12 less less ADV bracis-33557 184 13 in in ADP bracis-33557 184 14 the the DET bracis-33557 184 15 original original ADJ bracis-33557 184 16 ihm ihm ADJ bracis-33557 184 17 formulation formulation NOUN bracis-33557 184 18 . . PUNCT bracis-33557 185 1 in in ADP bracis-33557 185 2 contrast contrast NOUN bracis-33557 185 3 , , PUNCT bracis-33557 185 4 the the DET bracis-33557 185 5 formulation formulation NOUN bracis-33557 185 6 of of ADP bracis-33557 185 7 the the DET bracis-33557 185 8 original original ADJ bracis-33557 185 9 cld cld PROPN bracis-33557 185 10 , , PUNCT bracis-33557 185 11 dcp dcp NOUN bracis-33557 185 12 and and CCONJ bracis-33557 185 13 kdn kdn NOUN bracis-33557 185 14 measures measure NOUN bracis-33557 185 15 is be AUX bracis-33557 185 16 highly highly ADV bracis-33557 185 17 influenced influence VERB bracis-33557 185 18 by by ADP bracis-33557 185 19 the the DET bracis-33557 185 20 label label NOUN bracis-33557 185 21 of of ADP bracis-33557 185 22 each each DET bracis-33557 185 23 instance instance NOUN bracis-33557 185 24 where where SCONJ bracis-33557 185 25 they they PRON bracis-33557 185 26 are be AUX bracis-33557 185 27 measured measure VERB bracis-33557 185 28 . . PUNCT bracis-33557 186 1 this this PRON bracis-33557 186 2 decreases decrease VERB bracis-33557 186 3 the the DET bracis-33557 186 4 correlations correlation NOUN bracis-33557 186 5 , , PUNCT bracis-33557 186 6 especially especially ADV bracis-33557 186 7 in in ADP bracis-33557 186 8 datasets dataset NOUN bracis-33557 186 9 with with ADP bracis-33557 186 10 many many ADJ bracis-33557 186 11 instances instance NOUN bracis-33557 186 12 with with ADP bracis-33557 186 13 feature feature NOUN bracis-33557 186 14 values value NOUN bracis-33557 186 15 akin akin ADJ bracis-33557 186 16 to to ADP bracis-33557 186 17 a a DET bracis-33557 186 18 class class NOUN bracis-33557 186 19 , , PUNCT bracis-33557 186 20 despite despite SCONJ bracis-33557 186 21 being be AUX bracis-33557 186 22 originally originally ADV bracis-33557 186 23 labeled label VERB bracis-33557 186 24 into into ADP bracis-33557 186 25 another another DET bracis-33557 186 26 class class NOUN bracis-33557 186 27 in in ADP bracis-33557 186 28 the the DET bracis-33557 186 29 dataset dataset NOUN bracis-33557 186 30 . . PUNCT bracis-33557 187 1 this this PRON bracis-33557 187 2 is be AUX bracis-33557 187 3 the the DET bracis-33557 187 4 case case NOUN bracis-33557 187 5 for for ADP bracis-33557 187 6 hospital hospital NOUN bracis-33557 187 7 1 1 NUM bracis-33557 187 8 and and CCONJ bracis-33557 187 9 2 2 NUM bracis-33557 187 10 datasets dataset NOUN bracis-33557 187 11 , , PUNCT bracis-33557 187 12 where where SCONJ bracis-33557 187 13 situations situation NOUN bracis-33557 187 14 such such ADJ bracis-33557 187 15 as as ADP bracis-33557 187 16 instances instance NOUN bracis-33557 187 17 wrongly wrongly ADV bracis-33557 187 18 labeled label VERB bracis-33557 187 19 or or CCONJ bracis-33557 187 20 with with ADP bracis-33557 187 21 overlapping overlap VERB bracis-33557 187 22 feature feature NOUN bracis-33557 187 23 values value NOUN bracis-33557 187 24 are be AUX bracis-33557 187 25 more more ADV bracis-33557 187 26 common common ADJ bracis-33557 187 27 . . PUNCT bracis-33557 188 1 table table NOUN bracis-33557 188 2 3 3 NUM bracis-33557 188 3 . . X bracis-33557 188 4 spearman spearman NOUN bracis-33557 188 5 coefficient coefficient NOUN bracis-33557 188 6 obtained obtain VERB bracis-33557 188 7 for for ADP bracis-33557 188 8 the the DET bracis-33557 188 9 rf rf ADJ bracis-33557 188 10 algorithm algorithm NOUN bracis-33557 188 11 concerning concern VERB bracis-33557 188 12 predicting predict VERB bracis-33557 188 13 the the DET bracis-33557 188 14 ihms ihm NOUN bracis-33557 188 15 to to ADP bracis-33557 188 16 different different ADJ bracis-33557 188 17 datasets.full datasets.full NUM bracis-33557 188 18 size size NOUN bracis-33557 188 19 table table NOUN bracis-33557 188 20 table table NOUN bracis-33557 188 21 4 4 NUM bracis-33557 188 22 . . PUNCT bracis-33557 188 23 spearman spearman NOUN bracis-33557 188 24 coefficient coefficient NOUN bracis-33557 188 25 obtained obtain VERB bracis-33557 188 26 for for ADP bracis-33557 188 27 adjusted adjust VERB bracis-33557 188 28 measures measure NOUN bracis-33557 188 29 compared compare VERB bracis-33557 188 30 to to ADP bracis-33557 188 31 the the DET bracis-33557 188 32 original original ADJ bracis-33557 188 33 ihms ihm NOUN bracis-33557 188 34 in in ADP bracis-33557 188 35 different different ADJ bracis-33557 188 36 datasets.full datasets.full NUM bracis-33557 188 37 size size NOUN bracis-33557 188 38 table table NOUN bracis-33557 188 39 table table NOUN bracis-33557 188 40     SPACE bracis-33557 188 41 4 4 NUM bracis-33557 188 42 presents present VERB bracis-33557 188 43 the the DET bracis-33557 188 44 same same ADJ bracis-33557 188 45 results result NOUN bracis-33557 188 46 for for ADP bracis-33557 188 47 the the DET bracis-33557 188 48 adjusted adjusted ADJ bracis-33557 188 49 ihms ihm NOUN bracis-33557 188 50 : : PUNCT bracis-33557 188 51 their their PRON bracis-33557 188 52 spearman spearman NOUN bracis-33557 188 53 correlation correlation NOUN bracis-33557 188 54 to to ADP bracis-33557 188 55 the the DET bracis-33557 188 56 original original ADJ bracis-33557 188 57 ihms ihm NOUN bracis-33557 188 58 . . PUNCT bracis-33557 189 1 as as ADP bracis-33557 189 2 in in ADP bracis-33557 189 3 table table NOUN bracis-33557 189 4     SPACE bracis-33557 189 5 3 3 NUM bracis-33557 189 6 , , PUNCT bracis-33557 189 7 values value NOUN bracis-33557 189 8 higher high ADJ bracis-33557 189 9 than than ADP bracis-33557 189 10 0.5 0.5 NUM bracis-33557 189 11 are be AUX bracis-33557 189 12 boldfaced boldface VERB bracis-33557 189 13 . . PUNCT bracis-33557 190 1 more more ADV bracis-33557 190 2 boldfaced boldfaced ADJ bracis-33557 190 3 correlations correlation NOUN bracis-33557 190 4 are be AUX bracis-33557 190 5 observed observe VERB bracis-33557 190 6 here here ADV bracis-33557 190 7 . . PUNCT bracis-33557 191 1 similar similar ADJ bracis-33557 191 2 observations observation NOUN bracis-33557 191 3 concerning concern VERB bracis-33557 191 4 the the DET bracis-33557 191 5 higher high ADJ bracis-33557 191 6 correlation correlation NOUN bracis-33557 191 7 values value NOUN bracis-33557 191 8 for for ADP bracis-33557 191 9 tree tree NOUN bracis-33557 191 10 depth depth NOUN bracis-33557 191 11 - - PUNCT bracis-33557 191 12 based base VERB bracis-33557 191 13 measures measure NOUN bracis-33557 191 14 are be AUX bracis-33557 191 15 observed observe VERB bracis-33557 191 16 in in ADP bracis-33557 191 17 table table NOUN bracis-33557 191 18     SPACE bracis-33557 191 19 4 4 NUM bracis-33557 191 20 too too ADV bracis-33557 191 21 . . PUNCT bracis-33557 192 1 the the DET bracis-33557 192 2 correlations correlation NOUN bracis-33557 192 3 observed observe VERB bracis-33557 192 4 for for ADP bracis-33557 192 5 the the DET bracis-33557 192 6 adjusted adjust VERB bracis-33557 192 7 measures measure NOUN bracis-33557 192 8 are be AUX bracis-33557 192 9 generally generally ADV bracis-33557 192 10 higher high ADJ bracis-33557 192 11 than than ADP bracis-33557 192 12 those those PRON bracis-33557 192 13 observed observe VERB bracis-33557 192 14 for for ADP bracis-33557 192 15 the the DET bracis-33557 192 16 measures measure NOUN bracis-33557 192 17 estimated estimate VERB bracis-33557 192 18 by by ADP bracis-33557 192 19 the the DET bracis-33557 192 20 meta meta NOUN bracis-33557 192 21 - - PUNCT bracis-33557 192 22 regressors regressor NOUN bracis-33557 192 23 . . PUNCT bracis-33557 193 1 to to PART bracis-33557 193 2 make make VERB bracis-33557 193 3 the the DET bracis-33557 193 4 differences difference NOUN bracis-33557 193 5 clearer clear ADJ bracis-33557 193 6 , , PUNCT bracis-33557 193 7 fig fig NOUN bracis-33557 193 8 . . PUNCT bracis-33557 193 9     SPACE bracis-33557 194 1 3 3 NUM bracis-33557 194 2 plots plot VERB bracis-33557 194 3 the the DET bracis-33557 194 4 spearman spearman NOUN bracis-33557 194 5 ’s ’s PART bracis-33557 194 6 correlations correlation NOUN bracis-33557 194 7 for for ADP bracis-33557 194 8 the the DET bracis-33557 194 9 adjusted adjust VERB bracis-33557 194 10 ihm ihm NOUN bracis-33557 194 11 and and CCONJ bracis-33557 194 12 the the DET bracis-33557 194 13 meta meta NOUN bracis-33557 194 14 - - PUNCT bracis-33557 194 15 models model NOUN bracis-33557 194 16 compared compare VERB bracis-33557 194 17 to to ADP bracis-33557 194 18 the the DET bracis-33557 194 19 original original ADJ bracis-33557 194 20 values value NOUN bracis-33557 194 21 . . PUNCT bracis-33557 195 1 blue blue ADJ bracis-33557 195 2 bars bar NOUN bracis-33557 195 3 represent represent VERB bracis-33557 195 4 the the DET bracis-33557 195 5 correlation correlation NOUN bracis-33557 195 6 of of ADP bracis-33557 195 7 the the DET bracis-33557 195 8 adjusted adjust VERB bracis-33557 195 9 measures measure NOUN bracis-33557 195 10 , , PUNCT bracis-33557 195 11 while while SCONJ bracis-33557 195 12 orange orange PROPN bracis-33557 195 13 bars bar NOUN bracis-33557 195 14 denote denote VERB bracis-33557 195 15 the the DET bracis-33557 195 16 meta meta ADV bracis-33557 195 17 - - PUNCT bracis-33557 195 18 learning learn VERB bracis-33557 195 19 approach approach NOUN bracis-33557 195 20 . . PUNCT bracis-33557 196 1 only only ADV bracis-33557 196 2 for for SCONJ bracis-33557 196 3 the the DET bracis-33557 196 4 blobs blob NOUN bracis-33557 196 5 dataset dataset VERB bracis-33557 196 6 and and CCONJ bracis-33557 196 7 for for ADP bracis-33557 196 8 the the DET bracis-33557 196 9 dcp dcp NOUN bracis-33557 196 10 - - PUNCT bracis-33557 196 11 diabetes diabetes NOUN bracis-33557 196 12 combination combination NOUN bracis-33557 196 13 were be AUX bracis-33557 196 14 correlations correlation NOUN bracis-33557 196 15 of of ADP bracis-33557 196 16 the the DET bracis-33557 196 17 meta meta NOUN bracis-33557 196 18 - - PUNCT bracis-33557 196 19 models model NOUN bracis-33557 196 20 higher high ADJ bracis-33557 196 21 than than ADP bracis-33557 196 22 those those PRON bracis-33557 196 23 of of ADP bracis-33557 196 24 the the DET bracis-33557 196 25 adjusted adjust VERB bracis-33557 196 26 ihms ihm NOUN bracis-33557 196 27 . . PUNCT bracis-33557 197 1 the the DET bracis-33557 197 2 blobs blob NOUN bracis-33557 197 3 dataset dataset VERB bracis-33557 197 4 has have VERB bracis-33557 197 5 difficult difficult ADJ bracis-33557 197 6 instances instance NOUN bracis-33557 197 7 concentrated concentrate VERB bracis-33557 197 8 on on ADP bracis-33557 197 9 the the DET bracis-33557 197 10 border border NOUN bracis-33557 197 11 of of ADP bracis-33557 197 12 the the DET bracis-33557 197 13 classes class NOUN bracis-33557 197 14 , , PUNCT bracis-33557 197 15 while while SCONJ bracis-33557 197 16 the the DET bracis-33557 197 17 other other ADJ bracis-33557 197 18 datasets dataset NOUN bracis-33557 197 19 may may AUX bracis-33557 197 20 pose pose VERB bracis-33557 197 21 other other ADJ bracis-33557 197 22 sources source NOUN bracis-33557 197 23 of of ADP bracis-33557 197 24 difficulties difficulty NOUN bracis-33557 197 25 which which PRON bracis-33557 197 26 are be AUX bracis-33557 197 27 not not PART bracis-33557 197 28 captured capture VERB bracis-33557 197 29 when when SCONJ bracis-33557 197 30 the the DET bracis-33557 197 31 labels label NOUN bracis-33557 197 32 are be AUX bracis-33557 197 33 absent absent ADJ bracis-33557 197 34 , , PUNCT bracis-33557 197 35 such such ADJ bracis-33557 197 36 as as ADP bracis-33557 197 37 label label NOUN bracis-33557 197 38 noise noise NOUN bracis-33557 197 39 . . PUNCT bracis-33557 198 1 fig fig NOUN bracis-33557 198 2 . . PUNCT bracis-33557 199 1 3 3 X bracis-33557 199 2 . . X bracis-33557 199 3 spearman spearman NOUN bracis-33557 199 4 ’s ’s PART bracis-33557 199 5 correlation correlation NOUN bracis-33557 199 6 applied apply VERB bracis-33557 199 7 to to ADP bracis-33557 199 8 the the DET bracis-33557 199 9 adjusted adjust VERB bracis-33557 199 10 ihm ihm NOUN bracis-33557 199 11 vs. vs. ADP bracis-33557 199 12 the the DET bracis-33557 199 13 original original ADJ bracis-33557 199 14 ihm ihm NOUN bracis-33557 199 15 ( ( PUNCT bracis-33557 199 16 blue blue ADJ bracis-33557 199 17 bars bar NOUN bracis-33557 199 18 ) ) PUNCT bracis-33557 199 19 and and CCONJ bracis-33557 199 20 the the DET bracis-33557 199 21 predicted predict VERB bracis-33557 199 22 vs. vs. ADP bracis-33557 199 23 expected expect VERB bracis-33557 199 24 values value NOUN bracis-33557 199 25 from from ADP bracis-33557 199 26 mtl mtl PROPN bracis-33557 199 27 ( ( PUNCT bracis-33557 199 28 orange orange PROPN bracis-33557 199 29 bars bar NOUN bracis-33557 199 30 ) ) PUNCT bracis-33557 199 31 . . PUNCT bracis-33557 200 1 ( ( PUNCT bracis-33557 200 2 color color NOUN bracis-33557 200 3 figure figure NOUN bracis-33557 200 4 online online ADV bracis-33557 200 5 ) ) PUNCT bracis-33557 200 6 full full ADJ bracis-33557 200 7 size size NOUN bracis-33557 200 8 image image NOUN bracis-33557 200 9 figure figure NOUN bracis-33557 200 10     SPACE bracis-33557 200 11 4 4 NUM bracis-33557 200 12 shows show VERB bracis-33557 200 13 the the DET bracis-33557 200 14 instances instance NOUN bracis-33557 200 15 in in ADP bracis-33557 200 16 the the DET bracis-33557 200 17 blobs blob NOUN bracis-33557 200 18 dataset dataset VERB bracis-33557 200 19 colored color VERB bracis-33557 200 20 by by ADP bracis-33557 200 21 the the DET bracis-33557 200 22 hardness hardness NOUN bracis-33557 200 23 of of ADP bracis-33557 200 24 the the DET bracis-33557 200 25 original original ADJ bracis-33557 200 26 ihm ihm NOUN bracis-33557 200 27 ( ( PUNCT bracis-33557 200 28 in in ADP bracis-33557 200 29 the the DET bracis-33557 200 30 left left NOUN bracis-33557 200 31 ) ) PUNCT bracis-33557 200 32 followed follow VERB bracis-33557 200 33 by by ADP bracis-33557 200 34 the the DET bracis-33557 200 35 adjusted adjust VERB bracis-33557 200 36 ihm ihm NOUN bracis-33557 200 37 ( ( PUNCT bracis-33557 200 38 in in ADP bracis-33557 200 39 the the DET bracis-33557 200 40 center center NOUN bracis-33557 200 41 ) ) PUNCT bracis-33557 200 42 and and CCONJ bracis-33557 200 43 the the DET bracis-33557 200 44 meta meta ADJ bracis-33557 200 45 - - PUNCT bracis-33557 200 46 learning learn VERB bracis-33557 200 47 approach approach NOUN bracis-33557 200 48 ( ( PUNCT bracis-33557 200 49 in in ADP bracis-33557 200 50 the the DET bracis-33557 200 51 right right NOUN bracis-33557 200 52 ) ) PUNCT bracis-33557 200 53 . . PUNCT bracis-33557 201 1 this this PRON bracis-33557 201 2 can can AUX bracis-33557 201 3 be be AUX bracis-33557 201 4 done do VERB bracis-33557 201 5 for for ADP bracis-33557 201 6 this this DET bracis-33557 201 7 dataset dataset NOUN bracis-33557 201 8 , , PUNCT bracis-33557 201 9 as as SCONJ bracis-33557 201 10 it it PRON bracis-33557 201 11 is be AUX bracis-33557 201 12 bi bi ADJ bracis-33557 201 13 - - ADJ bracis-33557 201 14 dimensional dimensional ADJ bracis-33557 201 15 . . PUNCT bracis-33557 202 1 the the PRON bracis-33557 202 2 harder hard ADJ bracis-33557 202 3 the the DET bracis-33557 202 4 instance instance NOUN bracis-33557 202 5 is be AUX bracis-33557 202 6 to to PART bracis-33557 202 7 classify classify VERB bracis-33557 202 8 , , PUNCT bracis-33557 202 9 the the PRON bracis-33557 202 10 more more ADV bracis-33557 202 11 intense intense ADJ bracis-33557 202 12 it it PRON bracis-33557 202 13 is be AUX bracis-33557 202 14 colored color VERB bracis-33557 202 15 in in ADP bracis-33557 202 16 red red PROPN bracis-33557 202 17 . . PUNCT bracis-33557 203 1 in in ADP bracis-33557 203 2 contrast contrast NOUN bracis-33557 203 3 , , PUNCT bracis-33557 203 4 instances instance NOUN bracis-33557 203 5 that that PRON bracis-33557 203 6 are be AUX bracis-33557 203 7 easier easy ADJ bracis-33557 203 8 to to PART bracis-33557 203 9 classify classify VERB bracis-33557 203 10 are be AUX bracis-33557 203 11 filled fill VERB bracis-33557 203 12 with with ADP bracis-33557 203 13 darker dark ADJ bracis-33557 203 14 blue blue NOUN bracis-33557 203 15 . . PUNCT bracis-33557 204 1 the the DET bracis-33557 204 2 central central ADJ bracis-33557 204 3 areas area NOUN bracis-33557 204 4 of of ADP bracis-33557 204 5 the the DET bracis-33557 204 6 plots plot NOUN bracis-33557 204 7 contain contain VERB bracis-33557 204 8 the the DET bracis-33557 204 9 overlapping overlap VERB bracis-33557 204 10 region region NOUN bracis-33557 204 11 between between ADP bracis-33557 204 12 the the DET bracis-33557 204 13 three three NUM bracis-33557 204 14 classes class NOUN bracis-33557 204 15 ( ( PUNCT bracis-33557 204 16 see see VERB bracis-33557 204 17 fig fig NOUN bracis-33557 204 18 . . PUNCT bracis-33557 204 19     SPACE bracis-33557 204 20 2 2 X bracis-33557 204 21 ) ) PUNCT bracis-33557 204 22 and and CCONJ bracis-33557 204 23 , , PUNCT bracis-33557 204 24 therefore therefore ADV bracis-33557 204 25 , , PUNCT bracis-33557 204 26 are be AUX bracis-33557 204 27 harder hard ADJ bracis-33557 204 28 to to PART bracis-33557 204 29 classify classify VERB bracis-33557 204 30 . . PUNCT bracis-33557 205 1 the the DET bracis-33557 205 2 first first ADJ bracis-33557 205 3 row row NOUN bracis-33557 205 4 corresponds correspond VERB bracis-33557 205 5 to to ADP bracis-33557 205 6 the the DET bracis-33557 205 7 kdn kdn PROPN bracis-33557 205 8 measure measure NOUN bracis-33557 205 9 , , PUNCT bracis-33557 205 10 while while SCONJ bracis-33557 205 11 the the DET bracis-33557 205 12 second second NOUN bracis-33557 205 13 is be AUX bracis-33557 205 14 the the DET bracis-33557 205 15 td\(_\text td\(_\text ADJ bracis-33557 205 16 { { PUNCT bracis-33557 205 17 u}\ u}\ PROPN bracis-33557 205 18 ) ) PUNCT bracis-33557 205 19 measure measure NOUN bracis-33557 205 20 . . PUNCT bracis-33557 206 1 for for ADP bracis-33557 206 2 kdn kdn NOUN bracis-33557 206 3 , , PUNCT bracis-33557 206 4 it it PRON bracis-33557 206 5 is be AUX bracis-33557 206 6 clear clear ADJ bracis-33557 206 7 that that SCONJ bracis-33557 206 8 the the DET bracis-33557 206 9 hardest hard ADJ bracis-33557 206 10 instances instance NOUN bracis-33557 206 11 are be AUX bracis-33557 206 12 those those PRON bracis-33557 206 13 in in ADP bracis-33557 206 14 the the DET bracis-33557 206 15 border border NOUN bracis-33557 206 16 of of ADP bracis-33557 206 17 the the DET bracis-33557 206 18 classes class NOUN bracis-33557 206 19 . . PUNCT bracis-33557 207 1 for for ADP bracis-33557 207 2 td\(_\text td\(_\text X bracis-33557 207 3 { { PUNCT bracis-33557 207 4 u}\ u}\ PROPN bracis-33557 207 5 ) ) PUNCT bracis-33557 207 6 the the DET bracis-33557 207 7 pattern pattern NOUN bracis-33557 207 8 observed observe VERB bracis-33557 207 9 in in ADP bracis-33557 207 10 the the DET bracis-33557 207 11 three three NUM bracis-33557 207 12 approaches approach NOUN bracis-33557 207 13 shows show VERB bracis-33557 207 14 that that SCONJ bracis-33557 207 15 the the DET bracis-33557 207 16 harness harness ADJ bracis-33557 207 17 level level NOUN bracis-33557 207 18 is be AUX bracis-33557 207 19 related relate VERB bracis-33557 207 20 to to ADP bracis-33557 207 21 partition partition NOUN bracis-33557 207 22 derived derive VERB bracis-33557 207 23 from from ADP bracis-33557 207 24 the the DET bracis-33557 207 25 decision decision NOUN bracis-33557 207 26 tree tree NOUN bracis-33557 207 27 classification classification NOUN bracis-33557 207 28 . . PUNCT bracis-33557 208 1 all all DET bracis-33557 208 2 measures measure NOUN bracis-33557 208 3 show show VERB bracis-33557 208 4 similar similar ADJ bracis-33557 208 5 behaviors behavior NOUN bracis-33557 208 6 . . PUNCT bracis-33557 209 1 however however ADV bracis-33557 209 2 , , PUNCT bracis-33557 209 3 for for ADP bracis-33557 209 4 the the DET bracis-33557 209 5 adjusted adjust VERB bracis-33557 209 6 kdn kdn NOUN bracis-33557 209 7 measure measure NOUN bracis-33557 209 8 , , PUNCT bracis-33557 209 9 more more ADJ bracis-33557 209 10 central central ADJ bracis-33557 209 11 instances instance NOUN bracis-33557 209 12 have have VERB bracis-33557 209 13 higher high ADJ bracis-33557 209 14 ihm ihm ADJ bracis-33557 209 15 values value NOUN bracis-33557 209 16 compared compare VERB bracis-33557 209 17 to to ADP bracis-33557 209 18 the the DET bracis-33557 209 19 other other ADJ bracis-33557 209 20 measures measure NOUN bracis-33557 209 21 . . PUNCT bracis-33557 210 1 it it PRON bracis-33557 210 2 is be AUX bracis-33557 210 3 important important ADJ bracis-33557 210 4 to to PART bracis-33557 210 5 note note VERB bracis-33557 210 6 that that SCONJ bracis-33557 210 7 since since SCONJ bracis-33557 210 8 the the DET bracis-33557 210 9 adjusted adjust VERB bracis-33557 210 10 measures measure NOUN bracis-33557 210 11 can can AUX bracis-33557 210 12 vary vary VERB bracis-33557 210 13 on on ADP bracis-33557 210 14 a a DET bracis-33557 210 15 different different ADJ bracis-33557 210 16 scale scale NOUN bracis-33557 210 17 from from ADP bracis-33557 210 18 the the DET bracis-33557 210 19 original original ADJ bracis-33557 210 20 ihm ihm NOUN bracis-33557 210 21 , , PUNCT bracis-33557 210 22 the the DET bracis-33557 210 23 results result NOUN bracis-33557 210 24 presented present VERB bracis-33557 210 25 in in ADP bracis-33557 210 26 the the DET bracis-33557 210 27 plots plot NOUN bracis-33557 210 28 were be AUX bracis-33557 210 29 normalized normalize VERB bracis-33557 210 30 between between ADP bracis-33557 210 31 0 0 NUM bracis-33557 210 32 and and CCONJ bracis-33557 210 33 1 1 NUM bracis-33557 210 34 to to PART bracis-33557 210 35 allow allow VERB bracis-33557 210 36 a a DET bracis-33557 210 37 direct direct ADJ bracis-33557 210 38 comparison comparison NOUN bracis-33557 210 39 . . PUNCT bracis-33557 211 1 fig fig NOUN bracis-33557 211 2 . . PUNCT bracis-33557 212 1 4 4 X bracis-33557 212 2 . . X bracis-33557 212 3 visualization visualization NOUN bracis-33557 212 4 of of ADP bracis-33557 212 5 the the DET bracis-33557 212 6 measures measure NOUN bracis-33557 212 7 kdn kdn NOUN bracis-33557 212 8 ( ( PUNCT bracis-33557 212 9 top top NOUN bracis-33557 212 10 ) ) PUNCT bracis-33557 212 11 and and CCONJ bracis-33557 212 12 td\(_\text td\(_\text ADJ bracis-33557 212 13 { { PUNCT bracis-33557 212 14 u u NOUN bracis-33557 212 15 } } PUNCT bracis-33557 212 16 \ \ NOUN bracis-33557 212 17 ) ) PUNCT bracis-33557 212 18 ( ( PUNCT bracis-33557 212 19 bottom bottom ADJ bracis-33557 212 20 ): ): PUNCT bracis-33557 212 21 original original ADJ bracis-33557 212 22 ihm ihm NOUN bracis-33557 212 23 ( ( PUNCT bracis-33557 212 24 left left ADV bracis-33557 212 25 ) ) PUNCT bracis-33557 212 26 , , PUNCT bracis-33557 212 27 adjusted adjust VERB bracis-33557 212 28 ihm ihm PROPN bracis-33557 212 29 ( ( PUNCT bracis-33557 212 30 middle middle NOUN bracis-33557 212 31 ) ) PUNCT bracis-33557 212 32 and and CCONJ bracis-33557 212 33 meta meta ADJ bracis-33557 212 34 - - PUNCT bracis-33557 212 35 learning learn VERB bracis-33557 212 36 approach approach NOUN bracis-33557 212 37 ( ( PUNCT bracis-33557 212 38 right right ADJ bracis-33557 212 39 ) ) PUNCT bracis-33557 212 40 for for SCONJ bracis-33557 212 41 the the DET bracis-33557 212 42 blobs blob NOUN bracis-33557 212 43 dataset dataset VERB bracis-33557 212 44 . . PUNCT bracis-33557 213 1 ( ( PUNCT bracis-33557 213 2 color color NOUN bracis-33557 213 3 figure figure NOUN bracis-33557 213 4 online online ADV bracis-33557 213 5 ) ) PUNCT bracis-33557 213 6 full full ADJ bracis-33557 213 7 size size NOUN bracis-33557 213 8 image image NOUN bracis-33557 213 9 4.3 4.3 NUM bracis-33557 213 10 discussion discussion NOUN bracis-33557 213 11 considering consider VERB bracis-33557 213 12 the the DET bracis-33557 213 13 difference difference NOUN bracis-33557 213 14 in in ADP bracis-33557 213 15 the the DET bracis-33557 213 16 nature nature NOUN bracis-33557 213 17 of of ADP bracis-33557 213 18 the the DET bracis-33557 213 19 datasets dataset NOUN bracis-33557 213 20 , , PUNCT bracis-33557 213 21 where where SCONJ bracis-33557 213 22 the the DET bracis-33557 213 23 blobs blob NOUN bracis-33557 213 24 were be AUX bracis-33557 213 25 artificially artificially ADV bracis-33557 213 26 designed design VERB bracis-33557 213 27 with with ADP bracis-33557 213 28 three three NUM bracis-33557 213 29 classes class NOUN bracis-33557 213 30 and and CCONJ bracis-33557 213 31 the the DET bracis-33557 213 32 other other ADJ bracis-33557 213 33 are be AUX bracis-33557 213 34 real real ADJ bracis-33557 213 35 - - PUNCT bracis-33557 213 36 world world NOUN bracis-33557 213 37 health health NOUN bracis-33557 213 38 data datum NOUN bracis-33557 213 39 , , PUNCT bracis-33557 213 40 the the DET bracis-33557 213 41 spearman spearman NOUN bracis-33557 213 42 ’s ’s PART bracis-33557 213 43 correlation correlation NOUN bracis-33557 213 44 in in ADP bracis-33557 213 45 fig fig NOUN bracis-33557 213 46 . . PUNCT bracis-33557 213 47     SPACE bracis-33557 214 1 3 3 NUM bracis-33557 214 2 shows show VERB bracis-33557 214 3 that that SCONJ bracis-33557 214 4 the the DET bracis-33557 214 5 mtl mtl PROPN bracis-33557 214 6 achieves achieve VERB bracis-33557 214 7 more more ADJ bracis-33557 214 8 convergent convergent ADJ bracis-33557 214 9 result result NOUN bracis-33557 214 10 than than ADP bracis-33557 214 11 the the DET bracis-33557 214 12 adjusted adjust VERB bracis-33557 214 13 ihm ihm NOUN bracis-33557 214 14 in in ADP bracis-33557 214 15 the the DET bracis-33557 214 16 blobs blob NOUN bracis-33557 214 17 dataset dataset VERB bracis-33557 214 18 for for ADP bracis-33557 214 19 great great ADJ bracis-33557 214 20 part part NOUN bracis-33557 214 21 of of ADP bracis-33557 214 22 the the DET bracis-33557 214 23 measures measure NOUN bracis-33557 214 24 . . PUNCT bracis-33557 215 1 conversely conversely ADV bracis-33557 215 2 , , PUNCT bracis-33557 215 3 for for ADP bracis-33557 215 4 real real ADJ bracis-33557 215 5 datasets dataset NOUN bracis-33557 215 6 , , PUNCT bracis-33557 215 7 the the DET bracis-33557 215 8 adjusted adjust VERB bracis-33557 215 9 ihm ihm NOUN bracis-33557 215 10 is be AUX bracis-33557 215 11 more more ADV bracis-33557 215 12 associated associate VERB bracis-33557 215 13 with with ADP bracis-33557 215 14 the the DET bracis-33557 215 15 original original ADJ bracis-33557 215 16 measure measure NOUN bracis-33557 215 17 for for ADP bracis-33557 215 18 almost almost ADV bracis-33557 215 19 all all PRON bracis-33557 215 20 measures measure NOUN bracis-33557 215 21 . . PUNCT bracis-33557 216 1 the the DET bracis-33557 216 2 tree tree NOUN bracis-33557 216 3 - - PUNCT bracis-33557 216 4 depth depth NOUN bracis-33557 216 5 measures measure NOUN bracis-33557 216 6 had have VERB bracis-33557 216 7 the the DET bracis-33557 216 8 highest high ADJ bracis-33557 216 9 correlations correlation NOUN bracis-33557 216 10 to to ADP bracis-33557 216 11 the the DET bracis-33557 216 12 original original ADJ bracis-33557 216 13 measures measure NOUN bracis-33557 216 14 for for ADP bracis-33557 216 15 the the DET bracis-33557 216 16 adjusted adjust VERB bracis-33557 216 17 ihm ihm NOUN bracis-33557 216 18 and and CCONJ bracis-33557 216 19 the the DET bracis-33557 216 20 mtl mtl PROPN bracis-33557 216 21 approach approach NOUN bracis-33557 216 22 . . PUNCT bracis-33557 217 1 this this PRON bracis-33557 217 2 is be AUX bracis-33557 217 3 mostly mostly ADV bracis-33557 217 4 related relate VERB bracis-33557 217 5 to to ADP bracis-33557 217 6 the the DET bracis-33557 217 7 fact fact NOUN bracis-33557 217 8 that that SCONJ bracis-33557 217 9 the the DET bracis-33557 217 10 original original ADJ bracis-33557 217 11 tree tree NOUN bracis-33557 217 12 - - PUNCT bracis-33557 217 13 depth depth NOUN bracis-33557 217 14 measures measure NOUN bracis-33557 217 15 do do AUX bracis-33557 217 16 not not PART bracis-33557 217 17 depend depend VERB bracis-33557 217 18 so so ADV bracis-33557 217 19 directly directly ADV bracis-33557 217 20 on on ADP bracis-33557 217 21 the the DET bracis-33557 217 22 label label NOUN bracis-33557 217 23 of of ADP bracis-33557 217 24 the the DET bracis-33557 217 25 instances instance NOUN bracis-33557 217 26 . . PUNCT bracis-33557 218 1 the the DET bracis-33557 218 2 other other ADJ bracis-33557 218 3 measures measure NOUN bracis-33557 218 4 all all PRON bracis-33557 218 5 regard regard VERB bracis-33557 218 6 whether whether SCONJ bracis-33557 218 7 the the DET bracis-33557 218 8 labels label NOUN bracis-33557 218 9 of of ADP bracis-33557 218 10 some some DET bracis-33557 218 11 vicinity vicinity NOUN bracis-33557 218 12 are be AUX bracis-33557 218 13 in in ADP bracis-33557 218 14 accordance accordance NOUN bracis-33557 218 15 with with ADP bracis-33557 218 16 the the DET bracis-33557 218 17 registered register VERB bracis-33557 218 18 label label NOUN bracis-33557 218 19 of of ADP bracis-33557 218 20 the the DET bracis-33557 218 21 instance instance NOUN bracis-33557 218 22 . . PUNCT bracis-33557 219 1 this this PRON bracis-33557 219 2 makes make VERB bracis-33557 219 3 them they PRON bracis-33557 219 4 deviate deviate VERB bracis-33557 219 5 more more ADJ bracis-33557 219 6 for for ADP bracis-33557 219 7 instances instance NOUN bracis-33557 219 8 that that PRON bracis-33557 219 9 are be AUX bracis-33557 219 10 mislabeled mislabele VERB bracis-33557 219 11 , , PUNCT bracis-33557 219 12 for for ADP bracis-33557 219 13 instance instance NOUN bracis-33557 219 14 . . PUNCT bracis-33557 220 1 this this PRON bracis-33557 220 2 can can AUX bracis-33557 220 3 be be AUX bracis-33557 220 4 observed observe VERB bracis-33557 220 5 in in ADP bracis-33557 220 6 fig fig NOUN bracis-33557 220 7 . . PUNCT bracis-33557 220 8     SPACE bracis-33557 221 1 5 5 NUM bracis-33557 221 2 , , PUNCT bracis-33557 221 3 where where SCONJ bracis-33557 221 4 the the DET bracis-33557 221 5 original original ADJ bracis-33557 221 6 and and CCONJ bracis-33557 221 7 estimated estimated ADJ bracis-33557 221 8 ihms ihm NOUN bracis-33557 221 9 kdn kdn PROPN bracis-33557 221 10 ( ( PUNCT bracis-33557 221 11 in in ADP bracis-33557 221 12 the the DET bracis-33557 221 13 top top NOUN bracis-33557 221 14 ) ) PUNCT bracis-33557 221 15 and and CCONJ bracis-33557 221 16 td\(_\text td\(_\text NOUN bracis-33557 221 17 { { PUNCT bracis-33557 221 18 u}\ u}\ PROPN bracis-33557 221 19 ) ) PUNCT bracis-33557 221 20 ( ( PUNCT bracis-33557 221 21 in in ADP bracis-33557 221 22 the the DET bracis-33557 221 23 bottom bottom NOUN bracis-33557 221 24 ) ) PUNCT bracis-33557 221 25 are be AUX bracis-33557 221 26 contraposed contrapose VERB bracis-33557 221 27 for for ADP bracis-33557 221 28 all all DET bracis-33557 221 29 instances instance NOUN bracis-33557 221 30 of of ADP bracis-33557 221 31 the the DET bracis-33557 221 32 blobs blob NOUN bracis-33557 221 33 dataset dataset VERB bracis-33557 221 34 . . PUNCT bracis-33557 222 1 in in ADP bracis-33557 222 2 the the DET bracis-33557 222 3 x x NOUN bracis-33557 222 4 - - NOUN bracis-33557 222 5 axis axis ADJ bracis-33557 222 6 , , PUNCT bracis-33557 222 7 we we PRON bracis-33557 222 8 have have VERB bracis-33557 222 9 the the DET bracis-33557 222 10 original original ADJ bracis-33557 222 11 measures measure NOUN bracis-33557 222 12 , , PUNCT bracis-33557 222 13 whilst whilst ADV bracis-33557 222 14 , , PUNCT bracis-33557 222 15 in in ADP bracis-33557 222 16 the the DET bracis-33557 222 17 y y NOUN bracis-33557 222 18 - - PUNCT bracis-33557 222 19 axis axis NOUN bracis-33557 222 20 , , PUNCT bracis-33557 222 21 the the DET bracis-33557 222 22 proposed propose VERB bracis-33557 222 23 counterparts counterpart NOUN bracis-33557 222 24 are be AUX bracis-33557 222 25 taken take VERB bracis-33557 222 26 . . PUNCT bracis-33557 223 1 the the DET bracis-33557 223 2 adjusted adjust VERB bracis-33557 223 3 kdn kdn NOUN bracis-33557 223 4 is be AUX bracis-33557 223 5 normalized normalize VERB bracis-33557 223 6 between between ADP bracis-33557 223 7 0 0 NUM bracis-33557 223 8 and and CCONJ bracis-33557 223 9 1 1 NUM bracis-33557 223 10 for for ADP bracis-33557 223 11 direct direct ADJ bracis-33557 223 12 comparison comparison NOUN bracis-33557 223 13 . . PUNCT bracis-33557 224 1 fig fig NOUN bracis-33557 224 2 . . PUNCT bracis-33557 225 1 5 5 X bracis-33557 225 2 . . PUNCT bracis-33557 225 3 the the DET bracis-33557 225 4 adjusted adjusted ADJ bracis-33557 225 5 ihm ihm NOUN bracis-33557 225 6 vs vs ADP bracis-33557 225 7 the the DET bracis-33557 225 8 original original ADJ bracis-33557 225 9 ihm ihm NOUN bracis-33557 225 10 and and CCONJ bracis-33557 225 11 meta meta ADV bracis-33557 225 12 - - PUNCT bracis-33557 225 13 learning learn VERB bracis-33557 225 14 prediction prediction NOUN bracis-33557 225 15 vs vs ADP bracis-33557 225 16 original original ADJ bracis-33557 225 17 ihm ihm NOUN bracis-33557 225 18 for for SCONJ bracis-33557 225 19 the the DET bracis-33557 225 20 blobs blob NOUN bracis-33557 225 21 dataset dataset VERB bracis-33557 225 22 . . PUNCT bracis-33557 226 1 full full ADJ bracis-33557 226 2 size size NOUN bracis-33557 226 3 image image NOUN bracis-33557 226 4 for for ADP bracis-33557 226 5 the the DET bracis-33557 226 6 measure measure NOUN bracis-33557 226 7 kdn kdn PROPN bracis-33557 226 8 , , PUNCT bracis-33557 226 9 one one PRON bracis-33557 226 10 can can AUX bracis-33557 226 11 observe observe VERB bracis-33557 226 12 that that SCONJ bracis-33557 226 13 as as SCONJ bracis-33557 226 14 the the DET bracis-33557 226 15 hardness hardness NOUN bracis-33557 226 16 to to PART bracis-33557 226 17 classify classify VERB bracis-33557 226 18 the the DET bracis-33557 226 19 instances instance NOUN bracis-33557 226 20 grows grow VERB bracis-33557 226 21 , , PUNCT bracis-33557 226 22 both both CCONJ bracis-33557 226 23 the the DET bracis-33557 226 24 adjusted adjust VERB bracis-33557 226 25 and and CCONJ bracis-33557 226 26 the the DET bracis-33557 226 27 original original ADJ bracis-33557 226 28 ihm ihm NOUN bracis-33557 226 29 increase increase VERB bracis-33557 226 30 their their PRON bracis-33557 226 31 values value NOUN bracis-33557 226 32 , , PUNCT bracis-33557 226 33 reaching reach VERB bracis-33557 226 34 their their PRON bracis-33557 226 35 peak peak NOUN bracis-33557 226 36 in in ADP bracis-33557 226 37 the the DET bracis-33557 226 38 middle middle NOUN bracis-33557 226 39 of of ADP bracis-33557 226 40 the the DET bracis-33557 226 41 scale scale NOUN bracis-33557 226 42 . . PUNCT bracis-33557 227 1 after after ADP bracis-33557 227 2 that that PRON bracis-33557 227 3 , , PUNCT bracis-33557 227 4 the the DET bracis-33557 227 5 value value NOUN bracis-33557 227 6 of of ADP bracis-33557 227 7 the the DET bracis-33557 227 8 estimated estimate VERB bracis-33557 227 9 ihm ihm NOUN bracis-33557 227 10 assumes assume VERB bracis-33557 227 11 the the DET bracis-33557 227 12 opposite opposite ADJ bracis-33557 227 13 direction direction NOUN bracis-33557 227 14 of of ADP bracis-33557 227 15 the the DET bracis-33557 227 16 original original ADJ bracis-33557 227 17 measure measure NOUN bracis-33557 227 18 . . PUNCT bracis-33557 228 1 this this DET bracis-33557 228 2 result result NOUN bracis-33557 228 3 is be AUX bracis-33557 228 4 expected expect VERB bracis-33557 228 5 considering consider VERB bracis-33557 228 6 the the DET bracis-33557 228 7 unlabeled unlabeled ADJ bracis-33557 228 8 data datum NOUN bracis-33557 228 9 , , PUNCT bracis-33557 228 10 given give VERB bracis-33557 228 11 that that SCONJ bracis-33557 228 12 instances instance NOUN bracis-33557 228 13 harder hard ADV bracis-33557 228 14 to to PART bracis-33557 228 15 classify classify VERB bracis-33557 228 16 without without ADP bracis-33557 228 17 a a DET bracis-33557 228 18 label label NOUN bracis-33557 228 19 will will AUX bracis-33557 228 20 be be AUX bracis-33557 228 21 predicted predict VERB bracis-33557 228 22 as as ADP bracis-33557 228 23 belonging belong VERB bracis-33557 228 24 to to ADP bracis-33557 228 25 any any DET bracis-33557 228 26 other other ADJ bracis-33557 228 27 class class NOUN bracis-33557 228 28 rather rather ADV bracis-33557 228 29 than than ADP bracis-33557 228 30 being be AUX bracis-33557 228 31 an an DET bracis-33557 228 32 outlier outlier NOUN bracis-33557 228 33 from from ADP bracis-33557 228 34 a a DET bracis-33557 228 35 specific specific ADJ bracis-33557 228 36 class class NOUN bracis-33557 228 37 . . PUNCT bracis-33557 229 1 conversely conversely ADV bracis-33557 229 2 , , PUNCT bracis-33557 229 3 in in ADP bracis-33557 229 4 the the DET bracis-33557 229 5 td td NOUN bracis-33557 229 6 unpruned unpruned ADJ bracis-33557 229 7 graphs graph NOUN bracis-33557 229 8 , , PUNCT bracis-33557 229 9 the the DET bracis-33557 229 10 results result NOUN bracis-33557 229 11 indicate indicate VERB bracis-33557 229 12 that that SCONJ bracis-33557 229 13 the the DET bracis-33557 229 14 hardness hardness NOUN bracis-33557 229 15 in in ADP bracis-33557 229 16 classification classification NOUN bracis-33557 229 17 is be AUX bracis-33557 229 18 independent independent ADJ bracis-33557 229 19 of of ADP bracis-33557 229 20 the the DET bracis-33557 229 21 class class NOUN bracis-33557 229 22 being be AUX bracis-33557 229 23 known know VERB bracis-33557 229 24 or or CCONJ bracis-33557 229 25 not not PART bracis-33557 229 26 . . PUNCT bracis-33557 230 1 for for ADP bracis-33557 230 2 both both DET bracis-33557 230 3 adjusted adjust VERB bracis-33557 230 4 ihm ihm NOUN bracis-33557 230 5 and and CCONJ bracis-33557 230 6 meta meta ADV bracis-33557 230 7 - - PUNCT bracis-33557 230 8 learning learn VERB bracis-33557 230 9 approaches approach NOUN bracis-33557 230 10 , , PUNCT bracis-33557 230 11 there there PRON bracis-33557 230 12 is be VERB bracis-33557 230 13 some some DET bracis-33557 230 14 linearity linearity NOUN bracis-33557 230 15 between between ADP bracis-33557 230 16 those those DET bracis-33557 230 17 measures measure NOUN bracis-33557 230 18 and and CCONJ bracis-33557 230 19 the the DET bracis-33557 230 20 original original ADJ bracis-33557 230 21 ihm ihm NOUN bracis-33557 230 22 . . PUNCT bracis-33557 231 1 it it PRON bracis-33557 231 2 means mean VERB bracis-33557 231 3 that that SCONJ bracis-33557 231 4 the the DET bracis-33557 231 5 proposed propose VERB bracis-33557 231 6 measures measure NOUN bracis-33557 231 7 for for ADP bracis-33557 231 8 unlabeled unlabeled ADJ bracis-33557 231 9 data datum NOUN bracis-33557 231 10 capture capture VERB bracis-33557 231 11 the the DET bracis-33557 231 12 increase increase NOUN bracis-33557 231 13 of of ADP bracis-33557 231 14 hardness hardness NOUN bracis-33557 231 15 for for ADP bracis-33557 231 16 classification classification NOUN bracis-33557 231 17 problems problem NOUN bracis-33557 231 18 equivalent equivalent ADJ bracis-33557 231 19 to to ADP bracis-33557 231 20 the the DET bracis-33557 231 21 increase increase NOUN bracis-33557 231 22 in in ADP bracis-33557 231 23 hardness hardness NOUN bracis-33557 231 24 when when SCONJ bracis-33557 231 25 the the DET bracis-33557 231 26 class class NOUN bracis-33557 231 27 is be AUX bracis-33557 231 28 given give VERB bracis-33557 231 29 for for ADP bracis-33557 231 30 this this DET bracis-33557 231 31 measure measure NOUN bracis-33557 231 32 . . PUNCT bracis-33557 232 1 this this DET bracis-33557 232 2 result result NOUN bracis-33557 232 3 can can AUX bracis-33557 232 4 be be AUX bracis-33557 232 5 expected expect VERB bracis-33557 232 6 considering consider VERB bracis-33557 232 7 the the DET bracis-33557 232 8 nature nature NOUN bracis-33557 232 9 of of ADP bracis-33557 232 10 the the DET bracis-33557 232 11 measure measure NOUN bracis-33557 232 12 . . PUNCT bracis-33557 233 1 the the DET bracis-33557 233 2 cld cld PROPN bracis-33557 233 3 measure measure VERB bracis-33557 233 4 , , PUNCT bracis-33557 233 5 the the DET bracis-33557 233 6 only only ADJ bracis-33557 233 7 measure measure NOUN bracis-33557 233 8 using use VERB bracis-33557 233 9 likelihood likelihood NOUN bracis-33557 233 10 as as ADP bracis-33557 233 11 the the DET bracis-33557 233 12 metric metric NOUN bracis-33557 233 13 , , PUNCT bracis-33557 233 14 performed perform VERB bracis-33557 233 15 more more ADV bracis-33557 233 16 closely closely ADV bracis-33557 233 17 to to ADP bracis-33557 233 18 the the DET bracis-33557 233 19 original original ADJ bracis-33557 233 20 measure measure NOUN bracis-33557 233 21 with with ADP bracis-33557 233 22 the the DET bracis-33557 233 23 adjusted adjust VERB bracis-33557 233 24 ihm ihm NOUN bracis-33557 233 25 for for ADP bracis-33557 233 26 all all DET bracis-33557 233 27 datasets dataset NOUN bracis-33557 233 28 . . PUNCT bracis-33557 234 1 especially especially ADV bracis-33557 234 2 for for ADP bracis-33557 234 3 datasets dataset NOUN bracis-33557 234 4 with with ADP bracis-33557 234 5 two two NUM bracis-33557 234 6 classes class NOUN bracis-33557 234 7 , , PUNCT bracis-33557 234 8 in in ADP bracis-33557 234 9 many many ADJ bracis-33557 234 10 cases case NOUN bracis-33557 234 11 , , PUNCT bracis-33557 234 12 both both DET bracis-33557 234 13 estimates estimate NOUN bracis-33557 234 14 might might AUX bracis-33557 234 15 agree agree VERB bracis-33557 234 16 when when SCONJ bracis-33557 234 17 the the DET bracis-33557 234 18 first first ADJ bracis-33557 234 19 and and CCONJ bracis-33557 234 20 second second ADJ bracis-33557 234 21 classes class NOUN bracis-33557 234 22 with with ADP bracis-33557 234 23 maximum maximum ADJ bracis-33557 234 24 likelihoods likelihood NOUN bracis-33557 234 25 are be AUX bracis-33557 234 26 the the DET bracis-33557 234 27 same same ADJ bracis-33557 234 28 . . PUNCT bracis-33557 235 1 overall overall ADJ bracis-33557 235 2 , , PUNCT bracis-33557 235 3 adjusted adjusted ADJ bracis-33557 235 4 and and CCONJ bracis-33557 235 5 meta meta ADV bracis-33557 235 6 - - PUNCT bracis-33557 235 7 learning learn VERB bracis-33557 235 8 ihms ihm NOUN bracis-33557 235 9 were be AUX bracis-33557 235 10 able able ADJ bracis-33557 235 11 to to PART bracis-33557 235 12 assess assess VERB bracis-33557 235 13 the the DET bracis-33557 235 14 hardness hardness ADJ bracis-33557 235 15 level level NOUN bracis-33557 235 16 of of ADP bracis-33557 235 17 the the DET bracis-33557 235 18 unlabeled unlabele VERB bracis-33557 235 19 instances instance NOUN bracis-33557 235 20 , , PUNCT bracis-33557 235 21 with with ADP bracis-33557 235 22 some some DET bracis-33557 235 23 prominence prominence NOUN bracis-33557 235 24 of of ADP bracis-33557 235 25 the the DET bracis-33557 235 26 adjusted adjust VERB bracis-33557 235 27 measures measure NOUN bracis-33557 235 28 , , PUNCT bracis-33557 235 29 which which PRON bracis-33557 235 30 showed show VERB bracis-33557 235 31 larger large ADJ bracis-33557 235 32 correlations correlation NOUN bracis-33557 235 33 to to ADP bracis-33557 235 34 the the DET bracis-33557 235 35 original original ADJ bracis-33557 235 36 measures measure NOUN bracis-33557 235 37 in in ADP bracis-33557 235 38 most most ADJ bracis-33557 235 39 cases case NOUN bracis-33557 235 40 . . PUNCT bracis-33557 236 1 they they PRON bracis-33557 236 2 are be AUX bracis-33557 236 3 also also ADV bracis-33557 236 4 simpler simple ADJ bracis-33557 236 5 to to PART bracis-33557 236 6 compute compute VERB bracis-33557 236 7 , , PUNCT bracis-33557 236 8 as as SCONJ bracis-33557 236 9 they they PRON bracis-33557 236 10 do do AUX bracis-33557 236 11 not not PART bracis-33557 236 12 need need VERB bracis-33557 236 13 to to PART bracis-33557 236 14 induce induce VERB bracis-33557 236 15 a a DET bracis-33557 236 16 ml ml NOUN bracis-33557 236 17 model model NOUN bracis-33557 236 18 as as ADP bracis-33557 236 19 in in ADP bracis-33557 236 20 the the DET bracis-33557 236 21 meta meta ADJ bracis-33557 236 22 - - PUNCT bracis-33557 236 23 learning learn VERB bracis-33557 236 24 approach approach NOUN bracis-33557 236 25 . . PUNCT bracis-33557 237 1 in in ADP bracis-33557 237 2 the the DET bracis-33557 237 3 absence absence NOUN bracis-33557 237 4 of of ADP bracis-33557 237 5 labels label NOUN bracis-33557 237 6 , , PUNCT bracis-33557 237 7 most most ADJ bracis-33557 237 8 measures measure NOUN bracis-33557 237 9 are be AUX bracis-33557 237 10 more more ADV bracis-33557 237 11 effective effective ADJ bracis-33557 237 12 in in ADP bracis-33557 237 13 pointing point VERB bracis-33557 237 14 borderline borderline NOUN bracis-33557 237 15 instances instance NOUN bracis-33557 237 16 as as ADP bracis-33557 237 17 posing pose VERB bracis-33557 237 18 a a DET bracis-33557 237 19 higher high ADJ bracis-33557 237 20 difficulty difficulty NOUN bracis-33557 237 21 of of ADP bracis-33557 237 22 posterior posterior ADJ bracis-33557 237 23 classification classification NOUN bracis-33557 237 24 . . PUNCT bracis-33557 238 1 5 5 NUM bracis-33557 238 2 conclusion conclusion NOUN bracis-33557 238 3 and and CCONJ bracis-33557 238 4     SPACE bracis-33557 238 5 future future ADJ bracis-33557 238 6 work work NOUN bracis-33557 238 7 this this DET bracis-33557 238 8 research research NOUN bracis-33557 238 9 analyzed analyze VERB bracis-33557 238 10 alternative alternative ADJ bracis-33557 238 11 ways way NOUN bracis-33557 238 12 to to PART bracis-33557 238 13 measure measure VERB bracis-33557 238 14 the the DET bracis-33557 238 15 hardness hardness NOUN bracis-33557 238 16 of of ADP bracis-33557 238 17 instances instance NOUN bracis-33557 238 18 for for ADP bracis-33557 238 19 classification classification NOUN bracis-33557 238 20 problems problem NOUN bracis-33557 238 21 in in ADP bracis-33557 238 22 scenarios scenario NOUN bracis-33557 238 23 where where SCONJ bracis-33557 238 24 the the DET bracis-33557 238 25 label label NOUN bracis-33557 238 26 of of ADP bracis-33557 238 27 an an DET bracis-33557 238 28 instance instance NOUN bracis-33557 238 29 is be AUX bracis-33557 238 30 unknown unknown ADJ bracis-33557 238 31 , , PUNCT bracis-33557 238 32 that that ADV bracis-33557 238 33 is is ADV bracis-33557 238 34 , , PUNCT bracis-33557 238 35 in in ADP bracis-33557 238 36 the the DET bracis-33557 238 37 deployment deployment NOUN bracis-33557 238 38 stage stage NOUN bracis-33557 238 39 . . PUNCT bracis-33557 239 1 standard standard ADJ bracis-33557 239 2 ihms ihm NOUN bracis-33557 239 3 from from ADP bracis-33557 239 4 the the DET bracis-33557 239 5 literature literature NOUN bracis-33557 239 6 were be AUX bracis-33557 239 7 adapted adapt VERB bracis-33557 239 8 to to ADP bracis-33557 239 9 this this DET bracis-33557 239 10 scenario scenario NOUN bracis-33557 239 11 . . PUNCT bracis-33557 240 1 their their PRON bracis-33557 240 2 results result NOUN bracis-33557 240 3 were be AUX bracis-33557 240 4 compared compare VERB bracis-33557 240 5 to to ADP bracis-33557 240 6 the the DET bracis-33557 240 7 alternative alternative NOUN bracis-33557 240 8 of of ADP bracis-33557 240 9 generating generate VERB bracis-33557 240 10 regression regression NOUN bracis-33557 240 11 meta meta NOUN bracis-33557 240 12 - - PUNCT bracis-33557 240 13 models model NOUN bracis-33557 240 14 to to PART bracis-33557 240 15 predict predict VERB bracis-33557 240 16 the the DET bracis-33557 240 17 ihm ihm ADJ bracis-33557 240 18 values value NOUN bracis-33557 240 19 . . PUNCT bracis-33557 241 1 both both DET bracis-33557 241 2 alternatives alternative NOUN bracis-33557 241 3 were be AUX bracis-33557 241 4 effective effective ADJ bracis-33557 241 5 on on ADP bracis-33557 241 6 their their PRON bracis-33557 241 7 behalf behalf NOUN bracis-33557 241 8 , , PUNCT bracis-33557 241 9 correlating correlate VERB bracis-33557 241 10 to to ADP bracis-33557 241 11 the the DET bracis-33557 241 12 original original ADJ bracis-33557 241 13 ihms ihm NOUN bracis-33557 241 14 that that PRON bracis-33557 241 15 need need VERB bracis-33557 241 16 to to PART bracis-33557 241 17 know know VERB bracis-33557 241 18 the the DET bracis-33557 241 19 label label NOUN bracis-33557 241 20 of of ADP bracis-33557 241 21 each each DET bracis-33557 241 22 instance instance NOUN bracis-33557 241 23 . . PUNCT bracis-33557 242 1 the the DET bracis-33557 242 2 correlations correlation NOUN bracis-33557 242 3 were be AUX bracis-33557 242 4 higher high ADJ bracis-33557 242 5 for for ADP bracis-33557 242 6 some some DET bracis-33557 242 7 measures measure NOUN bracis-33557 242 8 that that PRON bracis-33557 242 9 do do AUX bracis-33557 242 10 not not PART bracis-33557 242 11 rely rely VERB bracis-33557 242 12 as as ADV bracis-33557 242 13 much much ADJ bracis-33557 242 14 on on ADP bracis-33557 242 15 the the DET bracis-33557 242 16 labels label NOUN bracis-33557 242 17 , , PUNCT bracis-33557 242 18 but but CCONJ bracis-33557 242 19 the the DET bracis-33557 242 20 results result NOUN bracis-33557 242 21 for for ADP bracis-33557 242 22 other other ADJ bracis-33557 242 23 measures measure NOUN bracis-33557 242 24 are be AUX bracis-33557 242 25 expected expect VERB bracis-33557 242 26 as as SCONJ bracis-33557 242 27 their their PRON bracis-33557 242 28 original original ADJ bracis-33557 242 29 formulation formulation NOUN bracis-33557 242 30 allows allow VERB bracis-33557 242 31 one one NOUN bracis-33557 242 32 to to PART bracis-33557 242 33 identify identify VERB bracis-33557 242 34 noise noise NOUN bracis-33557 242 35 and and CCONJ bracis-33557 242 36 outliers outlier NOUN bracis-33557 242 37 on on ADP bracis-33557 242 38 data datum NOUN bracis-33557 242 39 regarding regard VERB bracis-33557 242 40 their their PRON bracis-33557 242 41 labels label NOUN bracis-33557 242 42 . . PUNCT bracis-33557 243 1 the the DET bracis-33557 243 2 results result NOUN bracis-33557 243 3 encourages encourage VERB bracis-33557 243 4 the the DET bracis-33557 243 5 usage usage NOUN bracis-33557 243 6 of of ADP bracis-33557 243 7 the the DET bracis-33557 243 8 adjusted adjust VERB bracis-33557 243 9 measures measure NOUN bracis-33557 243 10 in in ADP bracis-33557 243 11 the the DET bracis-33557 243 12 deployment deployment NOUN bracis-33557 243 13 of of ADP bracis-33557 243 14 ml ml NOUN bracis-33557 243 15 models model NOUN bracis-33557 243 16 , , PUNCT bracis-33557 243 17 allowing allow VERB bracis-33557 243 18 the the DET bracis-33557 243 19 identification identification NOUN bracis-33557 243 20 of of ADP bracis-33557 243 21 instances instance NOUN bracis-33557 243 22 that that PRON bracis-33557 243 23 ml ml NOUN bracis-33557 243 24 models model NOUN bracis-33557 243 25 might might AUX bracis-33557 243 26 struggle struggle VERB bracis-33557 243 27 to to PART bracis-33557 243 28 classify classify VERB bracis-33557 243 29 . . PUNCT bracis-33557 244 1 in in ADP bracis-33557 244 2 future future ADJ bracis-33557 244 3 work work NOUN bracis-33557 244 4 , , PUNCT bracis-33557 244 5 we we PRON bracis-33557 244 6 will will AUX bracis-33557 244 7 explore explore VERB bracis-33557 244 8 the the DET bracis-33557 244 9 patterns pattern NOUN bracis-33557 244 10 found find VERB bracis-33557 244 11 in in ADP bracis-33557 244 12 the the DET bracis-33557 244 13 comparisons comparison NOUN bracis-33557 244 14 between between ADP bracis-33557 244 15 the the DET bracis-33557 244 16 original original ADJ bracis-33557 244 17 and and CCONJ bracis-33557 244 18 adjusted adjust VERB bracis-33557 244 19 ihm ihm NOUN bracis-33557 244 20 not not PART bracis-33557 244 21 presented present VERB bracis-33557 244 22 in in ADP bracis-33557 244 23 this this DET bracis-33557 244 24 work work NOUN bracis-33557 244 25 and and CCONJ bracis-33557 244 26 alternative alternative ADJ bracis-33557 244 27 measures measure NOUN bracis-33557 244 28 for for ADP bracis-33557 244 29 unlabeled unlabeled ADJ bracis-33557 244 30 data datum NOUN bracis-33557 244 31 not not PART bracis-33557 244 32 addressed address VERB bracis-33557 244 33 in in ADP bracis-33557 244 34 this this DET bracis-33557 244 35 research research NOUN bracis-33557 244 36 . . PUNCT bracis-33557 245 1 we we PRON bracis-33557 245 2 will will AUX bracis-33557 245 3 expand expand VERB bracis-33557 245 4 the the DET bracis-33557 245 5 application application NOUN bracis-33557 245 6 of of ADP bracis-33557 245 7 the the DET bracis-33557 245 8 adjusted adjust VERB bracis-33557 245 9 measures measure NOUN bracis-33557 245 10 and and CCONJ bracis-33557 245 11 meta meta NOUN bracis-33557 245 12 - - PUNCT bracis-33557 245 13 learning learning NOUN bracis-33557 245 14 to to ADP bracis-33557 245 15 more more ADJ bracis-33557 245 16 datasets dataset NOUN bracis-33557 245 17 , , PUNCT bracis-33557 245 18 and and CCONJ bracis-33557 245 19 tuning tune VERB bracis-33557 245 20 the the DET bracis-33557 245 21 meta meta NOUN bracis-33557 245 22 - - PUNCT bracis-33557 245 23 models model NOUN bracis-33557 245 24 could could AUX bracis-33557 245 25 lead lead VERB bracis-33557 245 26 to to ADP bracis-33557 245 27 new new ADJ bracis-33557 245 28 findings finding NOUN bracis-33557 245 29 about about ADP bracis-33557 245 30 the the DET bracis-33557 245 31 characteristics characteristic NOUN bracis-33557 245 32 of of ADP bracis-33557 245 33 the the DET bracis-33557 245 34 instances instance NOUN bracis-33557 245 35 . . PUNCT bracis-33557 246 1 another another DET bracis-33557 246 2 fruitful fruitful ADJ bracis-33557 246 3 direction direction NOUN bracis-33557 246 4 will will AUX bracis-33557 246 5 be be AUX bracis-33557 246 6 to to PART bracis-33557 246 7 explore explore VERB bracis-33557 246 8 the the DET bracis-33557 246 9 usage usage NOUN bracis-33557 246 10 of of ADP bracis-33557 246 11 the the DET bracis-33557 246 12 adjusted adjust VERB bracis-33557 246 13 measures measure NOUN bracis-33557 246 14 for for ADP bracis-33557 246 15 designing design VERB bracis-33557 246 16 classification classification NOUN bracis-33557 246 17 rejection rejection NOUN bracis-33557 246 18 options option NOUN bracis-33557 246 19 . . PUNCT bracis-33557 247 1 references reference NOUN bracis-33557 247 2 al al PROPN bracis-33557 247 3     SPACE bracis-33557 247 4 hosni hosni PROPN bracis-33557 247 5 , , PUNCT bracis-33557 247 6 o.s o.s PROPN bracis-33557 247 7 . . PROPN bracis-33557 247 8 , , PUNCT bracis-33557 247 9 starkey starkey PROPN bracis-33557 247 10 , , PUNCT bracis-33557 247 11 a. a. NOUN bracis-33557 247 12 : : PUNCT bracis-33557 247 13 investigating investigate VERB bracis-33557 247 14 the the DET bracis-33557 247 15 performance performance NOUN bracis-33557 247 16 of of ADP bracis-33557 247 17 data datum NOUN bracis-33557 247 18 complexity complexity NOUN bracis-33557 247 19 & & CCONJ bracis-33557 247 20 instance instance PROPN bracis-33557 247 21 hardness hardness NOUN bracis-33557 247 22 measures measure NOUN bracis-33557 247 23 as as ADP bracis-33557 247 24 a a DET bracis-33557 247 25 meta meta ADJ bracis-33557 247 26 - - PUNCT bracis-33557 247 27 feature feature NOUN bracis-33557 247 28 in in ADP bracis-33557 247 29 overlapping overlap VERB bracis-33557 247 30 classes class NOUN bracis-33557 247 31 problem problem NOUN bracis-33557 247 32 . . PUNCT bracis-33557 248 1 in in ADP bracis-33557 248 2 : : PUNCT bracis-33557 248 3 iccbdc iccbdc NOUN bracis-33557 248 4 2023 2023 NUM bracis-33557 248 5 , , PUNCT bracis-33557 248 6 manchester manchester PROPN bracis-33557 248 7 , , PUNCT bracis-33557 248 8 united united ADJ bracis-33557 248 9 kingdom kingdom PROPN bracis-33557 248 10 ( ( PUNCT bracis-33557 248 11 2023 2023 NUM bracis-33557 248 12 ) ) PUNCT bracis-33557 248 13 . . PUNCT bracis-33557 249 1 https://doi.org/10.1145/3616131.3616132 https://doi.org/10.1145/3616131.3616132 X bracis-33557 249 2 brazdil brazdil NOUN bracis-33557 249 3 , , PUNCT bracis-33557 249 4 p. p. PROPN bracis-33557 249 5 , , PUNCT bracis-33557 249 6 van van PROPN bracis-33557 249 7 rijn rijn PROPN bracis-33557 249 8 , , PUNCT bracis-33557 249 9 j.n j.n PROPN bracis-33557 249 10 . . PROPN bracis-33557 249 11 , , PUNCT bracis-33557 249 12 soares soares PROPN bracis-33557 249 13 , , PUNCT bracis-33557 249 14 c. c. PROPN bracis-33557 249 15 , , PUNCT bracis-33557 249 16 vanschoren vanschoren PROPN bracis-33557 249 17 , , PUNCT bracis-33557 249 18 j. j. PROPN bracis-33557 249 19 : : PUNCT bracis-33557 249 20 metalearning metalearning NOUN bracis-33557 249 21 : : PUNCT bracis-33557 249 22 applications application NOUN bracis-33557 249 23 to to ADP bracis-33557 249 24 automated automate VERB bracis-33557 249 25 machine machine NOUN bracis-33557 249 26 learning learning NOUN bracis-33557 249 27 and and CCONJ bracis-33557 249 28 data datum NOUN bracis-33557 249 29 mining mining NOUN bracis-33557 249 30 . . PUNCT bracis-33557 250 1 springer springer NOUN bracis-33557 250 2 , , PUNCT bracis-33557 250 3 cham cham PROPN bracis-33557 250 4 ( ( PUNCT bracis-33557 250 5 2022 2022 NUM bracis-33557 250 6 ) ) PUNCT bracis-33557 250 7 . . PUNCT bracis-33557 251 1 https://doi.org/10.1007/978-3-030-67024-5 https://doi.org/10.1007/978-3-030-67024-5 NOUN bracis-33557 251 2 book book NOUN bracis-33557 251 3     SPACE bracis-33557 251 4 math math NOUN bracis-33557 251 5     SPACE bracis-33557 251 6 google google PROPN bracis-33557 251 7 scholar scholar NOUN bracis-33557 251 8     SPACE bracis-33557 251 9 fapesp fapesp NOUN bracis-33557 251 10 : : PUNCT bracis-33557 251 11 fapesp fapesp PROPN bracis-33557 251 12 covid-19 covid-19 PROPN bracis-33557 251 13 data datum NOUN bracis-33557 251 14 sharing sharing NOUN bracis-33557 251 15 / / SYM bracis-33557 251 16 br br PROPN bracis-33557 251 17 ( ( PUNCT bracis-33557 251 18 2020 2020 NUM bracis-33557 251 19 ) ) PUNCT bracis-33557 251 20 . . PUNCT bracis-33557 252 1 https://repositoriodatasharingfapesp.uspdigital.usp.br https://repositoriodatasharingfapesp.uspdigital.usp.br ADJ bracis-33557 252 2 franc franc NOUN bracis-33557 252 3 , , PUNCT bracis-33557 252 4 v. v. PROPN bracis-33557 252 5 , , PUNCT bracis-33557 252 6 prusa prusa PROPN bracis-33557 252 7 , , PUNCT bracis-33557 252 8 d. d. PROPN bracis-33557 252 9 , , PUNCT bracis-33557 252 10 voracek voracek PROPN bracis-33557 252 11 , , PUNCT bracis-33557 252 12 v. v. ADV bracis-33557 252 13 : : PUNCT bracis-33557 252 14 optimal optimal ADJ bracis-33557 252 15 strategies strategy NOUN bracis-33557 252 16 for for ADP bracis-33557 252 17 reject reject NOUN bracis-33557 252 18 option option NOUN bracis-33557 252 19 classifiers classifier NOUN bracis-33557 252 20 . . PUNCT bracis-33557 253 1 j. j. PROPN bracis-33557 253 2 mach mach PROPN bracis-33557 253 3 . . PUNCT bracis-33557 254 1 learn learn VERB bracis-33557 254 2 . . PUNCT bracis-33557 255 1 res re NOUN bracis-33557 255 2 . . PUNCT bracis-33557 256 1 24(11 24(11 NUM bracis-33557 256 2 ) ) PUNCT bracis-33557 256 3 , , PUNCT bracis-33557 256 4 1–49 1–49 PROPN bracis-33557 256 5 ( ( PUNCT bracis-33557 256 6 2023 2023 NUM bracis-33557 256 7 ) ) PUNCT bracis-33557 256 8 mathscinet mathscinet NOUN bracis-33557 256 9     SPACE bracis-33557 256 10 math math NOUN bracis-33557 256 11     SPACE bracis-33557 256 12 google google PROPN bracis-33557 256 13 scholar scholar NOUN bracis-33557 256 14     SPACE bracis-33557 256 15 ho ho PROPN bracis-33557 256 16 , , PUNCT bracis-33557 256 17 t.k t.k PROPN bracis-33557 256 18 . . PROPN bracis-33557 256 19 , , PUNCT bracis-33557 256 20 basu basu PROPN bracis-33557 256 21 , , PUNCT bracis-33557 256 22 m. m. NOUN bracis-33557 256 23 : : PUNCT bracis-33557 256 24 complexity complexity NOUN bracis-33557 256 25 measures measure NOUN bracis-33557 256 26 of of ADP bracis-33557 256 27 supervised supervised ADJ bracis-33557 256 28 classification classification NOUN bracis-33557 256 29 problems problem NOUN bracis-33557 256 30 . . PUNCT bracis-33557 257 1 ieee ieee PROPN bracis-33557 257 2 trans trans PROPN bracis-33557 257 3 . . PUNCT bracis-33557 257 4 pattern pattern PROPN bracis-33557 257 5 anal anal PROPN bracis-33557 257 6 . . PUNCT bracis-33557 258 1 mach mach PROPN bracis-33557 258 2 . . PUNCT bracis-33557 259 1 intell intell PROPN bracis-33557 259 2 . . PUNCT bracis-33557 260 1 24(3 24(3 X bracis-33557 260 2 ) ) PUNCT bracis-33557 260 3 , , PUNCT bracis-33557 260 4 289–300 289–300 NUM bracis-33557 260 5 ( ( PUNCT bracis-33557 260 6 2002 2002 NUM bracis-33557 260 7 ) ) PUNCT bracis-33557 260 8 . . PUNCT bracis-33557 261 1 https://doi.org/10.1109/34.990132 https://doi.org/10.1109/34.990132 PROPN bracis-33557 261 2 article article PROPN bracis-33557 261 3     SPACE bracis-33557 261 4 math math NOUN bracis-33557 261 5     SPACE bracis-33557 261 6 google google PROPN bracis-33557 261 7 scholar scholar NOUN bracis-33557 261 8     SPACE bracis-33557 261 9 kelly kelly PROPN bracis-33557 261 10 , , PUNCT bracis-33557 261 11 m. m. NOUN bracis-33557 261 12 , , PUNCT bracis-33557 261 13 longjohn longjohn PROPN bracis-33557 261 14 , , PUNCT bracis-33557 261 15 r. r. PROPN bracis-33557 261 16 , , PUNCT bracis-33557 261 17 nottingham nottingham PROPN bracis-33557 261 18 , , PUNCT bracis-33557 261 19 k. k. PROPN bracis-33557 261 20 : : PUNCT bracis-33557 261 21 the the DET bracis-33557 261 22 uci uci PROPN bracis-33557 261 23 machine machine NOUN bracis-33557 261 24 learning learn VERB bracis-33557 261 25 repository repository NOUN bracis-33557 261 26 ( ( PUNCT bracis-33557 261 27 2023 2023 NUM bracis-33557 261 28 ) ) PUNCT bracis-33557 261 29 . . PUNCT bracis-33557 262 1 https://archive.ics.uci.edu https://archive.ics.uci.edu X bracis-33557 263 1 lorena lorena PROPN bracis-33557 263 2 , , PUNCT bracis-33557 263 3 a.c a.c PROPN bracis-33557 263 4 . . PROPN bracis-33557 263 5 , , PUNCT bracis-33557 263 6 paiva paiva PROPN bracis-33557 263 7 , , PUNCT bracis-33557 263 8 p.y p.y PROPN bracis-33557 263 9 . . PROPN bracis-33557 263 10 , , PUNCT bracis-33557 263 11 prudêncio prudêncio PROPN bracis-33557 263 12 , , PUNCT bracis-33557 263 13 r.b r.b PROPN bracis-33557 263 14 . . PROPN bracis-33557 263 15 : : PUNCT bracis-33557 263 16 trusting trust VERB bracis-33557 263 17 my my PRON bracis-33557 263 18 predictions prediction NOUN bracis-33557 263 19 : : PUNCT bracis-33557 263 20 on on ADP bracis-33557 263 21 the the DET bracis-33557 263 22 value value NOUN bracis-33557 263 23 of of ADP bracis-33557 263 24 instance instance NOUN bracis-33557 263 25 - - PUNCT bracis-33557 263 26 level level NOUN bracis-33557 263 27 analysis analysis NOUN bracis-33557 263 28 . . PUNCT bracis-33557 263 29 acm acm PROPN bracis-33557 263 30 comput comput NOUN bracis-33557 263 31 . . PUNCT bracis-33557 264 1 surv surv NOUN bracis-33557 264 2 . . PUNCT bracis-33557 265 1 56(7 56(7 X bracis-33557 265 2 ) ) PUNCT bracis-33557 265 3 , , PUNCT bracis-33557 265 4 1–28 1–28 PROPN bracis-33557 265 5 ( ( PUNCT bracis-33557 265 6 2024 2024 NUM bracis-33557 265 7 ) ) PUNCT bracis-33557 265 8 article article NOUN bracis-33557 265 9     SPACE bracis-33557 265 10 math math NOUN bracis-33557 265 11     SPACE bracis-33557 265 12 google google PROPN bracis-33557 265 13 scholar scholar NOUN bracis-33557 265 14     SPACE bracis-33557 265 15 martínez martínez PROPN bracis-33557 265 16 - - PUNCT bracis-33557 265 17 plumed plume VERB bracis-33557 265 18 , , PUNCT bracis-33557 265 19 f. f. PROPN bracis-33557 265 20 , , PUNCT bracis-33557 265 21 prudêncio prudêncio PROPN bracis-33557 265 22 , , PUNCT bracis-33557 265 23 r.b r.b PROPN bracis-33557 265 24 . . PROPN bracis-33557 265 25 , , PUNCT bracis-33557 265 26 martínez martínez NOUN bracis-33557 265 27 - - PUNCT bracis-33557 265 28 usó usó ADJ bracis-33557 265 29 , , PUNCT bracis-33557 265 30 a. a. NOUN bracis-33557 265 31 , , PUNCT bracis-33557 265 32 hernández hernández NOUN bracis-33557 265 33 - - PUNCT bracis-33557 265 34 orallo orallo NOUN bracis-33557 265 35 , , PUNCT bracis-33557 265 36 j. j. PROPN bracis-33557 265 37 : : PUNCT bracis-33557 265 38 item item NOUN bracis-33557 265 39 response response NOUN bracis-33557 265 40 theory theory NOUN bracis-33557 265 41 in in ADP bracis-33557 265 42 ai ai NOUN bracis-33557 265 43 : : PUNCT bracis-33557 265 44 analysing analyse VERB bracis-33557 265 45 machine machine NOUN bracis-33557 265 46 learning learn VERB bracis-33557 265 47 classifiers classifier NOUN bracis-33557 265 48 at at ADP bracis-33557 265 49 the the DET bracis-33557 265 50 instance instance NOUN bracis-33557 265 51 level level NOUN bracis-33557 265 52 . . PUNCT bracis-33557 266 1 artif artif INTJ bracis-33557 266 2 . . PUNCT bracis-33557 267 1 intell intell PROPN bracis-33557 267 2 . . PUNCT bracis-33557 268 1 271 271 NUM bracis-33557 268 2 , , PUNCT bracis-33557 268 3 18–42 18–42 NUM bracis-33557 268 4 ( ( PUNCT bracis-33557 268 5 2019 2019 NUM bracis-33557 268 6 ) ) PUNCT bracis-33557 268 7 article article NOUN bracis-33557 268 8     SPACE bracis-33557 268 9 mathscinet mathscinet NOUN bracis-33557 268 10     SPACE bracis-33557 268 11 math math NOUN bracis-33557 268 12     SPACE bracis-33557 268 13 google google PROPN bracis-33557 268 14 scholar scholar NOUN bracis-33557 268 15     SPACE bracis-33557 268 16 paiva paiva NOUN bracis-33557 268 17 , , PUNCT bracis-33557 268 18 p.y.a p.y.a NOUN bracis-33557 268 19 . . PROPN bracis-33557 268 20 , , PUNCT bracis-33557 268 21 moreno moreno PROPN bracis-33557 268 22 , , PUNCT bracis-33557 268 23 c.c c.c PROPN bracis-33557 268 24 . . PROPN bracis-33557 268 25 , , PUNCT bracis-33557 268 26 smith smith PROPN bracis-33557 268 27 - - PUNCT bracis-33557 268 28 miles mile NOUN bracis-33557 268 29 , , PUNCT bracis-33557 268 30 k. k. PROPN bracis-33557 268 31 , , PUNCT bracis-33557 268 32 valeriano valeriano PROPN bracis-33557 268 33 , , PUNCT bracis-33557 268 34 m.g m.g PROPN bracis-33557 268 35 . . PROPN bracis-33557 268 36 , , PUNCT bracis-33557 268 37 lorena lorena PROPN bracis-33557 268 38 , , PUNCT bracis-33557 268 39 a.c a.c PROPN bracis-33557 268 40 . . PROPN bracis-33557 268 41 : : PUNCT bracis-33557 269 1 relating relate VERB bracis-33557 269 2 instance instance NOUN bracis-33557 269 3 hardness hardness NOUN bracis-33557 269 4 to to ADP bracis-33557 269 5 classification classification NOUN bracis-33557 269 6 performance performance NOUN bracis-33557 269 7 in in ADP bracis-33557 269 8 a a DET bracis-33557 269 9 dataset dataset NOUN bracis-33557 269 10 : : PUNCT bracis-33557 269 11 a a DET bracis-33557 269 12 visual visual ADJ bracis-33557 269 13 approach approach NOUN bracis-33557 269 14 . . PUNCT bracis-33557 270 1 mach mach NOUN bracis-33557 270 2 . . PUNCT bracis-33557 271 1 learn learn VERB bracis-33557 271 2 . . PUNCT bracis-33557 272 1 111(8 111(8 NUM bracis-33557 272 2 ) ) PUNCT bracis-33557 272 3 , , PUNCT bracis-33557 272 4 3085–3123 3085–3123 NUM bracis-33557 272 5 ( ( PUNCT bracis-33557 272 6 2022 2022 NUM bracis-33557 272 7 ) ) PUNCT bracis-33557 272 8 article article NOUN bracis-33557 272 9     SPACE bracis-33557 272 10 mathscinet mathscinet NOUN bracis-33557 272 11     SPACE bracis-33557 272 12 math math NOUN bracis-33557 272 13     SPACE bracis-33557 272 14 google google PROPN bracis-33557 272 15 scholar scholar NOUN bracis-33557 272 16     SPACE bracis-33557 272 17 pedregosa pedregosa PROPN bracis-33557 272 18 , , PUNCT bracis-33557 272 19 f. f. PROPN bracis-33557 272 20 , , PUNCT bracis-33557 272 21 et et PROPN bracis-33557 272 22     SPACE bracis-33557 272 23 al al PROPN bracis-33557 272 24 . . PROPN bracis-33557 272 25 : : PUNCT bracis-33557 273 1 scikit scikit NOUN bracis-33557 273 2 - - PUNCT bracis-33557 273 3 learn learn VERB bracis-33557 273 4 : : PUNCT bracis-33557 273 5 machine machine NOUN bracis-33557 273 6 learning learning NOUN bracis-33557 273 7 in in ADP bracis-33557 273 8 python python PROPN bracis-33557 273 9 . . PUNCT bracis-33557 274 1 j. j. PROPN bracis-33557 274 2 mach mach PROPN bracis-33557 274 3 . . PUNCT bracis-33557 275 1 learn learn VERB bracis-33557 275 2 . . PUNCT bracis-33557 276 1 res re NOUN bracis-33557 276 2 . . PROPN bracis-33557 277 1 12 12 NUM bracis-33557 277 2 , , PUNCT bracis-33557 277 3 2825–2830 2825–2830 NUM bracis-33557 277 4 ( ( PUNCT bracis-33557 277 5 2011 2011 NUM bracis-33557 277 6 ) ) PUNCT bracis-33557 277 7 google google PROPN bracis-33557 277 8 scholar scholar NOUN bracis-33557 277 9     SPACE bracis-33557 277 10 prudêncio prudêncio NOUN bracis-33557 277 11 , , PUNCT bracis-33557 277 12 r.b r.b PROPN bracis-33557 277 13 . . PROPN bracis-33557 277 14 , , PUNCT bracis-33557 277 15 silva silva PROPN bracis-33557 277 16 filho filho PROPN bracis-33557 277 17 , , PUNCT bracis-33557 277 18 t.m t.m PROPN bracis-33557 277 19 . . PUNCT bracis-33557 277 20 : : PUNCT bracis-33557 278 1 explaining explain VERB bracis-33557 278 2 learning learn VERB bracis-33557 278 3 performance performance NOUN bracis-33557 278 4 with with ADP bracis-33557 278 5 local local ADJ bracis-33557 278 6 performance performance NOUN bracis-33557 278 7 regions region NOUN bracis-33557 278 8 and and CCONJ bracis-33557 278 9 maximally maximally ADV bracis-33557 278 10 relevant relevant ADJ bracis-33557 278 11 meta meta NOUN bracis-33557 278 12 - - PUNCT bracis-33557 278 13 rules rule NOUN bracis-33557 278 14 . . PUNCT bracis-33557 279 1 in in ADP bracis-33557 279 2 : : PUNCT bracis-33557 279 3 xavier xavier NOUN bracis-33557 279 4 - - PUNCT bracis-33557 279 5 junior junior PROPN bracis-33557 279 6 , , PUNCT bracis-33557 279 7 j.c j.c PROPN bracis-33557 279 8 . . PROPN bracis-33557 279 9 , , PUNCT bracis-33557 279 10 rios rios PROPN bracis-33557 279 11 , , PUNCT bracis-33557 279 12 r.a r.a PROPN bracis-33557 279 13 . . PROPN bracis-33557 279 14 ( ( PUNCT bracis-33557 279 15 eds ed NOUN bracis-33557 279 16 . . PUNCT bracis-33557 279 17 ) ) PUNCT bracis-33557 279 18 bracis bracis PROPN bracis-33557 279 19 2022 2022 NUM bracis-33557 279 20 . . PUNCT bracis-33557 280 1 lncs lncs PROPN bracis-33557 280 2 , , PUNCT bracis-33557 280 3 vol vol NOUN bracis-33557 280 4 . . PROPN bracis-33557 280 5 13653 13653 NUM bracis-33557 280 6 , , PUNCT bracis-33557 280 7 pp pp ADP bracis-33557 280 8 . . PUNCT bracis-33557 281 1 550–564 550–564 NUM bracis-33557 281 2 . . PUNCT bracis-33557 282 1 springer springer NOUN bracis-33557 282 2 , , PUNCT bracis-33557 282 3 cham cham PROPN bracis-33557 282 4 ( ( PUNCT bracis-33557 282 5 2022 2022 NUM bracis-33557 282 6 ) ) PUNCT bracis-33557 282 7 . . PUNCT bracis-33557 283 1 https://doi.org/10.1007/978-3-031-21686-2_38 https://doi.org/10.1007/978-3-031-21686-2_38 PROPN bracis-33557 283 2 chapter chapter NOUN bracis-33557 283 3     SPACE bracis-33557 283 4 math math NOUN bracis-33557 283 5     SPACE bracis-33557 283 6 google google PROPN bracis-33557 283 7 scholar scholar NOUN bracis-33557 283 8     SPACE bracis-33557 283 9 rivolli rivolli NOUN bracis-33557 283 10 , , PUNCT bracis-33557 283 11 a. a. NOUN bracis-33557 283 12 , , PUNCT bracis-33557 283 13 garcia garcia PROPN bracis-33557 283 14 , , PUNCT bracis-33557 283 15 l.p l.p PROPN bracis-33557 283 16 . . PROPN bracis-33557 283 17 , , PUNCT bracis-33557 283 18 soares soares PROPN bracis-33557 283 19 , , PUNCT bracis-33557 283 20 c. c. PROPN bracis-33557 283 21 , , PUNCT bracis-33557 283 22 vanschoren vanschoren PROPN bracis-33557 283 23 , , PUNCT bracis-33557 283 24 j. j. PROPN bracis-33557 283 25 , , PUNCT bracis-33557 283 26 de de ADJ bracis-33557 283 27     SPACE bracis-33557 283 28 carvalho carvalho NOUN bracis-33557 283 29 , , PUNCT bracis-33557 283 30 a.c a.c PROPN bracis-33557 283 31 . . PROPN bracis-33557 283 32 : : PUNCT bracis-33557 284 1 meta meta VERB bracis-33557 284 2 - - PUNCT bracis-33557 284 3 features feature VERB bracis-33557 284 4 for for ADP bracis-33557 284 5 meta meta NOUN bracis-33557 284 6 - - PUNCT bracis-33557 284 7 learning learning NOUN bracis-33557 284 8 . . PUNCT bracis-33557 285 1 knowl.-based knowl.-based PROPN bracis-33557 285 2 syst syst PROPN bracis-33557 285 3 . . PUNCT bracis-33557 286 1 108101 108101 NUM bracis-33557 286 2 ( ( PUNCT bracis-33557 286 3 2022 2022 NUM bracis-33557 286 4 ) ) PUNCT bracis-33557 286 5 google google PROPN bracis-33557 286 6 scholar scholar NOUN bracis-33557 286 7     SPACE bracis-33557 286 8 schweighofer schweighofer NOUN bracis-33557 286 9 , , PUNCT bracis-33557 286 10 e. e. PROPN bracis-33557 286 11 : : PUNCT bracis-33557 286 12 data data NOUN bracis-33557 286 13 - - PUNCT bracis-33557 286 14 centric centric NOUN bracis-33557 286 15 machine machine NOUN bracis-33557 286 16 learning learning NOUN bracis-33557 286 17 : : PUNCT bracis-33557 286 18 improving improve VERB bracis-33557 286 19 model model NOUN bracis-33557 286 20 performance performance NOUN bracis-33557 286 21 and and CCONJ bracis-33557 286 22 understanding understanding NOUN bracis-33557 286 23 through through ADP bracis-33557 286 24 dataset dataset ADJ bracis-33557 286 25 analysis analysis NOUN bracis-33557 286 26 . . PUNCT bracis-33557 287 1 in in ADP bracis-33557 287 2 : : PUNCT bracis-33557 287 3 legal legal ADJ bracis-33557 287 4 knowledge knowledge NOUN bracis-33557 287 5 and and CCONJ bracis-33557 287 6 information information NOUN bracis-33557 287 7 systems system NOUN bracis-33557 287 8 : : PUNCT bracis-33557 287 9 jurix jurix NOUN bracis-33557 287 10 2021 2021 NUM bracis-33557 287 11 , , PUNCT bracis-33557 287 12 vol vol NOUN bracis-33557 287 13 . . PUNCT bracis-33557 287 14     SPACE bracis-33557 287 15 346 346 NUM bracis-33557 287 16 , , PUNCT bracis-33557 287 17 p. p. NOUN bracis-33557 287 18     SPACE bracis-33557 287 19 54 54 NUM bracis-33557 287 20 . . PUNCT bracis-33557 288 1 ios ios PROPN bracis-33557 288 2 press press NOUN bracis-33557 288 3 ( ( PUNCT bracis-33557 288 4 2021 2021 NUM bracis-33557 288 5 ) ) PUNCT bracis-33557 288 6 google google PROPN bracis-33557 288 7 scholar scholar NOUN bracis-33557 288 8     SPACE bracis-33557 288 9 smith smith PROPN bracis-33557 288 10 , , PUNCT bracis-33557 288 11 m.r m.r PROPN bracis-33557 288 12 . . PROPN bracis-33557 288 13 , , PUNCT bracis-33557 288 14 martinez martinez PROPN bracis-33557 288 15 , , PUNCT bracis-33557 288 16 t. t. PROPN bracis-33557 288 17 , , PUNCT bracis-33557 288 18 giraud giraud NOUN bracis-33557 288 19 - - PUNCT bracis-33557 288 20 carrier carrier NOUN bracis-33557 288 21 , , PUNCT bracis-33557 288 22 c. c. PROPN bracis-33557 288 23 : : PUNCT bracis-33557 288 24 an an DET bracis-33557 288 25 instance instance NOUN bracis-33557 288 26 level level NOUN bracis-33557 288 27 analysis analysis NOUN bracis-33557 288 28 of of ADP bracis-33557 288 29 data datum NOUN bracis-33557 288 30 complexity complexity NOUN bracis-33557 288 31 . . PUNCT bracis-33557 289 1 mach mach NOUN bracis-33557 289 2 . . PUNCT bracis-33557 290 1 learn learn VERB bracis-33557 290 2 . . PUNCT bracis-33557 291 1 95(2 95(2 NOUN bracis-33557 291 2 ) ) PUNCT bracis-33557 291 3 , , PUNCT bracis-33557 291 4 225–256 225–256 NUM bracis-33557 291 5 ( ( PUNCT bracis-33557 291 6 2014 2014 NUM bracis-33557 291 7 ) ) PUNCT bracis-33557 291 8 article article NOUN bracis-33557 291 9     SPACE bracis-33557 291 10 mathscinet mathscinet NOUN bracis-33557 291 11     SPACE bracis-33557 291 12 math math NOUN bracis-33557 291 13     SPACE bracis-33557 291 14 google google PROPN bracis-33557 291 15 scholar scholar NOUN bracis-33557 291 16     SPACE bracis-33557 291 17 valeriano valeriano PROPN bracis-33557 291 18 , , PUNCT bracis-33557 291 19 m.g m.g PROPN bracis-33557 291 20 . . PROPN bracis-33557 291 21 , , PUNCT bracis-33557 291 22 et et PROPN bracis-33557 291 23 al al PROPN bracis-33557 291 24 . . PUNCT bracis-33557 291 25 : : PUNCT bracis-33557 292 1 let let VERB bracis-33557 292 2 the the DET bracis-33557 292 3 data datum NOUN bracis-33557 292 4 speak speak VERB bracis-33557 292 5 : : PUNCT bracis-33557 292 6 analysing analyse VERB bracis-33557 292 7 data datum NOUN bracis-33557 292 8 from from ADP bracis-33557 292 9 multiple multiple ADJ bracis-33557 292 10 health health NOUN bracis-33557 292 11 centers center NOUN bracis-33557 292 12 of of ADP bracis-33557 292 13 the the DET bracis-33557 292 14 são são PROPN bracis-33557 292 15 paulo paulo PROPN bracis-33557 292 16 metropolitan metropolitan PROPN bracis-33557 292 17 area area NOUN bracis-33557 292 18 for for ADP bracis-33557 292 19 covid-19 covid-19 PROPN bracis-33557 292 20 clinical clinical ADJ bracis-33557 292 21 deterioration deterioration NOUN bracis-33557 292 22 prediction prediction NOUN bracis-33557 292 23 . . PUNCT bracis-33557 293 1 in in ADP bracis-33557 293 2 : : PUNCT bracis-33557 293 3 2022 2022 NUM bracis-33557 293 4 22nd 22nd NOUN bracis-33557 293 5 ieee ieee PROPN bracis-33557 293 6 international international ADJ bracis-33557 293 7 symposium symposium NOUN bracis-33557 293 8 on on ADP bracis-33557 293 9 cluster cluster NOUN bracis-33557 293 10 , , PUNCT bracis-33557 293 11 cloud cloud NOUN bracis-33557 293 12 and and CCONJ bracis-33557 293 13 internet internet NOUN bracis-33557 293 14 computing computing NOUN bracis-33557 293 15 ( ( PUNCT bracis-33557 293 16 ccgrid ccgrid NOUN bracis-33557 293 17 ) ) PUNCT bracis-33557 293 18 , , PUNCT bracis-33557 293 19 pp pp ADP bracis-33557 293 20 . . PUNCT bracis-33557 294 1 948–951 948–951 NUM bracis-33557 294 2 . . PUNCT bracis-33557 294 3 ieee ieee NOUN bracis-33557 294 4 ( ( PUNCT bracis-33557 294 5 2022 2022 NUM bracis-33557 294 6 ) ) PUNCT bracis-33557 294 7 google google PROPN bracis-33557 294 8 scholar scholar NOUN bracis-33557 294 9     SPACE bracis-33557 294 10 download download NOUN bracis-33557 294 11 references reference NOUN bracis-33557 294 12 acknowledgements acknowledgement NOUN bracis-33557 294 13 this this DET bracis-33557 294 14 study study NOUN bracis-33557 294 15 was be AUX bracis-33557 294 16 financed finance VERB bracis-33557 294 17 in in ADP bracis-33557 294 18 part part NOUN bracis-33557 294 19 by by ADP bracis-33557 294 20 the the DET bracis-33557 294 21 coordenação coordenação PROPN bracis-33557 294 22 de de PROPN bracis-33557 294 23 aperfeiçoamento aperfeiçoamento PROPN bracis-33557 294 24 de de X bracis-33557 294 25 pessoal pessoal PROPN bracis-33557 294 26 de de PROPN bracis-33557 294 27 nível nível PROPN bracis-33557 294 28 superior superior PROPN bracis-33557 294 29 — — PUNCT bracis-33557 294 30 brasil brasil NOUN bracis-33557 294 31 ( ( PUNCT bracis-33557 294 32 capes)—finance capes)—finance NOUN bracis-33557 294 33 code code PROPN bracis-33557 294 34 001 001 NUM bracis-33557 294 35 . . PUNCT bracis-33557 295 1 the the DET bracis-33557 295 2 authors author NOUN bracis-33557 295 3 thank thank VERB bracis-33557 295 4 fapesp fapesp NOUN bracis-33557 295 5 for for ADP bracis-33557 295 6 its its PRON bracis-33557 295 7 support support NOUN bracis-33557 295 8 under under ADP bracis-33557 295 9 grant grant NOUN bracis-33557 295 10 2021/06870 2021/06870 NUM bracis-33557 295 11 - - SYM bracis-33557 295 12 3 3 NUM bracis-33557 295 13 . . PUNCT bracis-33557 296 1 author author NOUN bracis-33557 296 2 information information NOUN bracis-33557 296 3 authors author NOUN bracis-33557 296 4 and and CCONJ bracis-33557 296 5 affiliations affiliation NOUN bracis-33557 296 6 universidade universidade PROPN bracis-33557 296 7 tecnológica tecnológica PROPN bracis-33557 296 8 federal federal PROPN bracis-33557 296 9 do do AUX bracis-33557 296 10 paraná paraná PROPN bracis-33557 296 11 , , PUNCT bracis-33557 296 12 cornélio cornélio NOUN bracis-33557 296 13 procópio procópio PROPN bracis-33557 296 14 , , PUNCT bracis-33557 296 15 brazil brazil PROPN bracis-33557 296 16 patricia patricia PROPN bracis-33557 296 17 s. s. PROPN bracis-33557 296 18 m. m. PROPN bracis-33557 296 19 ueda ueda PROPN bracis-33557 296 20     SPACE bracis-33557 296 21 & & CCONJ bracis-33557 296 22     SPACE bracis-33557 296 23 ana ana PROPN bracis-33557 296 24 carolina carolina PROPN bracis-33557 296 25 lorena lorena PROPN bracis-33557 296 26 instituto instituto PROPN bracis-33557 296 27 tecnológico tecnológico PROPN bracis-33557 296 28 de de PROPN bracis-33557 296 29 aeronáutica aeronáutica PROPN bracis-33557 296 30 , , PUNCT bracis-33557 296 31 são são PROPN bracis-33557 296 32 josé josé PROPN bracis-33557 296 33 dos dos PROPN bracis-33557 296 34 campos campos PROPN bracis-33557 296 35 , , PUNCT bracis-33557 296 36 brazil brazil PROPN bracis-33557 296 37 adriano adriano PROPN bracis-33557 296 38 rivolli rivolli PROPN bracis-33557 296 39 authors author NOUN bracis-33557 296 40 patricia patricia PROPN bracis-33557 296 41 s. s. PROPN bracis-33557 296 42 m. m. PROPN bracis-33557 296 43 uedaview uedaview PROPN bracis-33557 296 44 author author NOUN bracis-33557 296 45 publications publication VERB bracis-33557 296 46 search search NOUN bracis-33557 296 47 author author NOUN bracis-33557 296 48 on on ADP bracis-33557 296 49 : : PUNCT bracis-33557 296 50 pubmed pubmed PROPN bracis-33557 296 51     SPACE bracis-33557 296 52 google google PROPN bracis-33557 296 53 scholar scholar NOUN bracis-33557 296 54 adriano adriano PROPN bracis-33557 296 55 rivolliview rivolliview NOUN bracis-33557 296 56 author author NOUN bracis-33557 296 57 publications publication NOUN bracis-33557 296 58 search search NOUN bracis-33557 296 59 author author NOUN bracis-33557 296 60 on on ADP bracis-33557 296 61 : : PUNCT bracis-33557 296 62 pubmed pubmed PROPN bracis-33557 296 63     SPACE bracis-33557 296 64 google google PROPN bracis-33557 296 65 scholar scholar PROPN bracis-33557 296 66 ana ana PROPN bracis-33557 296 67 carolina carolina PROPN bracis-33557 296 68 lorenaview lorenaview PROPN bracis-33557 296 69 author author NOUN bracis-33557 296 70 publications publication NOUN bracis-33557 296 71 search search NOUN bracis-33557 296 72 author author NOUN bracis-33557 296 73 on on ADP bracis-33557 296 74 : : PUNCT bracis-33557 296 75 pubmed pubmed PROPN bracis-33557 296 76     SPACE bracis-33557 296 77 google google PROPN bracis-33557 296 78 scholar scholar NOUN bracis-33557 296 79 corresponding correspond VERB bracis-33557 296 80 author author NOUN bracis-33557 296 81 correspondence correspondence NOUN bracis-33557 296 82 to to ADP bracis-33557 296 83 patricia patricia PROPN bracis-33557 296 84 s. s. PROPN bracis-33557 296 85 m. m. PROPN bracis-33557 296 86 ueda ueda PROPN bracis-33557 296 87 . . PUNCT bracis-33557 297 1 editor editor NOUN bracis-33557 297 2 information information NOUN bracis-33557 297 3 editors editor NOUN bracis-33557 297 4 and and CCONJ bracis-33557 297 5 affiliations affiliation NOUN bracis-33557 297 6 universidade universidade VERB bracis-33557 297 7 federal federal PROPN bracis-33557 297 8 fluminense fluminense PROPN bracis-33557 297 9 , , PUNCT bracis-33557 297 10 niterói niterói PROPN bracis-33557 297 11 , , PUNCT bracis-33557 297 12 brazil brazil PROPN bracis-33557 297 13 aline aline PROPN bracis-33557 297 14 paes paes PROPN bracis-33557 297 15 instituto instituto PROPN bracis-33557 297 16 tecnológico tecnológico PROPN bracis-33557 297 17 de de PROPN bracis-33557 297 18 aeronáutica aeronáutica PROPN bracis-33557 297 19 , , PUNCT bracis-33557 297 20 são são PROPN bracis-33557 297 21 josé josé PROPN bracis-33557 297 22 dos dos PROPN bracis-33557 297 23 campos campos PROPN bracis-33557 297 24 , , PUNCT bracis-33557 297 25 brazil brazil PROPN bracis-33557 297 26 filipe filipe PROPN bracis-33557 297 27 a. a. PROPN bracis-33557 297 28 n. n. PROPN bracis-33557 297 29 verri verri PROPN bracis-33557 297 30 rights right NOUN bracis-33557 297 31 and and CCONJ bracis-33557 297 32 permissions permission NOUN bracis-33557 297 33 reprints reprint NOUN bracis-33557 297 34 and and CCONJ bracis-33557 297 35 permissions permission VERB bracis-33557 297 36 copyright copyright NOUN bracis-33557 297 37 information information NOUN bracis-33557 297 38 © © ADP bracis-33557 297 39 2025 2025 NUM bracis-33557 297 40 the the DET bracis-33557 297 41 author(s author(s NOUN bracis-33557 297 42 ) ) PUNCT bracis-33557 297 43 , , PUNCT bracis-33557 297 44 under under ADP bracis-33557 297 45 exclusive exclusive ADJ bracis-33557 297 46 license license NOUN bracis-33557 297 47 to to ADP bracis-33557 297 48 springer springer NOUN bracis-33557 297 49 nature nature PROPN bracis-33557 297 50 switzerland switzerland PROPN bracis-33557 297 51 ag ag PROPN bracis-33557 297 52 about about ADP bracis-33557 297 53 this this DET bracis-33557 297 54 paper paper NOUN bracis-33557 297 55 cite cite VERB bracis-33557 297 56 this this DET bracis-33557 297 57 paper paper NOUN bracis-33557 297 58 ueda ueda PROPN bracis-33557 297 59 , , PUNCT bracis-33557 297 60 p.s.m p.s.m PROPN bracis-33557 297 61 . . PROPN bracis-33557 297 62 , , PUNCT bracis-33557 297 63 rivolli rivolli PROPN bracis-33557 297 64 , , PUNCT bracis-33557 297 65 a. a. PROPN bracis-33557 297 66 , , PUNCT bracis-33557 297 67 lorena lorena PROPN bracis-33557 297 68 , , PUNCT bracis-33557 297 69 a.c a.c PROPN bracis-33557 297 70 . . PROPN bracis-33557 297 71 ( ( PUNCT bracis-33557 297 72 2025 2025 NUM bracis-33557 297 73 ) ) PUNCT bracis-33557 297 74 . . PUNCT bracis-33557 298 1 an an DET bracis-33557 298 2 instance instance NOUN bracis-33557 298 3 level level NOUN bracis-33557 298 4 analysis analysis NOUN bracis-33557 298 5 of of ADP bracis-33557 298 6     SPACE bracis-33557 298 7 classification classification NOUN bracis-33557 298 8 difficulty difficulty NOUN bracis-33557 298 9 for for ADP bracis-33557 298 10     SPACE bracis-33557 298 11 unlabeled unlabele VERB bracis-33557 298 12 data datum NOUN bracis-33557 298 13 . . PUNCT bracis-33557 299 1 in in ADP bracis-33557 299 2 : : PUNCT bracis-33557 299 3 paes paes PROPN bracis-33557 299 4 , , PUNCT bracis-33557 299 5 a. a. PROPN bracis-33557 299 6 , , PUNCT bracis-33557 299 7 verri verri NOUN bracis-33557 299 8 , , PUNCT bracis-33557 299 9 f.a.n f.a.n ADJ bracis-33557 299 10 . . PUNCT bracis-33557 299 11 ( ( PUNCT bracis-33557 299 12 eds ed NOUN bracis-33557 299 13 ) ) PUNCT bracis-33557 299 14 intelligent intelligent ADJ bracis-33557 299 15 systems system NOUN bracis-33557 299 16 . . PUNCT bracis-33557 299 17 bracis bracis PROPN bracis-33557 299 18 2024 2024 NUM bracis-33557 299 19 . . PUNCT bracis-33557 300 1 lecture lecture NOUN bracis-33557 300 2 notes note NOUN bracis-33557 300 3 in in ADP bracis-33557 300 4 computer computer NOUN bracis-33557 300 5 science science NOUN bracis-33557 300 6 ( ( PUNCT bracis-33557 300 7 ) ) PUNCT bracis-33557 300 8 , , PUNCT bracis-33557 300 9 vol vol NOUN bracis-33557 300 10 15412 15412 NUM bracis-33557 300 11 . . PUNCT bracis-33557 301 1 springer springer NOUN bracis-33557 301 2 , , PUNCT bracis-33557 301 3 cham cham PROPN bracis-33557 301 4 . . PUNCT bracis-33557 302 1 https://doi.org/10.1007/978-3-031-79029-4_10 https://doi.org/10.1007/978-3-031-79029-4_10 PROPN bracis-33557 302 2 download download NOUN bracis-33557 302 3 citation citation NOUN bracis-33557 302 4 .ris .ris PUNCT bracis-33557 303 1 .enw .enw PROPN bracis-33557 303 2 .bib .bib PUNCT bracis-33557 304 1 doi doi PROPN bracis-33557 304 2 : : PUNCT bracis-33557 304 3 https://doi.org/10.1007/978-3-031-79029-4_10 https://doi.org/10.1007/978-3-031-79029-4_10 AUX bracis-33557 304 4 published publish VERB bracis-33557 304 5 : : PUNCT bracis-33557 304 6 30 30 NUM bracis-33557 304 7 january january PROPN bracis-33557 304 8 2025 2025 NUM bracis-33557 304 9 publisher publisher NOUN bracis-33557 304 10 name name NOUN bracis-33557 304 11 : : PUNCT bracis-33557 304 12 springer springer NOUN bracis-33557 304 13 , , PUNCT bracis-33557 304 14 cham cham PROPN bracis-33557 304 15 print print PROPN bracis-33557 304 16 isbn isbn PROPN bracis-33557 304 17 : : PUNCT bracis-33557 304 18 978 978 NUM bracis-33557 304 19 - - SYM bracis-33557 304 20 3 3 NUM bracis-33557 304 21 - - PUNCT bracis-33557 304 22 031 031 NUM bracis-33557 304 23 - - PUNCT bracis-33557 304 24 79028 79028 NUM bracis-33557 304 25 - - SYM bracis-33557 304 26 7 7 NUM bracis-33557 304 27 online online ADJ bracis-33557 304 28 isbn isbn NOUN bracis-33557 304 29 : : PUNCT bracis-33557 304 30 978 978 NUM bracis-33557 304 31 - - SYM bracis-33557 304 32 3 3 NUM bracis-33557 304 33 - - PUNCT bracis-33557 304 34 031 031 NUM bracis-33557 304 35 - - PUNCT bracis-33557 304 36 79029 79029 NUM bracis-33557 304 37 - - SYM bracis-33557 304 38 4 4 NUM bracis-33557 304 39 ebook ebook NOUN bracis-33557 304 40 packages package NOUN bracis-33557 304 41 : : PUNCT bracis-33557 304 42 computer computer NOUN bracis-33557 304 43 sciencecomputer sciencecomputer NOUN bracis-33557 304 44 science science NOUN bracis-33557 304 45 ( ( PUNCT bracis-33557 304 46 r0 r0 NOUN bracis-33557 304 47 ) ) PUNCT bracis-33557 304 48 share share VERB bracis-33557 304 49 this this DET bracis-33557 304 50 paper paper NOUN bracis-33557 304 51 anyone anyone PRON bracis-33557 304 52 you you PRON bracis-33557 304 53 share share VERB bracis-33557 304 54 the the DET bracis-33557 304 55 following follow VERB bracis-33557 304 56 link link NOUN bracis-33557 304 57 with with ADP bracis-33557 304 58 will will AUX bracis-33557 304 59 be be AUX bracis-33557 304 60 able able ADJ bracis-33557 304 61 to to PART bracis-33557 304 62 read read VERB bracis-33557 304 63 this this DET bracis-33557 304 64 content content NOUN bracis-33557 304 65 : : PUNCT bracis-33557 304 66 get get VERB bracis-33557 304 67 shareable shareable ADJ bracis-33557 304 68 linksorry linksorry NOUN bracis-33557 304 69 , , PUNCT bracis-33557 304 70 a a DET bracis-33557 304 71 shareable shareable ADJ bracis-33557 304 72 link link NOUN bracis-33557 304 73 is be AUX bracis-33557 304 74 not not PART bracis-33557 304 75 currently currently ADV bracis-33557 304 76 available available ADJ bracis-33557 304 77 for for ADP bracis-33557 304 78 this this DET bracis-33557 304 79 article article NOUN bracis-33557 304 80 . . PUNCT bracis-33557 305 1 copy copy VERB bracis-33557 305 2 shareable shareable ADJ bracis-33557 305 3 link link NOUN bracis-33557 305 4 to to PART bracis-33557 305 5 clipboard clipboard NOUN bracis-33557 305 6 provided provide VERB bracis-33557 305 7 by by ADP bracis-33557 305 8 the the DET bracis-33557 305 9 springer springer NOUN bracis-33557 305 10 nature nature PROPN bracis-33557 305 11 sharedit sharedit PROPN bracis-33557 305 12 content content NOUN bracis-33557 305 13 - - PUNCT bracis-33557 305 14 sharing share VERB bracis-33557 305 15 initiative initiative NOUN bracis-33557 305 16 keywords keyword NOUN bracis-33557 305 17 machine machine NOUN bracis-33557 305 18 learning learn VERB bracis-33557 305 19 instance instance NOUN bracis-33557 305 20 hardness hardness NOUN bracis-33557 305 21 measures measure NOUN bracis-33557 305 22 unlabeled unlabele VERB bracis-33557 305 23 data data NOUN bracis-33557 305 24 deployment deployment NOUN bracis-33557 305 25 of of ADP bracis-33557 305 26 models model NOUN bracis-33557 305 27 publish publish VERB bracis-33557 305 28 with with ADP bracis-33557 305 29 us us PROPN bracis-33557 305 30 policies policy NOUN bracis-33557 305 31 and and CCONJ bracis-33557 305 32 ethics ethic NOUN bracis-33557 305 33 profiles profile NOUN bracis-33557 305 34 ana ana PROPN bracis-33557 305 35 carolina carolina PROPN bracis-33557 305 36 lorena lorena PROPN bracis-33557 305 37 view view PROPN bracis-33557 305 38 author author NOUN bracis-33557 305 39 profile profile NOUN bracis-33557 305 40 search search NOUN bracis-33557 305 41 search search NOUN bracis-33557 305 42 by by ADP bracis-33557 305 43 keyword keyword NOUN bracis-33557 305 44 or or CCONJ bracis-33557 305 45 author author NOUN bracis-33557 305 46 search search NOUN bracis-33557 305 47 navigation navigation NOUN bracis-33557 305 48 find find VERB bracis-33557 305 49 a a DET bracis-33557 305 50 journal journal NOUN bracis-33557 305 51 publish publish VERB bracis-33557 305 52 with with ADP bracis-33557 305 53 us we PRON bracis-33557 305 54 track track VERB bracis-33557 305 55 your your PRON bracis-33557 305 56 research research NOUN bracis-33557 305 57 discover discover VERB bracis-33557 305 58 content content NOUN bracis-33557 305 59 journals journal NOUN bracis-33557 305 60 a a DET bracis-33557 305 61 - - PUNCT bracis-33557 305 62 z z NOUN bracis-33557 305 63 books book NOUN bracis-33557 305 64 a a DET bracis-33557 305 65 - - PUNCT bracis-33557 305 66 z z NOUN bracis-33557 305 67 publish publish NOUN bracis-33557 305 68 with with ADP bracis-33557 305 69 us us PROPN bracis-33557 305 70 journal journal PROPN bracis-33557 305 71 finder finder PROPN bracis-33557 305 72 publish publish VERB bracis-33557 305 73 your your PRON bracis-33557 305 74 research research NOUN bracis-33557 305 75 language language NOUN bracis-33557 305 76 editing edit VERB bracis-33557 305 77 open open ADJ bracis-33557 305 78 access access NOUN bracis-33557 305 79 publishing publishing NOUN bracis-33557 305 80 products product NOUN bracis-33557 305 81 and and CCONJ bracis-33557 305 82 services service NOUN bracis-33557 305 83 our our PRON bracis-33557 305 84 products product NOUN bracis-33557 305 85 librarians librarian VERB bracis-33557 305 86 societies society NOUN bracis-33557 305 87 partners partner NOUN bracis-33557 305 88 and and CCONJ bracis-33557 305 89 advertisers advertiser NOUN bracis-33557 305 90 our our PRON bracis-33557 305 91 brands brand NOUN bracis-33557 305 92 springer springer NOUN bracis-33557 305 93 nature nature PROPN bracis-33557 305 94 portfolio portfolio PROPN bracis-33557 305 95 bmc bmc PROPN bracis-33557 305 96 palgrave palgrave PROPN bracis-33557 305 97 macmillan macmillan PROPN bracis-33557 305 98 apress apress PROPN bracis-33557 305 99 discover discover VERB bracis-33557 305 100 your your PRON bracis-33557 305 101 privacy privacy NOUN bracis-33557 305 102 choices choice NOUN bracis-33557 305 103 / / SYM bracis-33557 305 104 manage manage NOUN bracis-33557 305 105 cookies cookie NOUN bracis-33557 305 106 your your PRON bracis-33557 305 107 us us PROPN bracis-33557 306 1 state state NOUN bracis-33557 306 2 privacy privacy NOUN bracis-33557 306 3 rights right NOUN bracis-33557 306 4 accessibility accessibility NOUN bracis-33557 306 5 statement statement NOUN bracis-33557 306 6 terms term NOUN bracis-33557 306 7 and and CCONJ bracis-33557 306 8 conditions condition NOUN bracis-33557 306 9 privacy privacy NOUN bracis-33557 306 10 policy policy NOUN bracis-33557 306 11 help help NOUN bracis-33557 306 12 and and CCONJ bracis-33557 306 13 support support VERB bracis-33557 306 14 legal legal ADJ bracis-33557 306 15 notice notice NOUN bracis-33557 306 16 cancel cancel VERB bracis-33557 306 17 contracts contract NOUN bracis-33557 306 18 here here ADV bracis-33557 306 19 129.74.145.123 129.74.145.123 NUM bracis-33557 306 20 hesburgh hesburgh PROPN bracis-33557 306 21 library library PROPN bracis-33557 306 22 er er INTJ bracis-33557 306 23 unit unit NOUN bracis-33557 306 24 ( ( PUNCT bracis-33557 306 25 3005732405 3005732405 NUM bracis-33557 306 26 ) ) PUNCT bracis-33557 306 27 northeast northeast ADJ bracis-33557 306 28 research research NOUN bracis-33557 306 29 libraries library NOUN bracis-33557 306 30 ( ( PUNCT bracis-33557 306 31 nerl nerl PROPN bracis-33557 306 32 ) ) PUNCT bracis-33557 306 33 ( ( PUNCT bracis-33557 306 34 8200828607 8200828607 NUM bracis-33557 306 35 ) ) PUNCT bracis-33557 306 36 nerl nerl VERB bracis-33557 306 37 ta ta X bracis-33557 306 38 account account NOUN bracis-33557 306 39 ( ( PUNCT bracis-33557 306 40 3006206169 3006206169 NUM bracis-33557 306 41 ) ) PUNCT bracis-33557 306 42 university university NOUN bracis-33557 306 43 of of ADP bracis-33557 306 44 notre notre PROPN bracis-33557 306 45 dame dame PROPN bracis-33557 306 46 hesburgh hesburgh PROPN bracis-33557 306 47 library library NOUN bracis-33557 306 48 ( ( PUNCT bracis-33557 306 49 3000184373 3000184373 NUM bracis-33557 306 50 ) ) PUNCT bracis-33557 307 1 © © ADP bracis-33557 307 2 2025 2025 NUM bracis-33557 307 3 springer springer NOUN bracis-33557 307 4 nature nature NOUN