Upgrade to Pro — share decks privately, control downloads, hide ads and more …

クックパッドと機械学習（短縮版）/MLTBP

Avatar for j.harashima j.harashima
March 28, 2019
0
9k

 クックパッドと機械学習（短縮版）/MLTBP

Avatar for j.harashima

j.harashima

March 28, 2019
Tweet

More Decks by j.harashima

See All by j.harashima
日本語レシピデータセットの継続的な構築と複合的な利用/JED2022
Avatar for j.harashima junharashima
0
12k
自然言語処理の基礎と応用 〜 料理と医療を題材として 〜 /JADI2021
Avatar for j.harashima junharashima
0
4.7k
企業での研究開発の楽しさと苦労/WAP-Tech-Talk
Avatar for j.harashima junharashima
1
11k
クックパッドにおける研究開発/HCG2020
Avatar for j.harashima junharashima
0
4k
Calorie Estimation in a Real-World Recipe Service/iaai-20
Avatar for j.harashima junharashima
0
14k
クックパッドと NLP/CV/nlpaper-challenge
Avatar for j.harashima junharashima
0
110
AI 部門の実戦投入/repro-tech-meetup
Avatar for j.harashima junharashima
0
4k
クックパッドと機械学習/uec-aix-seminar
Avatar for j.harashima junharashima
0
11k
実データにふれる自然言語処理インターンシップ/internship2018
Avatar for j.harashima junharashima
1
15k

Featured

See All Featured
Designing for humans not robots
Avatar for Tammie Lister tammielis
253
25k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
Avatar for Rebecca Miller-Webster rmw
34
3.1k
Imperfection Machines: The Place of Print at Facebook
Avatar for Scott Boms scottboms
268
13k
XXLCSS - How to scale CSS and keep your sanity
Avatar for Zaharenia Atzitzikaki sugarenia
248
1.3M
Typedesign – Prime Four
Avatar for Hannes Fritz hannesfritz
42
2.8k
The Success of Rails: Ensuring Growth for the Next 100 Years
Avatar for Eileen M. Uchitelle eileencodes
46
7.6k
The Straight Up "How To Draw Better" Workshop
Avatar for Dennis Kardys denniskardys
236
140k
The Pragmatic Product Professional
Avatar for Laura Van Doore lauravandoore
36
6.8k
The Art of Delivering Value - GDevCon NA Keynote
Avatar for David Neal reverentgeek
15
1.6k
Six Lessons from altMBA
Avatar for Skipper Chong Warson skipperchong
28
4k
How STYLIGHT went responsive
Avatar for nonsquared nonsquared
100
5.7k
Documentation Writing (for coders)
Avatar for Carmen Chung carmenintech
73
5k

Transcript

  1. .BDIJOF-FBSOJOH5FBN#VJMEJOH1JUDI ΫοΫύουͱػցֶशɹ ΫοΫύουגࣜձࣾݪౡ७ ʢ୹ॖ൛ʣ

  2. ೥݄ ೥݄ ژ౎େֶࠇڮɾՏݪݚڀࣨ഑ଐ ΫοΫύουೖࣾɺαʔϏε։ൃ෦໳഑ଐ ത࢜ʢ৘ใֶʣऔಘ ΤϯδχΞʢ3VCZPO3BJMTػցֶशʣ ೥݄ Ϧαʔνϟʔʢࣗવݴޠॲཧ৘ใݕࡧʣ ݪౡ७ ೥݄

    Ϛωʔδϟʔʢਓࣄ޿ใࡒ຿ʜʣ ݚڀ։ൃ෦໳৽ઃɺಉ෦໳഑ଐ

  3. ΫοΫύου Πϯλʔωοτ্ͰϨγϐͷ౤ߘɾݕࡧ౳͕Ͱ͖Δ ೔ຊ࠷େͷྉཧϨγϐαʔϏε

  4. Ԋֵ ೥݄ ༗ݶձࣾίΠϯʢݱΫοΫύουגࣜձࣾʣઃཱ ೥݄ ೥݄ ೥݄ ೥݄ ೥݄ ೥݄ Ϩγϐͷ౤ߘɾݕࡧαʔϏεLJUDIFO!DPJO։࢝

    ΫοΫύουʹαʔϏε໊มߋ ೥݄ ϓϨϛΞϜαʔϏε։࢝ Ϛβʔζ্৔ ౦ূҰ෦ʹࢢ৔มߋ ೥݄ ւ֎ల։Λຊ֨Խ ઃཱप೥ ϓϨϛΞϜαʔϏεձһ਺ສਓಥഁ

  5.  ݴޠΧࠃ

  6.  ݴޠΧࠃ

  7. ϓϨϛΞϜձһ਺ ݄ؒར༻ऀ਺ʢࠃ಺ʣ ݄ؒར༻ऀ਺ʢւ֎ʣ Ϩγϐ਺ ສਓ ສ඼ ສ඼  ສਓ 

    ສਓ

  8. Ԋֵ ೥݄ ༗ݶձࣾίΠϯʢݱΫοΫύουגࣜձࣾʣઃཱ ೥݄ ೥݄ ೥݄ ೥݄ ೥݄ ೥݄ Ϩγϐͷ౤ߘɾݕࡧαʔϏεLJUDIFO!DPJO։࢝

    ΫοΫύουʹαʔϏε໊มߋ ೥݄ ϓϨϛΞϜαʔϏε։࢝ Ϛβʔζ্৔ ౦ূҰ෦ʹࢢ৔มߋ ϓϨϛΞϜαʔϏεձһ਺ສਓಥഁ ೥݄ ւ֎ల։Λຊ֨Խ ઃཱप೥ ೥݄ ݚڀ։ൃ෦໳৽ઃ

  9. എܠ ʕʕ·ͣ͸ɺݚڀ։ൃ෦͕ੜ·Εͨഎܠʹ͍ͭͯڭ͍͑ͯͩ͘͞ɻ ɹݚڀ։ൃ෦্ཱ͕͕͖͔͚ͪͬͨͬ͸ɺ೥͝Ζʹࣾ಺ͰػցֶशͳͲʹڵຯΛ࣋ͬͨࣾһ͕਺໊ू·ͬͯݚڀ ։ൃνʔϜΛ࡞Γɺۀ຿֎Ͱࣗओతʹٕज़ͷݚڀΛ͸͡Ίͨͱ͜Ζʹ͔͞ͷ΅Γ·͢ɻΫοΫύου͸αʔϏεͷӡӦ Λ௨ͯ͡ɺ͞·͟·ͳϢχʔΫͳσʔλΛอ༗͍ͯͨ͠ͷͰɺͦΕΒͱػցֶशΛ૊Έ߹ΘͤΔ͜ͱͰ໘ന͍αʔϏε ͕࡞Εͳ͍͔ͱߟ͍͑ͯͨͷͰ͢ɻ ɹ౰࣌ɺͦΕ͸ձࣾʹͱͬͯ͸͋͘·Ͱlඇެࣜzͳ׆ಈͩͬͨͷͰ͕͢ɺͨͱ͑͹ΫοΫύουͷϏοάσʔλΛݚڀ ໨తͱͯ͠େֶͷݚڀࣨʹఏڙ͢Δͱ͍ͬͨ׆ಈΛଓ͚ΔதͰɺࣾ಺Ͱ΋஫໨͞ΕΔΑ͏ʹͳΓ·ͨ͠ɻ͢Δͱɺ ೥͝Ζ͔Β͸ੈͷதͰػցֶशʹେ͖ͳ஫໨͕ू·ΔΑ͏ʹͳΓɺͦΕʹͱ΋ͳͬͯࣾ಺ͷ͞·͟·ͳ෦ॺ͔ ΒɺࢲͨͪʹػցֶशΛ׆༻ͨ͠αʔϏε΁ͷ։ൃࢧԉͷ੠ֻ͕͔ΔΑ͏ʹͳͬͨͷͰ͢ɻ

    ɹͦ͏ͳΔͱɺػցֶश͕ಘҙͳΤϯδχΞ͕ධՁ͞ΕΔ૊৫͕ࣾ಺ʹͳ͍ͱ͍͏՝୊͕ੜ·Εͯ͠·͍·͢ɻ౰࣌ɺ ͦ͏͍͏ΤϯδχΞ͸͞·͟·ͳ෦ॺʹϔϧϓʹ͍ͬͯ׆ಈ͍ͯͨ͠ͷͰɻͦ͏ͨ͠૊৫తͳ՝୊ͱɺੈͷதͷػց ֶशʹؔ͢ΔϜʔϒϝϯτͳͲΛഎܠʹɺ೥݄ʹݚڀ։ൃࣨͱͯ͠ਖ਼ࣜͳࣾ಺νʔϜ͕Ͱ͖ɺ݄ʹ෦ʹঢ֨ ͠·ͨ͠ɻ

  10. ؔ࿈αʔϏε ͓·͔ͤ੔ཧ ०ͷ໺ࡊϨγϐ΍ ϝΠϯʹ࢖͑ΔϨγϐ͕ ͙͢ʹݟ͔ͭΔ "MFYBεΩϧ ࢖͍͍ͨࡐྉ͚ͩͰ ͙͢ʹ࡞ΕΔਓؾͷϨγϐఏҊ ྉཧ͖Ζ͘ ࡱΔ͚ͩͰه࿥ʹʂ

    ͋ͳͨͷ͓ྉཧࣸਅΛ ࣗಈͰ੔ཧ Ϩγϐͷͦͷઌ΁ ͓͍͍͠εϚʔτΩονϯ 0J$Z

  11. ਓ਺ ʢׅހ಺͸.-ܥͷਓ਺ʣ        

         ਓ ਓ ਓ ਓ ਓ ਓ ϋʔυ΢ΣΞܥͷνʔϜ͕ ผ෦໳ͱͯ͠εϐϯΞ΢τ

  12. .-ܥͷτϐοΫʢҰྫʣ ࣗવݴޠॲཧ ɾ&ODPEFS%FDPEFSʹΑΔࡐྉ໊ͷਖ਼نԽ ɾ-45.ʹΑΔඇखॱͷݕग़ ɾGBTU5FYUʹΑΔࡐྉͷϨίϝϯυ ը૾ೝࣝ ɾ*ODFQUJPOWʹΑΔྉཧࣸਅͷநग़΍෼ྨ ɾ.PCJMF/FUʹΑΔݟӫ͑͢Δྉཧࣸਅͷநग़ ɾ43("/ʹΑΔྉཧࣸਅͷ௒ղ૾ .-0QT

    ɾ4MBDL$IBU#PUʹΑΔ&$Πϯελϯεͷ؅ཧ ɾ%PDLFSͱDPPLJFDVUUFSʹΑΔϓϩδΣΫτͷ؅ཧ ɾ&$4ʹΑΔΞϓϦͷσϓϩΠ΍%8)ʹΑΔσʔλͷ؅ཧ ΦʔϓϯαΠΤϯε ɾݚڀ༻Ϩγϐσʔλͷެ։ʢXࠃཱ৘ใֶݚڀॴʣ ɾ֤छίϯϖςΟγϣϯͷ࣮ࢪʢX಺ֳ෎ɺਓ޻஌ೳֶձɺʜʣ ɾ࿦จͷࣥචʢ೥Ͱຊ͕࠾࿥ʣ .-ܥ͸ਓ

  13. ৄ͘͠͸ BOENPSF

  14. ͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠ɻ