Generative Adversarial Networks by Hasanov Vagif - TMLS #4

Slide 1

Slide 1 text

July 2016 (FOFSBUJWF"EWFSTBSJBM /FUXPSLT

Slide 2

Slide 2 text

Tokyo Machine Learning Society Presentation Reactive (FOFSBUFOBUVSBMMPPLJOHJNBHFT %PTFNJTVQFSWJTFEMFBSOJOHXJUIGFXMBCFMT CVUNBOZVOMBCFMFETBNQMFT (FOFSBUFJNBHFTPGQMBOFTXJUICJSEMFHT 8IZ(FOFSBUJWF"EWFSTBSJBM/FUXPSLT

Slide 3

Slide 3 text

Tokyo Machine Learning Society Presentation Reactive 5JNFMJOFPGTFMFDUFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLTSFTFBSDI

Slide 4

Slide 4 text

Tokyo Machine Learning Society Presentation Reactive (FOFSBUJWF"EWFSTBSJBM/FUXPSLT *BO(PPEGFMMPXFUBM VQMPBEFEJO+VOF 5JNFMJOFPGTFMFDUFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLTSFTFBSDI

Slide 5

Slide 5 text

Tokyo Machine Learning Society Presentation Reactive 3BOEPN WFDUPS (FOFSBUPS NPEFM 3BOEPNTBNQMF GSPNUIFEBUBTFU %JTDSJNJOBUPS NPEFM QSPCBCJMJUZ QSPCBCJMJUZ JGSFBM JGHFOFSBUFE (FOFSBUPSJTUSZJOHUPNJOJNJ[FUIFNBYJNVNQPTTJCMF QFSGPSNBODFPGEJTDSJNJOBUPS (FOFSBUJWF"EWFSTBSJBM/FUXPSL

Slide 6

Slide 6 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQTQBQFSTOJQTDDQBQFSHFOFSBUJWFBEWFSTBSJBMOFUTQEG /PJTF %JTUSJCVUJPOPG SFBMEBUB %JTUSJCVUJPOPG HFOFSBUFEEBUB %JTDSJNJOBUPS PVUQVU &YBNQMFGPSPOFEJNFOTJPOBMEBUB

Slide 7

Slide 7 text

Slide 8

Slide 8 text

Slide 9

Slide 9 text

Slide 10

Slide 10 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQTQBQFSTOJQTDDQBQFSHFOFSBUJWFBEWFSTBSJBMOFUTQEG (FOFSBUFEJNBHFT

Slide 11

Slide 11 text

Slide 12

Slide 12 text

Tokyo Machine Learning Society Presentation Reactive #SFBLUIFDPNQMFYHFOFSBUPSGVODUJPOJOUPBTFSJFTPGMFTTDPNQMFY GVODUJPOT .BJOJEFB

Slide 13

Slide 13 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQBSYJWPSHQEGWQEG (FOFSBUFEEJ⒎FSFODF JNBHF "WFDUPSPGSBOEPN OVNCFSTJO< > (FOFSBUPSOFUXPSL %JTDSJNJOBUPS OFUXPSL "DUVBMEJ⒎FSFODF JNBHF 5SBJOJOHUIFNPEFM

Slide 14

Slide 14 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQBSYJWPSHQEGWQEG 5SBJOJOHUIFNPEFM

Slide 15

Slide 15 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQXXXDTOZVFEVdEFOUPOQBQFSTMBQHBO@TVQQQEG

Slide 16

Slide 16 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQBSYJWPSHQEGWQEG (FOFSBUJOHJNBHFT

Slide 17

Slide 17 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDF JNBHFqJQQFEIPSJ[POUBMMZ IUUQBSYJWPSHQEGWQEG $PBSTFUPpOFJNBHFHFOFSBUJPOQSPDFTT

Slide 18

Slide 18 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQXXXDTOZVFEVdEFOUPOQBQFSTMBQHBO@TMJEFTQEG (FOFSBUFEDIVSDIFT

Slide 19

Slide 19 text

Tokyo Machine Learning Society Presentation Reactive (FOFSBUJWF"EWFSTBSJBM/FUXPSLT *BO(PPEGFMMPXFUBM VQMPBEFEJO+VOF %FFQHFOFSBUJWFJNBHFNPEFMTVTJOHB -BQMBDJBOQZSBNJEPGBEWFSTBSJBMOFUXPSLT &NJMZ%FOUPOFUBM VQMPBEFEJO+VOF 6OTVQFSWJTFESFQSFTFOUBUJPOMFBSOJOHXJUI EFFQDPOWPMVUJPOBMHFOFSBUJWFBEWFSTBSJBMOFUXPSLT "MFD3BEGPSEFUBM VQMPBEFEJO/PWFNCFS 5JNFMJOFPGTFMFDUFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLTSFTFBSDI

Slide 20

Slide 20 text

Tokyo Machine Learning Society Presentation Reactive %FFQDPOWPMVUJPOBMHFOFSBUJWFBEWFSTBSJBMOFUXPSLT %$("/ ("/XJUICFUUFSOFUXPSLBSDIJUFDUVSF

Slide 21

Slide 21 text

Slide 22

Slide 22 text

Tokyo Machine Learning Society Presentation Reactive 5SJDLT 3FQMBDFQPPMJOHXJUITUSJEFEDPOWPMVUJPOTJOEJTDSJNJOBUPS 6TFGSBDUJPOBMTUSJEFEDPOWPMVUJPOT EFDPOWPMVUJPOT JOHFOFSBUPS 6TFCBUDIOPSNBMJ[BUJPOJOCPUIUIFHFOFSBUPSBOEUIFEJTDSJNJOBUPS 3FNPWFGVMMZDPOOFDUFEIJEEFOMBZFSTGPSEFFQFSBSDIJUFDUVSFT 6TF3F-6BDUJWBUJPOJOHFOFSBUPSGPSBMMMBZFSTFYDFQUGPSUIFPVUQVU XIJDIVTFT5BOI 6TF-FBLZ3F-6BDUJWBUJPOJOUIFEJTDSJNJOBUPSGPSBMMMBZFST 6TF"%".PQUJNJ[BUJPONFUIPEGPSPQUJNJ[BUJPO BOEUXFBLUIF IZQFSQBSBNFUFST

Slide 23

Slide 23 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG (FOFSBUFECFESPPNJNBHFT

Slide 24

Slide 24 text

Tokyo Machine Learning Society Presentation Reactive .PSFJNBHFT JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG

Slide 25

Slide 25 text

Tokyo Machine Learning Society Presentation Reactive .PSFJNBHFT 0VSGSJFOE JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG

Slide 26

Slide 26 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG "SJUINFUJDTXJUIWFDUPST

Slide 27

Slide 27 text

Tokyo Machine Learning Society Presentation Reactive (FOFSBUJWF"EWFSTBSJBM/FUXPSLT *BO(PPEGFMMPXFUBM VQMPBEFEJO+VOF %FFQHFOFSBUJWFJNBHFNPEFMTVTJOHB -BQMBDJBOQZSBNJEPGBEWFSTBSJBMOFUXPSLT &NJMZ%FOUPOFUBM VQMPBEFEJO+VOF 6OTVQFSWJTFESFQSFTFOUBUJPOMFBSOJOHXJUI EFFQDPOWPMVUJPOBMHFOFSBUJWFBEWFSTBSJBMOFUXPSLT "MFD3BEGPSEFUBM VQMPBEFEJO/PWFNCFS *NQSPWFE5FDIOJRVFTGPS5SBJOJOH("/T 5JN4BMJNBOTFUBM 6QMPBEFE+VOF 5JNFMJOFPGTFMFDUFE(FOFSBUJWF"EWFSTBSJBM/FUXPSLTSFTFBSDI

Slide 28

Slide 28 text

Tokyo Machine Learning Society Presentation Reactive .PSFUSJDLTGPSUSBJOJOH("/T 'FBUVSFNBUDIJOH (FOFSBUPSPCKFDUJWF

Slide 29

Slide 29 text

Tokyo Machine Learning Society Presentation Reactive .PSFUSJDLTGPSUSBJOJOH("/T 'FBUVSFNBUDIJOH .JOJCBUDIEJTDSJNJOBUJPOQBTTUIFJOGPSNBUJPOBCPVUUIFTJNJMBSJUZ CFUXFFOTBNQMFTXJUIJONJOJCBUDIUPUIFEJTDSJNJOBUPSOFUXPSL

Slide 30

Slide 30 text

Tokyo Machine Learning Society Presentation Reactive .PSFUSJDLTGPSUSBJOJOH("/T 'FBUVSFNBUDIJOH .JOJCBUDIEJTDSJNJOBUJPO )JTUPSJDBMBWFSBHJOHQFOBMJ[FMBSHFTVEEFODIBOHFTJOUIF QBSBNFUFSTCZBEEJOHUPUIFDPTUGVODUJPOPGCPUI %BOE(

Slide 31

Slide 31 text

Tokyo Machine Learning Society Presentation Reactive .PSFUSJDLTGPSUSBJOJOH("/T 'FBUVSFNBUDIJOH .JOJCBUDIEJTDSJNJOBUJPO )JTUPSJDBMBWFSBHJOH 0OFTJEFEMBCFMTNPPUIJOHSFQMBDFUIFIBSEMBCFMTCZTPGUFS MBCFMT

Slide 32

Slide 32 text

Tokyo Machine Learning Society Presentation Reactive .PSFUSJDLTGPSUSBJOJOH("/T 'FBUVSFNBUDIJOH .JOJCBUDIEJTDSJNJOBUJPO )JTUPSJDBMBWFSBHJOH 0OFTJEFEMBCFMTNPPUIJOH 7JSUVBMCBUDIOPSNBMJ[BUJPOVTFBpYFESFGFSFODFCBUDIGPSUIF CBUDIOPSNBMJ[BUJPOTUBUJTUJDTDBMDVMBUJPO5IJTQSFWFOUTUIFOFUXPSL PVUQVUGPSBHJWFOTBNQMFCFJOHEFQFOEFOUPOUIFPUIFSTBNQMFT XJUIJONJOJCBUDI

Slide 33

Slide 33 text

Tokyo Machine Learning Society Presentation Reactive (FOFSBUFETNBMMJNBHFTBSFPGHPPERVBMJUZ JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG

Slide 34

Slide 34 text

Tokyo Machine Learning Society Presentation Reactive JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG 3FBMIPSTF 3FBMDBS (FOFSBUFETNBMMJNBHFTBSFPGHPPERVBMJUZ

Slide 35

Slide 35 text

Tokyo Machine Learning Society Presentation Reactive +VOF +VOF 1SPHSFTTPGUIFQBTUUXPZFBST "NB[JOH JNBHFT`TPVSDFT IUUQTQBQFSTOJQTDDQBQFSHFOFSBUJWFBEWFSTBSJBMOFUTQEG IUUQTBSYJWPSHQEGWQEG

Slide 36

Slide 36 text

Tokyo Machine Learning Society Presentation Reactive %$("/ *NQSPWFE("/ -BSHFBOEDPNQMFYJNBHFHFOFSBUJPOJTTUJMMBOPQFOQSPCMFN JNBHFTPVSDFIUUQTBSYJWPSHQEGWQEG

Slide 37

Slide 37 text

Tokyo Machine Learning Society Presentation Reactive 4FNJTVQFSWJTFEMFBSOJOHXJUI("/T 6TFDMBTTJpFSBTBEJTDSJNJOBUPSCZBEEJOHBlHFOFSBUFEEBUBz , DMBTTUPJUTMJTUPGDMBTTFTGPSDMBTTJpDBUJPO 0QUJNJ[FTVQFSWJTFEBOEVOTVQFSWJTFEMPTTFTKPJOUMZ "DIJFWFTTUBUFPGUIFBSUGPSTFNJTVQFSWJTFEMFBSOJOHPO ./*45 $*'"3 47)/

Slide 38

Slide 38 text

Tokyo Machine Learning Society Presentation Reactive $PODMVTJPO 6TF("/TGPSHFOFSBUJOHTNBMMJNBHFT PSMBSHFCVUOPU DPNQMFYJNBHFT 6TF("/TUPUSBJODMBTTJpFSTXIFOZPVIBWFTNBMMBNPVOU PGMBCFMFEJNBHFT BOEMBSHFBNPVOUPGVOMBCFMFEJNBHFT

Slide 39

Slide 39 text

5IBOLZPV

Slide 40

Slide 40 text

Tokyo Machine Learning Society Presentation Reactive • Generative adversarial nets   (Ian J. Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, Yoshua Bengio, NIPS, 2014) • Deep generative image models using a Laplacian pyramid of adversarial networks   (Emily Denton, Soumith Chintala, Arthur Szlam, Rob Fergus, NIPS, 2015) • Unsupervised representation learning with deep convolutional generative adversarial networks  (Alec Radford, Luke Metz, Soumith Chintala, ICLR, 2016) • Improved Techniques for Training GANs   (Tim Salimans, Ian J. Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, Xi Chen, arXiv: 1606.03498, 2016) 3FGFSFODFT