Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
[計算機構論] Learning Models of Individual Behavior ...
Search
mei28
January 10, 2023
0
69
[計算機構論] Learning Models of Individual Behavior in Chess
計算機構論の資料
Learning Models of Individual Behavior in Chess(KDD2022)
mei28
January 10, 2023
Tweet
Share
More Decks by mei28
See All by mei28
[読み会] “Are You Really Sure?” Understanding the Effects of Human Self-Confidence Calibration in AI-Assisted Decision Making
mei28
0
57
[JSAI'24] 人間の判断根拠は文脈によって異なるのか?〜信頼されるXAIに向けた人間の判断根拠理解〜
mei28
1
410
[CHI'24] Fair Machine Guidance to Enhance Fair Decision Making in Biased People
mei28
0
42
[DEIM2024] 卓球の得点予測における重要要素の分析
mei28
0
30
[Human-AI Decision Making勉強会] 意思決定 with AIは個人vsグループで変わるの?
mei28
0
190
[読み会] Words are All You Need? Language as an Approximation for Human Similality Judgements
mei28
0
29
[参加報告] AAAI'23
mei28
0
83
[計算機構論] Why do tree-based models still outperform deep learning on tabular data?
mei28
0
47
チーム開発と機械学習
mei28
0
50
Featured
See All Featured
It's Worth the Effort
3n
183
28k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
32
2.7k
Principles of Awesome APIs and How to Build Them.
keavy
126
17k
Gamification - CAS2011
davidbonilla
80
5.1k
GitHub's CSS Performance
jonrohan
1030
460k
RailsConf 2023
tenderlove
29
940
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
95
17k
BBQ
matthewcrist
85
9.4k
Building Flexible Design Systems
yeseniaperezcruz
327
38k
The Cult of Friendly URLs
andyhume
78
6.1k
How to Think Like a Performance Engineer
csswizardry
22
1.2k
jQuery: Nuts, Bolts and Bling
dougneiner
61
7.5k
Transcript
ܭࢉػߏ! ༶໌ -FBSOJOH.PEFMTPG*OEJWJEVBM #FIBWJPSJO$IFTT
จใ ,%% • બΜͩཧ༝ɿ • ूஂϨϕϧͰ฿͢Δػցֶशଘࡏ͢Δ͕ɼݸਓϨϕϧ Ͱ฿͢Δํ๏ʹ͍ͭͯڵຯ͕͔͋ͬͨΒ • νΣεʢήʔϜʣΛςʔϚʹͨ͠ͷͰ໘നͦ͏͔ͩͬͨ
Β
ຊݚڀͷ֓ཁ • νΣεΛࡐʹɼݸਓϨϕϧͰ฿͢Δ Ϟσϧͷֶशɼ׆༻ʹ͍ͭͯߟ͑Δ •
ݚڀഎܠ • ػցֶशϞσϧ͕ਓؒͷೳྗΛΔ͔ʹ্ճΔΑ͏ʹ • ਓؒɼʮػց͔ΒʢͱʣֶͿʯΑ͏ͳ׆༻Λ࢝͠Ίͯ ͍Δ • ػցͷৼΔ͍ਓؒͱҟͳΔͨΊɼ
ػց͔ΒֶͿͷ୭ͰͰ͖Δ͜ͱͰͳ͍ • ˠػցλεΫΛ͜ͳͨ͢Ίʹɼ࠷దͳํࡦΛͱΔ͕ɼ ɹͦͷํࡦࣗମ͕ɼਓ͕ؒཧղͰ͖ΔͷͱݶΒͳ͍
Ͳ͏ػցΛ׆༻͢Δͷ͔ʁ • ػցͷߦಈͱਓؒͷཧղͷΪϟοϓΛຒΊΔͨΊʹɼ ػցଆ͕λεΫʹରͯ͠࠷దͳํࡦΛۙࣅ͢ΔͷͰͳ ͘ɼਓؒͷํࡦʹ͚ۙͮΔํ๏Λߟ͑Δ • ຊݚڀͰɼ࠷దͳػցΛਓֶ͕ؒͿελϯεͰͳ ͘ɼػցଆ͕ਓؒΒ͠͞ʹدͤΔํ๏Λߟ͑Δ
νΣεΛࡐʹਓؒʹدͤΔ • ʹ*#.ͷʮ%FFQ#MVFʯ͕࣌ͷνΣεͷ ੈքԦऀʹରͯ͠ɼউརΛऩΊͨ • Ԧऀ͕ཧղͰ͖ͳ͔ͬͨҰख͕ɼ ࣮%FFQ#MVFଆͷόάͩͬͨ͋ΔΈ͍ͨ
• νΣεΦϯϥΠϯͰΜʹ༡Εͯɼ େྔʹσʔλ͕͋Δ
طଘͷνΣεϞσϧͳ͍ͷʁ • .BJBͱݺΕΔ"MQIB;FSPΛͱʹͨ͠ϑϨʔϜϫʔΫ ͕ଘࡏ͍ͯ͠Δ • ΦϯϥΠϯͷϓϨΠσʔλΛͱʹɼʮਓؒΒ͍͠ʯࢦ ͠ํΛߦ͏͜ͱ͕Մೳɽ • Ϩϕϧ͚Λͨ͠ڧ͞ͳΒଧͯΔ͕ɼ
ݸਓϨϕϧͷ฿ʹࢸ͍ͬͯͳ͍
ࣅͨΑ͏ͳݚڀΛͬ͘͟Γ • ࠓճɼసҠֶशͷจ຺ΛऔΓೖΕΔ • ˠଞʹ฿ֶशɼυϝΠϯద༻ɼϝλϥʔχϯάɼϚ ϧνλεΫֶशͳͲࣅͨ৭ʑ͋Δ • ຊདྷͷసҠֶशͰɼग़དྷΔ͚ͩগͳ͍αϯϓϧͰੑ ೳ্Λࢦ͢
• ຊݚڀͰɼେྔͷσʔλΛ༻͍ͯɼݸผʹϑΟοτ͢ Δ͜ͱΛඪʹ͍ͯ͠Δɽ
ݸผϞσϧʹಛԽͨ͠ݚڀ͋Δͷ͔ʁ • ݸผϞσϧΛಛఆ͢ΔͨΊʹɼطଘݚڀͰϓϨΠϠʔͷ ࣝผ͔͠ߦΘΕ͍ͯͳ͍ • ຒΊࠐΈΛհͯ͠ͷΈͰ͔͠ಈ࡞͠ͳ͍ • ϨϕϧΛ฿͢Δํ๏ͱͯ͠ɼطଘͷνΣεΤϯδϯͰ ɼڧ͞Λམͱ͢͜ͱͰ࠶ݱ͍ͯͨ͠
• ˠ͜ΕͰ฿͍ͯ͠Δͱݴ͑ͳ͍ΑͶ • ຊݚڀͰɼ ݸผϞσϧͷֶश ݸผͷಛఆ͕Ͱ͖ΔͰҧ͏
σʔληοτʹ͍ͭͯ • νΣεͷΦʔϓϯιʔεϓϥοτϑΥʔϜͷ -JDIFTTΛ༻͍Δ • େମԯͷରઓσʔλ͕͋Γɼଟ༷ͳϓϨΠϠʔɼࢼ߹ ܗ͕ࣜଘࡏ •
#MJU[ΧςΰϦ ͷήʔϜ Λରʹߦ͏
͏Ϟσϧͷ • -FFMB$IFTT;FSPɿ"MQIB;FSPͷਂڧԽֶशͷ044 • .BJB$IFTTɿ-FFMB$IFTT;FSP ڭࢣΛՃɽ ਓؒͷ࣍ͷҰखΛ༧ଌ͢Δͷͷ4P5" •
ఏҊख๏Ͱ.BJBϞσϧΛϕʔεʹ͢Δɽ • $// 3F-6ͱεΫΠʔζͷؒʹɼ ࣌ܥྻࠩଓΛՃʢϓʔϦϯάͳ͠ʣ
ϋΠύϥɼϞσϧͷॏΈॳظͱ͔Ͳ͏͢Δʁ • طଘͷ.BJBϞσϧ͔ΒసҠֶश͢Δ͚Ͳɼֶशલͷύϥ ϝʔλઃఆʹ͍ͭͯߟ͑Δ • ̎ஈ֊ͰॏΈͷॳظԽɼݸผԽΛߦ͏ • ॳظԽɿෳਓɼେྔͷࢼ߹Ͱେ·͔ͳֶशΛߦ͏ •
ݸผԽɿಛఆͷϓϨΠϠʔͷࢼ߹Ͱ fi OFUVOJOH • ॳظԽͷͱ͖ͲΜͳਓͷࢼ߹Λ͍͍͑ͷ͔ʁ
༧උ࣮ݧɿॳظͲΜͳਓͰֶश͢Δ͖͔ʁ • طଘݚڀͷ.BJBͰɼࣅͨΑ͏ͳϨϕϧͰֶशͨ͠Ϟσϧ ͷ΄͏͕ɼࣅͨϨϕϧͷҰखΛ༧ଌਫ਼͕ߴ͔ͬͨɽ • ͔͜͜Βɼ࣍ͷҰखΛߟ͑Δͱ͖Ϩϕϧʹґଘ͢Δͱ ࢥ͍ͬͯͨɽ • ͔͠͠ɼಛఆͷϓϨΠϠʔͷҰख༧ଌʹ͍ͭͯɼͲͷϨ
ϕϧ͔Βֶशͯ͋͠·ΓมΘΒͳ͔ͬͨ • ˠຊݚڀͰɼϨϕϧ͕ߴ͍ूஂͰֶशͨ͠Ϟσϧ͔Βݸ ผԽΛߦ͍ͬͯ͘ɽ
ॏΈͷݻఆͲ͏͢Δ͔ʁ • సҠֶश fi OFUVOJOHͰɼதؒදݱͷݻఆͯ͠ɼ ࠷ऴ͚ۙͩΛௐ͢Δͷ͕Ұൠత • ຊݚڀͰɼதؒදݱͷΛݻఆͤͣʹɼॳظͱͯ͠༩
͑ͨޙɼҰॹʹॏΈߋ৽͞ΕΔ • ༧උ࣮ݧతʹੑೳ͕མͪͨͨΊ
ॳظԽͷεςοϓͷֶशͷઓུ • εςοϓͷֶशͷதͰͷݕূޡࠩΛ͍࣋ͪͯɼ ֶशͷεέδϡʔϦϯάΛߟ͑Δɽ • ͙Β͍͔Βݕূޡ͕ࠩେ͖͘աֶश࢝͠Ίͨɽ
• ˠ ɾ ɾ ͰֶशΛʹͨ͠ɽ • ࠷ऴతͳલσʔλͰͬͯɼ ༧උ࣮ݧͱಉ͡Α͏ͳΈΒΕͨ
ֶशαϯϓϧબʹ͢Δ • αϯϓϧબҰ༷ʹΔͷͰΑ͘ͳ͍ • νΣεͷং൫ͱத൫ऴ൫Ͱߦͬͯ༧ଌͷқ͕ҧ͏ • ং൫͋Δఔఆੴ͕͋ΔͷͰ༧ଌ͕؆୯ • த൫ऴ൫Λॏతʹֶश͍ͨ͠ͷͰɼϕʔλΛ༻͍
ͯɼ͏͔Ͳ͏͔ͷબΛߦ͏ɽ • ˠ͜Εʹؔͯ͠ɼॳظԽͰݕূਫ਼্͕ͨ͠ɼ ɹݸผԽʹؔͯ͠ɼΈΒΕͳ͔ͬͨ
ఏҊख๏ͷϙΠϯτ·ͱΊ • ॳظԽɼݸผԽͷ̎ஈ֊ͰϞσϧͷֶशΛߦ͏ • ݸผԽ͢ΔࡍʹɼॳظԽͷॏΈΛͱʹશͯߋ৽ʹ͏ • ֶश࣌ͷֶशεςοϓʹԠͯ͡ݮਰ͠աֶश༧
• αϯϓϦϯάؔɼத൫ऴ൫ʹॏΛஔ͘Α͏ʹ
࣮ݧ • ࣮ݧͰߦ͏λεΫ̎ͭ • ϝΠϯλεΫɿಛఆͷϓϨΠϠʔͷ࣍ͷҰखͷ༧ଌ • αϒλεΫɿϓϨΠϠʔͷखΛ༩͑ͨ࣌ʹ୭͔ΛͯΔ • ର߅ख๏ɿ
• ҟͳΔϨϕϧͷ.BJBΛͦͷ··༻͍Δ
ఏҊख๏ͷ΄͏͕࣍ͷҰखͷਫ਼͕ߴ͔ͬͨ • ͷਫ਼্͕ݟΒΕͨɽϨϕϧࠩͷӨڹগͳ͍
ͳΜͰਫ਼͕ߴ͍ͷ͔ʁʢͳͥ.BJB͕ऑ͍ͷ͔ ʣ • த൫ɾऴ൫ʹ͔͚ͯ༧ଌ͕͍͠ .BJB͋Δ͔࣌ΒΨΫοͱམͪΔ
ࢼ߹ͱϞσϧͷؔʁ • ࢼ߹Ҏ্͋Δͱɼ.BJBͷϕʔεϞσϧΑΓੑೳ͕ߴ͍
ݸผϞσϧຊʹݸਓΛଊ͍͑ͯΔͷ͔ʁ • ߴ͍ਫ਼Λୡ͚ͨ͠ͲɼͲͷ͘Β͍·ͰಛΛͱΒ͑ͯ ͍Δ͔ؾʹͳΔʁ • ಛఆͷϓϨΠϠʔͷಈ͖Λ༩͑ͯɼ୭ͷ͔Λ༧ଌ͢Δλε ΫΛߦ͏ • ˠQMZޙͷσʔλΛೖྗ͢ΔͱͯΕͨ
1MZͷDVUP ff ͱήʔϜͷؔΛΈ͍ͨ • $VUP ff ʢং൫ͷಈ͖ΛͲΕ͚ͩলུ͢Δʣ͕େ͖͍ͱ ং൫ͷಈ͖͕ͳ͍͔ΒλεΫͱ͍ͯ͠͠
• ࢼ߹͕ͦͦ͋͜͜Δͱेͳਫ਼͕ग़Δ ϥϯμϜͩͱ͔ͩΒݸਓΛଊ͑ΒΕ͍ͯΔͷ
ϓϨΠϠʔಛ༗ͷϛεͬͯଊ͑ΒΕΔʁ • ϓϨΠϠʔ͕ى͜͢ಛ༗ͷϛεʹ͍ͭͯΈΔɽ • ϛεࠓճɼউΛҎ্Լ͛ͯ͠·͏ߦಈͱఆٛ • ୭ͷϛε͔ʁʹ͍ͭͯಉఆ͢ΔλεΫΛߦͳͬͨͱ͜Ζ ਫ਼͕ۇ͔ʹ্ͨ͠
• ˠϛεݸਓΛಛ͚ͮΔେ͖ͳཁҼʹͳΓಘΔͷͰ ʁ
·ͱΊ • ઌߦݚڀͷ.BJBΛ͍ɼେྔͷσʔλͱసҠֶशΛΈ ߹ΘͤΔ͜ͱͰɼݸਓϨϕϧͷ฿Ϟσϧ͕࡞Ͱ͖ͨɽ • ϓϨΠϠʔͷಈ͖Λ༩͑Δ͚ͩͰɼͩΕͷಈ͖͔Λಉఆ Ͱ͖Δ͜ͱΛ֬ೝͨ͠ •
ݸਓϨϕϧͷϞσϧΛ͏͜ͱͰɼਓؒػցͱ ڠௐ͢Δ͜ͱΛଅਐ͞ΕΔ͜ͱΛظ
ॴײ • ݸਓϨϕϧΛ฿͢ΔϞσϧັྗత͕ͩͬͨɼ ݁ہେྔͷσʔλ͕͋Εͬͱ͍͏ͱ͜Ζ͕͕͔ͬΓ • Ϟσϧࣗମͪΐͬͱมߋ͚ͨ͠ͲɼͲ͏ͯͦ͠͏͔ͨ͠ ॻ͔Ε͍ͯͳ͍͔Β೦ɽ
• ݸਓͷಛΛܾΊΔҰͭͷཁҼʹϛε͕͋Δͷͪΐͬ ͱ໘ന͍ͱࢥͬͨ
DISPNFͷখٕ • 63-ΞυϨεόʔʹ ʮFݕࡧΩʔϫʔυʯΛೖΕΔ ͱɼೳಈతʹݴޠࢦఆݕࡧՄೳ • ӳ୯ޠ͚ͩͲຊޠهࣄΛ
ώοτ͍ͤͨ࣌͞ͱ͔ • ຊޠهࣄ ݟ͍ͨ͘ͳ͍࣌ͱ͔
Ԡ༻͍Ζ͍Ζ • %FFQMͩͱ63-ʹೖΕΔ͚ͩͰ༁ͯ͘͠ΕΔ ʢஈམۭനআ͞ΕΔʣ • ݕࡧΫΤϦΛ63-ͰࢦఆͰ͖Δͷશ෦ͰԠ༻Ͱ͖Δʂͣ