game dev light

635e53b96114c922fa5486b418895960?s=47 Fadis
January 11, 2019

game dev light

635e53b96114c922fa5486b418895960?s=128

Fadis

January 11, 2019
Tweet

Transcript

  1. 4.

    IUUQTXXXVOSFBMFOHJOFDPNKBCMPHXFMDPNFUPVOSFBMFOHJOF GGXΛඪ४γΣʔμͱ͢ΔUnreal Engine 4͕ϦϦʔε͞ΕΔ ࣍ʹϏσΦήʔϜ࢈ۀʹ޿·Δٕज़͸ಡΊΔ GGXʹؔ͢Δ࠷ॳͷ࿦จ Bruce Walter, Stephen R.

    Marschner, Hongsong Li, and Kenneth E. Torrance. 2007. Microfacet models for refraction through rough surfaces. In Proceedings of the 18th Eurographics conference on Rendering Techniques (EGSR'07), Jan Kautz and Sumanta Pattanaik (Eds.). Eurographics Association, Aire-la-Ville, Switzerland, Switzerland, 195-206. DOI=http:// dx.doi.org/10.2312/EGWR/EGSR07/195-206 2007೥ Stephen McAuley, Stephen Hill, Naty Hoffman, Yoshiharu Gotanda, Brian Smits, Brent Burley, and Adam Martinez. 2012. Practical physically-based shading in film and game production. In ACM SIGGRAPH 2012 Courses (SIGGRAPH '12). ACM, New York, NY, USA, Article 10, 7 pages. DOI: https://doi.org/10.1145/2343483.2343493 σΟεχʔ͕GGXΛ࢖͍ͬͯΔࣄΛެද 2012೥ 2014೥ Unity͕ඪ४γΣʔμΛGGXʹมߋ IUUQTVOJUZEDPNKQVOJUZXIBUTOFXVOJUZ 2015೥
  2. 9.

    Coins $ - 7 gold pieces ---------- ------------ Armor #-........|

    ##...........+####### ---------- a - an uncursed +2 pair of leather gloves (being worn) |.........#####|.....)....| #####|........| b - an uncursed +1 robe (being worn) .........| ####...........-# #|........| Comestibles |........| # ------------# #|<.......| g - 3 uncursed food rations |........| # # #-........- h - 6 uncursed apples ----------## # #|........| i - 8 uncursed oranges #### # #---------- j - 4 uncursed fortune cookies # # ########### l - a food ration # # # # m - 2 jackal corpses ####### # # p - a kobold corpse ## ## ## Scrolls -------# # # d - an uncursed scroll of identify |.f.@.-# # ## n - a scroll labeled ZLORFIK |...>.|# #------------# Spellbooks |.....| #|..........|# c - a blessed spellbook of protection |...... #...........-# Potions |.....| |..........| e - 2 uncursed potions of healing ------- --------|--- f - a blessed potion of healing # Tools Foo the Candidate St:15 Dx:14 Co:12 In:9 Wi:1 k - a magic marker (0:43) Dlvl:1 $:7 HP:14(14) Pw:5(5) AC:4 Xp:1 Stressed Gems/Stones o - 6 rocks (end) େ͖ͳ෺Λখ͘͞࡞Δʹ͸ ࣗಈੜ੒ ೥୅ॳ಄ʹ͸طʹ μϯδϣϯΛࣗಈੜ੒͢ΔήʔϜ͕࡞ΒΕ͍ͯͨ
  3. 23.

    ৚݅෇͖ఢରతੜ੒ωοτϫʔΫΛ༻͍ͨ ຊ෺Β͍͠஍ܗͷੜ੒ Éric Guérin, Julie Digne, Éric Galin, Adrien Peytavie,

    Christian Wolf, Bedrich Benes, and Benoît Martinez. 2017. Interactive example-based terrain authoring with conditional generative adversarial networks. ACM Trans. Graph. 36, 6, Article 228 (November 2017), 13 pages. DOI: https://doi.org/10.1145/3130800.3130804 ෼Ͱඳ͔Εͨεέον͔Βੜ੒͞Εͨ஍ܗ
  4. 26.

    ܗࣜχϡʔϩϯ ೖྗ1 ೖྗ2 ೖྗ3 ೖྗ4 ೖྗ5 w1 w2 w3 w4

    w5 શͯͷೖྗʹ͋ΔॏΈwΛ ͔͚ͨ΋ͷͷ࿨ΛٻΊͯ ͜ΕΛແ਺ʹܨ͛ͨ΋ͷ͸ ॏΈ࣍ୈͰ ೚ҙͷؔ਺ΛۙࣅͰ͖Δ ReLU ग़ྗ ׆ੑԽؔ਺(ReLU) ʹ௨ͨ͠஋Λग़ྗ Formal Neuron
  5. 29.

    w0k w1k w2k w3k w4k yk x0 x1 x2 x3

    x4 yk = R (∑ n wnk xn) YͱX͕෼͔͍ͬͯͯZΛٻΊΔํ๏ ∂E ∂wjk = ∂E ∂xk R′ k (∑ n wnk xn) xj XͷมԽ͕ଛࣦؔ਺ʹ༩͑ΔӨڹXʹΑΔภඍ෼ R′ k͸3F-6ΛLʹ͍ͭͯඍ෼ͨ͠΋ͷ R͸׆ੑԽؔ਺3F-6
  6. 30.

    ∂E ∂wij = ∑ k ∂E ∂xk R′ k (∑

    n wnk xn) wjk R′ j (∑ n wnj xn) xi i j k ࠷ޙ͔Β૚໨ͷภඍ෼͸ ߹੒ؔ਺ͷඍ෼Λ࢖ͬͯٻΊΔ ઌ΄ͲٻΊͨ࠷ޙͷ૚ͷภඍ෼ͷ஋Λ࢖͑Δ ࿈࠯཯ Chain Rule
  7. 34.

    ∂E ∂wij = ∑ k ∂E ∂xk R′ k (∑

    n wnk xn) wjk R′ j (∑ n wnj xn) xi ஈḪΔຖʹ׆ੑԽؔ਺ͷඍ෼ֻ͕͔Δ ͜ͷ஋͕খ͍͞ͱ্ͷ૚·ͰXͷमਖ਼͕ಧ͔ͳ͍ ReLU ReLUͷඍ෼ ඍ෼ͯ͠΋஋͕খ͘͞ͳΒͳ͍ ׆ੑԽؔ਺3F-6ͷൃݟʹΑΓ ௕͍ωοτϫʔΫ͕࣮༻తʹͳͬͨ σΟʔϓϥʔχϯά Deep Learning
  8. 36.

    w13 w23 w33 w43 w53 y3 x1 x2 x3 x4

    x5 w93 ʜ ͷΑ͏ͳԕ͘ͷϐΫηϧͱͷ઀ଓ͸ ແବʹͳΔՄೳੑ͕ߴ͍ w93
  9. 46.

    ຊ෺ͬΆ͍஍ܗ࡞Δ ຊ෺ͷ஍ܗ ੜ੒ͨ͠஍ܗ ஍ܗ͕ຊ෺͔൑ผ͢Δ΋͠൑ผʹࣦഊͨ͠ͳΒ ੜ੒ͨ͠஍ܗ͸ ຊ෺Β͍͠஍ܗͰ͋Δ ఢରతੜ੒ωοτϫʔΫ Generative Adversarial Network

    I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. C. Courville, and Y. Bengio. Generative adversarial nets. In Proceedings of NIPS, pages 2672– 2680, 2014. papers.nips.cc/paper/5423- generativeadversarial-nets.pdf.
  10. 47.

    ఢରతੜ੒ωοτϫʔΫ Generative Adversarial Network ຊ෺ͬΆ͍஍ܗ࡞Δ ஍ܗ͕ຊ෺͔൑ผ͢Δ ຊ෺ͷ஍ܗ ੜ੒ͨ͠஍ܗ ΋͠൑ผʹࣦഊͨ͠ͳΒ ੜ੒ͨ͠஍ܗ͸

    ຊ෺Β͍͠஍ܗͰ͋Δ ্ԼͷωοτϫʔΫͷXΛަޓʹमਖ਼͢Δ ্ͷωοτϫʔΫͷXΛमਖ਼͢Δࡍʹ͸ԼͷωοτϫʔΫ͸ݻఆ ԼͷωοτϫʔΫ͕ ΑΓ൑ผͰ͖ͳ͍஍ܗ͕ੜ੒͞ΕΔΑ͏ʹमਖ਼Λߦ͏
  11. 50.

    ৚݅෇͖ఢରతੜ੒ωοτϫʔΫΛ༻͍ͨ ຊ෺Β͍͠஍ܗͷੜ੒ Éric Guérin, Julie Digne, Éric Galin, Adrien Peytavie,

    Christian Wolf, Bedrich Benes, and Benoît Martinez. 2017. Interactive example-based terrain authoring with conditional generative adversarial networks. ACM Trans. Graph. 36, 6, Article 228 (November 2017), 13 pages. DOI: https://doi.org/10.1145/3130800.3130804 ෼Ͱඳ͔Εͨεέον͔Βੜ੒͞Εͨ஍ܗ
  12. 51.

    ("/Λ࢖ͬͨ%00.ͷεςʔδͷੜ੒ (JBDPNFMMP &EPBSEPFUBMl%00.-FWFM(FOFSBUJPO6TJOH(FOFSBUJWF "EWFSTBSJBM/FUXPSLTz*&&&(BNFT &OUFSUBJONFOU .FEJB$POGFSFODF (&.   

    ("/Ͱ ࣮ࡍͷ%ήʔϜͷεςʔδΛ ຊ෺Β͘͠ ੜ੒ͨ͠ࣄྫ ιʔείʔυ https://github.com/DanieleLoiacono/DoomGAN ࿦จ https://arxiv.org/abs/1804.09154
  13. 55.

    $ZDMF("/ΛԠ༻ͨ͠υοτֆͷੜ੒ Chu Han, Qiang Wen, Shengfeng He, Qianshu Zhu, Yinjie

    Tan, Guoqiang Han, and Tien- Tsin Wong. 2018. Deep unsupervised pixelization. In SIGGRAPH Asia 2018 Technical Papers (SIGGRAPH Asia '18). ACM, New York, NY, USA, Article 243, 11 pages. DOI: https://doi.org/10.1145/3272127.3275082 http://www.shengfenghe.com/deep- unsupervised-pixelization.html ࣸਅ΍ΠϥετΛ ಥͬࠐΜͩΒ υοτֆʹͳͬͯग़ͯ͘Δ ͨͩॖখ͢ΔͷͰ͸ͳ͘ ಛ௃Λ͖ͬ͘Γ࢒ͯ͠ ϑϥοτͳ৭࢖͍ʹͳΔ ͜Ε͸ ਓ͕ඳ͍ͨ৔߹
  14. 57.

    $ZDMF("/ "Λ#ʹม׵ ຊ෺ͷ#ͱࠨͷੜ੒෺Λࣝผ #Λ"ʹม׵ ຊ෺ͷ"ͱӈͷੜ੒෺Λࣝผ A B ຊ෺ͷ#ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ ຊ෺ͷ"ͱੜ੒෺Λ

    ࣝผͰ͖ͳ͍ఔྑ͍ #ˠ"ˠ#Ͱ ݩͱಉ͡ʹͳΔఔྑ͍ "ˠ#ˠ"Ͱ ݩͱಉ͡ʹͳΔఔྑ͍ Jun-Yan Zhu*, Taesung Park*, Phillip Isola, and Alexei A. Efros. "Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks", in IEEE International Conference on Computer Vision (ICCV), 2017. (* indicates equal contributions)
  15. 58.

    $ZDMF("/ "Λ#ʹม׵ ຊ෺ͷ#ͱࠨͷੜ੒෺Λࣝผ #Λ"ʹม׵ ຊ෺ͷ"ͱӈͷੜ੒෺Λࣝผ A B #ˠ"ˠ#Ͱ ݩͱಉ͡ʹͳΔఔྑ͍ ຊ෺ͷ#ͱੜ੒෺Λ

    ࣝผͰ͖ͳ͍ఔྑ͍ ຊ෺ͷ"ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ "ˠ#ˠ"Ͱ ݩͱಉ͡ʹͳΔఔྑ͍ ຊ෺ͷ"ͱຊ෺ͷ#͸ରͰରԠ͍ͯ͠Δඞཁ͸ͳ͍ "ͷྫͱ#ͷྫ͕୔ࢁ͋Ε͹૒ํ޲ͷม׵͕ಘΒΕΔ
  16. 59.

    υοτֆͷੜ੒ ֦େ ຊ෺ͷ"ͱӈͷੜ੒෺Λࣝผ A B ֦େˠॖখˠλονͰ ݩͱಉ͡ʹͳΔఔྑ͍ ॖখˠλονˠ֦େͰ ݩͱಉ͡ʹͳΔఔྑ͍ ࣸਅ΍Πϥετ

    υοτֆ ݩͷ৭ͱڥքΛ อ͍ͯͯΔఔྑ͍ ॖখ υοτֆͬΆ͍ λονʹ͢Δ ຊ෺ͷυοτֆͱੜ੒෺Λࣝผ͢Δ ຊ෺ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ ਅٯͷૢ࡞ʹ ͳ͍ͬͯΔఔྑ͍ http://www.shengfenghe.com/pixelization.html ͜͜ʹੜ੒ྫ͕୔ࢁ͋Δ
  17. 61.

    ("/Λ୔ࢁ૊Έ߹Θͤͨ ౎ࢢͷσΟςΟʔϧͷੜ੒ Tom Kelly, Paul Guerrero, Anthony Steed, Peter Wonka,

    and Niloy J. Mitra. 2018. FrankenGAN: guided detail synthesis for building mass models using style-synchonized GANs. In SIGGRAPH Asia 2018 Technical Papers (SIGGRAPH Asia '18). ACM, New York, NY, USA, Article 216, 14 pages. DOI: https://doi.org/10.1145/3272127.3275065 ೖྗ ग़ྗ http://geometry.cs.ucl.ac.uk/projects/2018/frankengan/
  18. 68.

    ཚ਺͔Β"Λੜ੒ A' A ຊ෺ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ Z ຊ෺ͷ"ͱੜ੒͞Εͨ"Λࣝผ *OGP("/ C ఢରతੜ੒ωοτϫʔΫ

    C' ࣅ͍ͯΔఔྑ͍ $͕"ͷॏཁͳಛ௃ʹؔ܎͍ͯ͠ΔͳΒ $ ; ˠ"ˠ$Ͱݩͷ$ʹ໭ΕΔഺ ΦʔτΤϯίʔμ $͕"ͷಛ௃Λίϯτϩʔϧ͢Δ ύϥϝʔλʹͳΔ "͔Β$Λ෮ݩ X. Chen, Y. Duan, R. Houthooft, J. Schulman, I. Sutskever, and P. Abbeel. Infogan: interpretable representation learning by information maximizing generative adversarial nets. In NIPS, 2016.
  19. 70.

    ;͔Β"Λੜ੒ A' A ຊ෺ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ ຊ෺ͷ"ͱੜ੒͞Εͨ"Λࣝผ 7"&("/ ఢରతੜ੒ωοτϫʔΫ ΦʔτΤϯίʔμ ࣅ͍ͯΔఔྑ͍

    ੜ੒ث͕;Λ࢖͍ͬͯΔͳΒ "ˠ;ˠ"Ͱݩͷ"ʹ໭ΕΔഺ Ϟʔυ่յΛى͜͢ͱ ΦʔτΤϯίʔμଆͷଛࣦ͕૿͑Δ "ͷॏཁͳಛ௃ΛؚΉ;Λ࡞Δ Rosca, Mihaela & Lakshminarayanan, Balaji & Warde-Farley, David & Mohamed, Shakir. (2017). Variational Approaches for Auto- Encoding Generative Adversarial Networks. Z
  20. 71.

    Z A' A ຊ෺ͱੜ੒෺Λ ࣝผͰ͖ͳ͍ఔྑ͍ #JDZDMF("/ C' C Jun-Yan Zhu,

    Richard Zhang, Deepak Pathak, Trevor Darrell, Alexei A. Efros, Oliver Wang: “Toward Multimodal Image-to-Image Translation”, 2017; arXiv:1711.11586 ࣅ͍ͯΔఔྑ͍ ࣅ͍ͯΔఔྑ͍ D("/Ͱ৚݅Λࢦఆ *OGP("/͕ ੜ੒෺ͷόϦΤʔγϣϯΛ࡞Γ 7"&("/͕Ϟʔυ่յΛ๷͙
  21. 73.

    ্͔Βݟͨ ܗঢ় େࡶ೺ͳ ݐ෺ͷܗ ࣮ࡍͷ นͷը૾ ࣮ࡍͷ ԰ࠜͷը૾ ૭౳ͷ഑ஔͷநग़ Ԏಥ౳ͷ഑ஔͷநग़

    ԣ͔Β ݟͨܗঢ় ԰ࠜͷԜತ ԰ࠜͷ໛༷ ૭ͷ഑ஔ นͷ໛༷ นͷԜತ ߴղ૾౓ͳ ԰ࠜͷ໛༷ ߴղ૾౓ͳ ૭ͷ໛༷ ૭࿮ͷܗঢ় ߴղ૾౓ͳ นͷ໛༷ ԰ࠜͷԜತ นͷԜತ ৄࡉͳ ݐ෺ͷ ܗͱ໛༷ FrankenGAN
  22. 80.

    2ֶश Q Learning  ͋Δঢ়ଶ͔Β͋ΔߦಈΛߦͳͬͨ৔߹ͷՁ஋ͷද     

       ঢ়ଶ ߦಈ Ձ஋                   ʜ ঢ়ଶ΍ߦಈͷ਺͕ଟ͍໰୊Ͱ͸ ͜ͷද͸େ͖͘ͳΓ͗ͯ͢هԱͰ͖ͳ͍
  23. 81.

    /FVSBM'JUUFE2*UFSBUJPO ͦͷߦಈʹ͸ ͜Ε͚ͩͷ Ձ஋͕͋Δ શͯͷߦಈͷՁ஋ΛهԱ͓ͯ͘͠୅ΘΓʹ ͋Δঢ়ଶ͔Β֤ߦಈΛߦͳͬͨ৔߹ͷՁ஋Λਪఆ͢Δ χϡʔϥϧωοτϫʔΫͷֶशΛߦ͏ ࠓ͜͏͍͏ঢ়ଶͰ ͜͏͍͏ߦಈΛ ͨ͠৔߹

    Riedmiller, Martin. Neural fitted Q iteration–first experiences with a data efficient neural reinforcement learning method. In: European Conference on Machine Learning. Springer, Berlin, Heidelberg, 2005. p. 317-328.
  24. 82.

    ਪఆ ࣮ࡍ͸ ਪఆ ࣮ࡍ͸ ਪఆ ࣮ࡍ͸ ਪఆ ࣮ࡍ͸ Ձ஋ ʹḷΓண͚ͨ

    Ͱ΋ߦಈͷՁ஋͸ ͷ༧૝ͱҧͬͨ ࠓ"ͷঢ়ଶ ͜͏͍͏ଛࣦؔ਺ͰֶशΛߦ͏ ʹߦ͘ͱ͘Β͍ͷྑ͞ ͷྑ͞͸Ͳ͜ΖͰ͸ͳ͔ͬͨ ࣮ࡍͱͷ ޡࠩ "
  25. 84.

    Q(st , at ) ← Q(st , at ) +

    α (R(st , at ) + γmax (Q(st+1 , at+1 )) − Q(st , at )) XΛগ͠मਖ਼͢Δࡍͷগ͠۩߹ ࠓͷߦಈͰಘΒΕΔใु ଍ݩʹ͕͋Ε͹ͦͷՁ஋ मਖ਼લͷχϡʔϥϧωοτϫʔΫ͕ਪఆͨ͠ ࣍ͷߦಈΛߦͳͬͯભҠͨ͠ઌͷߦಈͷՁ஋ मਖ਼લͷχϡʔϥϧωοτϫʔΫ͕ਪఆͨ͠ ࠓͷߦಈͷՁ஋ ͜Εͱ͜ΕͷࠩΛখ͘͢͞ΔΑ͏ʹXΛमਖ਼͢Δ
  26. 85.

    Q(st , at ) ← Q(st , at ) +

    α (R(st , at ) + γmax (Q(st+1 , at+1 )) − Q(st , at )) ޡࠩΛܭࢉ͢ΔաఔʹχϡʔϥϧωοτϫʔΫͷग़ྗ͕ඞཁ ֶशத͸XΛαϯϓϧຖʹมߋ͍ͯ͠Δҝ͜ͷ஋͕҆ఆ͠ͳ͍ ਪఆͨ͠Ձ஋͕ ٻΊͨՁ஋ʹ͍ۙ΄Ͳྑ͍ ࣌ʑίϐʔ Xͷमਖ਼͕Ձ஋ͷਪఆʹ ௚͙ʹ൓ө͞Εͳ͍Α͏ʹͯ͠ ଛࣦؔ਺Λ҆ఆͤ͞Δ ਂ૚2ωοτϫʔΫ Deep Q Network Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, Martin Riedmiller Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602, 2013.
  27. 86.

    Vlad Firoiu, William F. Whitney, Joshua B. Tenenbaum Beating the

    World's Best at Super Smash Bros. with Deep Reinforcement Learning. arXiv preprint arXiv:1702.06230, 2017. ਂ૚QωοτϫʔΫΛ༻͍ͨ େཚಆεϚογϡϒϥβʔζDXͷAIͷ࡞੒ https://youtu.be/dXJUlqBsZtE ڧԽֶश ϑΝϧίϯ ϓϩήʔϚʔ ϑΝϧίϯ "*͕ϓϩͱઓͬͯউͬͨࣄΛใ͡Δهࣄ https://www.businessinsider.com/super-smash-bros- pro-slox-mafia-artificial-intelligence-2017-2
  28. 89.

    Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou,

    Daan Wierstra, Martin Riedmiller Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602, 2013. Atari 2600ͷը໘Λೖྗͱͯ͠ ద੾ͳߦಈΛબ୒͢ΔڧԽֶश
  29. 90.

    Guillaume Lample, Devendra Singh Chaplot Playing FPS Games with Deep

    Reinforcement Learning. In: AAAI. 2017. p. 2140-2146. ը໘Λೖྗͱͯ͠ DOOMΛ߈ུ͢ΔڧԽֶश https://github.com/fhennecker/deepdoom https://youtu.be/L_9F9bO3-mI ࿦จͷஶऀʹΑΔσϞ ஶऀͱ͸ผͷਓʹΑΔ࣮૷
  30. 91.

    Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou,

    Daan Wierstra, Martin Riedmiller Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602, 2013. ը໘͔Β͙͢ʹ࣍͢΂͖ࣄ͕Θ͔ΔήʔϜͰ͸ ڧԽֶशͷείΞ͸ਓؒͷείΞΛ௒͑Δ ௕ظతͳઓུΛཁ͢ΔήʔϜͰ͸ਓ͕ؒѹউ͢Δ
  31. 93.

    Daniel Holden, Taku Komura, and Jun Saito. 2017. Phase-functioned neural

    networks for character control. ACM Trans. Graph. 36, 4, Article 42 (July 2017), 13 pages. DOI: https:// doi.org/10.1145/3072959.3073663 http://theorangeduck.com/page/phase-functioned-neural-networks-character-control ϞʔγϣϯΩϟϓνϟͱ஍ܗͷϖΞͰ χϡʔϥϧωοτϫʔΫͷֶशΛߦ͍ ༩͑ΒΕͨ஍ܗΛࣗવʹา͘ಈ͖Λ ੜ੒͢Δ ஍ܗ ஍ܗʹԊͬͨ า͖ํ
  32. 94.

    ࣍ʹલʹग़͢΂͖଍͕पظతʹʹมԽ͍ͯ͠Δ + + = = ಉ͡஍ܗͰ΋ λΠϛϯάʹΑͬͯ ద੾ͳಈ͖ํ͸มΘΔ w(0) w(1)

    w(2) w(3) पظతʹ੾ΓସΘΔͭͷXΛ࣋ͭωοτϫʔΫͰਪఆ 1IBTF'VODUJPOFE/FVSBM/FUXPSL
  33. 95.

    = = λΠϛϯάʹΑͬͯ ద੾ͳಈ͖ํ͸มΘΔ w(0) w(1) w(2) w(3) पظతʹ੾ΓସΘΔͭͷXΛ࣋ͭωοτϫʔΫͰਪఆ े෼ͳྔͷֶशσʔλΛूΊΔͨΊʹ

    ࣮ࡍʹ༷ʑͳ஍ܗΛ࣌ؒʹΘͨͬͯา͖ճͬͨ৔߹ͷಈ͖Λ ϞʔγϣϯΩϟϓνϟͰूΊΔ ΋͏ͪΐͬͱָͰ͖ͳ͍͔
  34. 96.

    ڧԽֶशʹΑΔาߦϞʔγϣϯͷੜ੒ Xue Bin Peng, Glen Berseth, Kangkang Yin, and Michiel

    Van De Panne. 2017. DeepLoco: dynamic locomotion skills using hierarchical deep reinforcement learning. ACM Trans. Graph. 36, 4, Article 41 (July 2017), 13 pages. DOI: https://doi.org/ 10.1145/3072959.3073602 https://www.cs.ubc.ca/~van/papers/2017-TOG-deepLoco/
  35. 99.

    ݱࡏ஍ ஍ܗ ໨ඪ ࢟੎ ԿॲΛ౿Ή͔ ࣍ͷҰาͷҝʹͲ͏ಈ͔͘ ࣍ͷҰาΛ ࢦఆ͞ΕͨҐஔͰ౿ΊͨΒ ใु ໨ඪҐஔ·Ͱ

    า͘͜ͱ͕Ͱ͖ͨΒ ใु ҰาΛਖ਼͘͠า͘͜ͱΛֶश͢ΔڧԽֶशͱ ໨ඪ஍఺·ͰͲͷΑ͏ͳܦ࿏Ͱา͔͘Λֶश͢ΔڧԽֶशΛ ૊Έ߹ΘͤΔ͜ͱͰ ໨ඪ஍఺·Ͱา͘ҝͷಈ͖ํΛशಘͤ͞Δ ඵʹճ ඵʹճ
  36. 100.

    ڧԽֶशʹΑΔө૾͔Βͷಈ͖ͷशಘ Xue Bin Peng, Angjoo Kanazawa, Jitendra Malik, Pieter Abbeel,

    and Sergey Levine. 2018. SFV: reinforcement learning of physical skills from videos. In SIGGRAPH Asia 2018 Technical Papers (SIGGRAPH Asia '18). ACM, New York, NY, USA, Article 178, 14 pages. DOI: https://doi.org/10.1145/3272127.3275014
  37. 101.

    Xue Bin Peng, Angjoo Kanazawa, Jitendra Malik, Pieter Abbeel, and

    Sergey Levine. 2018. SFV: reinforcement learning of physical skills from videos. In SIGGRAPH Asia 2018 Technical Papers (SIGGRAPH Asia '18). ACM, New York, NY, USA, Article 178, 14 pages. DOI: https://doi.org/10.1145/3272127.3275014 ө૾͔Β࢟੎Λਪఆ͠ ͓खຊͱͳΔಈ͖Λ࡞Δ ΩϟϥΫλʔ͕෺ཧతʹՄೳͳൣғͰ͓खຊʹ͍ۙಈ͖Λֶश͢Δ ෺ཧγϛϡϨʔγϣϯԼͰ ಈ͘ΩϟϥΫλʔ͕ ֤ϑϨʔϜʹ͓͍ͯ ͓खຊʹ͍ۙϙʔζΛ͍ͯ͠Δఔ ߴ͍ใु
  38. 102.

    ڧԽֶशʹΑΔ෰ΛணΔಈ͖ͷशಘ Alexander Clegg, Wenhao Yu, Jie Tan, C. Karen Liu,

    and Greg Turk. 2018. Learning to dress: synthesizing human dressing motion via deep reinforcement learning. In SIGGRAPH Asia 2018 Technical Papers (SIGGRAPH Asia '18). ACM, New York, NY, USA, Article 179, 10 pages. DOI: https://doi.org/10.1145/3272127.3275048 ෰ʹ͸෍γϛϡϨʔγϣϯΛߦ͍͍ͨ ࣄલʹΩϟϓνϟͨ͠ಈ͖Ͱ͸ ର৅ͷΩϟϥΫλʔ͸ ਖ਼͘͠෰ΛணΕͳ͍͔΋͠Εͳ͍ https://www.cc.gatech.edu/~aclegg3/projects/LearningToDress.html ڧԽֶशͰΩϟϥΫλʔʹ ࣗྗͰ෰ΛணΔํ๏Λֶशͯ͠΋Β͏
  39. 103.

    r(s) = w1 rp (s) + w2 rd (s) +

    w3 rg (s) + w4 rt (s) + w5 rr (s) ෰ΛணΔڧԽֶशͷใु rp (s) ਐḿใुକΛ௨աͨ͠࿹΍಄ͷ௕͞ʹൺྫ͢Δใु rd (s) rg (s) rr (s) ͱΓ͋͑ͣକʹ࿹Λ௨ͦ͏ͱ͢ΔΑ͏ʹ༠ಋ มܗใु෰͕ҾͬுΒΕͯ৳ͼ͍ͯΔ΄Ͳݮগ͢Δใु ෰Λഁ͘Α͏ͳಈ͖Λආ͚ΔΑ͏ʹ༠ಋ
  40. 104.

    r(s) = w1 rp (s) + w2 rd (s) +

    w3 rg (s) + w4 rt (s) + w5 rr (s) ෰ΛணΔڧԽֶशͷใु rp (s) ਐḿใुକΛ௨աͨ͠࿹΍಄ͷ௕͞ʹൺྫ͢Δใु rd (s) มܗใु෰͕ҾͬுΒΕͯ৳ͼ͍ͯΔ΄Ͳݮগ͢Δใु rg (s) ଌ஍ใु෰͕ഽʹ઀৮͍ͯ͠Δׂ߹ʹൺྫ͢Δใु rr (s) ͱΓ͋͑ͣକʹ࿹Λ௨ͦ͏ͱ͢ΔΑ͏ʹ༠ಋ ෰Λഁ͘Α͏ͳಈ͖Λආ͚ΔΑ͏ʹ༠ಋ ෰Λ਎ʹ͚͍ͭͯΔঢ়ଶΛ໨ࢦ͢Α͏ʹ༠ಋ ΰʔϧใु෰Λணͨঢ়ଶʹࣅ͍ͯΔ΄Ͳ૿Ճ͢Δใु କ͔Β಄Λग़ͦ͏ͱͨ͠Γ͠ͳ͍Α͏ʹ༠ಋ ෰ͷछྨʹΑͬͯͷׂ߹Λௐ੔ w
  41. 107.

    ϊΠζͩΒ͚ͷϨϯμϦϯά݁ՌΛΦʔτΤϯίʔμʹ௨ͯ͠ ϊΠζ͕ແ͔ͬͨ৔߹ͷը૾Λਪఆͤ͞Δ Chakravarty R. Alla Chaitanya, Anton S. Kaplanyan, Christoph

    Schied, Marco Salvi, Aaron Lefohn, Derek Nowrouzezahrai, and Timo Aila. 2017. Interactive reconstruction of Monte Carlo image sequences using a recurrent denoising autoencoder. ACM Trans. Graph. 36, 4, Article 98 (July 2017), 12 pages. DOI: https://doi.org/10.1145/3072959.3073601 https://research.nvidia.com/publication/interactive-reconstruction-monte-carlo-image-sequences-using-recurrent-denoising