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
iclr2020deepsemi-supervisedanomalydetectionyama...
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Yamato.OKAMOTO
June 14, 2020
Technology
160
0
Share
iclr2020deepsemi-supervisedanomalydetectionyamatookamoto-200531022507.pdf
Yamato.OKAMOTO
June 14, 2020
More Decks by Yamato.OKAMOTO
See All by Yamato.OKAMOTO
いまAI組織が求める企画開発エンジニアとは?
roadroller
2
1.6k
Slide ICCV2023 Constructing Image Text Pair Dataset from Books
roadroller
0
140
第11回 全日本コンピュータビジョン勉強会 CVPR2022 "A Self-Supervised Descriptor for Image Copy Detection"
roadroller
0
670
第9回 全日本コンピュータビジョン勉強会 発表資料
roadroller
0
670
第七回全日本コンピュータビジョン勉強会 A Multiplexed Network for End-to-End, Multilingual OCR
roadroller
1
990
部下のマネジメントはAI開発に学べ
roadroller
0
190
Domain Generalization via Model-Agnostic Learning of Semantic Features NeurIPS’19 読み会 in 京都
roadroller
0
310
ICML’2019 読み会in京都 Federated Learningの研究動向
roadroller
0
140
CVPR2019@Long Beach 参加速報(本会議)
roadroller
0
160
Other Decks in Technology
See All in Technology
AIが盛んな時代に 技術記事を書き始めて起きた私の中での小さな変化
peintangos
0
260
20260428_Product Management Summit_Loglass_JoeHirose
loglassjoe
3
4k
COBOL婆さんの伝説
poropinai1966
0
110
ネットワーク運用を楽にするAWS DevOps Agent活用法!! / 20260421 Masaki Okuda
shift_evolve
PRO
2
240
Do Vibe Coding ao LLM em Produção para Busca Agêntica - TDC 2026 - Summit IA - São Paulo
jpbonson
3
160
AIはハッカーを減らすのか、増やすのか?──現役ホワイトハッカーから見るAI時代のリアル【MEGU-Meet】
cscengineer
PRO
0
220
Revisiting [CLS] and Patch Token Interaction in Vision Transformers
yu4u
0
400
AIが自律的に働く時代へ Amazon Quick で実現するAIエージェント紹介
koheiyoshikawa
0
140
「誰一人取り残されない」 AIエージェント時代のプロダクト設計思想 Product Management Summit 2026
mizushimac
1
1.8k
【技術書典20】OpenFOAM(自宅で深める流体解析)流れと熱移動(2)
kamakiri1225
0
280
目的ファーストのハーネス設計 ~ハーネスの変更容易性を高めるための優先順位~
gotalab555
8
2.5k
Agents CLI と Gemini Enterprise Agent Platform で マルチエージェント開発が楽しくなる!
kaz1437
0
170
Featured
See All Featured
Utilizing Notion as your number one productivity tool
mfonobong
4
290
The MySQL Ecosystem @ GitHub 2015
samlambert
251
13k
Effective software design: The role of men in debugging patriarchy in IT @ Voxxed Days AMS
baasie
0
300
Data-driven link building: lessons from a $708K investment (BrightonSEO talk)
szymonslowik
1
1k
Visualizing Your Data: Incorporating Mongo into Loggly Infrastructure
mongodb
49
9.9k
For a Future-Friendly Web
brad_frost
183
10k
[Rails World 2023 - Day 1 Closing Keynote] - The Magic of Rails
eileencodes
38
2.8k
The Illustrated Guide to Node.js - THAT Conference 2024
reverentgeek
1
340
Tell your own story through comics
letsgokoyo
1
900
How to Ace a Technical Interview
jacobian
281
24k
Optimising Largest Contentful Paint
csswizardry
37
3.7k
How to build an LLM SEO readiness audit: a practical framework
nmsamuel
1
720
Transcript
2020/6/14 Yamato OKAMOTO ICLRΦϯϥΠϯಡΈձ Deep Semi-supervised Anomaly Detection
ࣗݾհʢ͘!!ʣ ɹԬຊେʢ͓͔ͱ·ͱʣ • ژେֶඒೱݚڀࣨͰύλʔϯೝࣝΛݚڀͯ͠म࢜՝ఔमྃ • ΦϜϩϯͰ৽نࣄۀΛܦݧޙɺ͍·ࣾձγεςϜࣄۀ෦ͷݚڀॴϦʔμʔ • ເژΛϙετɾγϦίϯόϨʔʹ͢Δ͜ͱɺؔͷίϛϡχςΟΛڧԽ͍ͨ͠ ɹ㱺 ژͷมਓύϫʔΛੈքʹΒ͠Ί͍ͨ
Twitter : RoadRoller_DESU ҆৺҆શͳࣾձͷ࣮ݱʹ͚ͯɺ ࠷ۙ Anomaly Detection ʹڵຯΞϦ
Anomaly Detection ͋Δ͋Δ ఆٛࠔ • ҟৗʹ༷ʑͳόϦΤʔγϣϯ͕͋Δ • ҟৗݕग़͍͚ͨ͠ͲʮWhat is ҟৗʁʯ͕ఆٛͰ͖ͳ͍
ֶशσʔλ͕ೖखࠔ • ҟৗ໓ଟʹൃੜ͠ͳ͍ʢ※ සൟʹൃੜ͢ΔΠϕϯτҟৗ͡Όͳ༷ͯ͘ʣ • ѹతʹҟৗσʔλ͕ෆͯ͠ػցֶश͕ࠔ ैདྷख๏ɿਖ਼ৗΛఆٛ͢Δ • ʮWhat is ҟৗʁʯͷఆٛΛఘΊΔɺҟৗσʔλͷֶशఘΊΔ • ͦͷΘΓʮWhat is ਖ਼ৗʁʯͷఆٛΛֶशͯ͠ɺʮNot ਖ਼ৗʯΛҟৗͱఆ͢Δ
Anomaly Detection ͷैདྷݚڀ Deep One-Class Classification (ICML’18) • ਖ਼ৗσʔλͷΈΛ༻͍ͯɺClassifierͳΓAutoEncoderͳΓΛैདྷ௨Γʹֶश •
͜ͷͱ͖ɺಛྔ͕࣍ݩ෦ۭؒʹऩଋ͢ΔΑ͏LOSSΛՃ͑Δ • ਖ਼ৗσʔλͳΒٿʹ͢ΔͣͳͷͰɺٿ͔Β֎ΕͨσʔλΛҟৗͱఆ͢Δ ୈҰ߲ʹΑͬͯٿʹ͕ԡ͠ࠐ·ΕΔ cɿ ٿͷத৺ʢͨͩ͠≠0ʣ nɿֶश͢Δਖ਼ৗσʔλͷ
Anomaly Detection ͷධՁ؍ ͲΕ͚ͩਖ਼֬ʹҟৗΛݕͰ͖͔ͨʁ • ਖ਼ৗσʔλΛਖ਼ৗͱఆͯ͠ɺҟৗσʔλΛҟৗͱఆ͢Δਫ਼ ԼྲྀλεΫΛअຐ͠ͳ͍͔ʁ • ԼྲྀλεΫ͕͋Δ߹ɺҟৗݕػೳͷՃʹΑͬͯѱӨڹ͕ͳ͍͔Ͳ͏͔ •
ྫ͑ɺ10ΫϥεͷࣈࣝผثʹɺਤܗͳͲࣈҎ֎͕ೖྗ͞Εͨͱ͖ҟৗͱఆ͢Δػ ೳΛ͚Ճ͍͑ͨͤͰɺैདྷͷ10Ϋϥεࣝผੑೳ͕Լ͢ΔͱࠔΔ ad-hoc͔post-hoc͔ʁ • ҟৗݕ͢ΔͨΊʹϞσϧߏֶशํ๏·Ͱม͑Δඞཁ͕͋Δ͔ʁ • ·ͨɺLOSSΛޙ͔Β͚͚̍ͭͩͯ͠Ճֶश͢Δ͚ͩͰOK͔ʁ • ͲͪΒ͕ྑ͍ѱ͍ͳͲҰ֓ʹݴ͑ͳ͍͕ɺpost-hocͷํ͕ѻ͍͍͢ɻ
հจͷ֓ཁ ʮSemi-supervisedʹֶश͠Α͏ʂʯ Anomaly Detection ͷݚڀUnsupervised͕ओྲྀͷΑ͏ͩ Ͱɺֶश༻ͷҟৗσʔλ͕ೖखࠔͩͱͯ͠ɺ ӡ༻Λଓ͚ͯͨΒҟৗσʔλʹ͍ͣΕग़ձ͏ͣ ͳΒɺͦΕΒগྔͷҟৗσʔλΛͬͯɺ Semi-supervisedʹֶशͨ͠ํ͕ྑ͍ͷͰʁ ※Semi-supervisedͷAnomaly
Detectionݚڀඇৗʹগͳ͍
ఏҊख๏ ʮLOSSʹ߲Λ̍ͭՃ͠·ͨ͠ʯ Deep One-Class Classification (ICML’18) ͷLOSSʹSemi-supervisedͷ߲Λ̍ͭՃ • ࣮ಉ͡ஶऀͰͨ͠ɻࣗͷݚڀΛࣗͰΞοϓσʔτͨ͠ܗʹͳΔɻ ͠ҟৗσʔλʹग़ձͬͨΒɺ
ٿͷ֎ଆʹߦ͘Α͏ֶश͢Δ mɿsemi-supervisedʹֶश͢Δσʔλ yj ɿਖ਼ৗorҟৗͷϥϕϧ
࣮ݧ݁Ռ ॎ࣠ɿҟৗσʔλͷݕग़ੑೳ ʢHigher is Betterʣ Unsupervised Semi-supervised ԣ࣠ɿSemi-supervisedͰڭࢣ͖ͷҟৗσʔλΛֶशׂͨ͠߹ ఏҊख๏ MNISTɺFashion-MNISTɺCIFAR-10ͷσʔληοτͰධՁ
• ̍Ϋϥεͱਖ਼ৗͱఆٛͯ͠ɺAutoEncoderʴఏҊख๏ͰಛྔදݱΛֶश • Γͷ̕ΫϥεΛೖྗͨ͠ͱ͖ɺҟৗͱఆͰ͖Δ͔Ͳ͏͔ධՁ ੑೳվળΛ֬ೝ
·ͱΊͱߟ ਂֶशʹΑΔ Semi-supervised ͳ Anomaly Detection ख๏ΛఏҊ • ॳΊͯͰͳ͍ͱࢥ͏͕ɺਂֶशʹΑΔAnomaly DetectionͰsemi-supervised͍͠
• ͔ͨ͠ʹࣾձ࣮Λߟ͑Δͱɺ͜ͷઃఆద • ख๏γϯϓϧͰɺpost-hocͳͷͰѻ͍͍͢ • ࠓճԼྲྀλεΫ͕AE͕ͩͬͨɺClassificationͩͱͲ͏ͳΔ͔ʁ • Anomaly DetectionͷධՁσʔληοτͬͯଞʹͳ͍ͷ͔ͳɺɺɺɺ ʢ͍ͭ·ͰMNISTʹΑΔධՁ͕ଓ͘ͷͩΖ͏͔ʣ