Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
【WSSIT2019】食材名の分散表現学習を用いた料理レシピの栄養推定手法
Search
umeco
March 08, 2019
Research
650
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
【WSSIT2019】食材名の分散表現学習を用いた料理レシピの栄養推定手法
WSSIT2019で発表した研究のスライドです
umeco
March 08, 2019
More Decks by umeco
See All by umeco
Clineプロンプト徹底解剖
umeco
0
890
LLMでの多言語対応どうする問題
umeco
0
300
大生成AI時代の新規事業戦略を考える
umeco
0
190
Cookpad_R&D_internship_2018_byumeco
umeco
0
520
Distributed prioritized experience replay
umeco
0
570
【WSSIT2018】料理レシピの分散表現を用いた代替食材の発見手法
umeco
2
720
Using an Artificial Financial Market for studying a Cryptocurrency Market
umeco
0
650
【WSSIT2017】過去の変動に対する類似検索を用いた短時間USD/JPY為替レート予測
umeco
0
560
Other Decks in Research
See All in Research
MIRU2026 チュートリアル講演2:三次元データ処理の動向
nnchiba
8
5.3k
Anthropic が提案する LLM の内部状態を自然言語で説明可能にした Natural Language Autoencoders / Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations
shunk031
0
330
2026年 オープンキャンパス 研究室紹介
junkurihara
0
230
Bilan travail en élevage : première restitution des enquêtes 2025-2026
institutdelelevage
PRO
0
150
データサイエンティストの就労意識~2015 → 2026 一般(個人)会員アンケートより
datascientistsociety
PRO
0
990
SAKURAONE:An Open Ethernet-based AI HPC System And Its Observed Workload Dynamicsin a Single-Tenant LLM Development Environment
yuukit
1
660
Fukui Shibiten 39 - AI Art
butchi
0
260
論文読み会 SNLP2026 Tau2-Bench: Evaluating Conversational Agents in a Dual-Control Environment
s_mizuki_nlp
0
300
20260624 NLP colloquium: 単一のhubテキストがCLIPを壊す:hubnessによる埋め込みの脆弱性特定
de9uch1
2
290
Kaggle|AI Agent Security — しくじり先生、俺みたいになるな
pomcho555
1
120
ふとした出会いで生まれたSkillが、 社内利用1位になるまで
mikimhk
22
24k
[最先端NLP勉強会2026] Agentic Rubrics as Contextual Verifiers for SWE Agents
rfujii
1
380
Featured
See All Featured
Facilitating Awesome Meetings
lara
57
7.2k
Into the Great Unknown - MozCon
thekraken
41
2.8k
How To Speak Unicorn (iThemes Webinar)
marktimemedia
1
590
Documentation Writing (for coders)
carmenintech
77
5.6k
Navigating Algorithm Shifts & AI Overviews - #SMXNext
aleyda
1
1.6k
Imperfection Machines: The Place of Print at Facebook
scottboms
270
14k
Data-driven link building: lessons from a $708K investment (BrightonSEO talk)
szymonslowik
1
1.4k
AI Search: Implications for SEO and How to Move Forward - #ShenzhenSEOConference
aleyda
1
1.4k
Stewardship and Sustainability of Urban and Community Forests
pwiseman
0
540
Building a Modern Day E-commerce SEO Strategy
aleyda
45
9.2k
Odyssey Design
rkendrick25
PRO
2
860
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
123
22k
Transcript
৯ࡐ໊ͷࢄදݱֶशΛ༻͍ͨ ྉཧϨγϐͷӫཆਪఆख๏ കຊɼ๛ాɼେݪ߶ࡾ ੨ࢁֶӃେֶେֶӃ ཧֶઐ߈
ݚڀഎܠ n ۙɼ8&#্ʹ͓͍ͯྉཧϨγϐͷڞ༗͕׆ൃ n ݈߁ͷ্ͷͨΊʹྉཧϨγϐΛར༻͢Δݚڀ͕Μ n ӫཆૉΛߟྀ͢Δ߹ɼӫཆૉྔͷܭࢉ͕ඞཁ ྉཧϨγϐͷදతͳӫཆૉྔͷਪఆख๏·ͩͳ͍
৯ࡐͷӫཆૉใ Ұൠతʹຊ৯ඪ४ද ͕༻͍ΒΕΔ ৯දͷྫ kcal g
g ຊͰৗ༻͞ΕΔ৯ࡐ Hதͷӫཆૉྔ͕هࡌ
ྉཧϨγϐͷӫཆૉྔͷܭࢉํ๏ ɾɾɾ ϒϩοίϦʔ ຊʢHʣ ɾɾɾ ྉཧϨγϐ
৯ද͔Β৯ࡐʹ ରԠ͢Δ߲Λબ ৯ࡐͷάϥϜॏྔΛܭࢉ ߲ͷ֤ͱάϥϜॏྔ ͔ΒӫཆૉྔΛܭࢉ ֤৯ࡐͷӫཆૉྔΛ߹ܭ ΤωϧΪʔɿ ⁄ 33 #$%& 100) ∗ 180) = 59.4 #$%& ਫɿ ⁄ 89 ) 100) ∗ 180) = 160.2 ) λϯύΫ࣭ɿ ⁄ 4.3 ) 100) ∗ 180) = 7.74 )
ӫཆૉྔࣗಈܭࢉʹ͓͚ΔͭͷλεΫ ৯ࡐ໊͔Βਖ਼͍͠৯ද߲Λਪఆʢ߲ਪఆʣ άϥϜॏྔΛਪఆʢॏྔਪఆʣ γνϡʔ༻ͷڇϒϩοΫ ͏͠ ੜ
දه༳Ε͕͋Δ τϚτ େݸ H άϥϜදهͰͳ͍߹ਪఆ
ؔ࿈ݚڀ n ۄాΒͷݚڀ<> ରσʔλɿʮϨγϐେඦՊʯσʔλ ߲ਪఆ๏ɿจࣈྻͷશϚονϯά ॏྔਪఆ๏ɿਓखʹΑΔॏྔมࣙॻͷߏங n ןถΒͷݚڀ<> ରσʔλɿʮΩϡʔϐʔΫοΩϯάʯσʔλ ߲ਪఆ๏ɿಡΈԾ໊Ͱͷฤूڑൺֱ
ॏྔਪఆ๏ɿਓखʹΑΔॏྔมࣙॻͷߏங දه༳Ε͕େ͖͍ϢʔβߘϨγϐʹదԠ͕͍͠ hXi + 2#a,; 9m51 MImN'4O @Gn@ACK<2 _QCm3B :*JE @mLPD;7=H?3MKRbcdeYWX]ST gkjVQcYU[QRYWX]SVQ hYi)&%2 #$a,"FmMPA>.6 8/-O@Gn @ACK2!*0( b2 gkjVQ`YTQfkVQ^TQllVQ`^\–`_ZQRYWW`SVQ
ݚڀత ಛ n ࢄදݱΛར༻͢Δ͜ͱͰදهΏΕʹରԠ Ϩγϐσʔλ͔ΒྉཧΧςΰϦ༧ଌΛ࡞͠ɼ 'PPEOBNF&ODPEFS '& Λֶश n ඪ४ॏྔࣙॻͷࣗಈߏங๏ͷఏҊ
දهΏΕ͕େ͖͍ϢʔβߘϨγϐʹରԠͰ͖Δ ؤ݈ͳӫཆૉྔਪఆख๏ͷఏҊ
ఏҊख๏ දه༳ΕʹରԠ͢ΔͨΊ৯ࡐ໊ͷࢄදݱΛར༻ ௐཧखॱʹXPSEWFDΛద༻͢Δ͜ͱͰ֫ಘՄೳ<> ৯ࡐ໊ͷࢄදݱԽʢʣ ಘΒΕΔࢄදݱܗଶૉ͝ͱʹ༩͑ΒΕΔ ಲόϥ ࢄදݱ ʢଟ࣍ݩϕΫτϧʣ ྫ
V 0 3 N ) I E . - 2 214C 6 ( 24
ఏҊख๏ ৯ࡐ໊ܗଶૉ͕ͭʹͳΔͱݶΒͳ͍ ৯ࡐ໊ͷࢄදݱԽʢʣ ̍ͭͷ߹ɿಲόϥ ಲόϥ ̎ͭͷ߹ɿಲʢόϥʣ ಲ όϥ ෳͷࢄදݱΛͭͷࢄදݱʹ·ͱΊΔ͜ͱ͕ඞཁ
ఏҊख๏ ྉཧϨγϐʹ͓͍ͯҎԼͷʹண λΠτϧ͔ΒྉཧΧςΰϦʢྉཧ໊ʣ͕நग़Մೳ ৯ࡐ͔ΒྉཧΧςΰϦ͕༧ଌՄೳ ྉཧΧςΰϦ༧ଌ ఆ൪ʂೱްΫϦʔϜγνϡʔ ৯ࡐ͔ΒྉཧΧςΰϦΛ༧ଌ͢ΔΛߏங ྫ
ྉཧΧςΰϦ ৯ࡐ໊ ࢄදݱ ྉཧΧςΰϦ
ఏҊख๏ ҎԼͷχϡʔϥϧωοτϫʔΫϞσϧͰֶश ྉཧΧςΰϦ༧ଌϞσϧ ྉཧΧςΰϦɿΫϦʔϜγνϡʔ ৯ࡐɿγνϡʔ༻ͷڇϒϩοΫɼਓࢀɼ ͡Ό͕͍ɼڇೕɼFUD (BUFE3FDVSSFOU6OJU (36 'VMM$POOFDUJPO '$
ఏҊख๏ ৯ࡐ໊Τϯίʔμ ֶशޙͷΤϯίʔμͰ৯ࡐ໊ΛࢄදݱԽ จࣈྻࣄલʹ XPSEWFDͰࢄදݱԽ ৯ࡐ໊Τϯίʔμ 'PPEOBNF&ODPEFS '& ৯ࡐ໊ͷࢄදݱ
ೖྗྫɿγνϡʔ༻ͷڇϒϩοΫ
ఏҊख๏ ࢄදݱͷڑʹج߲ͮ͘બ ಲʢόϥʣ ΤϯίʔμʹΑΔࢄදݱԽ Ϳͨ Β Ϳͨ
ίαΠϯྨࣅΛܭࢉ ৯ࡐ໊ʹ࠷ྨࣅ͢Δ߲Λબ
ఏҊख๏ ҎԼͷϧʔϧͰ৯ࡐͷॏྔΛਪఆ άϥϜදهͰ͋ΕͦΕΛ༻ ମੵදهʢେ͞͡ΧοϓʣͰ͋Εɼ 1 "# = 1
%ͱͯ͠άϥϜදهม ͦΕҎ֎ͷ߹ඪ४ॏྔࣙॻͷΛ༻ ৯ࡐॏྔͷਪఆํ๏
ఏҊख๏ طଘݚڀ< >ͰਓखͰࣙॻ͕࡞͞Ε͓ͯΓೖखෆՄ ຊݚڀͰྉཧϨγϐσʔλ͔Β ҎԼͷఆٛʹج͖ͮࣗಈతʹඪ४ॏྔࣙॻΛߏங ඪ४ॏྔࣙॻͷߏஙʢ̍ʣ શϨγϐʹ͓͚Δ֤৯ࡐͷάϥϜදهͷதԝ hXi + 2#a,;
9m51 MImN'4O @Gn@ACK<2 _QCm3B :*JE @mLPD;7=H?3MKRbcdeYWX]ST gkjVQcYU[QRYWX]SVQ hYi)&%2 #$a,"FmMPA>.6 8/-O@Gn @ACK2!*0( b2 gkjVQ`YTQfkVQ^TQllVQ`^\–`_ZQRYWW`SVQ
ఏҊख๏ ҎԼͷϧʔϧͰඪ४ॏྔࣙॻΛߏங ྉཧϨγϐσʔλ͔Β৯ࡐ໊ͱॏྔͷϖΞΛநग़ ॏྔ͕άϥϜදهͷ߹ɼ৯ࡐ໊Τϯίʔμʔͷ ग़ྗͱͳΔࢄදݱʹରԠ͢ΔϦετʹॏྔΛՃ ֤ϦετͷதԝΛରԠ͢Δ৯ࡐͷඪ४తͳ ॏ͞ͱ͢Δ
ඪ४ॏྔࣙॻͷߏஙʢʣ
ධՁ࣮ݧ ࣮ݧσʔλ $00,1"%͕ఏڙ͢ΔྉཧϨγϐσʔλ n Ϩγϐɿສ݅ఔ n ΧςΰϦ༧ଌσʔλɿສ݅ఔ ධՁσʔλ $00,1"%্Ͱެ։͞Ε͍ͯΔྉཧϨγϐ n
Ϩγϐɿ݅ n ؚ·ΕΔ৯ࡐɿ݅ ৯දΛ༻͍ͯਓखͰӫཆૉྔΛܭࢉ ࣮ݧσʔλͱධՁσʔλ
ධՁ࣮ݧ ৯ද߲ͷਪఆਫ਼ ߲ͷΈਪఆͨ࣌͠ͷӫཆૉྔਪఆਫ਼ ߲ͱॏྔΛਪఆͨ࣌͠ͷӫཆૉྔਪఆਫ਼ ධՁରͱධՁࢦඪ ධՁࢦඪɿ5PQ! QSFDJTJPOʢ!ݸͷީิʹਖ਼ղ͕͋Δ֬ʣ
ؔ࿈ݚڀ<>Ͱ༻͍ΒΕ͍ͯΔධՁࢦඪ ฏۉ૬ରޡࠩ ฏۉઈରޡࠩ ૬ؔ ૬ରޡࠩҎׂ߹ ૬ରޡࠩதԝ ઈରޡࠩதԝ Ճͨ͠ධՁࢦඪ , ", :CNN !)% # $+ # ) *( '& D,Vol. 101, No. 8, pp. 1099–1109 (2018).
ධՁ࣮ݧ XPSEWFDͰಘͨࢄදݱΛ༻͍߲ͯΛਪఆ͢Δख๏ XPSEWFD NFBO ๏ XPSEWFD UPQ ๏
ൺֱख๏ʢXPSEWFDʣ ෳͷࢄදݱ ৯ࡐ໊ͷ֤ࢄදݱͱ߲ͷࢄදݱͷڑΛෳܭࢉ ࠷খ͍͞ڑΛදతͳڑͱ͢Δख๏ ʢఏҊख๏ʣෳͷࢄදݱ ࢄදݱ ࢄදݱ ฏۉ '&
ධՁ࣮ݧ ฤूڑΛ༻͍߲ͯΛਪఆ͢Δख๏ &EJUEJTUBODF๏ &EJUEJTUBODF OPSN ๏ ൺֱख๏ʢฤूڑʣ ৯ࡐ໊ͱ৯දͷ߲ؒͷฤूڑ͕࠷
ͱͳΔ߲Λબ͢Δख๏ ͷख๏ʹ͓͚ΔฤूڑΛ͍ํͷจࣈྻͰׂͬͨ ฤूڑΛ༻͍Δख๏
࣮ݧ݁Ռͱߟ ৯ද߲ਪఆਫ਼ͷൺֱ ! -! ӫཆਪఆͰॏཁͳ5PQͰͷਫ਼ͰɼఏҊख๏ߴ͍༏Ґੑ
࣮ݧ݁Ռͱߟ ߲ਪఆਫ਼ͷൺֱ -! ! ฤूڑ ݸ͔ΒީิΛ૿ͯ͠ਫ਼্͕ঢͮ͠Β͔ͬͨ
࣮ݧ݁Ռͱߟ ఏҊख๏͕ਖ਼ղɼฤूڑ͕ෆਖ਼ղͩͬͨྫ ߲ਪఆͰͷఏҊख๏ͷ༏Ґੑ !) " %( ' ,
* * $+ &+ # # จࣈྻͱͯ͠ҟͳΔ͕ɼྨࣅ͢Δ֓೦ͷ৯ࡐΛબՄೳ ฤूڑͰจࣈྻͱͯ͠ҟͳΔ߹ਖ਼ղ͕ࠔ
࣮ݧ݁Ռ ߲ͷΈਪఆͨ݁͠ՌʢΧϩϦʔʣ 2 5 4 . ) . ) 1%0
4 ) ) 1%0 . ( ) 3 શͯͷධՁࢦඪͰఏҊख๏͕༏Ґ
࣮ݧ݁Ռ ߲ͷΈਪఆͨ݁͠ՌʢΧϩϦʔʣ 2 5 4 . ) . ) 1%0
4 ) ) 1%0 . ( ) 3 ฏۉͱதԝʹେ͖ͳ͕ࠩ͋Γɼ֎Εతͳαϯϓϧ͕ଘࡏ
࣮ݧ݁Ռ ߲ͷΈਪఆͨ݁͠ՌʢΧϩϦʔʣ 2 5 4 . ) . ) 1%0
4 ) ) 1%0 . ( ) 3 ࢄදݱख๏͕ฤूڑख๏ΑΓߴ͍ਫ਼
࣮ݧ݁Ռ ߲ͱॏྔΛਪఆͨ݁͠ՌʢΧϩϦʔʣ 5% 8 .9 7)12 12 4 3 7)02
02 4 3 1( 6 ఏҊख๏ͷ༏Ґੑ͕খ͘͞ͳ͍ͬͯΔ
࣮ݧ݁Ռ ߲ͱॏྔΛਪఆͨ݁͠ՌʢΧϩϦʔʣ 5% 8 .9 7)12 12 4 3 7)02
02 4 3 1( 6 ॏྔͷΈਪఆʢ߲ਖ਼ղϥϕϧʣͨ͠߹Ͱਫ਼͕ѱ͍
࣮ݧ݁Ռ ߲ͱॏྔΛਪఆͨ݁͠ՌʢΧϩϦʔʣ 5% 8 .9 7)12 12 4 3 7)02
02 4 3 1( 6 ॏྔͷਪఆਫ਼͕ѱ͍͜ͱ͔Βɼ༏Ґੑ͕খ͘͞ͳͬͨ
·ͱΊ n ྉཧΧςΰϦ༧ଌΛ࡞͠ɼֶशͨ͠৯ࡐ໊ ΤϯίʔμΛ༻͍ͯ৯ද߲Λਪఆͨ͠ n ఏҊख๏࣮ݧʹΑΓɼ৯ද߲ͷ༧ଌʹ ͓͍ͯ༏ҐੑΛࣔͨ͠ nॏྔਪఆͷޡࠩʹΑͬͯɼશࣗಈͰͷӫཆૉྔͷ ਪఆޡࠩେ͖͘ͳͬͨ
ࠓޙͷ՝ n ॏྔඪ४ࣙॻͷߏஙํ๏Λݟ͠ɼ৯ࡐॏྔͷ ਪఆޡࠩΛখ͘͢͞Δ n ௐཧखॱ͔Β৯ࡐͷঢ়ଶʢੜɼΏͰɼᖱΊʣΛ ਪఆ͢Δ͜ͱͰɼӫཆૉྔͷਪఆޡࠩΛখ͘͢͞Δ n ྉཧΧςΰϦ༧ଌͰͷɼޡநग़ΧςΰϦͷ আڈʹΑΔఏҊख๏ͷਫ਼্
͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠
!"#$%&%'(@* = 1 ( - ./0 1 2. ∩
4.,0 , 4.,6 , … , 4.,8 (ධՁσʔλͷ 2.%൪ͷධՁσʔλͰͷਖ਼ղϥϕϧ 4.,9 %൪ͷධՁσʔλͰͷ:൪ͷީิ 5PQ* QSFDJTJPOͷఆٛࣜ
ଞͷӫཆૉͰͷਪఆޡࠩʢఏҊख๏ʣ 0 .1 2 ) % ) %
) ( %
ఏҊख๏͕ෆਖ਼ղɼฤूڑ͕ਖ਼ղͩͬͨྫ ఏҊख๏ͷ & #% $( !" !
! ' ' ' ' ৯ࡐͷΘΕํ͕ࠅࣅ͢Δ৯ࡐࢄදݱֶश͕͍͠