20221005_AI勉強会

https://deeplearning-b.connpass.com/event/259882/

Transcript

1. AI࠷৽࿦จಡΈձ2022೥10݄
   ᷂tech vein ழມ ॆԝ
   

   View Slide

2. ࣗݾ঺հ
   ழມ ॆԝ (͍ͷ·ͨ ΈͭͻΖ)
   
   
   גࣜձࣾ tech vein / DeepRad גࣜձࣾ
   
   
   ֤୅දऔక໾ ݉ σϕϩούʔ
   
   
   twitter: @ino2222
   

   View Slide

3. Facebook άϧʔϓͷ঺հ
   IUUQTXXXGBDFCPPLDPNHSPVQT
   

   View Slide

4. ΞδΣϯμ
   https://paperswithcode.com/ ʹΑΔɺ
   
   
   arxiv.org ͷաڈ1ϲ݄ؒͷ࿦จτοϓ10ຊΛ
   ঺հɻ
   
   
   

   View Slide

5. Papers with Code
   https://www.arxiv-sanity.com/top
   

   View Slide

6. ໨࣍
   

   View Slide

7. Top10
   1. Git Re-Basin: Merging Models modulo Permutation Symmetries
   
   
   2. QuestSim: Human Motion Tracking from Sparse Sensors with
   Simulated Avatars
   
   
   3. Understanding Diffusion Models: A Uni
   fi
   ed Perspective
   
   
   4. Transformers are Sample Ef
   fi
   cient World Models
   
   
   5. Brain Imaging Generation with Latent Diffusion Models
   
   
   6. Learning with Differentiable Algorithms
   
   
   7. Operationalizing Machine Learning: An Interview Study
   
   
   8. Decoding speech from non-invasive brain recordings
   
   
   9. Faithful Reasoning Using Large Language Models
   
   
   10. AudioLM: a Language Modeling Approach to Audio Generation
   

   View Slide

8. 1. Git Re-Basinɻฒ΂׵͑ରশੑʹΑΔϞσϧͷϚʔδ
   
   
   (ݪจ: Git Re-Basin: Merging Models modulo Permutation Symmetries)
   
   
   σΟʔϓϥʔχϯάͷ੒ޭ͸ɺ͋Δछͷڊେͳඇತ࠷దԽ໰୊Λൺֱత༰қʹղ͘͜ͱ͕Ͱ͖ΔΑ͏ʹ
   ͳ͓͔ͬͨ͛Ͱ͋Δɻඇತ࠷దԽ͸NPࠔ೉Ͱ͋Δʹ΋͔͔ΘΒͣɺ୯७ͳΞϧΰϦζϜʢ͠͹͠͹֬཰
   తޯ഑߱Լͷมछʣ͸ɺ࣮ࡍʹେن໛ͳχϡʔϥϧωοτϫʔΫʹద߹͢Δࡍʹڻ͘΂͖༗ޮੑΛࣔ
   ͢ɻզʑ͸ɺχϡʔϥϧωοτϫʔΫͷଛࣦϥϯυεέʔϓ͸ɺӅΕϢχοτͷ͢΂ͯͷՄೳͳॱྻର
   শੑΛߟྀͨ͠ޙɺʢ΄΅ʣ୯ҰͷྲྀҬ(basin)ΛؚΉ͜ͱΛओு͢Δɻզʑ͸ɺ͋ΔϞσϧͷϢχο
   τΛฒ΂ସ͑ͯɺࢀরϞσϧͷϢχοτͱҰகͤ͞ΔͨΊͷ3ͭͷΞϧΰϦζϜΛ঺հ͢Δɻ͜ͷม׵
   ʹΑΓɺࢀরϞσϧʹ͍ۙತຍ஍ʹҐஔ͢Δɺػೳతʹ౳ՁͳॏΈͷू߹͕ੜ੒͞ΕΔɻ࣮ݧతʹ͸ɺ
   CIFAR-10ͱCIFAR-100Ͱಠཱʹֶशͨ͠ResNetϞσϧؒͷθϩόϦΞઢܗϞʔυ઀ଓΛॳΊ࣮ͯূ͢
   ΔͳͲɺ༷ʑͳϞσϧΞʔΩςΫνϟͱσʔληοτͰ୯ҰྲྀҬݱ৅Λ࣮ূ͠·ͨ͠ɻ͞Βʹɺ༷ʑͳ
   Ϟσϧ΍σʔληοτʹ͓͍ͯɺϞσϧ෯΍ֶश࣌ؒͱϞʔυ઀ଓੑͱͷؒʹڵຯਂ͍ݱ৅͕͋Δ͜ͱ
   Λ໌Β͔ʹ͠·ͨ͠ɻ࠷ޙʹɺ୯ҰྲྀҬཧ࿦ͷܽ఺ʹ͍ͭͯɺઢܗϞʔυ݁߹Ծઆͷ൓ྫΛؚΊͯٞ࿦
   ͢Δɻ
   w ໨తɿϞσϧ࠷దԽɾ࿈߹ֶशɾΞϯαϯϒϧֶशͷͨΊͷجૅݚڀ
   w ੒Ռɿ$*'"3
   Ͱݸผʹֶशͨͭ͠ͷ3FT/FUϞσϧΛθϩόϦΞઢܗϞʔυ઀ଓ͢Δ͜
   ͱ͕Ͱ͖ͨ
   w ํ๏ɿϞσϧϢχοτͷม׵ΞϧΰϦζϜͷಋೖ
   w ݻ༗໊ɿ(JU3F#BTJO
   w ஶऀॴଐɿϫγϯτϯେֶίϯϐϡʔλʔαΠΤϯεɾΤϯδχΞϦϯάֶ෦
   https://arxiv.org/abs/2209.04836v1
   

   View Slide

9. Loss Landscape
   
   
   →Lossؔ਺ͷՄࢹԽख๏
   IUUQTBSYJWPSHBCT
   

   View Slide

10. ଛࣦϥϯυεέʔϓ͕Single Basin
    ʢ୯Ұͷ͘΅Έʣ
    • SGDͰ࠷దղʹͨͲΓண͖΍͍͢ɻ
    

    View Slide

11. Linear Mode Connectivity (ઢܗϞʔυ઀ଓ)
    
    
    Ҿ༻࿦จ: Linear Mode Connectivity and the Lottery
    Ticket Hypothesis
    IUUQTBSYJWPSHBCT
    

    View Slide

12. Lottery Ticket Hypothesis
    
    
    (͘͡Ҿ͖Ծઆ)
    • େن໛ͳχϡʔϥϧωοτϫʔΫ͕ߴੑೳΛୡ੒͠΍͢
    ͍ͷ͸ɼ͘͡Λͨ͘͞ΜҾ͍͍ͯΔ͔ΒͰ͸ͳ͍͔આɻ
    
    
    • Ϟσϧ಺ʹ౰ͨΓ͘͡ܦ࿏͕͋Δ͔ΒߴੑೳʹͳΔɺͭ
    ·Γ౰ͨΓ͘͡Ҏ֎͸מΓࠐΜͰ΋݁ՌʹӨڹ͕ͳ͍͸
    ͣͱ͍͏࿦ࢫɻ
    
    
    • ʮઢܗิؒ(linear interpolation)ʹΑΔSGDϊΠζʹର
    ͯ҆͠ఆ͔ʁʯͰמΓࠐΈޮ཰Λௐ΂Δ͜ͱ͕Ͱ͖Δɻ
    ҆ఆ͞͸ෆ҆ఆੑղੳ(࣍ϖʔδ)Ͱௐ΂Δɻ
    

    View Slide

13. Linear instability(ઢܗෆ҆ఆ౓)
    • ͋ΔϞσϧ͕SDGϊΠζ(ϥϯμϜֶश࣌ͷมԽ)ʹରͯ҆͠ఆੑ͔Λௐ΂
    Δํ๏ɻ
    
    
    • ಛఆͷ࣌఺ͷॏΈʹ͍ͭͯෆ҆ఆ౓ΛଌΔʹ͸ɺݸผʹϥϯμϜͳσʔλ
    (αϯϓϧɾaugumentation)ͰεςοϓT·Ͱֶशͨ͠ॏΈW1
    T
    ͱW2
    T
    Λઢ
    ܗʹมԽͤͨ࣌͞ͷଛࣦؔ਺(Τϥʔؔ਺)ͷ࠷େ஋ΛInstabilityͱ͢Δɻ͞
    Βʹ k εςοϓ·Ͱֶशͯ͠ಉ༷ʹෆ҆ఆ౓ΛଌΓɺෆ҆ఆ౓͕૿͑ͳ͚
    Ε͹҆ఆͱ͢Δɻ
    

    View Slide

14. Linear mode Connectivity
    • ̎ͭͷωοτϫʔΫ͕ޡࠩোนͷߴ͞(ϞσϧAɾBͷॏΈ
    WaʙWbΛઢܗิؒ࣌ͨ࣌͠ͷଛࣦؔ਺ͷ࠷େ஋)͕΄΅
    0(2%ҎԼ)ʹͳΔ(θϩόϦΞͷ)ܦ࿏͕͋Δ৔߹ʹϞʔυ઀ଓ
    ͞Ε͍ͯΔͱ͍͏ɻઢܗϞʔυ઀ଓ͞Ε͍ͯΔ෦෼ʢʹๅ͘
    ͡ܦ࿏ʣҎ֎ΛמΓࠐΜͰ΋ਫ਼౓Λҡ࣋Ͱ͖ΔͷͰɺϞσϧ
    αΠζΛѹॖͰ͖Δɻ
    

    View Slide

15. Ҿ༻࿦จͰ͸ɺઢܗิؒͨ࣌͠
    ͷؒͷਫ਼౓͸௿͍
    • ෳࡶͳ՝୊(ImageNet)ɾϞσϧ(ResNet)ͳ΄Ͳinstabilityͷน(όϦΞ)͸ߴ
    ΊʹͳΔ ɻ
    
    
    • όϦΞ͕௿͍ʹๅ͘͡ޮՌ͕ߴ͍(מΓࠐΈѹॖͰ͖Δ)
    

    View Slide

16. ࠓճͷ࿦จख๏Ͱ͸ɺઢܗิؒ࣌ͷਫ਼
    ౓্͕͕͍ͬͯΔ(ଛࣦ͕ݮ͍ͬͯΔ)
    • αΠζѹॖޮՌ͕ߴ͘ͳΔ
    

    View Slide

17. ޻෉: ϞσϧBͷॏΈΛϞσϧAͷॏΈʹҰக
    ͢ΔΑ͏ɺಛఆͷϧʔϧͰฒͼସ͍͑ͯΔ
    • Activation matching
    

    ׆ੑ౓ʹΑΔҰக౓
    
    
    • Weight matching
    

    ॏΈʹΑΔҰக౓
    
    
    • STE(Straight-Through Estimator) matching
    

    ετϨʔτεϧʔਪఆྔʢSTEʣΛ࢖ͬͨҰக
    ౓
    

    View Slide

18. 2. QuestSim: ໛ٖΞόλʔʹΑΔૄͳηϯαʔ͔Βͷਓମӡಈ௥੻
    
    
    (ݪจ: QuestSim: Human Motion Tracking from Sparse Sensors with Simulated Avatars)
    
    
    AR/VRʹ͓͚ΔΠϯλϥΫςΟϒͰ຅ೖײͷ͋Δମݧʹ͸ɺਓମͷಈ͖ΛϦΞϧλΠϜʹτϥο
    Ωϯά͢Δ͜ͱ͕ॏཁͰ͋Δɻ͔͠͠ɺHMDʢHead Mounted Devicesʣ΍ARάϥεͳͲͷ୯
    ମͷ΢ΣΞϥϒϧσόΠε͔ΒಘΒΕΔ਎ମʹؔ͢Δηϯασʔλ͸ඇৗʹݶΒΕͨ΋ͷͰ͋
    ΔɻຊݚڀͰ͸ɺHMDͱ2ͭͷίϯτϩʔϥ͔Βͷૄͳ৴߸ΛऔΓࠐΈɺ΋ͬͱ΋Β͘͠ɺ෺
    ཧతʹଥ౰ͳશ਎ӡಈΛγϛϡϨʔτ͢ΔڧԽֶशͷϑϨʔϜϫʔΫΛఏࣔ͢ΔɻຊݚڀͰ
    ͸ɺHMDͱ2ͭͷίϯτϩʔϥ͔Βͷૄͳ৴߸Λೖྗͱ͠ɺ෺ཧతʹଥ౰ͳશ਎ӡಈΛγϛϡ
    Ϩʔτ͢ΔڧԽֶशϑϨʔϜϫʔΫΛఏҊ͢Δɻͦͷ݁ՌɺHMDͷ6࣍ݩมܗͷΈΛೖྗͱ͠
    ͨ৔߹Ͱ΋ɺԼ൒਎ΛҰ੾؍ଌ͢Δ͜ͱͳ͘ɺڻ͘΄Ͳਅ࣮ͱಉ͡Α͏ͳ٭ͷಈ͖Λ͢Δ͜ͱ
    Λ࣮ূͨ͠ɻ·ͨɺ୯ҰͷϙϦγʔͰɺଟ༷ͳӡಈελΠϧɺҟͳΔମ֨ɺ৽نͷ؀ڥʹର͠
    ͯϩόετͰ͋Δ͜ͱΛࣔ͢ɻ
    w ໨తɿ"373ͷ຅ೖײվળ
    w ੒Ռɿ).%ʴ̎ίϯτϩʔϥ͚ͩͰશ਎ӡಈΛγϛϡϨʔτ͢ΔֶशϑϨʔϜϫʔΫΛఏҊ
    w ํ๏ɿ෺ཧۭؒ .FUB2VFTU
    ΍Ծ૝ۭؒ *TTBDHZN
    ͰͭͷσόΠεͷ࣠৴߸ೖྗΛ༻͍
    ͯڧԽֶश
    w ݻ༗໊ɿ2VFTU4JN
    w ஶऀॴଐɿ.FUB
    https://arxiv.org/abs/2209.09391v1
    

    View Slide

19. HMD(ͱίϯτϩʔϥ)ͷ
    

    3࣍ݩҐஔ(3࣠)+֯౓(3࣠)ͷ࠲ඪͱ଎౓͔Β࢟੎Λਪఆ
    • ࠨ: ̏σόΠε(಄+྆ख) / ӈ:̍σόΠε(಄)
    

    View Slide

20. View Slide

21. View Slide

22. ֶशσʔλɾֶश؀ڥ
    • σʔληοτ:ඃݧऀ172໊ͷ߹ܭ8࣌ؒͷϞʔ
    γϣϯΫϦοϓ(ө૾ͱMetaQuestͷ࠲ඪɾ଎
    ౓৘ใ)
    
    
    • ֶश؀ڥ: Nvidia PhysXͱRLֶशϑϨʔϜϫʔ
    ΫIssac gym
    

    View Slide

23. ಄ͷηϯαʔ͚ͩͰԼ൒਎ͷಈ
    ࡞ΛೝࣝͰ͖͍ͯΔ
    

    View Slide

24. YoutubeσϞϏσΦ
    
    
    https://www.youtube.com/watch?v=CkTHsz6Ldas
    
    
    

    View Slide

25. Limitations
    • ະֶशͷಈ࡞͸׬શʹτϥοΩϯάͰ͖ͳ͍ɻ
    ෳࡶͳಈ࡞͸ਫ਼౓͕མͪͨΓϒϨͨΓ͢Δɻ
    
    
    • ະֶशͷμΠφϛοΫͳಈ࡞ʢϒϨΠΫμϯεɾ
    δϟϯϓͳͲʣ͸స౗ͯ͠͠·͏͜ͱ͕͋Δɻ
    
    
    • ্൒਎ͱԼ൒਎͕૬ؔͷͳ͍ಈ͖Ͱ͸ɺϢʔβ
    ͷϙʔζͱγϛϡϨʔγϣϯ͕Ұக͠ͳ͍Մೳੑ
    ͕͋Δɻ
    

    View Slide

26. 3. ֦ࢄϞσϧͷཧղɻ౷Ұతͳࢹ఺
    
    
    (ݪจ: Understanding Diffusion Models: A Uni
    fi
    ed Perspective)
    
    
    ֦ࢄϞσϧ͸ੜ੒Ϟσϧͱͯ͠ڻ͘΂͖ೳྗΛ͓ࣔͯ͠Γɺ࣮ࡍɺImagen ΍ DALL-E 2 ͳͲͷςΩε
    τ৚݅෇͖ը૾ੜ੒ʹؔ͢Δݱࡏͷ࠷ઌ୺ϞσϧΛࢧ͍͑ͯΔɻ͜ͷݚڀͰ͸ɺม෼๏ͱείΞϕʔ
    εͷ྆ํͷ؍఺͔Β֦ࢄϞσϧͷཧղΛݟ௚͠ɺṖΛղ͖ɺ౷Ұ͢Δɻ·ͣɺϚϧίϑܕ֊૚తม෼
    ΦʔτΤϯίʔμͷಛघͳέʔεͱͯ͠ม෼֦ࢄϞσϧʢVDMʣΛಋग़͠ɺ3ͭͷॏཁͳԾఆʹΑ
    ΓɺELBOͷѻ͍΍͍͢ܭࢉͱεέʔϥϒϧͳ࠷దԽΛՄೳʹ͠·͢ɻVDMΛ࠷దԽ͢Δ͜ͱ͸ɺ3ͭ
    ͷજࡏతͳ໨తͷ1ͭΛ༧ଌ͢ΔͨΊͷχϡʔϥϧωοτϫʔΫͷֶशʹؼண͢Δ͜ͱΛূ໌͠·͢ɻ
    ͦΕ͸ɺ೚ҙͷϊΠζԽ͞Εͨೖྗ͔ΒݩͷιʔεೖྗΛ༧ଌ͢Δ͜ͱɺ೚ҙͷϊΠζԽ͞Εͨೖྗ
    ͔ΒݩͷιʔεϊΠζΛ༧ଌ͢Δ͜ͱɺ೚ҙͷϊΠζϨϕϧʹ͓͍ͯϊΠζԽͨ͠ೖྗͷείΞؔ਺
    Λ༧ଌ͢Δ͜ͱɺͰ͢ɻ࣍ʹɺείΞؔ਺Λֶश͢Δ͜ͱͷҙຯΛਂ͘۷ΓԼ͛ɺTweedieͷࣜΛ௨
    ͯ͠ɺ֦ࢄϞσϧͷม෼తͳ؍఺ͱείΞϕʔεͷੜ੒ϞσϦϯάͷ؍఺Λ໌ࣔతʹ݁ͼ͚ͭΔɻ࠷
    ޙʹɺ֦ࢄϞσϧΛ༻͍ͨ৚݅෇͖෼෍ͷֶशํ๏ʹ͍ͭͯɺΨΠμϯεΛ௨ͯ͡આ໌͢Δɻ
    w ໨తɿ%J
    ff
    VTJPO.PEFMΛ౷Ұղऍ͢Δ
    w ੒Ռɿ&-#0ͷѻ͍΍͍͢ܭࢉͱεέʔϥϒϧͳ࠷దԽΛՄೳʹͨ͠
    w ํ๏ɿ
    w ݻ༗໊ɿ
    w ஶऀॴଐɿ(PPHMF3FTFBSDI #SBJO5FBN
    https://arxiv.org/abs/2208.11970v1
    

    View Slide

27. View Slide

28. ELBO; evidence lower bound
    • ม෼ਪ࿦Ϟσϧ͕؍ଌσʔλΛͲΕ͘Β͍આ
    ໌Ͱ͖͍ͯΔ͔ͷධՁࢦඪ
    

    View Slide

29. VAE
    
    
    ม෼ΦʔτΤϯίʔμ
    

    View Slide

30. HVAE
    
    
    ֊૚ܕม෼ΦʔτΤϯίʔμ
    • ଟ૚֦ுVAEΛҰൠԽͨ͠΋ͷ
    

    View Slide

31. VDM
    
    
    ม෼཭ࢄϞσϧ
    • ʹ੍ݶ͖ͭϚϧίϑܕHVAEʢMHVAEʣ
    

    View Slide

32. 4. τϥϯεϑΥʔϚʔ͸αϯϓϧޮ཰తͳੈքϞσϧ
    
    
    (ݪจ: Transformers are Sample Ef
    fi
    cient World Models)
    
    
    ਂ૚ڧԽֶशΤʔδΣϯτ͸αϯϓϧޮ཰͕ѱ͍͜ͱͰ༗໊Ͱ͋Γɺ࣮໰୊΁ͷద༻͕͔ͳΓ੍ݶ͞
    Ε͍ͯΔɻۙ೥ɺ͜ͷ໰୊Λղܾ͢ΔͨΊʹଟ͘ͷϞσϧϕʔεख๏͕ߟҊ͞ΕɺੈքϞσϧΛ૝૾
    ͠ͳ͕Βֶश͢Δํ๏͕࠷΋ݦஶͳΞϓϩʔνͷҰͭͰ͋Δɻ͔͠͠ɺγϛϡϨʔγϣϯ؀ڥͱͷࣄ
    ্࣮ແ੍ݶͷΠϯλϥΫγϣϯ͸ັྗతʹฉ͑͜Δ͕ɺੈքϞσϧ͸௕࣌ؒʹΘͨͬͯਖ਼֬Ͱͳ͚Ε
    ͹ͳΒͳ͍ɻτϥϯεϑΥʔϚʔ͕γʔέϯεϞσϦϯά՝୊Ͱ੒ޭͨ͜͠ͱʹಈػ͚ͮΒΕɺզʑ
    ͸཭ࢄΦʔτΤϯίʔμͱࣗݾճؼτϥϯεϑΥʔϚʔ͔ΒͳΔੈքϞσϧͰֶश͢Δσʔλޮ཰ͷ
    ྑ͍ΤʔδΣϯτɺIRISΛ঺հ͢ΔɻΞλϦ100kϕϯνϚʔΫͰΘ͔ͣ2࣌ؒͷήʔϜϓϨΠʹ૬౰
    ͢ΔIRIS͸ɺਓؒͷਖ਼نԽฏۉείΞ1.046Λୡ੒͠ɺ26ήʔϜத10ήʔϜͰਓؒΛ྇կ͢ΔੑೳΛ
    ൃش͢Δɻ͜ͷख๏͸ɺϧοΫϔου୳ࡧΛ༻͍ͳ͍ख๏ͱͯ͠৽ͨͳ஍ҐΛཱ֬͠ɺ͞ΒʹMuZero
    Λ྇կ͍ͯ͠·͢ɻαϯϓϧޮ཰ͷྑ͍ڧԽֶशͷͨΊͷTransformersͱੈքϞσϧʹؔ͢Δࠓޙͷ
    ݚڀΛଅਐ͢ΔͨΊɺࢲͨͪͷίʔυϕʔεΛhttps://github.com/eloialonso/iris Ͱެ։͠·͢ɻ
    w ໨తɿਂ૚ڧԽֶशͷֶशޮ཰ͷվળ
    w ੒Ռɿݱ࣮తͳֶश࣌ؒͰֶ΂Δσʔλޮ཰ͷྑ͍ΤʔδΣϯτΛ։ൃ
    w ํ๏ɿ཭ࢄΦʔτΤϯίʔμࣗݾճؼτϥϯεϑΥʔϚʔΛ࢖ͬͯɺ૝૾ੈքͰֶश͢Δ
    w ݻ༗໊ɿ*3*4
    w ஶऀॴଐɿδϡωʔϰେֶ εΠε
    
    https://arxiv.org/abs/2209.00588v1
    

    View Slide

33. View Slide

34. ݱ࣮(྘)͔ΒੈքϞσϧ(G)Λ
    
    
    ࢖ͬͯγϛϡϨʔγϣϯֶश
    

    View Slide

35. ੈքϞσϧ(G)ͷ༧ଌྫ
    
    
    ্:࣮؀ڥ, Լ:྘࿮͔Βਪ࿦ͨ݁͠Ռ
    

    View Slide

36. ੈքϞσϧͷ඼࣭޲্͕؊
    • ཭ࢄΦʔτΤϯίʔμͰɺήʔϜཁૉ(Ϙʔ
    ϧɾϓϨΠϠʔɾఢ)Λਖ਼͘͠࠶ߏங͢Δ
    
    
    • τϥϯεϑΥʔϚʔͰɺใुͷؼଐ΍Τϐ
    ιʔυͷऴྃΛଊ͑Δ
    

    View Slide

37. View Slide

38. 5. જࡏత֦ࢄϞσϧʹΑΔ೴ը૾ੜ੒
    
    
    (ݪจ: Brain Imaging Generation with Latent Diffusion Models)
    
    
    σΟʔϓχϡʔϥϧωοτϫʔΫ͸ɺҩྍը૾ղੳʹ໨֮·͍͠ϒϨʔΫεϧʔΛ΋ͨΒ͍ͯ͠
    ·͢ɻ͔͠͠ɺͦͷσʔλϋϯάϦʔͳੑ࣭͔Βɺҩྍը૾ϓϩδΣΫτʹ͓͚Δ߇͑Ίͳσʔ
    ληοταΠζ͸ɺͦͷજࡏೳྗΛे෼ʹൃش͢Δ๦͛ʹͳ͍ͬͯΔՄೳੑ͕͋Γ·͢ɻ߹੒
    σʔλͷੜ੒͸ɺֶशσʔληοτΛิ׬͠ɺΑΓେن໛ͳҩ༻ը૾ݚڀΛՄೳʹ͢Δ༗๬ͳ୅
    ସखஈΛఏڙ͠·͢ɻ֦ࢄϞσϧ͸࠷ۙɺϑΥτϦΞϦεςΟοΫͳ߹੒ը૾Λੜ੒͢Δ͜ͱͰ
    ίϯϐϡʔλϏδϣϯͷίϛϡχςΟͷ஫໨ΛूΊ͍ͯΔɻຊݚڀͰ͸ɼߴղ૾౓೴ը૾͔Β߹
    ੒ը૾Λੜ੒͢ΔͨΊʹɼજࡏత֦ࢄϞσϧΛར༻͢Δ͜ͱΛݕ౼͢ΔɽUK Biobankσʔλ
    ηοτʢN=31,740ʣͷT1w MRIը૾Λ༻͍ͯɺ೥ྸɺੑผɺ೴ߏ଄ମੵͳͲͷڞม਺Λ৚݅
    ͱͯ͠ɺ೴ը૾ͷ֬཰త෼෍Λֶश͢ΔϞσϧΛߏஙͨ͠ɻͦͷ݁ՌɼզʑͷϞσϧ͸ݱ࣮తͳ
    σʔλΛੜ੒͢Δ͜ͱ͕Ͱ͖ɼ৚݅෇͚ม਺Λ༻͍ͯσʔλੜ੒ΛޮՌతʹ੍ޚͰ͖Δ͜ͱ͕Θ
    ͔ͬͨɽ·ͨɼ10ສຕͷ೴ը૾͔ΒͳΔ߹੒σʔληοτΛ࡞੒͠ɼՊֶքʹެ։ͨ͠ɽ
    w ໨తɿֶश༻ҩྍը૾σʔληοτͷෆ଍Λิ͏
    w ੒Ռɿສຕͷ߹੒೴ը૾σʔληοτΛެ։
    w ํ๏ɿ೴ը૾ͷ֬཰෼෍Λֶश͢ΔϞσϧ -%.-BUFOU%J
    ff
    VTJPO.PEFMT
    Λߏங
    w ݻ༗໊ɿ
    w ஶऀॴଐɿΩϯάεɾΧϨοδɾϩϯυϯΞϝϦΧࠃཱਫ਼ਆӴੜݚڀॴͳͲ
    https://arxiv.org/abs/2209.07162v1
    

    View Slide

39. View Slide

40. ੜ੒ը૾ͷఆྔධՁ
    

    View Slide

41. ڞมྔͰ৚݅෇͚ͯ͠ը૾ੜ੒
    • ೥ྸ
    
    
    • ੑผ
    
    
    • ೴ߏ଄ମ༰ੵͳͲ
    
    
    • কདྷతʹ͸ɺը૾΍์ࣹઢϨϙʔτ΋ର৅ʹ
    ͢Δ༧ఆ
    

    View Slide

42. https://www.healthdatagateway.org/
    
    
    (ͳͲ)Ͱ೴ը૾σʔληοτެ։
    

    View Slide

43. 6. ඍ෼ՄೳͳΞϧΰϦζϜʹΑΔֶश
    
    
    (ݪจ: Learning with Differentiable Algorithms)
    
    
    ݹయతͳΞϧΰϦζϜͱχϡʔϥϧωοτϫʔΫͷΑ͏ͳػցֶशγεςϜ͸ɺͲͪΒ΋೔ৗੜ׆ʹᷓΕͯ
    ͍ΔɻݹయతͳίϯϐϡʔλαΠΤϯεͷΞϧΰϦζϜ͸ɺେ͖ͳάϥϑͷ࠷୹ܦ࿏ΛٻΊΔΑ͏ͳݫີʹ
    ఆٛ͞ΕͨλεΫͷਖ਼֬ͳ࣮ߦʹద͍ͯ͠Δ͕ɺχϡʔϥϧωοτϫʔΫ͸ɺਖ਼֬ͳΞϧΰϦζϜʹؐݩͰ
    ͖ͳ͍ը૾෼ྨͷΑ͏ͳΑΓෳࡶͳλεΫʹ͓͍ͯɺσʔλ͔Βֶशͯ͠࠷΋Մೳੑͷߴ͍౴͑Λ༧ଌ͢Δ
    ͜ͱ͕ՄೳͰ͋Δɻຊ࿦จͰ͸ɺ྆ऀͷ௕ॴΛੜ͔ͨ͢ΊʹɺΑΓؤ݈ͰɺΑΓߴੑೳͰɺΑΓղऍ͠΍͢
    ͘ɺΑΓܭࢉޮ཰ͷߴ͍ɺΑΓσʔλޮ཰ͷߴ͍ΞʔΩςΫνϟʹͭͳ͕Δ྆֓೦ͷ݁߹Λ୳ٻ͍ͯ͠Δɻ
    ຊ࿦จͰ͸ɺχϡʔϥϧωοτ͕ΞϧΰϦζϜ͔Βɺ͋Δ͍͸ΞϧΰϦζϜͱ࿈ܞֶͯ͠श͢Δ͜ͱΛՄೳ
    ʹ͢ΔΞϧΰϦζϜ؂ࢹͱ͍͏ߟ͑ํΛఆࣜԽ͠·͢ɻΞϧΰϦζϜΛχϡʔϥϧɾΞʔΩςΫνϟʹ૊Έ
    ࠐΉ৔߹ɺΞʔΩςΫνϟΛΤϯυπʔΤϯυͰֶशͤ͞ɺޯ഑ΛΞϧΰϦζϜʹ༗ҙٛʹ఻ൖͤ͞Δ͜ͱ
    ͕Ͱ͖ΔΑ͏ɺΞϧΰϦζϜ͕ඍ෼ՄೳͰ͋Δ͜ͱ͕ॏཁͰ͋Δɻຊ࿦จͰ͸ɺΞϧΰϦζϜΛඍ෼Մೳʹ
    ͢ΔͨΊʹɺม਺ʹઁಈΛ༩͑ɺดͨ͡ܗͰɺ͢ͳΘͪαϯϓϦϯάͳ͠Ͱظ଴஋Λۙࣅ͢Δ͜ͱʹΑͬ
    ͯɺΞϧΰϦζϜΛ࿈ଓతʹ؇࿨͢ΔҰൠతͳํ๏ΛఏҊ͢Δɻ͞Βʹɺඍ෼ՄೳͳιʔτωοτϫʔΫɺ
    ඍ෼ՄೳͳϨϯμϥʔɺඍ෼Մೳͳ࿦ཧήʔτωοτϫʔΫͱ͍ͬͨඍ෼ՄೳͳΞϧΰϦζϜΛఏҊ͢Δɻ
    ࠷ޙʹɺຊ࿦จ͸ΞϧΰϦζϜΛ༻ֶ͍ͨशͷͨΊͷ୅ସతͳֶशઓུΛఏࣔ͢Δɻ
    w ໨తɿݹయΞϧΰϦζϜΛχϡʔϥϧωοτϫʔΫʹ׆͔͍ͨ͠
    w ੒Ռɿඍ෼ՄೳͳΞϧΰϦζϜͷ։ൃͱϥΠϒϥϦఏڙ
    w ํ๏ɿΞϧΰϦζϜʹΑΔ؂ࢹ 4VQFSWJTPO
    ϑϨʔϜϫʔΫΛఆٛɾ։ൃ
    w ݻ༗໊ɿ
    w ஶऀॴଐɿίϯελϯπେֶ υΠπ
    
    https://arxiv.org/abs/2209.00616v1
    

    View Slide

44. ݹయΞϧΰϦζϜΛNNͷੈքʹ
    ૊ΈࠐΉ
    • ྫ: खॻ͖จࣈΛจࣈೝࣝͯ͠ιʔτ͢ΔϞσ
    ϧ
    

    View Slide

45. ίʔυྫ
    IUUQTHJUIVCDPN'FMJY1FUFSTFOBMHPWJTJPO
    

    View Slide

46. 7. ػցֶशͷӡ༻ԽɻΠϯλϏϡʔௐࠪ
    
    
    (ݪจ: Operationalizing Machine Learning: An Interview Study)
    
    
    اۀ͸MLΛӡ༻͢ΔͨΊʹػցֶशΤϯδχΞʢMLEʣʹཔ͍ͬͯ·͢ɻͭ·ΓɺMLύΠϓϥΠϯΛ
    ຊ൪؀ڥʹಋೖ͠ɺҡ࣋͢Δ͜ͱͰ͢ɻMLΛӡ༻͢ΔϓϩηεʢMLOpsʣ͸ɺʢiʣσʔλͷऩूͱϥ
    ϕϦϯάɺʢiiʣMLͷύϑΥʔϚϯεΛ޲্ͤ͞ΔͨΊͷ࣮ݧɺʢiiiʣଟஈ֊ͷల։ϓϩηεΛ௨ͯ͠ͷ
    ධՁɺʢivʣӡ༻தͷύϑΥʔϚϯε௿Լͷ؂ࢹɺͱ͍͏ܧଓతͳϧʔϓͰߏ੒͞Ε͍ͯ·͢ɻMLOps
    ΛͲͷΑ͏ʹߦ͏ͷ͔ɺະղܾͷ՝୊͸Կ͔ɺͦͯ͠πʔϧϏϧμʔʹͱͬͯͲͷΑ͏ͳҙຯ͕͋Δͷ
    ͔ʁࢲͨͪ͸ɺνϟοτϘοτɺࣗ཯૸ߦंɺۚ༥ͳͲɺ͞·͟·ͳΞϓϦέʔγϣϯͰ׆༂͢Δ18ਓ
    ͷMLEʹ൒ߏ଄ԽΤεϊάϥϑΟοΫΠϯλϏϡʔΛ࣮ࢪ͠·ͨ͠ɻΠϯλϏϡʔͰ͸ɺML ͷຊ൪ల
    ։ͷ੒ޭΛࢧ഑͢Δ 3 ͭͷม਺͕໌Β͔ʹͳΓ·ͨ͠ɻϕϩγςΟɺόϦσʔγϣϯɺόʔδϣχϯά
    Ͱ͢ɻզʑ͸ɺMLͷ࣮ݧɺσϓϩΠϝϯτɺͦͯ͠ຊ൪ύϑΥʔϚϯεΛҡ࣋͢ΔͨΊͷҰൠతͳϓϥ
    ΫςΟεΛཁ໿͍ͯ͠·͢ɻ࠷ޙʹɺΠϯλϏϡʔʹ౴͑ͯ͘ΕͨਓͨͪͷϖΠϯϙΠϯτ΍Ξϯνύ
    λʔϯʹ͍ͭͯɺπʔϧσβΠϯ΁ͷࣔࠦΛؚΊͯٞ࿦͠·͢ɻ
    w ໨తɿػցֶशͷӡ༻ .-0QT
    ʹ͍ͭͯͷΠϯλϏϡʔௐࠪ
    w ੒Ռɿ.-ಋೖͷͨΊͷϓϥΫςΟεΛཁ໿
    w ํ๏ɿػցֶशΤϯδχΞ໊΁ͷΠϯλϏϡʔ
    w ݻ༗໊ɿ
    w ஶऀॴଐɿΧϦϑΥϧχΞେֶόʔΫϨʔߍ
    https://arxiv.org/abs/2209.09125v1
    

    View Slide

47. MLOps ੒ޭͷൿ݃
    • ߴ଎
    

    ΞΠσΞͷ࣮ݱ଎౓ɺόάʹਝ଎ʹݕূͰ͖Δσόοά؀ڥΛ༻ҙ
    ͢ΔͳͲɻ
    
    
    • (ՄೳͳݶΓૣ͍ஈ֊Ͱͷ)ݕূ
    

    ॲཧίετΛԼ͛ΔͨΊɺͰ͖Δ͚ͩૣ͘มߋΛςετͯ͠ɺΞΠ
    σΞΛݕূͯ͠מΓࠐΉɻόά؂ࢹΛ͢Δɻ൓෮αΠΫϧΛߴ଎Խ
    ͢Δɻ
    
    
    • όʔδϣχϯά(όʔδϣϯຖͷҡ࣋؅ཧ)
    

    ੜ࢈ఀࢭ࣌ؒΛ࠷খݶʹ཈͑ΔͨΊɺෳ਺ͷόʔδϣϯͷຊ൪Ϟσ
    ϧΛҡ࣋ɾ؅ཧɾ੾Γସ͑ΒΕΔ࢓૊ΈΛ༻ҙ͢Δɻ
    

    View Slide

48. 8. ඇ৵ऻత೴ه࿥͔ΒͷԻ੠ղಡ
    
    
    (ݪจ: Decoding speech from non-invasive brain recordings)
    
    
    ೴׆ಈ͔ΒݴޠΛղಡ͢Δ͜ͱ͸ɺϔϧεέΞͱਆܦՊֶͷ྆෼໺ʹ͓͍ͯ଴ͪ๬·Ε͍ͯͨ໨ඪͰ͋Δɻجຊ
    తͳݴޠ՝୊ʹର͢Δ৵ऻతͳ೴൓Ԡ͔Β܇࿅͞Εͨඃݧऀݻ༗ͷύΠϓϥΠϯ͕ɺղऍՄೳͳಛ௃ʢจࣈɺ୯
    ޠɺεϖΫτϩάϥϜͳͲʣΛޮ཰తʹσίʔυ͢ΔΑ͏ʹͳͬͨͷͰ͢ɻ͔͠͠ɺ͜ͷΞϓϩʔνΛࣗવͳԻ
    ੠΍ඇ৵ऻతͳ೴ه࿥΁֦ு͢Δ͜ͱ͸ɺґવͱͯ͠େ͖ͳ՝୊Ͱ͋Δɻຊ࿦จͰ͸ɺࣗવԻ੠ͷࣗݾڭࢣ෇͖
    දݱΛ༧ଌ͢ΔͨΊʹɺେن໛ͳݸਓίϗʔτʹ͓͍ͯରরֶशͰ܇࿅͞Εͨ୯ҰͷΤϯυπʔΤϯυͷΞʔΩ
    ςΫνϟΛఏҊ͢Δɻզʑ͸ɺ169ਓͷϘϥϯςΟΞ͕ࣗવͳԻ੠Λฉ͖ͳ͕Β೴࣓ਤ΍೴೾Λه࿥ͨ͠4ͭͷ
    ެ։σʔληοτͰզʑͷϞσϧΛධՁͨ͠ɻͦͷ݁ՌɺզʑͷϞσϧ͸3ඵؒͷMEG৴߸͔Βɺ1,594ݸͷҟ
    ͳΔηάϝϯτͷ͏ͪ72.5%ͷ্Ґ10Ґ·Ͱͷਫ਼౓ʢ্Ґ1Ґ͸44%ʣͱɺEEGه࿥Ͱ͸2,604ݸͷηάϝϯτ
    ͷ͏ͪ19.1%ͷਫ਼౓ͰରԠ͢ΔԻ੠ηάϝϯτΛࣝผͰ͖Δ͜ͱ͕ࣔ͞Εͨʢֶ͕ͨͬͯ͠शηοτʹଘࡏ͠ͳ
    ͍ϑϨʔζΛ෮߸͢Δ͜ͱ͕ՄೳͱͳͬͨʣɻϞσϧൺֱͱΞϒϨʔγϣϯղੳʹΑΓɺ͜ΕΒͷੑೳ͸ɺզʑ
    ͷઃܭ্ͷબ୒ɺ͢ͳΘͪɺʢiʣରর໨తɺʢiiʣࣄલʹֶशͨ͠Ի੠දݱɺʢiiiʣෳ਺ͷࢀՃऀʹಉ࣌ʹֶश
    ͤͨ͞ڞ௨ͷ৞ΈࠐΈΞʔΩςΫνϟͷ࢖༻͔Β௚઀తʹརӹΛಘ͍ͯΔ͜ͱ͕ࣔ͞Εͨɻ͜ΕΒͷ݁Ռ͸ɺඇ
    ৵ऻతͳ೴׆ಈه࿥͔ΒࣗવݴޠॲཧΛϦΞϧλΠϜͰղಡ͢ΔͨΊͷ༗๬ͳಓےΛ໌Β͔ʹ͢Δ΋ͷͰ͋Δɻ
    w ໨తɿ೴׆ಈ͔Βͷݴޠղಡ
    w ํ๏ɿԻ੠ͱ೴࣓ਤ .&(
    ೴ిਤ &&(
    Λڭࢣσʔλͱͯ͠ɺ೴ͷ׆ಈΛֶश
    w ੒Ռɿඵͷ.&(&&(͔Βߴ͍ਫ਼౓ͰԻ੠ηάϝϯτΛࣝผͰ͖ͨ
    w ݻ༗໊ɿ
    w ஶऀॴଐɿ.FUB
    https://arxiv.org/abs/2208.12266v1
    

    View Slide

49. View Slide

50. View Slide

51. View Slide

52. Thank you for coming Ed.
    
    
    ͱௌ͍ͨ࣌ͷඃݧऀ(3ਓ)ͷ೴൓Ԡ
    

    View Slide

53. 9. େن໛ݴޠϞσϧʹΑΔ஧࣮ͳਪ࿦
    
    
    (ݪจ: Faithful Reasoning Using Large Language Models)
    
    
    ݱ୅ͷେن໛ݴޠϞσϧ(LM)͸ҹ৅తͳ࣭໰Ԡ౴ೳྗΛ͕ࣔ͢ɺͦͷճ౴͸௨ৗɺϞσϧ΁
    ͷ1ճͷݺͼग़͠ͷ݁ՌͰ͋Δɻ͜ͷͨΊɺಛʹຊ࣭తʹଟஈ֊ͷ໰୊ʹ͓͍ͯ͸ɺ޷·͘͠
    ͳ͍ෆಁ໌ੑΛ൐͍ɺੑೳΛ௿Լͤ͞Δɻ͜ΕΒͷ੍ݶʹରॲ͢ΔͨΊɺզʑ͸ɺҼՌߏ଄
    ͕໰୊ͷ࿦ཧߏ଄Λ൓ө͢ΔϓϩηεΛհͯ͠ɺLM͕஧࣮ͳϚϧνεςοϓਪ࿦Λ࣮ߦͰ͖
    ΔΑ͏ʹ͢Δํ๏Λࣔ͢ɻຊख๏͸ɺਪ࿦εςοϓΛ࿈࠯తʹ࣮ߦ͢Δɻ֤εςοϓ͸ɺ2ͭ
    ͷඍௐ੔͞ΕͨLMʢ1ͭ͸બ୒༻ɺ΋͏1ͭ͸ਪ࿦༻ʣͷݺͼग़͔͠Β੒Γɺ༗ޮͳਪ࿦τ
    ϨʔεΛੜ੒͢Δɻຊख๏͸ɺਪ࿦඼࣭Λ޲্ͤ͞ΔͨΊʹɺਪ࿦τϨʔεͷۭؒΛ௨ͯ͠
    ϏʔϜαʔνΛߦ͏ɻզʑ͸ɺଟஈ֊ͷ࿦ཧతਪ࿦ͱՊֶత࣭໰Ԡ౴ʹ͓͚ΔຊϞσϧͷ༗
    ޮੑΛ࣮ূ͠ɺ࠷ऴతͳճ౴ਫ਼౓ʹ͓͍ͯϕʔεϥΠϯΑΓ༏Ε͍ͯΔ͜ͱɺ·ͨɺϢʔβ
    ͕ଥ౰ੑΛνΣοΫͰ͖ΔਓؒతʹղऍՄೳͳਪ࿦τϨʔεΛੜ੒Ͱ͖Δ͜ͱΛࣔ͢ɻ
    w ໨తɿେن໛ݴޠϞσϧͷ࣭໰Ԡ౴ೳྗͷվྑ
    w ੒Ռɿଟஈ֊ͷ࿦ཧతਪ࿦ͱՊֶత࣭໰Ԡ౴ʹ͓͚ΔຊϞσϧͷ༗ޮੑΛ࣮ূ
    w ํ๏ɿਪ࿦εςοϓΛϚϧνεςοϓͰ࣮ߦ͢Δ
    w ݻ༗໊ɿ
    w ஶऀॴଐɿ%FFQ.JOE
    https://arxiv.org/abs/2208.14271v1
    

    View Slide

54. ContextΛݩʹબ୒ͱਪଌΛ܁Γฦ͠
    ͯ࠷ऴతʹQuestionͷ౴͑Λಋ͘ɻ
    

    View Slide

55. View Slide

56. View Slide

57. • SelectionϞσϧ…Context͚͔ͩΒཁૉબ୒
    
    
    • InferenceϞσϧ…Selection݁Ռ͚͔ͩΒਪ࿦
    
    
    • HalterϞσϧ…Infarance݁Ռʴ࣭໰͔Β౴͑
    Λग़͢ɻ·ͩෆ໌ͳΒUnknownɻ
    

    View Slide

58. Selection -> Inference -> Halter
    

    View Slide

59. 10. AudioLM: ݴޠϞσϦϯάʹΑΔԻ੠ੜ੒ͷΞϓϩʔν
    
    
    (ݪจ: AudioLM: a Language Modeling Approach to Audio Generation)
    
    
    ຊ࿦จͰ͸ɺ௕ظతͳҰ؏ੑΛอͬͨߴ඼࣭ͳԻ੠ੜ੒ͷͨΊͷϑϨʔϜϫʔΫͰ͋ΔAudioLMΛ঺հ
    ͢ΔɻAudioLM͸ೖྗԻ੠Λ཭ࢄతͳτʔΫϯͷྻʹରԠ෇͚ɺԻ੠ੜ੒Λ͜ͷදݱۭؒʹ͓͚Δݴޠ
    ϞσϦϯάλεΫͱͯ͠౤͔͚͛Δɻզʑ͸ɺطଘͷԻ੠τʔΫϯԽث͕࠶ߏ੒඼࣭ͱ௕ظߏ଄ͷؒͰ
    ͍͔ʹҟͳΔτϨʔυΦϑΛఏڙ͍ͯ͠Δ͔Λࣔ͠ɺ྆໨తΛୡ੒͢ΔͨΊͷϋΠϒϦουτʔΫφΠ
    θʔγϣϯεΩʔϜΛఏҊ͢Δɻ͢ͳΘͪɺԻ੠Ͱࣄલֶशͨ͠ϚεΫݴޠϞσϧͷ཭ࢄԽ׆ੑΛར༻
    ͯ͠௕ظߏ଄Λଊ͑ɺχϡʔϥϧԻ੠ίʔσοΫʹΑͬͯੜ੒͞ΕΔ཭ࢄԽίʔυΛར༻ͯ͠ߴ඼࣭ͳ
    ߹੒Λ࣮ݱ͢ΔɻAudioLM͸ɺେن໛ͳੜԻ੠ίʔύεͰֶश͢Δ͜ͱͰɺ୹͍ϓϩϯϓτͰࣗવͰҰ
    ؏ੑͷ͋Δ࿈ଓԻ੠Λੜ੒͢Δ͜ͱΛֶश͢Δɻ·ͨɺॻ͖ى͜͠΍஫ऍͷͳ͍Ի੠ʹରֶͯ͠शΛ
    ߦͬͨ৔߹ɺAudioLM͸ߏจతɾҙຯతʹଥ౰ͳԻ੠ͷ࿈ଓΛੜ੒͠ɺ͞Βʹະ஌ͷ࿩ऀʹରͯ͠΋࿩
    ऀͷಛఆͱӆ཯Λҡ࣋͢Δ͜ͱ͕Ͱ͖Δɻ͞Βʹɺզʑ͸ɺԻָͷ৅௃తͳදݱ͕ͳ͍ʹ΋͔͔ΘΒ
    ͣɺटඌҰ؏ͨ͠ϐΞϊԻָͷ࿈ଓΛੜ੒͢Δ͜ͱͰɺզʑͷΞϓϩʔν͕Ի੠Λ௒͑Δ͜ͱΛ࣮ূ͢
    Δɻ
    w ໨తɿߴ඼࣭ͳԻ੠ੜ੒Ϟσϧͷ࡞੒
    w ੒ՌɿԻ੠ੜ੒ϑϨʔϜϫʔΫ"VEJP-.ͷ։ൃ
    w ํ๏ɿҙຯτʔΫϯͱԻڹτʔΫϯΛ૊Έ߹Θͤͨ̏ஈ֊ͷݴޠϞσϧΛ઀ଓ
    w ݻ༗໊ɿ"VEJP-.
    w ஶऀॴଐɿ(PPHMF3FTFBSDI
    https://arxiv.org/abs/2209.03143v1
    

    View Slide

60. ֶशσʔλ͔ΒԻڹτʔΫϯɾ
    ҙຯτʔΫϯΛ࡞Δ
    

    View Slide

61. ࢀߟ: SoundStream
    • ϏοτϨʔτՄมͳχϡʔϥϧΦʔσΟΦ
    ίʔσοΫ
    

    View Slide

62. ԻڹτʔΫϯɾҙຯτʔΫϯΛ
    
    
    εςʔδʹΑͬͯ࢖͍෼͚Δ
    • ҙຯτʔΫϯ: ݴޠత಺༰΍Իָͷટ཯ɾϦζϜ
    
    
    • ߥ͍ԻڹτʔΫϯ: ࿩ऀͷࣝผ΍࿥Ի৚݅ͳͲͷԻڹಛੑ
    
    
    • ਫ਼ີͳԻڹτʔΫϯ: ߴ඼࣭ͷԻ੠߹੒
    

    View Slide

63. σϞαΠτ
    IUUQTHPPHMFSFTFBSDIHJUIVCJPTFBOFUBVEJPMNFYBNQMFT
    

    View Slide

64. σϞͷछྨ
    • εϐʔνͷ๯಄3ඵͷԻ੠͔ΒࣗવͳܧଓεϐʔνΛੜ੒
    
    
    • ࿩ऀ͚ͩม͑ͨಉ͡εϐʔνͷੜ੒
    
    
    • ϥϯμϜͳ࿩ऀɾݴޠɾ؀ڥͰແ৚݅ੜ੒
    
    
    • ҙຯτʔΫϯͳ͠Ͱͷੜ੒(ҙຯΛͳ͞ͳ͍εϐʔν)
    
    
    • SoundStreamͰͷԻ੠࠶ߏ੒࣌ͷྔࢠԽԻ࣭ൺֱ
    
    
    • ϐΞϊટ཯ͷ๯಄4ඵͷԻ੠͔Βࣗવͳટ཯Λੜ੒
    

    View Slide

65. ෆਖ਼ར༻Λݕग़͢ΔࢼΈ
    • ؆୯ʹߴ඼࣭ͳ߹੒Ի੠͕࡞Εͯ͠·͏ͷͰ๷ࢭࡦ
    Λ༻ҙ(Section IV-H)
    
    
    • ಛఆͷԻ੠σʔλ͕AudioLMͰ࡞ΒΕͨ߹੒σʔλ
    ͔Ͳ͏͔Λݕग़͢ΔػߏΛ૊ΈࠐΜͩɻ(ݩσʔλͱ
    ߹੒σʔλͰֶश)
    
    
    • ਓؒͷࣖʹ͸۠ผ͕͔ͭͳͯ͘΋ػցతʹ͸؆୯ʹ
    ൑ผͰ͖Δ͜ͱ͕Θ͔ͬͨ
    

    View Slide

66. DeepL Translator (deepl.com)
    https://www.deepl.com/en/translator
    

    View Slide