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
[DEIM2024] 卓球の得点予測における重要要素の分析
Search
mei28
March 01, 2024
0
9
[DEIM2024] 卓球の得点予測における重要要素の分析
DEIM2024の発表資料
卓球の得点予測における重要要素の分析
mei28
March 01, 2024
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
16
[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
0
290
[CHI'24] Fair Machine Guidance to Enhance Fair Decision Making in Biased People
mei28
0
26
[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
150
[読み会] Words are All You Need? Language as an Approximation for Human Similality Judgements
mei28
0
18
[参加報告] AAAI'23
mei28
0
65
[計算機構論] Learning Models of Individual Behavior in Chess
mei28
0
58
[計算機構論] Why do tree-based models still outperform deep learning on tabular data?
mei28
0
32
チーム開発と機械学習
mei28
0
38
Featured
See All Featured
Side Projects
sachag
451
41k
Optimising Largest Contentful Paint
csswizardry
14
2.6k
Art, The Web, and Tiny UX
lynnandtonic
290
19k
What's in a price? How to price your products and services
michaelherold
238
11k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
275
13k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
27
2.4k
GraphQLとの向き合い方2022年版
quramy
35
13k
Designing Dashboards & Data Visualisations in Web Apps
destraynor
226
51k
The Art of Programming - Codeland 2020
erikaheidi
46
12k
Making the Leap to Tech Lead
cromwellryan
126
8.7k
Optimizing for Happiness
mojombo
372
69k
VelocityConf: Rendering Performance Case Studies
addyosmani
321
23k
Transcript
ٿͷಘ༧ଌʹ͓͚Δॏཁཁૉͷੳ *1౦ژେֶɹ*2ΦϜϩϯαΠχοΫΤοΫεגࣜձࣾɹ*3खֶӃେֶ ༶໌*1, 2 ڮຊರ࢙*2 അՈℙ*2 ຊాलਔ*3 ాதҰහ*2 DEIM2024 ୈ16ճσʔλֶͱใϚωδϝϯτʹؔ͢ΔϑΥʔϥϜ
Track 5: ߴͳσʔλར׆༻ɾυϝΠϯԠ༻ʢҩྍใɺڭҭɺཧใʣ[T5-A-9-02]
എܠʛεϙʔπͷσʔλੳ͕Μ w༷ʑͳεϙʔπͰউརͷͨΊʹεϙʔπσʔλੳ͕ߦΘΕ͍ͯΔ 3 αοΧʔ ςχε • ಘঢ়گϓϨʔ༰͔Βಘ༧ଌ[2] B A A
B • બखͷҐஔ Ϙʔϧͷಈ͖͔ΒධՁ [1] <>%FDSPPT 5 "DUJPOTTQFBLMPVEFSUIBOHPBMT7BMVJOHQMBZFSBDUJPOTJOTPDDFS*O1SPDFFEJOHTPGUIFUI"$.4*(,%%JOUFSOBUJPOBMDPOGFSFODFPOLOPXMFEHFEJTDPWFSZEBUBNJOJOH <>.JDIBM4JQLP .BDIJOFMFBSOJOHGPSUIFQSFEJDUJPOPGQSPGFTTJPOBMUFOOJTNBUDIFT *NQFSJBM$PMMFHF-POEPO 7PM
എܠʛϓϨʔͨ͠ੳΛߦ͏ wٿɺҰఆͷಘΛઌʹऔΔͨΊɺ̍ϓϨʔ͝ͱͷಘ֫ಘ͕ॏཁ wςχεɿ wಘ͕૯͕େ͖͍ํ͕༗ར<>ɹ wࢼ߹શମͷઓज़ͷΞυόΠεʹ͔͠ͳΒͳ͍ɻ 4
എܠʛબखͷߦಈҙਤʹ 5 • ࢼ߹ͷಘঢ়گ • αʔϒݖͷ༗ແ • બखͷଧٿҐஔɹͳͲ طଘݚڀ →֎෦͔Β؍ଌՄೳͳಛྔΛར༻
طଘͷಛྔ ʴ બखͷߦಈҙਤ ຊݚڀ → બखͷ෦ঢ়ଶʹ બखͷ৺ཧঢ়ଶઓज़తͳҙਤ͕ϓϨʔʹө͞ΕΔ
ຊݚڀͷత wٿͷಘ֫ಘʹ͓͚Δॏཁཁૉͷੳ wػցֶशϞσϧΛར༻͠ɺσʔλ͔ΒԿ͕ॏཁͰ͔͋ͬͨΛੳ wબखͷߦಈҙਤΛಛྔͱͯ͠ར༻͠ɺಘʹӨڹΛ༩͍͑ͯΔͷ͔Λ֬ೝ 6
ར༻͢Δσʔληοτ wશຊٿબखݖͷࢼ߹ө૾ࢼ߹Λར༻ wө૾͔Βಘঢ়گͳͲͷಛྔΛ࡞ wαʔόʔΛىʹͨ͠ಛྔʹม 7 ಛྔ໊ ༰ 4FU ࢼ߹ͷԿήʔϜ͔ 1MBZ
ಉήʔϜͰԿຊͷϥϦʔ͔ /SBMMZ0SEFSFE ݱࡏͷଧٿ͕αʔϒ͔ΒԿଧ͔ 4DPSF4UBUVT'SPN4FSWFS ݱࡏͷαʔόʔͷಘঢ়گ 4DPSF4UBUVT'SPN3FDJFWFS ݱࡏͷϨγʔόʔͷಘঢ়گ 4DPSF4UBUVT%J⒎#JOBSZ αʔόʔͱϨγʔόͷಘࠩ "UUBDL-BCFMT'SPN4FSWFS ଧٿʹ߈ܸҙਤ͕͋Δ͔Ͳ͏͔
߈कϥϕϧʮϓϨʔͷҙਤ͕ಘͷͨΊʹ߈ܸతͰ͋Δ͔൱͔ʯ wࢼ߹ͷ͏ͪࢼ߹ΛΞϊςʔγϣϯɻͲͷબख͕߈Ί͍ͯΔ͔൱͔ͷ̐ΫϥεΛઃఆ wଧٿ̍ͭ̍ͭʹ༩͢ΔͨΊɺશͯʹΞϊςʔγϣϯΛ༩͢Δͷࠔ w̎ਓͷΞϊςʔγϣϯͷҰகͱߴ͍ wਓؒʹΑΔ߈कϥϕϧߴ͍Ұக͕ୡՄೳ 8 αʔϒΛͨ͠બखͷ ͜ͷଧٿ ಘͷͨΊʹ ߈Ί͍ͯΔ
ٖࣅ߈कϥϕϧͷ࡞ wΓͷࢼ߹ʹ͍ͭͯ-JHIU(#.ͷ༧ଌ݁ՌΛಛྔͱͯ͠ར༻ 9 ࢼ߹ঢ়گಛྔ ߈कϥϕϧ -JHIU(#. ਪఆͨ͠ ߈कϥϕϧ -JHIU(#. -JHIU(#.
-JHIU(#. -JHIU(#. ɾɾɾ ɾɾɾ Ұ෦ͷ߈कϥϕϧΛ༻͍ͯ ٖࣅతʹ߈कϥϕϧΛਪఆ ϥϦʔͷଧຖʹ Ϟσϧͷֶशͱ༧ଌ ֶशͨ͠ϞσϧʹΑΔ ಛྔॏཁ } ਓ͕ؒ࡞ͨ͠߈कϥϕϧʢ5ࢼ߹ʣ } ϞσϧʹͰਪఆٖͨ͠ࣅ߈कϥϕϧ ʢ40ࢼ߹ʣ
࣮ݧʛࢼ߹ঢ়گͱ߈कϥϕϧ͔Βಘ༧ଌ wత wಘ֫ಘͷͨΊʹ֤ϓϨʔͰͷॏཁཁૉͷੳ w߈कϥϕϧ͕ಘ֫ಘ༧ଌʹͯΔӨڹʹ͍ͭͯͷੳ wϞσϧʹ-JHIU(#.Λར༻ wߴ͍ਫ਼͕ୡՄೳ͔ͭɺಛྔॏཁͷࢉग़͕Մೳ wϞσϧͲͷબख͕ಘ͢Δ͔Λ༧ଌ 10 ಛྔ ߈कϥϕϧ
-JHIU(#. -JHIU(#. -JHIU(#. -JHIU(#. ɾɾɾ ɾɾɾ ͷ߈कϥϕϧΛ༻͍ͯ తʹ߈कϥϕϧΛਪఆ ϥϦʔͷଧຖʹ Ϟσϧͷֶशͱ༧ଌ ֶशͨ͠ϞσϧʹΑΔ ಛྔॏཁ
࣮ݧʛ̍ଧ͝ͱͷϓϨʔʹ wαʔϒ͔ΒԿଧͷϓϨʔͰ͋Δ͔ͰσʔληοτΛׂ w֤ଧ͝ͱʹϞσϧΛಠֶཱͯ͠श 11 ࢼ߹ঢ়گಛྔ ߈कϥϕϧ -JHIU(#. ਪఆͨ͠ ߈कϥϕϧ -JHIU(#.
-JHIU(#. -JHIU(#. -JHIU(#. ɾɾɾ ɾɾɾ Ұ෦ͷ߈कϥϕϧΛ༻͍ͯ ٖࣅతʹ߈कϥϕϧΛਪఆ ϥϦʔͷଧຖʹ Ϟσϧͷֶशͱ༧ଌ ֶशͨ͠ϞσϧʹΑΔ ಛྔॏཁ 1ଧͷΈ 6ଧҎ߱ͷۮଧ 5ଧҎ߱ͷحଧ 2ଧ ࢼ߹ใσʔληοτ
࣮ݧ݁Ռʛٖࣅ߈कϥϕϧͷਫ਼ wਓखʹΑΔ߈कϥϕϧΛਖ਼ղͱͯ͠ɺަࠩݕূͰ൚ԽੑೳΛධՁ w༧ଌਫ਼ɺ w߈कϥϕϧαʔόʔɺϨγʔόʔͷ߈कͷछྨͷͨΊ ϥϯμϜΑΓߴ͍༧ଌਫ਼ wҎ߱ͷ࣮ݧͰ༻͍Δ߈कϥϕϧɺ͜ͷϞσϧͷ༧ଌ݁ՌΛ༻͍Δɻ 12
࣮ݧ݁Ռʛ߈कϥϕϧͷ༗ແʹΑΔ༧ଌਫ਼ͷҧ͍ w߈कϥϕϧΛՃ͢Δ͜ͱͰɺ༧ଌਫ਼ͷ্Λ֬ೝ wϞσϧ༧ଌʹ͓͍ͯ߈कϥϕϧ͕ӨڹΛ༩͍͑ͯΔ 13 ༧ଌରͷଧʢ݅ʣ ߈कϥϕϧͳ͠ ߈कϥϕϧ͋Γ ଧٿɿαʔϒʢ݅ʣ
ଧٿɿϨγʔϒʢ݅ʣ ଧٿʢ݅ʣ ଧٿʢ݅ʣ ଧٿҎ߱ͷحଧʢ݅ʣ ଧٿҎ߱ͷۮଧʢ݅ʣ
࣮ݧ݁ՌʛϞσϧͷಛྔॏཁ wϨγʔόʔͷଧٿʢ̎ ଧʣͰॏཁʹͳΔ wˠϨγʔόϓϨʔͷੑ্࣭ෆརͳͨΊɺ߈ܸͷҙਤ͕ಘ֫ಘʹॏཁ 14 Ұଧ ೋଧʢϨγʔϒʣ ࡾଧʢࡾٿ߈ܸʣ ࢛ଧ ଧҎ߱ͷحଧ
ଧҎ߱ͷۮଧ ಛྔॏཁ ಛྔॏཁ ಛྔॏཁ ಛྔॏཁ ಛྔॏཁ ಛྔॏཁ ॏཁ͕ߴ͍ॱͷ߱ॱ
ຊݚڀͷ·ͱΊ wٿʹ͓͚Δಘ༧ଌͷͨΊͷಛྔͱͯ͠ ɹબखͷҙਤΛөͨ͠߈कϥϕϧʹ w߈कϥϕϧΛར༻͢Δ͜ͱͰɺϞσϧͷ༧ଌਫ਼ͷ্ʹ༗༻Ͱ͋ͬͨ wϨγʔόʔ͕߈ܸ͢ΔҙਤΛग़͢͜ͱɺಘ֫ಘͷॏཁཁૉͰ͋Δͱࣔࠦ wࠓޙɺબखͷҙਤ͚ͩͰͳ͘ɺબखͷٿઓུͱΈ߹Θͤͯ ɹಘ༧ଌʹର͢ΔӨڹΛௐࠪ 15
ຊݚڀͷ·ͱΊ wٿʹ͓͚Δಘ༧ଌͷͨΊͷಛྔͱͯ͠ ɹબखͷҙਤΛөͨ͠߈कϥϕϧʹ w߈कϥϕϧΛར༻͢Δ͜ͱͰɺϞσϧͷ༧ଌਫ਼ͷ্ʹ༗༻Ͱ͋ͬͨ wϨγʔόʔ͕߈ܸ͢ΔҙਤΛग़͢͜ͱɺಘ֫ಘͷॏཁཁૉͰ͋Δͱࣔࠦ wࠓޙɺબखͷҙਤ͚ͩͰͳ͘ɺબखͷٿઓུͱΈ߹Θͤͯ ɹಘ༧ଌʹର͢ΔӨڹΛௐࠪ 16
ิʛ߈कϥϕϧͷ࡞ํ๏ʹΑΔҧ͍ wϞσϧ༧ଌʹΑΔ༩ͷํ͕ɺ ɹ࠷ऴతͳಘ༧ଌʹྑ͍ӨڹΛ༩͑Δɻ wϥϯμϜܽଛͷ··Ͱ ɹ߈कϥϕϧແ͠ΑΓਫ਼͕ ɹߴ͘ͳΔ͕͋Δ 17