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
[読み会] ControlBurn- Feature Selection by Sparse ...
Search
mei28
June 07, 2022
0
26
[読み会] ControlBurn- Feature Selection by Sparse Forests
読み会資料
ControlBurn- Feature Selection by Sparse Forests (KDD2021)
mei28
June 07, 2022
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
50
[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
400
[CHI'24] Fair Machine Guidance to Enhance Fair Decision Making in Biased People
mei28
0
37
[DEIM2024] 卓球の得点予測における重要要素の分析
mei28
0
27
[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
180
[読み会] Words are All You Need? Language as an Approximation for Human Similality Judgements
mei28
0
28
[参加報告] AAAI'23
mei28
0
81
[計算機構論] Learning Models of Individual Behavior in Chess
mei28
0
69
[計算機構論] Why do tree-based models still outperform deep learning on tabular data?
mei28
0
44
Featured
See All Featured
For a Future-Friendly Web
brad_frost
175
9.4k
Templates, Plugins, & Blocks: Oh My! Creating the theme that thinks of everything
marktimemedia
26
2.1k
How to Create Impact in a Changing Tech Landscape [PerfNow 2023]
tammyeverts
47
2.1k
KATA
mclloyd
29
14k
Practical Orchestrator
shlominoach
186
10k
The Psychology of Web Performance [Beyond Tellerrand 2023]
tammyeverts
44
2.2k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
280
13k
Code Reviewing Like a Champion
maltzj
520
39k
Rails Girls Zürich Keynote
gr2m
94
13k
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
PRO
10
720
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
93
16k
Fontdeck: Realign not Redesign
paulrobertlloyd
82
5.2k
Transcript
݄!ಡΈձ ൃදऀɿ༶໌ $POUSPM#VSO 'FBUVSF4FMFDUJPOCZ4QBSTF'PSFTUT
จใͱબཧ༝ จใ બཧ༝ ಛྔબख๏આ໌ੑɼެฏੑʹେ͖ؔ͘ΘΔ͔Β ,BHHMFͷಛྔΛͱΓ͋͑ͣશ෦ͭͬ͜Ή͜ͱʹೲಘͰ͖ͳ͍
5SFF#BTFͷಛྔબख๏ΛఏҊ എܠɿܾఆʴΞϯαϯϒϧख๏ਫ਼ɼղऍੑͱʹ༏ल ɿಛྔಉ࢜ʹ૬͕ؔ͋ΔσʔλͰɼ ಛྔॏཁ͕ͯʹͳΒͳ͍ˠ૬ؔόΠΞεʹऑ͍ ఏҊɿಛྔબʹॏΈ͖-"440Λద༻ ݁Ռɿ૬͕ؔ͋ΔσʔλͰطଘͷಛྔબΑΓ 30$"6$͕ߴ͘ɼ͔ͭಉͷܭࢉίετʹ͑ͨ
-JHIU(#.ͷGFBUVSFJNQPSUBODFྫ ܾఆ૬ؔόΠΞεʹऑ͍ ܾఆϕʔεͷख๏ ਫ਼ͱղऍੑͷόϥϯε͕ྑ͍ Ξϯαϯϒϧϕʔεಛྔ͕ େ͖͍߹ͰΦʔόϑΟοτ͠ʹ͍͘ ϒʔεςΟϯάܥͷख๏Ͱಛྔॏཁ
ͷܭࢉ͕༰қ • -JHIU(#. 9(#PPTU 4DJLJUMFBSO ͷGFBUVSF@JNQPSUBODF
૬ؔόΠΞεʹΑͬͯಛྔॏཁ͕มԽ͢Δ ૬͕ؔ͋Δಛྔಉ࢜Ͱॏཁ͕ബ·ͬͯ͠·͏ 4FY͔Β XJHIU IFJHIU GFNVS "HF͔ΒUJCJBΛਓੜ
Θͳ͍ಛྔΛޙ͔ΒؒҾ͘ ࠷ॳʹܾఆʹΑΔਂ͍ʢΞϯαϯϒϧπϦʔʣΛ࡞Γ ͋ͱͰඞཁͳ͍ʢಛྔʣΛ೩͢🔥🌲🔥 Ԡ༻ઌ • આ໌ՄೳੑɿඞཁͳಛྔΛબ͢Δ͔Βղऍ͍͢͠ʂ • 0QUJNBMFYQFSJNFOUBMEFTJHO
• ಛྔΛऔಘ͢ΔͨΊͷίετΛݮͰ͖Δ • ҩྍͱ͔ͰΘ͟Θ͟σʔλΛͱΔ࣌ͳͲ
४උɿܾఆͷ͍Ζ͍Ζ ܾఆɿδχෆ७ྨޡࠩΛ࠷খԽ͢ΔΑ͏ʹֶश w ਂ͕͞ਂ͘ͳΔ΄Ͳදݱ͕ࡉ͔͘ͳΔʢ ʣ #BHHJOHɿCPPUTUSBQͨ͠σʔλͰ ɹɹɹɹɹֶशثΛಠཱʹֶशɼग़ྗΛ౷߹ͯ͠࠷ऴग़ྗ #PPTUJOHɿऑֶशثΛྻʹܨ͍Ͱֶश
࠷ऴతʹͦΕͧΕͷϞσϧΛ౷߹ͯ͠ग़ྗ 'FBUVSFJNQPSUBODF w δχෆ७ͳͲΛݩʹܭࢉ͢ΔˠNFBO%FDSFBTF*NQVSJUZ d → 2d
#BHHJOHɿόϦΞϯε͕খ͘͞ͳΓ͍͢ • ෳͰଟܾ͢Δ͔ΒόϦΞϯε͕͑ΒΕΔ ϒʔτετϥοϓɿॏෳ͋Γ෮ݩநग़
#PPTUJOHɿόΠΞε͕খ͘͞ͳΓ͍͢ • લͷϞσϧͷग़ྗΛݩʹ͢Δ͔ΒόΠΞε͕͑ΒΕΔ
४උɿಛྔબͷ͍Ζ͍Ζ σʔλͷதʹϞσϧੑೳʹӨڹ͠ͳ͍ಛྔ͕͋Δͣ Ϟσϧ͕ϊΠζʹରͯ͠ؤ݈ͳΒ͍͍͕ɼͦ͏͍͔ͳ͍ ˠඞཁͳಛྔΛબ͍ͨ͠ طଘख๏ύλʔϯ 'JMUFSCBTFɿσʔλ౷ܭ͔ΒಛྔΛධՁ 8SBQQFSɿػցֶशϞσϧΛͬͯධՁ &NCFEEJOHɿϞσϧֶशͱಉ࣌ʹධՁʢFH-BTTPʣ
ఏҊख๏ɿ-BTTPతͳ੍Ͱಛྔબ ɿOݸͷܾఆ ɿྨͷ Ϋϥεͷग़ྗ݁Ռ ɿQݸͷಛྔΛ͏͔Ͳ͏͔ͷϚεΫ ɿܾఆͷॏΈύϥϝʔλ ɿਖ਼ଇԽύϥϝʔλ ti GPSi
∈ {1,2,...,n} αi ∈ ℝm, A ∈ ℝm×n m gi ∈ {0,1}p, G ∈ ℝp×n w ∈ ℝn λ
-BTTPతͳ੍Ͱ࠷దԽͱ͢Δ (SPVQ-BTTPΈ͍ͨͳേଇ߲ ʹΑ͍͕͖ͬͯ·Δ ৭͖ϊʔυ͕༻͢Δಛྔˠ(ʹΑΓܾఆ w
ఏҊख๏ɿΛҭͯͯ ΛؒҾ͘ ΛҭͯΔɿܾఆͷֶशํ๏ΛͭఏҊ w *ODSFNFOUBMEFQUICBHHJOH w *ODSFNFOUBMEFQUICBHCPPTUJOH ΛؒҾ͘ɿલड़ͷ࠷దԽΛղ͘ ˠΛҭͯΔύʔτͱΛؒҾ͘ύʔτͦΕͧΕಠཱ͢Δ
*ODSFNFOUBMEFQUICBHHJOH ϋΠύϥͱͯ͠࠷େਂ͞ Λઃఆ ऩଋͨ͠Βͷਂ͞Λਂ͘͢Δ 5SBJOޡ͕ࠩेখ͍͞ˠCBHHJOH աֶश͠ʹ͍͘ ܭࢉ͕ૣ͍ dmax
*ODSFNFOUBMEFQUICBHCPPTUJOH • ᕓ͔ΒࣗಈௐͰ͖Δˠվળ͞ΕΔݶΓਂ͘ʂ dmax
࣮ݧɿఏҊͷಛྔબख๏ͷଥੑΛݟΔ 4FNJTZOUIFUJDɿ • ਓతʹ૬ؔͷ͋ΔಛྔΛՃ • ૬ؔόΠΞεΛ্खʹѻ͑Δ͔ΛݟΔ #FODINBSLɿ • ͷσʔληοτͰطଘख๏ͱ30$"6$ͷվળΛݟΔ
$BTFTUVEJFTɿ • ϦΞϧσʔληοτͰͲ͏ͳΔ͔ݟΔ
ϕϯνϚʔΫͱܭࢉྔධՁ ࣄલʹશͯͷσʔλΛͬͯϥϯμϜϑΥ ϨετΛֶश ಛྔॏཁΛݩʹ ্Ґ ݸͷಛྔΛબ બ͞Εͨಛྔ͚ͩͰ࠶ֶशɹ
ˠɹϕϯνϚʔΫͱ͢Δ 8SBQQFSCBTFख๏ͷ3FDVSTJWF 'FBUVSF&MJNJOBUJPOͱܭࢉྔൺֱ͢Δ k
݁Ռɿ૬ؔόΠΞεͷӨڹΛड͚ʹ͔ͬͨ͘ 3BOEPN'PSFTUͷಛྔॏཁ ্ҐʹϊΠζΛՃ͑ͯɼ৽ͨ ͳಛྔͱͯ͠Ճ ϊΠζಛྔ͕૿Ճͯ͠ Өڹ͕ड͚ͣʹؤ݈
#FODINBSLɿطଘख๏ΑΓਫ਼͕ߴΊ ಛྔΛ Λ ͬͨ30$"6$ͷ ϕϯνϚʔΫΑΓਫ਼্͕ ݟΒΕͨ k = 1,2,....,10
$BTF4UVEJFTɿ ϦΞϧσʔλͷϊΠζͰར༻Մೳͩͬͨ
ϦΞϧσʔληοτͰͷ૬ؔόΠΞε͕վળ ॎ࣠ଠࣈ͕ఏҊʹΑΔ ্Ґಛྔ ԣ͕࣠طଘख๏ͷ ্Ґಛྔ 'OMXHU͕ͳ্͔ͥҐʹདྷ ͍ͯΔͷ͕ফ͍͑ͯΔ
·ͱΊ • ॏΈ͖-BTTPϕʔεͷ ಛྔબΛߦ͏ΞϧΰϦζϜΛఏҊ • ͞·͟·ͳσʔληοτΛ༻͍ͯɼطଘख๏ͱൺ ૬ؔόΠΞεʹରͯ͠ؤ݈Ͱ͋Δ͜ͱΛࣔͨ͠
ײ • ಛྔબͷϞνϕʔγϣϯΘ͔Γ͘͢ɼख๏ͷΓͨ ͍͜ͱΘ͔͕ͬͨɼ࣮ࡍͷֶशํ๏͕গ͠ᐆດ • ੲ)VNBOCPPTUJOHಡΜ͕ͩɼ )VNBO4UBDLJOH )VNBO//ͱ͔ͬͯ͋Δͷ͔ؾʹͳͬͨ
• ެฏੑʹ͍ؔͯ͠͏ͱɼϨουϥΠϯޮՌʹରͯ͑ͦ͠͏ • ૬ؔͷ͋Δಛྔʹ͓͍ͯ Ͳ͕ͬͪຊ࣭͔ܾఆ͢Δͷͦ͠͏