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
Counterfactual learning to rank: introduction
Search
Daiki Tanaka
May 02, 2020
Research
0
670
Counterfactual learning to rank: introduction
一般的なランキング学習からcounterfactual LTRへの導入
Daiki Tanaka
May 02, 2020
Tweet
Share
More Decks by Daiki Tanaka
See All by Daiki Tanaka
カーネル法概観
daikitanak
0
520
カーネル法:正定値カーネルの理論
daikitanak
0
64
[Paper reading] L-SHAPLEY AND C-SHAPLEY: EFFICIENT MODEL INTERPRETATION FOR STRUCTURED DATA
daikitanak
1
180
[Paper Reading] Attention is All You Need
daikitanak
0
110
Interpretability of Machine Learning : Paper reading (LIME)
daikitanak
0
140
[Paper reading] Local Outlier Detection With Interpretation
daikitanak
0
63
Other Decks in Research
See All in Research
さくらインターネット研究所 アップデート2025年
matsumoto_r
PRO
0
660
Computational OT #1 - Monge and Kantorovitch
gpeyre
0
190
Mechanistic Interpretability:解釈可能性研究の新たな潮流
koshiro_aoki
1
310
NLP Colloquium
junokim
1
160
近似動的計画入門
mickey_kubo
4
980
RapidPen: AIエージェントによるペネトレーションテスト 初期侵入全自動化の研究
laysakura
0
1.6k
2025年度人工知能学会全国大会チュートリアル講演「深層基盤モデルの数理」
taiji_suzuki
24
15k
実行環境に中立なWebAssemblyライブマイグレーション機構/techtalk-2025spring
chikuwait
0
230
「エージェントって何?」から「実際の開発現場で役立つ考え方やベストプラクティス」まで
mickey_kubo
0
120
データxデジタルマップで拓く ミラノ発・地域共創最前線
mapconcierge4agu
0
180
MGDSS:慣性式モーションキャプチャを用いたジェスチャによるドローンの操作 / ec75-yamauchi
yumulab
0
250
CSP: Self-Supervised Contrastive Spatial Pre-Training for Geospatial-Visual Representations
satai
3
220
Featured
See All Featured
The Illustrated Children's Guide to Kubernetes
chrisshort
48
50k
Fashionably flexible responsive web design (full day workshop)
malarkey
407
66k
Making the Leap to Tech Lead
cromwellryan
134
9.4k
No one is an island. Learnings from fostering a developers community.
thoeni
21
3.4k
Testing 201, or: Great Expectations
jmmastey
43
7.6k
We Have a Design System, Now What?
morganepeng
53
7.7k
Rebuilding a faster, lazier Slack
samanthasiow
82
9.1k
Become a Pro
speakerdeck
PRO
29
5.4k
4 Signs Your Business is Dying
shpigford
184
22k
Reflections from 52 weeks, 52 projects
jeffersonlam
351
20k
The Web Performance Landscape in 2024 [PerfNow 2024]
tammyeverts
8
690
Understanding Cognitive Biases in Performance Measurement
bluesmoon
29
1.8k
Transcript
Unbiased Learning to Rank May 7, 2020
Learning to rank ઃఆ Supervised LTR Pointwise loss Pairwise loss
Listtwise loss Counterfactual Learning to Rank Counterfactual Evaluation Inverse Propensity Scoring Propensity-weighted Learning to Rank 2
Learning to rank: ઃఆ ೖྗɿ จॻͷू߹ D ग़ྗɿ จॻͷॱҐ R
= (R1; R2; R3:::) ͨͩ͠ɺ֤จॻʹϞσϧ f„ ʹΑͬͯείΞ͕͍͍ͭͯͯ f„ (R1) – f„ (R2) – f„ (R3) ::: ͱͳ͍ͬͯΔɻ(ߴ͍είΞ͕͚ΒΕΔ΄ͲॱҐ͕ߴ͍) Learning to Rank (LTR) ͷత࠷దͳॱҐΛग़ྗ͢ΔϞσϧ f„ ͷύϥϝʔλ „ Λ σʔλ͔ΒٻΊΔ͜ͱɻ 3
Supervised LTR ڭࢣ͋Γ LTR Ͱɺ › ݕࡧΫΤϦ › จॻू߹ ›
ॱҐͷϥϕϧ ΛؚΉσʔληοτΛͬͯϞσϧύϥϝʔλΛٻΊΔɻ ڭࢣ͋Γ LTR Ͱ༻͍ΒΕΔଛࣦओʹ 3 ͭɿ › Pointwise loss › Pairwise loss › Listwise loss y (d) ʹΑͬͯɺจॻ d ͷݕࡧΫΤϦͷؔ࿈Λද͢ͱ͢Δɻ(େ͖͍΄ͲॱҐͷ্Ґʹ ͖ͯཉ͍͠) 4
Pointwise loss Pointwise loss ɺॱҐͷਪఆΛྨɾճؼͱͯ͠ղ͘ɻྫ͑ɺ௨ৗͷճؼଛࣦ (squared loss) ͱͯ͠ҎԼͷΑ͏ʹ༩͑Δɿ Lpointwise :=
1 N N X i=1 (f„ (di) ` y (di))2 Pointwise loss ͷɺϞσϧͷग़ྗΛॱҐͱͯ͠͏͜ͱΛߟྀʹೖΕ͍ͯͳ͍͜ ͱɻLTR Ͱग़ྗͱͯ͠ಘΒΕΔείΞΛฒͼସ͑ͯಘΒΕΔॱҐʹͷΈؔ৺͕͋Δɻ 5
Pairwise loss Pairwise loss Ͱɺ2 ͭͷจॻؒͷ૬ରతͳείΞͷେখΛߟྀʹ͍ΕΔɻྫ͑ɺҎԼ ͷΑ͏ͳ hinge-loss Λ༩͑Δʀ Lpairwise
:= X y(di)>y(dj) max (0; 1 ` (f„ (di) ` f„ (di))): ॱҐ͕૬ରతʹߴ͍จॻείΞ͕ߴ͘ɺॱҐ͕͍จॻείΞΛ͘͢Δؾ࣋ͪɻ Pairwise loss ͷɺશͯͷهࣄϖΞΛಉ༷ʹѻ͏͜ͱɻ࣮ͦͯ͠༻্ top100 ͱ top10 ޙऀͷํ͕ॏࢹ͞ΕΔ͜ͱɻPairwise loss Ͱ top100 ͷԼͷํͷॱҐΛվળ ͤ͞ΔͨΊʹ্ҐͷॱҐΛ٘ਜ਼ʹ͢Δ͜ͱ͕͋Γ͑ͯ͠·͏ɻ 6
Listwise loss Listwise loss ͰॱҐࢦඪΛ࠷దԽ͢Δɻ՝ɺॱҐࢦඪ͕ඍՄೳͰͳ͍͜ͱɻ ྫ͑ɺDCG ɿ DCG = N
X i=1 y (di) log2 (rank (di) + 1) Ͱ͋Δ͕ɺlog2 (rank (di) + 1) ඍෆՄೳͰ͋Δɻ ͦͷͨΊʹ֬తۙࣅΛ༻͍Δํ๏ (ListNetɺListMLE) ɺώϡʔϦεςΟοΫॱҐ ࢦඪͷόϯυΛ࠷దԽ͢Δख๏͕͋Δɻ(LambdaRankɺLambdaLoss) ྫ͑ɺ LambdaRank ͷଛࣦ DCG ͷόϯυͱͳ͍ͬͯΔɿ LLambdaRank := X y(di)>y(dj) log (1 + exp (f„ (dj) ` f„ (di))) j´DCGj 7
ҼՌධՁ తɿ৽͍͠ϥϯΩϯάؔ f„ ΛɺผͷϥϯΩϯάؔ fdeploy ͷԼͰूΊΒΕͨաڈ ͷσʔλ (ΫϦοΫσʔλͳͲ) ΛͬͯධՁ͍ͨ͠ɻ ҎԼͷ
2 ͭͷ߹ʹ͍ͭͯߟ͑Δɻ › શͯͷจॻʹ͍ͭͯਅͷؔ࿈ y (di) ͕طͰ͋Δ࣌ › y (di) Θ͔Βͳ͍͕ɺΫϦοΫใͳͲͷ҉తͳϑΟʔυόοΫͷΈར༻Մೳͳ࣌ 8
ҼՌධՁɿϥϕϧ͕طͳΒશʹධՁ͕Ͱ͖Δ શͯͷจॻʹ͍ͭͯਅͷϥϕϧ y (di) ͕طͰ͋Δ࣌ɺIR(ใݕࡧ) ࢦඪΛܭࢉͰ͖Δɿ ´ (f„; D; y)
= X di2D – (rank (di j f„; D)) ´ y (di) ͜͜Ͱɺ– ॱҐॏΈ͚ؔͰ͋ͬͯɺྫ͑ɿ APR: – (r) = r DCG: – (r) = 1 log2 (1+r) ͳͲ͕༻͍ΒΕΔɻ 9
ҼՌධՁ y (di) Θ͔Βͳ͍͕ɺΫϦοΫใͳͲͷ҉తͳϑΟʔυόοΫͷΈར༻Մೳͳ࣌ɿ › ͋Δจॻʹର͢ΔΫϦοΫɺͦͷจॻ͕ؔ࿈͍ͯ͠Δ͜ͱΛࣔ͢ɺόΠΞεɾϊΠζ ͖ͭͷࢦඪʹͳ͍ͬͯΔɻ › ΫϦοΫ͞Εͳ͔͔ͬͨΒͱ͍ͬͯͦͷจॻ͕ؔͳ͍Θ͚Ͱͳ͍ɻ(จॻ͕ؔͳ ͍ɾϢʔβ͕จॻΛ؍ଌ͍ͯ͠ͳ͍ɾϥϯμϜཁૉʹΑΔͷ)
ଟ͘ͷ؍ଌσʔλʹ͍ͭͯฏۉΛऔΕϊΠζআڈͰ͖Δͱߟ͑ΒΕΔ͕ɺόΠΞεআ ڈͰ͖ͳ͍ɻ 10
ҼՌධՁɿ؍ଌɾΫϦοΫϞσϧ Ϣʔβͷ؍ଌٴͼจॻͷؔ࿈ͷΈΛߟྀʹೖΕΔͱɺϢʔβͷΫϦοΫҎԼͷΑ͏ʹϞ σϦϯάͰ͖ͦ͏ɿ › ϥϯΩϯά R ʹ͓͍ͯจॻ di ͕؍ଌ͞ΕΔ (oi
= 1 Ͱද͢) ֬ɺ P (oi = 1 j R; di) (؍ଌ͞ΕΔ֬ؔ࿈ʹؔͳ͍ͱԾఆ͍ͯ͠Δɻ) › ؔ࿈ y (di) ͱ؍ଌ oi ͕༩͑ΒΕͨ࣌ͷɺจॻ di ͕ΫϦοΫ͞ΕΔ֬ (ci = 1 Ͱද͢) ɺ P (ci = 1 j oi; y (di)) › ΫϦοΫ؍ଌ͞Εͨจॻʹ͔͠ى͜Βͳ͍ͨΊɺϥϯΩϯά R ʹ͓͍ͯΫϦοΫ͞ ΕΔ֬ɿ P (ci = 1 ^ oi = 1 j y (di) ; R) = P (ci = 1 j oi = 1; y (di)) ´ P (oi = 1 j R; di) 11
ҼՌධՁɿ´ (f„; D; y) ͷφΠʔϒਪఆ ´ (f„; D; y) ΛφΠʔϒʹਪఆ͢ΔʹɺΫϦοΫͷใ
(ci) Λਅͷؔ࿈ϥϕϧ (y (di)) ͷΘΓʹ͑Αͯ͘ɺ ´NAIVE (f„; D; c) := X di2D – (rank (di j f„; D)) ´ ci ͱͳΔɻ ΫϦοΫʹϊΠζ͕͍ͬͯͳ͍࣌ɺͭ·Γ P (ci = 1 j oi = 1; y (di)) = y (di) Ͱ͋Δ࣌Ͱ͑͞ɺφΠʔϒਪఆ؍ଌόΠΞεΛड͚͍ͯΔɿ Eo ˆ´NAIVE (f„; D; c)˜ = Eo 2 4 X di2D – (rank (di j f„; D)) ´ ci 3 5 = Eo 2 6 4 X di:oi=1^y(di)=1 – (rank (di j f„; D)) 3 7 5 = X di:y(di)=1 P (oi = 1 j R; di)– (rank (di j f„; D)) = X di2D P (oi = 1 j R; di)– (rank (di j f„; D)) ´ y (di) 12
ҼՌධՁɿ´ (f„; D; y) ͷφΠʔϒਪఆ φΠʔϒਪఆɿ Eo ˆ´NAIVE (f„; D;
c)˜ = X di:y(di)=1 P (oi = 1 j R; di)– (rank (di j f„; D)) ͰɺͦΕͧΕͷจॻͷɺϩάऩू࣌ͷϥϯΩϯά R Ͱͷ؍ଌ֬ͰॏΈͨ͠ਪఆʹͳͬ ͯ͠·͏ɻ ϥϯΩϯάͰɺߴॱҐͷจॻ΄Ͳ؍ଌ͞Ε͍͢ɿ͜ΕΛ position bias ͱݺͿɻϩάऩ ूͷࡍʹߴॱҐʹදࣔ͞Εͨจॻਅͷؔ࿈ΑΓؔ࿈͕͋ΔɺͱόΠΞεΛड͚ͯ͠· ͏ɻ όΠΞεΛআڈ͢ΔͨΊʹɺP (oi = 1 j R; di) Λਪఆ͠ɺิਖ਼ͯ͋͛͠Εྑͦ͞͏ ! είΞʹΑΔόΠΞεআڈ 13
είΞΛ༻͍ͨόΠΞεআڈ Inverse Propensity Scoring(IPS) ʹΑͬͯόΠΞεΛআڈ͢Δɿ ´IPS (f„; D; c) :=
X di2D – (rank (di j f„; D)) P (oi = 1 j R; di) ´ ci ͜͜ͰɺP (oi = 1 j R; di) ϩάऩूதʹදࣔ͞ΕͨϥϯΩϯά R Ͱจॻ di ͕؍ଌ͞ ΕΔ֬Ͱ͋Δɻ´IPS (f„; D; c) ΫϦοΫϊΠζ͕ͳ͍߹ɺͭ·Γ P (ci = 1 j oi = 1; y (di)) = y (di) Ͱ͋Δ࣌ʹ ´ (f„; D; y) ͷෆภਪఆྔͰ͋Δɿ Eo ˆ´IPS (f„; D; c)˜ = Eo 2 4 X di2D – (rank (di j f„; D)) P (oi = 1 j R; di) ´ ci 3 5 = Eo 2 6 4 X di:oi=1^y(di)=1 – (rank (di j f„; D)) P (oi = 1 j R; di) 3 7 5 = X di:y(di)=1 P (oi = 1 j R; di) ´ – (rank (di j f„; D)) P (oi = 1 j R; di) = X di2D – (rank (di j f„; D)) ´ y (di) = ´ (f„; D; y) : 14
Propensity-weighted LTR IPS ´ (f„; D; y) ͷෆภਪఆͰ͋ͬͨɻΑͬͯɺ࠷దͳϞσϧύϥϝʔλ „
IPS Λ ࠷దԽ͢Δ͜ͱͰٻΊΔ͜ͱ͕Ͱ͖ΔɻIPS Λ࠷దԽ͢ΔࡍɺϥϯΩϯάࢦඪ – (r) ͷඍ ෆՄೳੑʹରॲ͢ΔͨΊɺ– (r) ͷ bound Λར༻͢Δɻ Propensity-weighted LTR ͷྲྀΕɿ › ΫϦοΫͷείΞΛਪఆɿ P (oi = 1 j R; di) › ෆภਪఆྔ ´IPS (f„; D; c) ͷ bound ʹ͍ͭͯඍΛܭࢉɿ „0 = r„ "– (rank (di j f„; D)) P (oi = 1 j R; di) # › ϞσϧύϥϝʔλΛߋ৽ „new „old ` „0 15
References › https://ilps.github.io/webconf2020-tutorial-unbiased-ltr/ 16