id sid tid token lemma pos fcis-1701 1 1 frontiers frontier NOUN fcis-1701 1 2 in in ADP fcis-1701 1 3 computing computing NOUN fcis-1701 1 4 and and CCONJ fcis-1701 1 5 intelligent intelligent ADJ fcis-1701 1 6 systems system NOUN fcis-1701 1 7 issn issn VERB fcis-1701 1 8 : : PUNCT fcis-1701 1 9 2832 2832 NUM fcis-1701 1 10 - - SYM fcis-1701 1 11 6024 6024 NUM fcis-1701 1 12 | | NOUN fcis-1701 1 13 vol vol NOUN fcis-1701 1 14 . . PROPN fcis-1701 2 1 1 1 NUM fcis-1701 2 2 , , PUNCT fcis-1701 2 3 no no INTJ fcis-1701 2 4 . . NOUN fcis-1701 2 5 2 2 NUM fcis-1701 2 6 , , PUNCT fcis-1701 2 7 2022 2022 NUM fcis-1701 2 8 35 35 NUM fcis-1701 2 9 time time NOUN fcis-1701 2 10 series series PROPN fcis-1701 2 11 data data PROPN fcis-1701 2 12 anomaly anomaly PROPN fcis-1701 2 13 detection detection NOUN fcis-1701 2 14 based base VERB fcis-1701 2 15 on on ADP fcis-1701 2 16 lstmgan lstmgan PROPN fcis-1701 2 17 xiaofei xiaofei PROPN fcis-1701 2 18 pan pan PROPN fcis-1701 2 19 * * PROPN fcis-1701 2 20 school school NOUN fcis-1701 2 21 of of ADP fcis-1701 2 22 control control NOUN fcis-1701 2 23 and and CCONJ fcis-1701 2 24 computer computer NOUN fcis-1701 2 25 engineering engineering NOUN fcis-1701 2 26 , , PUNCT fcis-1701 2 27 north north PROPN fcis-1701 2 28 china china PROPN fcis-1701 2 29 electric electric PROPN fcis-1701 2 30 power power PROPN fcis-1701 2 31 university university PROPN fcis-1701 2 32 , , PUNCT fcis-1701 2 33 baoding baoding PROPN fcis-1701 2 34 071003 071003 NUM fcis-1701 2 35 , , PUNCT fcis-1701 2 36 hebei hebei PROPN fcis-1701 2 37 province province PROPN fcis-1701 2 38 , , PUNCT fcis-1701 2 39 china china PROPN fcis-1701 2 40 * * PUNCT fcis-1701 2 41 corresponding correspond VERB fcis-1701 2 42 author author NOUN fcis-1701 2 43 : : PUNCT fcis-1701 2 44 xiaofei xiaofei PROPN fcis-1701 2 45 pan pan PROPN fcis-1701 2 46 ( ( PUNCT fcis-1701 2 47 email email NOUN fcis-1701 2 48 : : PUNCT fcis-1701 2 49 pxf448265572@162.com pxf448265572@162.com X fcis-1701 2 50 ) ) PUNCT fcis-1701 2 51 . . PUNCT fcis-1701 3 1 abstract abstract ADV fcis-1701 3 2 : : PUNCT fcis-1701 3 3 with with ADP fcis-1701 3 4 the the DET fcis-1701 3 5 improvement improvement NOUN fcis-1701 3 6 of of ADP fcis-1701 3 7 modern modern ADJ fcis-1701 3 8 technology technology NOUN fcis-1701 3 9 , , PUNCT fcis-1701 3 10 a a DET fcis-1701 3 11 large large ADJ fcis-1701 3 12 number number NOUN fcis-1701 3 13 of of ADP fcis-1701 3 14 time time NOUN fcis-1701 3 15 series series PROPN fcis-1701 3 16 data datum NOUN fcis-1701 3 17 have have AUX fcis-1701 3 18 been be AUX fcis-1701 3 19 produced produce VERB fcis-1701 3 20 . . PUNCT fcis-1701 4 1 the the DET fcis-1701 4 2 anomaly anomaly NOUN fcis-1701 4 3 detection detection NOUN fcis-1701 4 4 of of ADP fcis-1701 4 5 time time NOUN fcis-1701 4 6 series series PROPN fcis-1701 4 7 data datum NOUN fcis-1701 4 8 can can AUX fcis-1701 4 9 provide provide VERB fcis-1701 4 10 relevant relevant ADJ fcis-1701 4 11 information information NOUN fcis-1701 4 12 for for ADP fcis-1701 4 13 key key ADJ fcis-1701 4 14 situations situation NOUN fcis-1701 4 15 faced face VERB fcis-1701 4 16 by by ADP fcis-1701 4 17 various various ADJ fcis-1701 4 18 fields field NOUN fcis-1701 4 19 . . PUNCT fcis-1701 5 1 this this DET fcis-1701 5 2 paper paper NOUN fcis-1701 5 3 proposed propose VERB fcis-1701 5 4 an an DET fcis-1701 5 5 unsupervised unsupervised ADJ fcis-1701 5 6 temporal temporal ADJ fcis-1701 5 7 anomaly anomaly NOUN fcis-1701 5 8 detection detection NOUN fcis-1701 5 9 method method NOUN fcis-1701 5 10 based base VERB fcis-1701 5 11 on on ADP fcis-1701 5 12 generation generation NOUN fcis-1701 5 13 countermeasure countermeasure NOUN fcis-1701 5 14 network network NOUN fcis-1701 5 15 . . PUNCT fcis-1701 6 1 in in ADP fcis-1701 6 2 this this DET fcis-1701 6 3 model model NOUN fcis-1701 6 4 , , PUNCT fcis-1701 6 5 wasserstein wasserstein NOUN fcis-1701 6 6 distance distance NOUN fcis-1701 6 7 is be AUX fcis-1701 6 8 used use VERB fcis-1701 6 9 instead instead ADV fcis-1701 6 10 of of ADP fcis-1701 6 11 the the DET fcis-1701 6 12 original original ADJ fcis-1701 6 13 measurement measurement NOUN fcis-1701 6 14 method method NOUN fcis-1701 6 15 , , PUNCT fcis-1701 6 16 and and CCONJ fcis-1701 6 17 lstm lstm NOUN fcis-1701 6 18 is be AUX fcis-1701 6 19 used use VERB fcis-1701 6 20 as as ADP fcis-1701 6 21 the the DET fcis-1701 6 22 basic basic ADJ fcis-1701 6 23 network network NOUN fcis-1701 6 24 of of ADP fcis-1701 6 25 gan gan PROPN fcis-1701 6 26 . . PUNCT fcis-1701 7 1 the the DET fcis-1701 7 2 model model NOUN fcis-1701 7 3 uses use VERB fcis-1701 7 4 the the DET fcis-1701 7 5 reconstruction reconstruction NOUN fcis-1701 7 6 loss loss NOUN fcis-1701 7 7 of of ADP fcis-1701 7 8 the the DET fcis-1701 7 9 generator generator NOUN fcis-1701 7 10 and and CCONJ fcis-1701 7 11 the the DET fcis-1701 7 12 loss loss NOUN fcis-1701 7 13 of of ADP fcis-1701 7 14 the the DET fcis-1701 7 15 discriminator discriminator NOUN fcis-1701 7 16 to to PART fcis-1701 7 17 define define VERB fcis-1701 7 18 the the DET fcis-1701 7 19 anomaly anomaly NOUN fcis-1701 7 20 function function NOUN fcis-1701 7 21 to to PART fcis-1701 7 22 judge judge VERB fcis-1701 7 23 the the DET fcis-1701 7 24 anomaly anomaly NOUN fcis-1701 7 25 . . PUNCT fcis-1701 8 1 this this DET fcis-1701 8 2 paper paper NOUN fcis-1701 8 3 uses use VERB fcis-1701 8 4 real real ADJ fcis-1701 8 5 world world NOUN fcis-1701 8 6 time time NOUN fcis-1701 8 7 series series PROPN fcis-1701 8 8 data data PROPN fcis-1701 8 9 sets set NOUN fcis-1701 8 10 involving involve VERB fcis-1701 8 11 various various ADJ fcis-1701 8 12 fields field NOUN fcis-1701 8 13 to to PART fcis-1701 8 14 evaluate evaluate VERB fcis-1701 8 15 the the DET fcis-1701 8 16 model model NOUN fcis-1701 8 17 . . PUNCT fcis-1701 9 1 experiments experiment NOUN fcis-1701 9 2 show show VERB fcis-1701 9 3 that that SCONJ fcis-1701 9 4 the the DET fcis-1701 9 5 model model NOUN fcis-1701 9 6 is be AUX fcis-1701 9 7 effective effective ADJ fcis-1701 9 8 in in ADP fcis-1701 9 9 anomaly anomaly NOUN fcis-1701 9 10 detection detection NOUN fcis-1701 9 11 of of ADP fcis-1701 9 12 time time NOUN fcis-1701 9 13 series series PROPN fcis-1701 9 14 data data PROPN fcis-1701 9 15 . . PUNCT fcis-1701 10 1 keywords keyword NOUN fcis-1701 10 2 : : PUNCT fcis-1701 10 3 time time NOUN fcis-1701 10 4 series series PROPN fcis-1701 10 5 data data PROPN fcis-1701 10 6 ; ; PUNCT fcis-1701 10 7 anomaly anomaly NOUN fcis-1701 10 8 detection detection NOUN fcis-1701 10 9 ; ; PUNCT fcis-1701 10 10 gan gan PROPN fcis-1701 10 11 . . PROPN fcis-1701 11 1 1 1 NUM fcis-1701 11 2 . . PUNCT fcis-1701 12 1 introduction introduction NOUN fcis-1701 12 2 exceptions exception NOUN fcis-1701 12 3 can can AUX fcis-1701 12 4 be be AUX fcis-1701 12 5 defined define VERB fcis-1701 12 6 as as ADP fcis-1701 12 7 abnormal abnormal ADJ fcis-1701 12 8 patterns pattern NOUN fcis-1701 12 9 that that PRON fcis-1701 12 10 do do AUX fcis-1701 12 11 not not PART fcis-1701 12 12 conform conform VERB fcis-1701 12 13 to to ADP fcis-1701 12 14 expected expect VERB fcis-1701 12 15 behavior behavior NOUN fcis-1701 12 16 . . PUNCT fcis-1701 13 1 [ [ X fcis-1701 13 2 1 1 X fcis-1701 13 3 ] ] PUNCT fcis-1701 13 4 with with ADP fcis-1701 13 5 the the DET fcis-1701 13 6 rapid rapid ADJ fcis-1701 13 7 growth growth NOUN fcis-1701 13 8 of of ADP fcis-1701 13 9 time time NOUN fcis-1701 13 10 series series PROPN fcis-1701 13 11 data data PROPN fcis-1701 13 12 , , PUNCT fcis-1701 13 13 abnormal abnormal ADJ fcis-1701 13 14 data datum NOUN fcis-1701 13 15 may may AUX fcis-1701 13 16 lead lead VERB fcis-1701 13 17 to to ADP fcis-1701 13 18 serious serious ADJ fcis-1701 13 19 failures failure NOUN fcis-1701 13 20 . . PUNCT fcis-1701 14 1 anomaly anomaly NOUN fcis-1701 14 2 detection detection NOUN fcis-1701 14 3 ( ( PUNCT fcis-1701 14 4 ad ad NOUN fcis-1701 14 5 ) ) PUNCT fcis-1701 14 6 refers refer VERB fcis-1701 14 7 to to ADP fcis-1701 14 8 automatic automatic ADJ fcis-1701 14 9 identification identification NOUN fcis-1701 14 10 of of ADP fcis-1701 14 11 abnormal abnormal ADJ fcis-1701 14 12 phenomena phenomenon NOUN fcis-1701 14 13 mixed mix VERB fcis-1701 14 14 with with ADP fcis-1701 14 15 a a DET fcis-1701 14 16 large large ADJ fcis-1701 14 17 number number NOUN fcis-1701 14 18 of of ADP fcis-1701 14 19 normal normal ADJ fcis-1701 14 20 data.[2]therefore data.[2]therefore NOUN fcis-1701 14 21 , , PUNCT fcis-1701 14 22 anomaly anomaly NOUN fcis-1701 14 23 detection detection NOUN fcis-1701 14 24 can can AUX fcis-1701 14 25 provide provide VERB fcis-1701 14 26 an an DET fcis-1701 14 27 opportunity opportunity NOUN fcis-1701 14 28 to to PART fcis-1701 14 29 take take VERB fcis-1701 14 30 action action NOUN fcis-1701 14 31 and and CCONJ fcis-1701 14 32 solve solve VERB fcis-1701 14 33 problems problem NOUN fcis-1701 14 34 before before SCONJ fcis-1701 14 35 potential potential ADJ fcis-1701 14 36 problems problem NOUN fcis-1701 14 37 cause cause VERB fcis-1701 14 38 disasters disaster NOUN fcis-1701 14 39 . . PUNCT fcis-1701 15 1 traditionally traditionally ADV fcis-1701 15 2 , , PUNCT fcis-1701 15 3 various various ADJ fcis-1701 15 4 statistical statistical ADJ fcis-1701 15 5 methods method NOUN fcis-1701 15 6 have have AUX fcis-1701 15 7 been be AUX fcis-1701 15 8 proposed propose VERB fcis-1701 15 9 to to PART fcis-1701 15 10 improve improve VERB fcis-1701 15 11 the the DET fcis-1701 15 12 threshold threshold NOUN fcis-1701 15 13 , , PUNCT fcis-1701 15 14 such such ADJ fcis-1701 15 15 as as ADP fcis-1701 15 16 spc spc PROPN fcis-1701 16 1 [ [ X fcis-1701 16 2 3 3 NUM fcis-1701 16 3 ] ] PUNCT fcis-1701 16 4 . . PUNCT fcis-1701 17 1 if if SCONJ fcis-1701 17 2 the the DET fcis-1701 17 3 monitoring monitoring NOUN fcis-1701 17 4 data datum NOUN fcis-1701 17 5 calculated calculate VERB fcis-1701 17 6 on on ADP fcis-1701 17 7 an an DET fcis-1701 17 8 instance instance NOUN fcis-1701 17 9 exceeds exceed VERB fcis-1701 17 10 the the DET fcis-1701 17 11 control control NOUN fcis-1701 17 12 limit limit NOUN fcis-1701 17 13 , , PUNCT fcis-1701 17 14 it it PRON fcis-1701 17 15 will will AUX fcis-1701 17 16 be be AUX fcis-1701 17 17 identified identify VERB fcis-1701 17 18 as as ADP fcis-1701 17 19 an an DET fcis-1701 17 20 exception exception NOUN fcis-1701 17 21 . . PUNCT fcis-1701 18 1 such such ADJ fcis-1701 18 2 methods method NOUN fcis-1701 18 3 require require VERB fcis-1701 18 4 a a DET fcis-1701 18 5 lot lot NOUN fcis-1701 18 6 of of ADP fcis-1701 18 7 human human ADJ fcis-1701 18 8 knowledge knowledge NOUN fcis-1701 18 9 to to PART fcis-1701 18 10 set set VERB fcis-1701 18 11 the the DET fcis-1701 18 12 prior prior ADJ fcis-1701 18 13 assumptions assumption NOUN fcis-1701 18 14 of of ADP fcis-1701 18 15 the the DET fcis-1701 18 16 model model NOUN fcis-1701 18 17 [ [ X fcis-1701 18 18 4 4 NUM fcis-1701 18 19 ] ] PUNCT fcis-1701 18 20 . . PUNCT fcis-1701 19 1 later later ADV fcis-1701 19 2 , , PUNCT fcis-1701 19 3 some some DET fcis-1701 19 4 unsupervised unsupervised ADJ fcis-1701 19 5 machine machine NOUN fcis-1701 19 6 learning learning NOUN fcis-1701 19 7 methods method NOUN fcis-1701 19 8 were be AUX fcis-1701 19 9 proposed propose VERB fcis-1701 19 10 . . PUNCT fcis-1701 20 1 a a DET fcis-1701 20 2 common common ADJ fcis-1701 20 3 method method NOUN fcis-1701 20 4 is be AUX fcis-1701 20 5 to to PART fcis-1701 20 6 divide divide VERB fcis-1701 20 7 time time NOUN fcis-1701 20 8 series series PROPN fcis-1701 20 9 data datum NOUN fcis-1701 20 10 into into ADP fcis-1701 20 11 subsequences subsequence NOUN fcis-1701 20 12 with with ADP fcis-1701 20 13 a a DET fcis-1701 20 14 certain certain ADJ fcis-1701 20 15 length length NOUN fcis-1701 20 16 , , PUNCT fcis-1701 20 17 and and CCONJ fcis-1701 20 18 use use VERB fcis-1701 20 19 clustering clustering NOUN fcis-1701 20 20 method method NOUN fcis-1701 20 21 to to PART fcis-1701 20 22 find find VERB fcis-1701 20 23 outliers outlier NOUN fcis-1701 20 24 . . PUNCT fcis-1701 21 1 the the DET fcis-1701 21 2 other other ADJ fcis-1701 21 3 is be AUX fcis-1701 21 4 to to PART fcis-1701 21 5 learn learn VERB fcis-1701 21 6 a a DET fcis-1701 21 7 model model NOUN fcis-1701 21 8 to to PART fcis-1701 21 9 predict predict VERB fcis-1701 21 10 or or CCONJ fcis-1701 21 11 reconstruct reconstruct VERB fcis-1701 21 12 time time NOUN fcis-1701 21 13 series series PROPN fcis-1701 21 14 data data PROPN fcis-1701 21 15 , , PUNCT fcis-1701 21 16 and and CCONJ fcis-1701 21 17 compare compare VERB fcis-1701 21 18 the the DET fcis-1701 21 19 actual actual ADJ fcis-1701 21 20 value value NOUN fcis-1701 21 21 with with ADP fcis-1701 21 22 the the DET fcis-1701 21 23 predicted predict VERB fcis-1701 21 24 value value NOUN fcis-1701 21 25 or or CCONJ fcis-1701 21 26 the the DET fcis-1701 21 27 reconstructed reconstructed ADJ fcis-1701 21 28 value value NOUN fcis-1701 21 29 . . PUNCT fcis-1701 22 1 a a DET fcis-1701 22 2 high high ADJ fcis-1701 22 3 prediction prediction NOUN fcis-1701 22 4 error error NOUN fcis-1701 22 5 or or CCONJ fcis-1701 22 6 reconstruction reconstruction NOUN fcis-1701 22 7 error error NOUN fcis-1701 22 8 calculated calculate VERB fcis-1701 22 9 indicates indicate VERB fcis-1701 22 10 the the DET fcis-1701 22 11 existence existence NOUN fcis-1701 22 12 of of ADP fcis-1701 22 13 anomalies anomaly NOUN fcis-1701 22 14 [ [ X fcis-1701 22 15 5 5 NUM fcis-1701 22 16 ] ] PUNCT fcis-1701 22 17 . . PUNCT fcis-1701 23 1 with with ADP fcis-1701 23 2 the the DET fcis-1701 23 3 rapid rapid ADJ fcis-1701 23 4 development development NOUN fcis-1701 23 5 of of ADP fcis-1701 23 6 artificial artificial ADJ fcis-1701 23 7 neural neural ADJ fcis-1701 23 8 networks network NOUN fcis-1701 23 9 in in ADP fcis-1701 23 10 various various ADJ fcis-1701 23 11 fields field NOUN fcis-1701 23 12 , , PUNCT fcis-1701 23 13 more more ADJ fcis-1701 23 14 and and CCONJ fcis-1701 23 15 more more ADJ fcis-1701 23 16 methods method NOUN fcis-1701 23 17 based base VERB fcis-1701 23 18 on on ADP fcis-1701 23 19 deep deep ADJ fcis-1701 23 20 learning learning NOUN fcis-1701 23 21 have have AUX fcis-1701 23 22 emerged emerge VERB fcis-1701 23 23 . . PUNCT fcis-1701 24 1 gan gan PROPN fcis-1701 24 2 is be AUX fcis-1701 24 3 widely widely ADV fcis-1701 24 4 used use VERB fcis-1701 24 5 in in ADP fcis-1701 24 6 many many ADJ fcis-1701 24 7 fields field NOUN fcis-1701 24 8 , , PUNCT fcis-1701 24 9 such such ADJ fcis-1701 24 10 as as ADP fcis-1701 24 11 image image NOUN fcis-1701 24 12 generation generation NOUN fcis-1701 24 13 , , PUNCT fcis-1701 24 14 data datum NOUN fcis-1701 24 15 balance balance NOUN fcis-1701 24 16 , , PUNCT fcis-1701 24 17 image image NOUN fcis-1701 24 18 recognition recognition NOUN fcis-1701 24 19 and and CCONJ fcis-1701 24 20 video video NOUN fcis-1701 24 21 processing processing NOUN fcis-1701 24 22 . . PUNCT fcis-1701 25 1 [ [ X fcis-1701 25 2 6 6 NUM fcis-1701 25 3 ] ] X fcis-1701 25 4 gan gan NOUN fcis-1701 25 5 can can AUX fcis-1701 25 6 learn learn VERB fcis-1701 25 7 the the DET fcis-1701 25 8 real real ADJ fcis-1701 25 9 data datum NOUN fcis-1701 25 10 distribution distribution NOUN fcis-1701 25 11 and and CCONJ fcis-1701 25 12 generate generate VERB fcis-1701 25 13 false false ADJ fcis-1701 25 14 data datum NOUN fcis-1701 25 15 similar similar ADJ fcis-1701 25 16 to to ADP fcis-1701 25 17 the the DET fcis-1701 25 18 real real ADJ fcis-1701 25 19 data datum NOUN fcis-1701 25 20 . . PUNCT fcis-1701 26 1 when when SCONJ fcis-1701 26 2 the the DET fcis-1701 26 3 input input NOUN fcis-1701 26 4 is be AUX fcis-1701 26 5 an an DET fcis-1701 26 6 abnormal abnormal ADJ fcis-1701 26 7 sample sample NOUN fcis-1701 26 8 , , PUNCT fcis-1701 26 9 the the DET fcis-1701 26 10 sample sample NOUN fcis-1701 26 11 generated generate VERB fcis-1701 26 12 by by ADP fcis-1701 26 13 gan gan PROPN fcis-1701 26 14 will will AUX fcis-1701 26 15 be be AUX fcis-1701 26 16 different different ADJ fcis-1701 26 17 from from ADP fcis-1701 26 18 the the DET fcis-1701 26 19 original original ADJ fcis-1701 26 20 real real ADJ fcis-1701 26 21 data datum NOUN fcis-1701 26 22 . . PUNCT fcis-1701 27 1 when when SCONJ fcis-1701 27 2 the the DET fcis-1701 27 3 difference difference NOUN fcis-1701 27 4 is be AUX fcis-1701 27 5 greater great ADJ fcis-1701 27 6 than than ADP fcis-1701 27 7 the the DET fcis-1701 27 8 threshold threshold NOUN fcis-1701 27 9 , , PUNCT fcis-1701 27 10 it it PRON fcis-1701 27 11 will will AUX fcis-1701 27 12 be be AUX fcis-1701 27 13 judged judge VERB fcis-1701 27 14 as as ADP fcis-1701 27 15 abnormal abnormal ADJ fcis-1701 27 16 . . PUNCT fcis-1701 28 1 in in ADP fcis-1701 28 2 this this DET fcis-1701 28 3 paper paper NOUN fcis-1701 28 4 , , PUNCT fcis-1701 28 5 a a DET fcis-1701 28 6 temporal temporal ADJ fcis-1701 28 7 data datum NOUN fcis-1701 28 8 anomaly anomaly NOUN fcis-1701 28 9 detection detection NOUN fcis-1701 28 10 model model NOUN fcis-1701 28 11 called call VERB fcis-1701 28 12 lstm lstm PROPN fcis-1701 28 13 - - PUNCT fcis-1701 28 14 gan gan PROPN fcis-1701 28 15 is be AUX fcis-1701 28 16 proposed propose VERB fcis-1701 28 17 . . PUNCT fcis-1701 29 1 the the DET fcis-1701 29 2 structure structure NOUN fcis-1701 29 3 of of ADP fcis-1701 29 4 the the DET fcis-1701 29 5 rest rest NOUN fcis-1701 29 6 of of ADP fcis-1701 29 7 this this DET fcis-1701 29 8 paper paper NOUN fcis-1701 29 9 is be AUX fcis-1701 29 10 as as SCONJ fcis-1701 29 11 follows follow VERB fcis-1701 29 12 : : PUNCT fcis-1701 29 13 in in ADP fcis-1701 29 14 section section NOUN fcis-1701 29 15 2 2 NUM fcis-1701 29 16 , , PUNCT fcis-1701 29 17 the the DET fcis-1701 29 18 relevant relevant ADJ fcis-1701 29 19 technologies technology NOUN fcis-1701 29 20 used use VERB fcis-1701 29 21 are be AUX fcis-1701 29 22 introduced introduce VERB fcis-1701 29 23 , , PUNCT fcis-1701 29 24 in in ADP fcis-1701 29 25 section section NOUN fcis-1701 29 26 3 3 NUM fcis-1701 29 27 , , PUNCT fcis-1701 29 28 the the DET fcis-1701 29 29 model model NOUN fcis-1701 29 30 structure structure NOUN fcis-1701 29 31 proposed propose VERB fcis-1701 29 32 in in ADP fcis-1701 29 33 this this DET fcis-1701 29 34 paper paper NOUN fcis-1701 29 35 is be AUX fcis-1701 29 36 introduced introduce VERB fcis-1701 29 37 , , PUNCT fcis-1701 29 38 in in ADP fcis-1701 29 39 section section NOUN fcis-1701 29 40 4 4 NUM fcis-1701 29 41 , , PUNCT fcis-1701 29 42 the the DET fcis-1701 29 43 relevant relevant ADJ fcis-1701 29 44 experiments experiment NOUN fcis-1701 29 45 conducted conduct VERB fcis-1701 29 46 with with ADP fcis-1701 29 47 the the DET fcis-1701 29 48 model model NOUN fcis-1701 29 49 proposed propose VERB fcis-1701 29 50 in in ADP fcis-1701 29 51 this this DET fcis-1701 29 52 paper paper NOUN fcis-1701 29 53 are be AUX fcis-1701 29 54 introduced introduce VERB fcis-1701 29 55 , , PUNCT fcis-1701 29 56 and and CCONJ fcis-1701 29 57 the the DET fcis-1701 29 58 last last ADJ fcis-1701 29 59 part part NOUN fcis-1701 29 60 is be AUX fcis-1701 29 61 summarized summarize VERB fcis-1701 29 62 . . PUNCT fcis-1701 30 1 2 2 X fcis-1701 30 2 . . X fcis-1701 30 3 related relate VERB fcis-1701 30 4 technologies technology NOUN fcis-1701 30 5 the the DET fcis-1701 30 6 generative generative ADJ fcis-1701 30 7 adversarial adversarial ADJ fcis-1701 30 8 network network NOUN fcis-1701 30 9 is be AUX fcis-1701 30 10 based base VERB fcis-1701 30 11 on on ADP fcis-1701 30 12 the the DET fcis-1701 30 13 scenario scenario NOUN fcis-1701 30 14 of of ADP fcis-1701 30 15 game game NOUN fcis-1701 30 16 theory theory NOUN fcis-1701 30 17 , , PUNCT fcis-1701 30 18 [ [ X fcis-1701 30 19 7 7 X fcis-1701 30 20 ] ] PUNCT fcis-1701 30 21 in in ADP fcis-1701 30 22 which which PRON fcis-1701 30 23 two two NUM fcis-1701 30 24 participants participant NOUN fcis-1701 30 25 compete compete VERB fcis-1701 30 26 with with ADP fcis-1701 30 27 each each DET fcis-1701 30 28 other other ADJ fcis-1701 30 29 to to PART fcis-1701 30 30 achieve achieve VERB fcis-1701 30 31 the the DET fcis-1701 30 32 goal goal NOUN fcis-1701 30 33 of of ADP fcis-1701 30 34 nash nash PROPN fcis-1701 30 35 equilibrium equilibrium NOUN fcis-1701 30 36 . . PUNCT fcis-1701 31 1 the the DET fcis-1701 31 2 gan gan PROPN fcis-1701 31 3 network network NOUN fcis-1701 31 4 flow flow NOUN fcis-1701 31 5 chart chart NOUN fcis-1701 31 6 is be AUX fcis-1701 31 7 as as SCONJ fcis-1701 31 8 follows follow VERB fcis-1701 31 9 : : PUNCT fcis-1701 31 10 figure figure NOUN fcis-1701 31 11 1 1 NUM fcis-1701 31 12 . . PUNCT fcis-1701 32 1 gan gan PROPN fcis-1701 32 2 network network PROPN fcis-1701 32 3 flow flow NOUN fcis-1701 32 4 chart chart NOUN fcis-1701 32 5 gan gan PROPN fcis-1701 32 6 is be AUX fcis-1701 32 7 composed compose VERB fcis-1701 32 8 of of ADP fcis-1701 32 9 a a DET fcis-1701 32 10 generator generator NOUN fcis-1701 32 11 g g NOUN fcis-1701 32 12 and and CCONJ fcis-1701 32 13 a a DET fcis-1701 32 14 discriminator discriminator NOUN fcis-1701 32 15 d. d. NOUN fcis-1701 32 16 the the DET fcis-1701 32 17 random random ADJ fcis-1701 32 18 variable variable NOUN fcis-1701 32 19 z z NOUN fcis-1701 32 20 from from ADP fcis-1701 32 21 potential potential ADJ fcis-1701 32 22 space space NOUN fcis-1701 32 23 z z NOUN fcis-1701 32 24 is be AUX fcis-1701 32 25 input input VERB fcis-1701 32 26 into into ADP fcis-1701 32 27 generator generator NOUN fcis-1701 32 28 g g PROPN fcis-1701 32 29 to to PART fcis-1701 32 30 learn learn VERB fcis-1701 32 31 the the DET fcis-1701 32 32 distribution distribution NOUN fcis-1701 32 33 on on ADP fcis-1701 32 34 real real ADJ fcis-1701 32 35 samples sample NOUN fcis-1701 32 36 by by ADP fcis-1701 32 37 mapping map VERB fcis-1701 32 38 g(z g(z PROPN fcis-1701 32 39 ) ) PUNCT fcis-1701 32 40 . . PUNCT fcis-1701 33 1 generator generator PROPN fcis-1701 33 2 g g PROPN fcis-1701 33 3 needs need VERB fcis-1701 33 4 to to PART fcis-1701 33 5 generate generate VERB fcis-1701 33 6 generated generate VERB fcis-1701 33 7 samples sample NOUN fcis-1701 33 8 that that PRON fcis-1701 33 9 are be AUX fcis-1701 33 10 as as ADV fcis-1701 33 11 similar similar ADJ fcis-1701 33 12 as as ADP fcis-1701 33 13 possible possible ADJ fcis-1701 33 14 to to ADP fcis-1701 33 15 the the DET fcis-1701 33 16 distribution distribution NOUN fcis-1701 33 17 of of ADP fcis-1701 33 18 real real ADJ fcis-1701 33 19 samples sample NOUN fcis-1701 33 20 . . PUNCT fcis-1701 34 1 the the DET fcis-1701 34 2 loss loss NOUN fcis-1701 34 3 function function NOUN fcis-1701 34 4 of of ADP fcis-1701 34 5 generator generator NOUN fcis-1701 34 6 g g PROPN fcis-1701 34 7 is be AUX fcis-1701 34 8 : : PUNCT fcis-1701 34 9 ( ( PUNCT fcis-1701 34 10 ) ) PUNCT fcis-1701 34 11 ~ ~ PUNCT fcis-1701 35 1 [ [ X fcis-1701 35 2 log(1 log(1 NOUN fcis-1701 35 3 ( ( PUNCT fcis-1701 35 4 ( ( PUNCT fcis-1701 35 5 ) ) PUNCT fcis-1701 35 6 ) ) PUNCT fcis-1701 35 7 ) ) PUNCT fcis-1701 35 8 ] ] PUNCT fcis-1701 36 1 z z NOUN fcis-1701 36 2 zg zg PROPN fcis-1701 36 3 z z NOUN fcis-1701 36 4 pl pl X fcis-1701 36 5 e e X fcis-1701 36 6 d d X fcis-1701 36 7 g g PROPN fcis-1701 36 8 z= z= PROPN fcis-1701 36 9 βˆ’ βˆ’ PROPN fcis-1701 36 10 ( ( PUNCT fcis-1701 36 11 1 1 NUM fcis-1701 36 12 ) ) PUNCT fcis-1701 36 13 input input NOUN fcis-1701 36 14 the the DET fcis-1701 36 15 real real ADJ fcis-1701 36 16 sample sample NOUN fcis-1701 36 17 x x PUNCT fcis-1701 36 18 or or CCONJ fcis-1701 36 19 generated generate VERB fcis-1701 36 20 sample sample NOUN fcis-1701 36 21 into into ADP fcis-1701 36 22 discriminator discriminator NOUN fcis-1701 36 23 d d PROPN fcis-1701 36 24 , , PUNCT fcis-1701 36 25 and and CCONJ fcis-1701 36 26 its its PRON fcis-1701 36 27 training training NOUN fcis-1701 36 28 goal goal NOUN fcis-1701 36 29 is be AUX fcis-1701 36 30 to to PART fcis-1701 36 31 correctly correctly ADV fcis-1701 36 32 distinguish distinguish VERB fcis-1701 36 33 between between ADP fcis-1701 36 34 the the DET fcis-1701 36 35 real real ADJ fcis-1701 36 36 sample sample NOUN fcis-1701 36 37 and and CCONJ fcis-1701 36 38 generated generate VERB fcis-1701 36 39 sample sample NOUN fcis-1701 36 40 . . PUNCT fcis-1701 37 1 generator generator NOUN fcis-1701 37 2 g g PROPN fcis-1701 37 3 and and CCONJ fcis-1701 37 4 discriminator discriminator NOUN fcis-1701 37 5 d d AUX fcis-1701 37 6 play play NOUN fcis-1701 37 7 games game NOUN fcis-1701 37 8 through through ADP fcis-1701 37 9 minmax minmax PROPN fcis-1701 37 10 function function NOUN fcis-1701 37 11 . . PUNCT fcis-1701 38 1 after after ADP fcis-1701 38 2 training training NOUN fcis-1701 38 3 , , PUNCT fcis-1701 38 4 generator generator NOUN fcis-1701 38 5 ( ( PUNCT fcis-1701 38 6 g g NOUN fcis-1701 38 7 ) ) PUNCT fcis-1701 38 8 has have VERB fcis-1701 38 9 the the DET fcis-1701 38 10 ability ability NOUN fcis-1701 38 11 to to PART fcis-1701 38 12 generate generate VERB fcis-1701 38 13 false false ADJ fcis-1701 38 14 data datum NOUN fcis-1701 38 15 discriminator discriminator NOUN fcis-1701 38 16 ( ( PUNCT fcis-1701 38 17 d d NOUN fcis-1701 38 18 ) ) PUNCT fcis-1701 38 19 has have VERB fcis-1701 38 20 the the DET fcis-1701 38 21 ability ability NOUN fcis-1701 38 22 to to PART fcis-1701 38 23 distinguish distinguish VERB fcis-1701 38 24 between between ADP fcis-1701 38 25 real real ADJ fcis-1701 38 26 data datum NOUN fcis-1701 38 27 and and CCONJ fcis-1701 38 28 false false ADJ fcis-1701 38 29 data datum NOUN fcis-1701 38 30 . . PUNCT fcis-1701 39 1 assuming assume VERB fcis-1701 39 2 π‘π‘‘π‘Žπ‘‘π‘Ž(π‘₯ π‘π‘‘π‘Žπ‘‘π‘Ž(π‘₯ PROPN fcis-1701 39 3 ) ) PUNCT fcis-1701 39 4 andπ‘π‘‘π‘Žπ‘‘π‘Ž(π‘₯ andπ‘π‘‘π‘Žπ‘‘π‘Ž(π‘₯ PROPN fcis-1701 39 5 ) ) PUNCT fcis-1701 39 6 are be AUX fcis-1701 39 7 known know VERB fcis-1701 39 8 , , PUNCT fcis-1701 39 9 the the DET fcis-1701 39 10 optimal optimal ADJ fcis-1701 39 11 discriminator discriminator NOUN fcis-1701 39 12 function function NOUN fcis-1701 39 13 is be AUX fcis-1701 39 14 : : PUNCT fcis-1701 39 15 36 36 NUM fcis-1701 39 16 π·βˆ—(π‘₯ π·βˆ—(π‘₯ NOUN fcis-1701 39 17 ) ) PUNCT fcis-1701 39 18 = = SYM fcis-1701 39 19 π‘π‘‘π‘Žπ‘‘π‘Ž(x π‘π‘‘π‘Žπ‘‘π‘Ž(x PROPN fcis-1701 39 20 ) ) PUNCT fcis-1701 39 21 π‘π‘‘π‘Žπ‘‘π‘Ž(x)+𝑝𝑔(x π‘π‘‘π‘Žπ‘‘π‘Ž(x)+𝑝𝑔(x NOUN fcis-1701 39 22 ) ) PUNCT fcis-1701 39 23 ( ( PUNCT fcis-1701 39 24 2 2 X fcis-1701 39 25 ) ) PUNCT fcis-1701 39 26 substitute substitute NOUN fcis-1701 39 27 the the DET fcis-1701 39 28 optimal optimal ADJ fcis-1701 39 29 discriminator discriminator NOUN fcis-1701 39 30 into into ADP fcis-1701 39 31 formula formula NOUN fcis-1701 39 32 2 2 NUM fcis-1701 39 33 , , PUNCT fcis-1701 39 34 and and CCONJ fcis-1701 39 35 the the DET fcis-1701 39 36 objective objective ADJ fcis-1701 39 37 function function NOUN fcis-1701 39 38 becomes become VERB fcis-1701 39 39 : : PUNCT fcis-1701 39 40 𝐿(𝐺|π·βˆ— 𝐿(𝐺|π·βˆ— X fcis-1701 39 41 ) ) PUNCT fcis-1701 39 42 = = NOUN fcis-1701 39 43 2𝐷𝐽𝑆(π‘π‘‘π‘Žπ‘‘π‘Ž 2𝐷𝐽𝑆(π‘π‘‘π‘Žπ‘‘π‘Ž NUM fcis-1701 39 44 βˆ₯ βˆ₯ NUM fcis-1701 39 45 𝑝𝑔 𝑝𝑔 NOUN fcis-1701 39 46 ) ) PUNCT fcis-1701 39 47 βˆ’ βˆ’ PROPN fcis-1701 39 48 2log2 2log2 NUM fcis-1701 39 49 ( ( PUNCT fcis-1701 39 50 3 3 NUM fcis-1701 39 51 ) ) PUNCT fcis-1701 39 52 when when SCONJ fcis-1701 39 53 the the DET fcis-1701 39 54 discriminator discriminator NOUN fcis-1701 39 55 is be AUX fcis-1701 39 56 optimal optimal ADJ fcis-1701 39 57 , , PUNCT fcis-1701 39 58 minimizing minimize VERB fcis-1701 39 59 the the DET fcis-1701 39 60 js js ADJ fcis-1701 39 61 difference difference NOUN fcis-1701 39 62 between between ADP fcis-1701 39 63 the the DET fcis-1701 39 64 actual actual ADJ fcis-1701 39 65 distribution distribution NOUN fcis-1701 39 66 and and CCONJ fcis-1701 39 67 the the DET fcis-1701 39 68 generated generate VERB fcis-1701 39 69 distribution distribution NOUN fcis-1701 39 70 is be AUX fcis-1701 39 71 taken take VERB fcis-1701 39 72 as as ADP fcis-1701 39 73 the the DET fcis-1701 39 74 optimization optimization NOUN fcis-1701 39 75 goal goal NOUN fcis-1701 39 76 of of ADP fcis-1701 39 77 the the DET fcis-1701 39 78 generator generator NOUN fcis-1701 39 79 . . PUNCT fcis-1701 40 1 it it PRON fcis-1701 40 2 will will AUX fcis-1701 40 3 cause cause VERB fcis-1701 40 4 the the DET fcis-1701 40 5 js js ADJ fcis-1701 40 6 divergence divergence NOUN fcis-1701 40 7 in in ADP fcis-1701 40 8 the the DET fcis-1701 40 9 generator generator NOUN fcis-1701 40 10 loss loss NOUN fcis-1701 40 11 function function NOUN fcis-1701 40 12 to to PART fcis-1701 40 13 be be AUX fcis-1701 40 14 equal equal ADJ fcis-1701 40 15 to to ADP fcis-1701 40 16 a a DET fcis-1701 40 17 constant constant ADJ fcis-1701 40 18 when when SCONJ fcis-1701 40 19 the the DET fcis-1701 40 20 real real ADJ fcis-1701 40 21 distribution distribution NOUN fcis-1701 40 22 and and CCONJ fcis-1701 40 23 the the DET fcis-1701 40 24 generated generate VERB fcis-1701 40 25 distribution distribution NOUN fcis-1701 40 26 do do AUX fcis-1701 40 27 not not PART fcis-1701 40 28 overlap overlap VERB fcis-1701 40 29 or or CCONJ fcis-1701 40 30 the the DET fcis-1701 40 31 overlapping overlap VERB fcis-1701 40 32 part part NOUN fcis-1701 40 33 can can AUX fcis-1701 40 34 be be AUX fcis-1701 40 35 ignored ignore VERB fcis-1701 40 36 , , PUNCT fcis-1701 40 37 and and CCONJ fcis-1701 40 38 the the DET fcis-1701 40 39 output output NOUN fcis-1701 40 40 of of ADP fcis-1701 40 41 the the DET fcis-1701 40 42 discriminant discriminant ADJ fcis-1701 40 43 network network NOUN fcis-1701 40 44 to to ADP fcis-1701 40 45 all all DET fcis-1701 40 46 generated generate VERB fcis-1701 40 47 data datum NOUN fcis-1701 40 48 is be AUX fcis-1701 40 49 0 0 NUM fcis-1701 40 50 , , PUNCT fcis-1701 40 51 the the DET fcis-1701 40 52 gradient gradient NOUN fcis-1701 40 53 disappears disappear VERB fcis-1701 40 54 . . PUNCT fcis-1701 41 1 the the DET fcis-1701 41 2 training training NOUN fcis-1701 41 3 process process NOUN fcis-1701 41 4 of of ADP fcis-1701 41 5 traditional traditional ADJ fcis-1701 41 6 generation generation NOUN fcis-1701 41 7 countermeasure countermeasure NOUN fcis-1701 41 8 network network NOUN fcis-1701 41 9 is be AUX fcis-1701 41 10 unstable unstable ADJ fcis-1701 41 11 and and CCONJ fcis-1701 41 12 mode mode ADJ fcis-1701 41 13 collapse collapse NOUN fcis-1701 41 14 may may AUX fcis-1701 41 15 occur occur VERB fcis-1701 41 16 , , PUNCT fcis-1701 41 17 which which PRON fcis-1701 41 18 makes make VERB fcis-1701 41 19 the the DET fcis-1701 41 20 training training NOUN fcis-1701 41 21 difficult difficult ADJ fcis-1701 41 22 and and CCONJ fcis-1701 41 23 takes take VERB fcis-1701 41 24 a a DET fcis-1701 41 25 long long ADJ fcis-1701 41 26 time time NOUN fcis-1701 41 27 . . PUNCT fcis-1701 42 1 in in ADP fcis-1701 42 2 view view NOUN fcis-1701 42 3 of of ADP fcis-1701 42 4 the the DET fcis-1701 42 5 problem problem NOUN fcis-1701 42 6 of of ADP fcis-1701 42 7 generator generator NOUN fcis-1701 42 8 and and CCONJ fcis-1701 42 9 discriminator discriminator NOUN fcis-1701 42 10 confronting confronting NOUN fcis-1701 42 11 instability instability NOUN fcis-1701 42 12 in in ADP fcis-1701 42 13 the the DET fcis-1701 42 14 training training NOUN fcis-1701 42 15 process process NOUN fcis-1701 42 16 of of ADP fcis-1701 42 17 gan gan PROPN fcis-1701 42 18 , , PUNCT fcis-1701 42 19 wasserstein wasserstein NOUN fcis-1701 42 20 distance distance NOUN fcis-1701 42 21 was be AUX fcis-1701 42 22 used use VERB fcis-1701 42 23 instead instead ADV fcis-1701 42 24 of of ADP fcis-1701 42 25 js js ADJ fcis-1701 42 26 divergence divergence NOUN fcis-1701 42 27 to to PART fcis-1701 42 28 optimize optimize VERB fcis-1701 42 29 training training NOUN fcis-1701 42 30 and and CCONJ fcis-1701 42 31 generate generate VERB fcis-1701 42 32 countermeasures countermeasure NOUN fcis-1701 42 33 network network NOUN fcis-1701 42 34 . . PUNCT fcis-1701 43 1 [ [ X fcis-1701 43 2 8]wasserstein 8]wasserstein NUM fcis-1701 43 3 distance distance NOUN fcis-1701 43 4 is be AUX fcis-1701 43 5 defined define VERB fcis-1701 43 6 as as SCONJ fcis-1701 43 7 follows follow VERB fcis-1701 43 8 : : PUNCT fcis-1701 43 9 ( ( PUNCT fcis-1701 43 10 4 4 NUM fcis-1701 43 11 ) ) PUNCT fcis-1701 43 12 since since SCONJ fcis-1701 43 13 it it PRON fcis-1701 43 14 is be AUX fcis-1701 43 15 difficult difficult ADJ fcis-1701 43 16 to to PART fcis-1701 43 17 directly directly ADV fcis-1701 43 18 calculate calculate VERB fcis-1701 43 19 the the DET fcis-1701 43 20 wasserstein wasserstein NOUN fcis-1701 43 21 distance distance NOUN fcis-1701 43 22 of of ADP fcis-1701 43 23 two two NUM fcis-1701 43 24 distributions distribution NOUN fcis-1701 43 25 , , PUNCT fcis-1701 43 26 the the DET fcis-1701 43 27 kantorovich kantorovich PROPN fcis-1701 43 28 rubinstein rubinstein PROPN fcis-1701 43 29 dual dual ADJ fcis-1701 43 30 theorem theorem NOUN fcis-1701 43 31 is be AUX fcis-1701 43 32 adopted adopt VERB fcis-1701 43 33 as as ADP fcis-1701 43 34 the the DET fcis-1701 43 35 calculation calculation NOUN fcis-1701 43 36 form form NOUN fcis-1701 43 37 : : PUNCT fcis-1701 43 38 ( ( PUNCT fcis-1701 43 39 5 5 X fcis-1701 43 40 ) ) PUNCT fcis-1701 43 41 wgan wgan VERB fcis-1701 43 42 uses use NOUN fcis-1701 43 43 critical critical ADJ fcis-1701 43 44 as as ADP fcis-1701 43 45 a a DET fcis-1701 43 46 differentiator differentiator NOUN fcis-1701 43 47 from from ADP fcis-1701 43 48 the the DET fcis-1701 43 49 discriminator discriminator NOUN fcis-1701 43 50 in in ADP fcis-1701 43 51 gan gan PROPN fcis-1701 43 52 . . PUNCT fcis-1701 44 1 the the DET fcis-1701 44 2 differences difference NOUN fcis-1701 44 3 between between ADP fcis-1701 44 4 the the DET fcis-1701 44 5 two two NUM fcis-1701 44 6 are be AUX fcis-1701 44 7 : : PUNCT fcis-1701 44 8 the the DET fcis-1701 44 9 final final ADJ fcis-1701 44 10 layer layer NOUN fcis-1701 44 11 of of ADP fcis-1701 44 12 critical critical ADJ fcis-1701 44 13 removes remove VERB fcis-1701 44 14 sigmoid sigmoid NOUN fcis-1701 44 15 . . PUNCT fcis-1701 45 1 there there PRON fcis-1701 45 2 is be VERB fcis-1701 45 3 no no DET fcis-1701 45 4 log log NOUN fcis-1701 45 5 item item NOUN fcis-1701 45 6 in in ADP fcis-1701 45 7 the the DET fcis-1701 45 8 target target NOUN fcis-1701 45 9 function function NOUN fcis-1701 45 10 of of ADP fcis-1701 45 11 critical critical ADJ fcis-1701 45 12 . . PUNCT fcis-1701 46 1 to to PART fcis-1701 46 2 ensure ensure VERB fcis-1701 46 3 lipschitz lipschitz NOUN fcis-1701 46 4 limits limit NOUN fcis-1701 46 5 , , PUNCT fcis-1701 46 6 critical critical ADJ fcis-1701 46 7 needs need NOUN fcis-1701 46 8 to to PART fcis-1701 46 9 truncate truncate VERB fcis-1701 46 10 the the DET fcis-1701 46 11 parameters parameter NOUN fcis-1701 46 12 to to ADP fcis-1701 46 13 a a DET fcis-1701 46 14 certain certain ADJ fcis-1701 46 15 range range NOUN fcis-1701 46 16 after after ADP fcis-1701 46 17 each each DET fcis-1701 46 18 update update NOUN fcis-1701 46 19 . . PUNCT fcis-1701 47 1 the the PRON fcis-1701 47 2 better well ADJ fcis-1701 47 3 the the DET fcis-1701 47 4 critical critical ADJ fcis-1701 47 5 training training NOUN fcis-1701 47 6 is be AUX fcis-1701 47 7 , , PUNCT fcis-1701 47 8 the the PRON fcis-1701 47 9 better well ADJ fcis-1701 47 10 the the DET fcis-1701 47 11 generator generator NOUN fcis-1701 47 12 will will AUX fcis-1701 47 13 be be AUX fcis-1701 47 14 . . PUNCT fcis-1701 48 1 3 3 X fcis-1701 48 2 . . X fcis-1701 48 3 lstm lstm NOUN fcis-1701 48 4 - - PUNCT fcis-1701 48 5 gan gin VERB fcis-1701 48 6 in in ADP fcis-1701 48 7 lstm lstm NOUN fcis-1701 48 8 - - PUNCT fcis-1701 48 9 gan gin VERB fcis-1701 48 10 anomaly anomaly NOUN fcis-1701 48 11 detection detection NOUN fcis-1701 48 12 model model NOUN fcis-1701 48 13 , , PUNCT fcis-1701 48 14 the the DET fcis-1701 48 15 generator generator NOUN fcis-1701 48 16 aims aim VERB fcis-1701 48 17 to to PART fcis-1701 48 18 generate generate VERB fcis-1701 48 19 a a DET fcis-1701 48 20 false false ADJ fcis-1701 48 21 data data NOUN fcis-1701 48 22 distribution distribution NOUN fcis-1701 48 23 similar similar ADJ fcis-1701 48 24 to to ADP fcis-1701 48 25 the the DET fcis-1701 48 26 real real ADJ fcis-1701 48 27 sample sample NOUN fcis-1701 48 28 to to PART fcis-1701 48 29 fool fool VERB fcis-1701 48 30 the the DET fcis-1701 48 31 discriminator discriminator NOUN fcis-1701 48 32 . . PUNCT fcis-1701 49 1 the the DET fcis-1701 49 2 generator generator NOUN fcis-1701 49 3 contains contain VERB fcis-1701 49 4 three three NUM fcis-1701 49 5 layers layer NOUN fcis-1701 49 6 of of ADP fcis-1701 49 7 lstm lstm NOUN fcis-1701 49 8 . . PUNCT fcis-1701 50 1 the the DET fcis-1701 50 2 structure structure NOUN fcis-1701 50 3 of of ADP fcis-1701 50 4 the the DET fcis-1701 50 5 generator generator NOUN fcis-1701 50 6 is be AUX fcis-1701 50 7 shown show VERB fcis-1701 50 8 in in ADP fcis-1701 50 9 the the DET fcis-1701 50 10 figure figure NOUN fcis-1701 50 11 : : PUNCT fcis-1701 50 12 figure figure NOUN fcis-1701 50 13 2 2 NUM fcis-1701 50 14 . . PUNCT fcis-1701 51 1 the the DET fcis-1701 51 2 structure structure NOUN fcis-1701 51 3 of of ADP fcis-1701 51 4 generator generator NOUN fcis-1701 51 5 the the DET fcis-1701 51 6 discriminator discriminator NOUN fcis-1701 51 7 needs need VERB fcis-1701 51 8 to to PART fcis-1701 51 9 distinguish distinguish VERB fcis-1701 51 10 between between ADP fcis-1701 51 11 real real ADJ fcis-1701 51 12 samples sample NOUN fcis-1701 51 13 and and CCONJ fcis-1701 51 14 false false ADJ fcis-1701 51 15 samples sample NOUN fcis-1701 51 16 . . PUNCT fcis-1701 52 1 since since SCONJ fcis-1701 52 2 it it PRON fcis-1701 52 3 is be AUX fcis-1701 52 4 necessary necessary ADJ fcis-1701 52 5 to to PART fcis-1701 52 6 fit fit VERB fcis-1701 52 7 wasserstein wasserstein NOUN fcis-1701 52 8 distance distance NOUN fcis-1701 52 9 , , PUNCT fcis-1701 52 10 the the DET fcis-1701 52 11 objective objective ADJ fcis-1701 52 12 function function NOUN fcis-1701 52 13 is be AUX fcis-1701 52 14 : : PUNCT fcis-1701 52 15 ( ( PUNCT fcis-1701 52 16 ) ) PUNCT fcis-1701 52 17 ( ( PUNCT fcis-1701 52 18 ) ) PUNCT fcis-1701 52 19 ~ ~ PUNCT fcis-1701 52 20 ~min ~min X fcis-1701 52 21 max max PROPN fcis-1701 52 22 ( ( PUNCT fcis-1701 52 23 , , PUNCT fcis-1701 52 24 ) ) PUNCT fcis-1701 52 25 ( ( PUNCT fcis-1701 52 26 ( ( PUNCT fcis-1701 52 27 ) ) PUNCT fcis-1701 52 28 ) ) PUNCT fcis-1701 53 1 ( ( PUNCT fcis-1701 53 2 1 1 NUM fcis-1701 53 3 ( ( PUNCT fcis-1701 53 4 ( ( PUNCT fcis-1701 53 5 ) ) PUNCT fcis-1701 53 6 ) ) PUNCT fcis-1701 53 7 ) ) PUNCT fcis-1701 53 8 data datum NOUN fcis-1701 53 9 x x SYM fcis-1701 53 10 z z NOUN fcis-1701 53 11 zx zx NUM fcis-1701 54 1 p p NOUN fcis-1701 54 2 z z NOUN fcis-1701 54 3 p p NOUN fcis-1701 54 4 g g PROPN fcis-1701 54 5 d d X fcis-1701 54 6 v v NOUN fcis-1701 54 7 c c NOUN fcis-1701 54 8 g g NOUN fcis-1701 54 9 e e NOUN fcis-1701 54 10 c c X fcis-1701 54 11 x x PUNCT fcis-1701 54 12 e e NOUN fcis-1701 54 13 d d X fcis-1701 54 14 g g PROPN fcis-1701 54 15 z= z= PROPN fcis-1701 55 1 + + CCONJ fcis-1701 55 2 βˆ’ βˆ’ PROPN fcis-1701 55 3 ( ( PUNCT fcis-1701 55 4 6 6 NUM fcis-1701 55 5 ) ) SYM fcis-1701 55 6 4 4 NUM fcis-1701 55 7 . . PUNCT fcis-1701 55 8 model model NOUN fcis-1701 55 9 evaluation evaluation NOUN fcis-1701 55 10 in in ADP fcis-1701 55 11 this this DET fcis-1701 55 12 paper paper NOUN fcis-1701 55 13 , , PUNCT fcis-1701 55 14 real real ADJ fcis-1701 55 15 world world NOUN fcis-1701 55 16 datasets dataset NOUN fcis-1701 55 17 from from ADP fcis-1701 55 18 nab nab ADJ fcis-1701 55 19 dataset dataset NOUN fcis-1701 55 20 are be AUX fcis-1701 55 21 used use VERB fcis-1701 55 22 in in ADP fcis-1701 55 23 the the DET fcis-1701 55 24 experiment experiment NOUN fcis-1701 55 25 . . PUNCT fcis-1701 56 1 since since SCONJ fcis-1701 56 2 f1 f1 NOUN fcis-1701 56 3 can can AUX fcis-1701 56 4 give give VERB fcis-1701 56 5 consideration consideration NOUN fcis-1701 56 6 to to ADP fcis-1701 56 7 both both DET fcis-1701 56 8 the the DET fcis-1701 56 9 accuracy accuracy NOUN fcis-1701 56 10 and and CCONJ fcis-1701 56 11 recall recall NOUN fcis-1701 56 12 of of ADP fcis-1701 56 13 the the DET fcis-1701 56 14 model model NOUN fcis-1701 56 15 , , PUNCT fcis-1701 56 16 it it PRON fcis-1701 56 17 serves serve VERB fcis-1701 56 18 as as ADP fcis-1701 56 19 the the DET fcis-1701 56 20 main main ADJ fcis-1701 56 21 indicator indicator NOUN fcis-1701 56 22 of of ADP fcis-1701 56 23 the the DET fcis-1701 56 24 evaluation evaluation NOUN fcis-1701 56 25 model model NOUN fcis-1701 56 26 . . PUNCT fcis-1701 57 1 the the DET fcis-1701 57 2 experimental experimental ADJ fcis-1701 57 3 results result NOUN fcis-1701 57 4 are be AUX fcis-1701 57 5 as as SCONJ fcis-1701 57 6 follows follow VERB fcis-1701 57 7 ： ： PUNCT fcis-1701 57 8 table table NOUN fcis-1701 57 9 1 1 NUM fcis-1701 57 10 . . PUNCT fcis-1701 58 1 experimental experimental ADJ fcis-1701 58 2 results result NOUN fcis-1701 58 3 methods method NOUN fcis-1701 58 4 art art NOUN fcis-1701 58 5 adex adex NOUN fcis-1701 58 6 aws aw NOUN fcis-1701 58 7 traf traf PROPN fcis-1701 58 8 tweets tweet NOUN fcis-1701 58 9 mean mean VERB fcis-1701 58 10 lstm lstm NOUN fcis-1701 58 11 0.375 0.375 NUM fcis-1701 58 12 0.538 0.538 NUM fcis-1701 58 13 0.474 0.474 NUM fcis-1701 58 14 0.643 0.643 NUM fcis-1701 58 15 0.543 0.543 NUM fcis-1701 58 16 0.509 0.509 NUM fcis-1701 58 17 arima arima NOUN fcis-1701 58 18 0.353 0.353 NUM fcis-1701 58 19 0.583 0.583 NUM fcis-1701 58 20 0.518 0.518 NUM fcis-1701 58 21 0.571 0.571 NUM fcis-1701 58 22 0.567 0.567 NUM fcis-1701 58 23 0.524 0.524 NUM fcis-1701 58 24 lstmgan lstmgan VERB fcis-1701 58 25 0.545 0.545 NUM fcis-1701 58 26 0.632 0.632 NUM fcis-1701 58 27 0.676 0.676 NUM fcis-1701 58 28 0.684 0.684 NUM fcis-1701 58 29 0.678 0.678 NUM fcis-1701 58 30 0.643 0.643 NUM fcis-1701 58 31 bold bold ADJ fcis-1701 58 32 font font NOUN fcis-1701 58 33 is be AUX fcis-1701 58 34 the the DET fcis-1701 58 35 highest high ADJ fcis-1701 58 36 value value NOUN fcis-1701 58 37 . . PUNCT fcis-1701 59 1 it it PRON fcis-1701 59 2 can can AUX fcis-1701 59 3 be be AUX fcis-1701 59 4 seen see VERB fcis-1701 59 5 from from ADP fcis-1701 59 6 the the DET fcis-1701 59 7 table table NOUN fcis-1701 59 8 that that SCONJ fcis-1701 59 9 the the DET fcis-1701 59 10 model model NOUN fcis-1701 59 11 proposed propose VERB fcis-1701 59 12 in in ADP fcis-1701 59 13 this this DET fcis-1701 59 14 paper paper NOUN fcis-1701 59 15 performs perform VERB fcis-1701 59 16 best well ADV fcis-1701 59 17 on on ADP fcis-1701 59 18 each each DET fcis-1701 59 19 dataset dataset NOUN fcis-1701 59 20 . . PUNCT fcis-1701 60 1 5 5 X fcis-1701 60 2 . . X fcis-1701 60 3 conclusion conclusion NOUN fcis-1701 60 4 this this DET fcis-1701 60 5 paper paper NOUN fcis-1701 60 6 proposes propose VERB fcis-1701 60 7 lstm lstm ADJ fcis-1701 60 8 - - PUNCT fcis-1701 60 9 gan gin VERB fcis-1701 60 10 model model NOUN fcis-1701 60 11 for for ADP fcis-1701 60 12 anomaly anomaly NOUN fcis-1701 60 13 detection detection NOUN fcis-1701 60 14 of of ADP fcis-1701 60 15 time time NOUN fcis-1701 60 16 series series PROPN fcis-1701 60 17 data data PROPN fcis-1701 60 18 . . PUNCT fcis-1701 61 1 this this DET fcis-1701 61 2 model model NOUN fcis-1701 61 3 uses use VERB fcis-1701 61 4 lstm lstm NOUN fcis-1701 61 5 as as ADP fcis-1701 61 6 the the DET fcis-1701 61 7 basic basic ADJ fcis-1701 61 8 network network NOUN fcis-1701 61 9 of of ADP fcis-1701 61 10 generator generator NOUN fcis-1701 61 11 and and CCONJ fcis-1701 61 12 discriminator discriminator NOUN fcis-1701 61 13 , , PUNCT fcis-1701 61 14 and and CCONJ fcis-1701 61 15 uses use VERB fcis-1701 61 16 wasserstein wasserstein NOUN fcis-1701 61 17 distance distance NOUN fcis-1701 61 18 to to PART fcis-1701 61 19 replace replace VERB fcis-1701 61 20 the the DET fcis-1701 61 21 original original ADJ fcis-1701 61 22 measurement measurement NOUN fcis-1701 61 23 method method NOUN fcis-1701 61 24 . . PUNCT fcis-1701 62 1 in in ADP fcis-1701 62 2 order order NOUN fcis-1701 62 3 to to PART fcis-1701 62 4 verify verify VERB fcis-1701 62 5 the the DET fcis-1701 62 6 performance performance NOUN fcis-1701 62 7 of of ADP fcis-1701 62 8 the the DET fcis-1701 62 9 model model NOUN fcis-1701 62 10 , , PUNCT fcis-1701 62 11 the the DET fcis-1701 62 12 model model NOUN fcis-1701 62 13 is be AUX fcis-1701 62 14 tested test VERB fcis-1701 62 15 on on ADP fcis-1701 62 16 several several ADJ fcis-1701 62 17 real real ADJ fcis-1701 62 18 - - PUNCT fcis-1701 62 19 world world NOUN fcis-1701 62 20 datasets dataset NOUN fcis-1701 62 21 , , PUNCT fcis-1701 62 22 and and CCONJ fcis-1701 62 23 the the DET fcis-1701 62 24 model model NOUN fcis-1701 62 25 proposed propose VERB fcis-1701 62 26 in in ADP fcis-1701 62 27 this this DET fcis-1701 62 28 paper paper NOUN fcis-1701 62 29 has have VERB fcis-1701 62 30 good good ADJ fcis-1701 62 31 anomaly anomaly NOUN fcis-1701 62 32 detection detection NOUN fcis-1701 62 33 results result NOUN fcis-1701 62 34 . . PUNCT fcis-1701 63 1 references reference NOUN fcis-1701 63 2 [ [ X fcis-1701 63 3 1 1 NUM fcis-1701 63 4 ] ] X fcis-1701 63 5 c. c. PROPN fcis-1701 63 6 -k -k PROPN fcis-1701 63 7 . . PUNCT fcis-1701 64 1 lee lee PROPN fcis-1701 64 2 , , PUNCT fcis-1701 64 3 y. y. PROPN fcis-1701 64 4 -j -j PROPN fcis-1701 64 5 . . PUNCT fcis-1701 65 1 cheon cheon PROPN fcis-1701 65 2 and and CCONJ fcis-1701 65 3 w. w. PROPN fcis-1701 65 4 -y -y PROPN fcis-1701 65 5 . . PUNCT fcis-1701 66 1 hwang hwang PROPN fcis-1701 66 2 , , PUNCT fcis-1701 66 3 " " PUNCT fcis-1701 66 4 studies study NOUN fcis-1701 66 5 on on ADP fcis-1701 66 6 the the DET fcis-1701 66 7 gan gan PROPN fcis-1701 66 8 - - PUNCT fcis-1701 66 9 based base VERB fcis-1701 66 10 anomaly anomaly NOUN fcis-1701 66 11 detection detection NOUN fcis-1701 66 12 methods method NOUN fcis-1701 66 13 for for ADP fcis-1701 66 14 the the DET fcis-1701 66 15 time time NOUN fcis-1701 66 16 series series PROPN fcis-1701 66 17 data data PROPN fcis-1701 66 18 , , PUNCT fcis-1701 66 19 " " PUNCT fcis-1701 66 20 in in ADP fcis-1701 66 21 ieee ieee NOUN fcis-1701 66 22 access access NOUN fcis-1701 66 23 , , PUNCT fcis-1701 66 24 vol vol NOUN fcis-1701 66 25 . . NOUN fcis-1701 66 26 9 9 NUM fcis-1701 66 27 , , PUNCT fcis-1701 66 28 pp pp ADJ fcis-1701 66 29 . . PUNCT fcis-1701 67 1 73201 73201 NUM fcis-1701 67 2 - - SYM fcis-1701 67 3 73215 73215 NUM fcis-1701 67 4 , , PUNCT fcis-1701 67 5 2021 2021 NUM fcis-1701 67 6 , , PUNCT fcis-1701 67 7 doi doi NOUN fcis-1701 67 8 : : PUNCT fcis-1701 67 9 10.1109 10.1109 NUM fcis-1701 67 10 / / SYM fcis-1701 67 11 access.2021.3078553 access.2021.3078553 NOUN fcis-1701 67 12 . . PUNCT fcis-1701 68 1 [ [ X fcis-1701 68 2 2 2 NUM fcis-1701 68 3 ] ] PUNCT fcis-1701 68 4 su su PROPN fcis-1701 68 5 y y PROPN fcis-1701 68 6 , , PUNCT fcis-1701 68 7 zhao zhao PROPN fcis-1701 68 8 y y PROPN fcis-1701 68 9 , , PUNCT fcis-1701 68 10 niu niu PROPN fcis-1701 68 11 c c PROPN fcis-1701 68 12 , , PUNCT fcis-1701 68 13 et et PROPN fcis-1701 68 14 al al PROPN fcis-1701 68 15 . . PROPN fcis-1701 68 16 robust robust ADJ fcis-1701 68 17 anomaly anomaly NOUN fcis-1701 68 18 detection detection NOUN fcis-1701 68 19 for for ADP fcis-1701 68 20 multivariate multivariate NOUN fcis-1701 68 21 time time NOUN fcis-1701 68 22 series series PROPN fcis-1701 68 23 through through ADP fcis-1701 68 24 stochastic stochastic ADJ fcis-1701 68 25 recurrent recurrent ADJ fcis-1701 68 26 neural neural NOUN fcis-1701 68 27 network[c]// network[c]// PROPN fcis-1701 68 28 the the DET fcis-1701 68 29 25th 25th ADJ fcis-1701 68 30 acm acm PROPN fcis-1701 68 31 sigkdd sigkdd NOUN fcis-1701 68 32 international international ADJ fcis-1701 68 33 conference conference NOUN fcis-1701 68 34 . . PUNCT fcis-1701 69 1 acm acm PROPN fcis-1701 69 2 , , PUNCT fcis-1701 69 3 2019 2019 NUM fcis-1701 69 4 . . PUNCT fcis-1701 70 1 [ [ X fcis-1701 70 2 3 3 NUM fcis-1701 70 3 ] ] X fcis-1701 70 4 tang tang PROPN fcis-1701 70 5 j j PROPN fcis-1701 70 6 , , PUNCT fcis-1701 70 7 chen chen PROPN fcis-1701 70 8 z z PROPN fcis-1701 70 9 , , PUNCT fcis-1701 70 10 fu fu PROPN fcis-1701 70 11 a a PRON fcis-1701 70 12 , , PUNCT fcis-1701 70 13 et et PROPN fcis-1701 70 14 al al PROPN fcis-1701 70 15 . . PROPN fcis-1701 70 16 enhancing enhance VERB fcis-1701 70 17 effectiveness effectiveness NOUN fcis-1701 70 18 of of ADP fcis-1701 70 19 outlier outlier NOUN fcis-1701 70 20 detections detection NOUN fcis-1701 70 21 for for ADP fcis-1701 70 22 low low ADJ fcis-1701 70 23 density density NOUN fcis-1701 70 24 patterns[j patterns[j PROPN fcis-1701 70 25 ] ] PUNCT fcis-1701 70 26 . . PUNCT fcis-1701 71 1 springerverlag springerverlag PROPN fcis-1701 71 2 , , PUNCT fcis-1701 71 3 2002 2002 NUM fcis-1701 71 4 . . PUNCT fcis-1701 72 1 [ [ X fcis-1701 72 2 4 4 X fcis-1701 72 3 ] ] X fcis-1701 72 4 ma ma PROPN fcis-1701 72 5 biao biao PROPN fcis-1701 72 6 , , PUNCT fcis-1701 72 7 jia jia PROPN fcis-1701 72 8 jun jun PROPN fcis-1701 72 9 - - PUNCT fcis-1701 72 10 cheng cheng PROPN fcis-1701 72 11 , , PUNCT fcis-1701 72 12 dong dong PROPN fcis-1701 72 13 guo guo PROPN fcis-1701 72 14 - - PUNCT fcis-1701 72 15 zhu zhu PROPN fcis-1701 72 16 et et PROPN fcis-1701 72 17 al al PROPN fcis-1701 72 18 . . PROPN fcis-1701 72 19 wgan wgan PROPN fcis-1701 72 20 : : PUNCT fcis-1701 72 21 industrial industrial ADJ fcis-1701 72 22 control control NOUN fcis-1701 72 23 sensor sensor PROPN fcis-1701 72 24 data datum NOUN fcis-1701 72 25 anomaly anomaly NOUN fcis-1701 72 26 detection detection NOUN fcis-1701 72 27 method method NOUN fcis-1701 72 28 based base VERB fcis-1701 72 29 on on ADP fcis-1701 72 30 wavelet wavelet NOUN fcis-1701 72 31 transform transform NOUN fcis-1701 72 32 and and CCONJ fcis-1701 72 33 attention attention NOUN fcis-1701 72 34 mechanism mechanism NOUN fcis-1701 72 35 . . PUNCT fcis-1701 73 1 [ [ X fcis-1701 73 2 j j X fcis-1701 73 3 / / SYM fcis-1701 73 4 ol ol PROPN fcis-1701 73 5 ] ] PUNCT fcis-1701 73 6 . . PUNCT fcis-1701 74 1 journal journal PROPN fcis-1701 74 2 of of ADP fcis-1701 74 3 chinese chinese ADJ fcis-1701 74 4 computer computer NOUN fcis-1701 74 5 systems,2021:1 systems,2021:1 NOUN fcis-1701 74 6 - - PUNCT fcis-1701 74 7 12 12 NUM fcis-1701 74 8 . . PUNCT fcis-1701 75 1 [ [ X fcis-1701 75 2 5 5 X fcis-1701 75 3 ] ] X fcis-1701 75 4 malhotra malhotra PROPN fcis-1701 75 5 p p PROPN fcis-1701 75 6 , , PUNCT fcis-1701 75 7 vig vig PROPN fcis-1701 75 8 l l NOUN fcis-1701 75 9 , , PUNCT fcis-1701 75 10 shroff shroff NOUN fcis-1701 75 11 g g PROPN fcis-1701 75 12 , , PUNCT fcis-1701 75 13 et et PROPN fcis-1701 75 14 al al PROPN fcis-1701 75 15 . . PUNCT fcis-1701 76 1 long long ADJ fcis-1701 76 2 short short ADJ fcis-1701 76 3 term term NOUN fcis-1701 76 4 memory memory NOUN fcis-1701 76 5 networks network NOUN fcis-1701 76 6 for for ADP fcis-1701 76 7 anomaly anomaly NOUN fcis-1701 76 8 detection detection NOUN fcis-1701 76 9 in in ADP fcis-1701 76 10 time time NOUN fcis-1701 76 11 series[c]// series[c]// PROPN fcis-1701 76 12 23rd 23rd ADJ fcis-1701 76 13 european european ADJ fcis-1701 76 14 symposium symposium NOUN fcis-1701 76 15 on on ADP fcis-1701 76 16 artificial artificial ADJ fcis-1701 76 17 neural neural ADJ fcis-1701 76 18 networks network NOUN fcis-1701 76 19 , , PUNCT fcis-1701 76 20 computational computational ADJ fcis-1701 76 21 intelligence intelligence NOUN fcis-1701 76 22 and and CCONJ fcis-1701 76 23 machine machine NOUN fcis-1701 76 24 learning learning NOUN fcis-1701 76 25 , , PUNCT fcis-1701 76 26 esann esann PROPN fcis-1701 76 27 2015 2015 NUM fcis-1701 76 28 . . PUNCT fcis-1701 77 1 2015 2015 NUM fcis-1701 77 2 . . PUNCT fcis-1701 78 1 [ [ X fcis-1701 78 2 6 6 NUM fcis-1701 78 3 ] ] PUNCT fcis-1701 78 4 r. r. PROPN fcis-1701 78 5 hsieh hsieh PROPN fcis-1701 78 6 , , PUNCT fcis-1701 78 7 j. j. PROPN fcis-1701 78 8 chou chou PROPN fcis-1701 78 9 and and CCONJ fcis-1701 78 10 c. c. PROPN fcis-1701 78 11 ho ho PROPN fcis-1701 78 12 , , PUNCT fcis-1701 78 13 " " PUNCT fcis-1701 78 14 unsupervised unsupervised ADJ fcis-1701 78 15 online online ADJ fcis-1701 78 16 anomaly anomaly NOUN fcis-1701 78 17 detection detection NOUN fcis-1701 78 18 on on ADP fcis-1701 78 19 multivariate multivariate NOUN fcis-1701 78 20 sensing sensing NOUN fcis-1701 78 21 time time NOUN fcis-1701 78 22 series series PROPN fcis-1701 78 23 data datum NOUN fcis-1701 78 24 for for ADP fcis-1701 78 25 smart smart ADJ fcis-1701 78 26 manufacturing manufacturing NOUN fcis-1701 78 27 , , PUNCT fcis-1701 78 28 " " PUNCT fcis-1701 78 29 2019 2019 NUM fcis-1701 78 30 ieee ieee NOUN fcis-1701 78 31 12th 12th NOUN fcis-1701 78 32 conference conference NOUN fcis-1701 78 33 on on ADP fcis-1701 78 34 serviceoriented serviceoriente VERB fcis-1701 78 35 computing computing NOUN fcis-1701 78 36 and and CCONJ fcis-1701 78 37 applications application NOUN fcis-1701 78 38 ( ( PUNCT fcis-1701 78 39 soca soca NOUN fcis-1701 78 40 ) ) PUNCT fcis-1701 78 41 , , PUNCT fcis-1701 78 42 2019 2019 NUM fcis-1701 78 43 , , PUNCT fcis-1701 78 44 pp pp ADJ fcis-1701 78 45 . . PUNCT fcis-1701 79 1 9097 9097 NUM fcis-1701 79 2 , , PUNCT fcis-1701 79 3 doi doi NOUN fcis-1701 79 4 : : PUNCT fcis-1701 79 5 10.1109 10.1109 NUM fcis-1701 79 6 / / SYM fcis-1701 79 7 soca.2019.00021 soca.2019.00021 NUM fcis-1701 79 8 . . PUNCT fcis-1701 80 1 37 37 NUM fcis-1701 81 1 [ [ X fcis-1701 81 2 7 7 NUM fcis-1701 81 3 ] ] X fcis-1701 81 4 goodfellow goodfellow PROPN fcis-1701 81 5 i i PROPN fcis-1701 81 6 j j PROPN fcis-1701 81 7 , , PUNCT fcis-1701 81 8 pouget pouget NOUN fcis-1701 81 9 - - PUNCT fcis-1701 81 10 abadie abadie NOUN fcis-1701 81 11 j j PROPN fcis-1701 81 12 , , PUNCT fcis-1701 81 13 mirza mirza PROPN fcis-1701 81 14 , , PUNCT fcis-1701 81 15 et et PROPN fcis-1701 81 16 al al PROPN fcis-1701 81 17 . . PROPN fcis-1701 81 18 generative generative PROPN fcis-1701 81 19 adversarial adversarial ADJ fcis-1701 81 20 networks[j networks[j PROPN fcis-1701 81 21 ] ] PUNCT fcis-1701 81 22 . . PUNCT fcis-1701 82 1 advances advance NOUN fcis-1701 82 2 in in ADP fcis-1701 82 3 neural neural ADJ fcis-1701 82 4 information information NOUN fcis-1701 82 5 processing processing NOUN fcis-1701 82 6 systems system NOUN fcis-1701 82 7 , , PUNCT fcis-1701 82 8 2014 2014 NUM fcis-1701 82 9 , , PUNCT fcis-1701 82 10 3:2672 3:2672 NUM fcis-1701 82 11 - - SYM fcis-1701 82 12 2680 2680 NUM fcis-1701 82 13 . . PUNCT fcis-1701 83 1 [ [ X fcis-1701 83 2 8 8 NUM fcis-1701 83 3 ] ] X fcis-1701 83 4 arjovsky arjovsky PROPN fcis-1701 83 5 m m PROPN fcis-1701 83 6 , , PUNCT fcis-1701 83 7 chintala chintala PROPN fcis-1701 83 8 s s PROPN fcis-1701 83 9 , , PUNCT fcis-1701 83 10 bottou bottou PROPN fcis-1701 83 11 l. l. PROPN fcis-1701 83 12 wasserstein wasserstein PROPN fcis-1701 83 13 gan[j gan[j PROPN fcis-1701 83 14 ] ] PUNCT fcis-1701 83 15 . . PUNCT fcis-1701 84 1 arxiv arxiv PROPN fcis-1701 84 2 preprint preprint VERB fcis-1701 84 3 arxiv:1701.07875 arxiv:1701.07875 PROPN fcis-1701 84 4 , , PUNCT fcis-1701 84 5 2017.m 2017.m X fcis-1701 84 6 . . PUNCT