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
クックパッドにおけるCloud AutoML事例
Search
chie8842
November 02, 2018
Technology
9
7.8k
クックパッドにおけるCloud AutoML事例
Cookpad Tech Kitchen #19 R&Dにおけるサービス開発者の仕事(
https://cookpad.connpass.com/event/104459/
)における発表資料です。
chie8842
November 02, 2018
Tweet
Share
More Decks by chie8842
See All by chie8842
MongoDB Atlas Search のご紹介
chie8842
2
1.3k
MongoDB Atlas Vectorsearchではじめる生成AIアプリ開発
chie8842
3
1.4k
AWS GlueとAWS Lake Formationではじめるデータマネジメント
chie8842
0
940
Distributed Processing in Python
chie8842
2
640
クックパッドにおける推薦(と検索)の取り組み
chie8842
21
7.9k
Understanding distributed processing in Python
chie8842
2
1.9k
Performance Tuning Tips of TensorFlow Inference
chie8842
1
720
Cookpad_Internship_MLOps_Lecture_2018
chie8842
35
16k
機械学習デプロイを支えるコンテナ技術(Machine Learning on Docker)
chie8842
14
8.2k
Other Decks in Technology
See All in Technology
rootlessコンテナのすゝめ - 研究室サーバーでもできる安全なコンテナ管理
kitsuya0828
3
380
VideoMamba: State Space Model for Efficient Video Understanding
chou500
0
190
RubyのWebアプリケーションを50倍速くする方法 / How to Make a Ruby Web Application 50 Times Faster
hogelog
3
940
【Pycon mini 東海 2024】Google Colaboratoryで試すVLM
kazuhitotakahashi
2
490
[CV勉強会@関東 ECCV2024 読み会] オンラインマッピング x トラッキング MapTracker: Tracking with Strided Memory Fusion for Consistent Vector HD Mapping (Chen+, ECCV24)
abemii
0
220
Making your applications cross-environment - OSCG 2024 NA
salaboy
0
180
障害対応指揮の意思決定と情報共有における価値観 / Waroom Meetup #2
arthur1
5
470
Amazon CloudWatch Network Monitor のススメ
yuki_ink
1
200
強いチームと開発生産性
onk
PRO
33
11k
Exadata Database Service on Dedicated Infrastructure(ExaDB-D) UI スクリーン・キャプチャ集
oracle4engineer
PRO
2
3.2k
いざ、BSC討伐の旅
nikinusu
2
780
iOSチームとAndroidチームでブランチ運用が違ったので整理してます
sansantech
PRO
0
130
Featured
See All Featured
Navigating Team Friction
lara
183
14k
Into the Great Unknown - MozCon
thekraken
32
1.5k
Docker and Python
trallard
40
3.1k
Ruby is Unlike a Banana
tanoku
97
11k
For a Future-Friendly Web
brad_frost
175
9.4k
Facilitating Awesome Meetings
lara
50
6.1k
Embracing the Ebb and Flow
colly
84
4.5k
Build The Right Thing And Hit Your Dates
maggiecrowley
33
2.4k
No one is an island. Learnings from fostering a developers community.
thoeni
19
3k
Documentation Writing (for coders)
carmenintech
65
4.4k
Building an army of robots
kneath
302
43k
Product Roadmaps are Hard
iamctodd
PRO
49
11k
Transcript
ΫοΫύουʹ͓͚Δ "VUP.- ݚڀ։ൃ෦ ػցֶशάϧʔϓ ྛాઍӯ !DIJF 2018/11/01 Cookpad Tech Kitchen
#19 R&D
$IJF)BZBTIJEB • ػցֶशνʔϜ ιϑτΣΞΤϯδχΞ • ػցֶशϞσϦϯάΔ͠ج൫Δ͠ΞϓϦॻ͘ • ͕͖ !DIJF 8FC
%# 1SFTTͰ ػցֶशΞϓϦ࡞ͷ ೖهࣄΛॻ͍ͨ 1Z$PO+1 %FW'FTU ͳͲͰొஃ
5-%3 • ,PNFSDPʢΫοΫύου৽نࣄۀʣʹ͓͚Δ (PPHMF $MPVE"VUP.-ࣄྫͷհ • αʔϏε։ൃͷಓ۩ͱͯ͠ػցֶशΛͬͯ ͍͜͏ʂͱ͍͏
,PNFSDPʹ͓͚Δ (PPHMF $MPVE "VUP.-ࣄྫ
• ΫοΫύουͷ&$৽نࣄۀ • ྉཧಓ۩ɺ͏ͭΘɺΧτϥϦʔɺϦωϯࡶ՟ͳͲͷ lྉཧָ͕͘͠ͳΔϞϊz͕ങ͑Δ ϚϧγΣΞϓϦ ͔ͭͬͯΈͯͶʂ
Ϟνϕʔγϣϯ ͱͱग़ऀͷࡋྔͰࣗ༝ʹλά͚ ͕ߦΘΕ͍͕ͯͨɺߪೖऀʹΑΔݕࡧ͠ ͢͞ͷͨΊɺΧςΰϦ͚Λߦ͍͍ͨ
ը૾Λ༻͍ͨΧςΰϦྨ • ը૾Λ༻͍ͯશΧςΰϦʹྨ͍ͨ͠ ۚଐͷث സ ΧτϥϦʔ Τϓϩϯ ϑϥΠύϯ ು แஸ
Ωονϯ πʔϧ ಃث ࣓ث ͷث Ψϥεͷث ࣫ث หശ Ωονϯ ϑΝϒϦοΫ
(PPHMF$MPVE"VUP.-7JTJPO • ػցֶशϞσϧΛτϨʔχϯάͯ͠ɺࣗͷఆٛ ͨ͠ϥϕϧʹैͬͯը૾Λྨ͢Δ͜ͱ͕Ͱ͖Δɺ ($1্ͷαʔϏε
(PPHMF$MPVE"VUP.-7JTJPO • ,PNFSDPج൫ͱͯ͠'JSFCBTFΛ࠾༻͓ͯ͠Γɺಉ͡ ($1্ͷαʔϏεͰ͋Δ$MPVE"VUP.-ͱ૬ੑ͕͍͍ • ػցֶशΤϯδχΞ͕͍ͳ͘ͱӡ༻͕Մೳ • ΨοͱࣗͰ࡞ͬͨ*ODFQUJPO7ϞσϧΑΓੑೳ͕ Αͦ͞͏ͩͬͨ •
ϞσϧͷαʔϏϯάͳͲͷख͕͍ؒΒͣɺΠχγϟϧ ͷར༻͕ߦ͍͍͢ • ίετ໘ͰػցֶशΤϯδχΞͷ(16 ΠϯελϯεΛར༻࣮ͨ͠ݧίετΛߟྀ͢Δͱ ༏ҐͰ͋Δ
ߟྀͨ͠ᶃ • ৽نαʔϏεͳͷͰɺը૾͕গͳ͍ΧςΰϦ ͕͋Δ • Πϯλʔωοτ্ͷը૾ΛՃֶͯ͠श • ΧςΰϦͷ͏ͪɺʮಃثʯͱʮ࣓ثʯͳͲɺࢹ ֮ใ͔Βผ͕͍͠ΧςΰϦ͕͋Δ •
ҰͭͷΧςΰϦͱͯ͠ਪఆ͠ɺग़ऀʹͲͪΒ͔બ ΜͰΒ͏
ߟྀͨ͠ᶄ • ෳͷΧςΰϦͷ͏ͪͲͪΒ͔·͍͠߹͕͋Δ • ͷหശˠΧςΰϦީิɿͷث PSหശ • ΧςΰϦͱ͍ͨͨ͠ΊɺείΞεϨογϣϧυ ˞ ΛԼ͛ͯෳͷΧςΰϦީิ͔Βग़ऀʹબͤΔ
• ࠓޙΧςΰϦใ͕มߋͱͳΔՄೳੑ͕͋Δ • ࠓճֶशͷͨΊͷϥϕϧ͚ࣾͰਓखͰߦ͕ͬͨɺ ࠓޙΧςΰϦͷՃɾมߋͳͲ͕ߟ͑ΒΕΔ • (PPHMF$MPVEͷ)VNBO-BCFMJOHαʔϏεͷར༻ͳͲ͕ ߟ͑ΒΕΔ ˞ "VUP.-ʹ͓͚Δਪఆ࣌ͷࢦඪɻ͜ͷΛߴ͘͢Δͱ৴པͷߴ͍ީิͷΈΛฦ٫͠ɺ ͘͢Δͱ৴པ͕͍ީิฦ٫͢Δ
͜Ε͔ΒͷαʔϏε ։ൃʹ͓͚Δػցֶश
ਓೳͷౙདྷͳ͍ʢͱ͍͏ਓ͍Δʣ https://www.wsj.com/articles/ai-guru-andrew-ng-on-the-job-market-of-tomorrow-1540562400 ػցֶशΛ༻͍ͨαʔϏε։ൃࠓޙٻΊΒΕ͍ͯ͘ (PPHMF#SBJOͷDPGPVOEFSɺ#BJEV 3FTFBSDIͷ"OESFX/Hͷهࣄ
ػցֶशͷར༻ύλʔϯ ֶशࡁΈϞσϧΛར༻͢Δ͚ͩͷύλʔϯ • Google Cloud Vision API • Amazon
Rekognition • Azure Cognitive Services ΧελϜϞσϧΛ࡞Δύλʔϯ • Google Cloud AutoML • "EB/FU r 5FOTPSGMPXϕʔεͷ"VUP.-ϑϨʔϜϫʔΫ • 5105 r TDJLJUMFBSOͷϋΠύύϥϝʔλνϡʔχϯάΛࣗಈԽ͢Δ πʔϧ ࠷৽ٕज़Λར༻ͯ͠ݻ༗ͷϞσϧΛ࡞Δ ύλʔϯ • 5FOTPSGMPXTDJLJUMFBSOΛར༻ͯࣗ͠Ͱ ϞσϧΛ࡞ΓɺσϓϩΠ͢Δ • Ͱղ͚ͳ͍ʹରͯ͜͠ͷํ๏͕ඞཁ easy difficult ͱ ػցֶशΤϯδχΞ͕͍ͳ͘ͱѻ͑Δ
easy difficulty ػցֶशͷར༻ύλʔϯ • ֶशࡁΈϞσϧΛར༻͢Δ͚ͩͷύλʔϯ • Google Cloud Vision API
• Amazon Rekognition • Azure Cognitive Services • ΧελϜϞσϧΛ࡞Δύλʔϯʢ"VUP.-ʣ • Google Cloud AutoML • "EB/FU r 5FOTPSGMPXϕʔεͷ"VUP.-ϑϨʔϜϫʔΫ • 5105 r TDJLJUMFBSOϕʔεͷػցֶशύΠϓϥΠϯͷ࠷దԽπʔϧ • ࠷৽ٕज़Λར༻ͯ͠ݻ༗ͷϞσϧΛ࡞Δύλʔϯ • Tensorflowscikit-learn • ػցֶशͷઐ͕ࣝඞཁͳΞϧΰϦζϜ࣮ɺϋΠύʔύϥϝʔλ νϡʔχϯάͳͲΛࣗಈͰߦͬͯ͘ΕΔ • ͜ΕΒΛ͏·͘͏͜ͱͰɺػցֶश͕Ͱ͖ͳͯ͘ɺαʔϏε։ൃͷ ෯͕͕Δ • Ϧαʔνʹ͓͍ͯ͞Ε͍ͯΔͷҰͭ
·ͱΊ • ػցֶशɺࠓޙػցֶशཧʹৄ͍͠ઐՈ Ͱͳͯ͘ѻ͏͜ͱ͕Ͱ͖ΔΑ͏ʹͳ͍ͬͯ͘ • (PPHMF$MPVE"VUP.-ศརʂ • ͱ͍͑ղ͚Δͱղ͚ͳ͍͕͋Δ • αʔϏε։ൃͷಓ۩ͱͯ͠ػցֶशΛ͏·͘
͍ͬͯ͜͏ʂ