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
3rd
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
udonmai
August 05, 2014
Research
59
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
3rd
2014_7_18
udonmai
August 05, 2014
More Decks by udonmai
See All by udonmai
ipsj_150318
udonmai
0
41
1st
udonmai
0
72
2nd
udonmai
0
59
Other Decks in Research
See All in Research
「行ける・行けない表」による地域公共交通の性能評価
bansousha
0
160
東京大学工学部計数工学科、計数工学特別講義の説明資料
kikuzo
0
500
Language and AI
ayaniwa
0
120
ブレグマン距離最小化に基づくリース表現量推定:バイアス除去学習の統一理論
masakat0
0
290
CyberAgent AI Lab研修 / Social Implementation Anti-Patterns in AI Lab
chck
7
4.7k
NLP colloquium: AI Safety Survey
kanekomasahiro
0
740
正規分布と最適化について
koide3
1
260
2026 東京科学大 情報通信系 研究室紹介 (すずかけ台)
icttitech
0
3.8k
敵対生成プロンプト同時探索による内省型プロンプト最適化
kinoue_smarthr
0
210
セマンティック通信勉強会 6Gに向けたデバイス間効率的な通信の技術紹介・課題・今後展望
satai
3
170
COFFEE-Japan PROJECT Impact Report(Uminomukou Coffee)
ontheslope
0
200
Dual Quadric表現を用いた動的物体追跡とRGB-D・IMU制約の密結合によるオドメトリ推定
nanoshimarobot
0
410
Featured
See All Featured
Intergalactic Javascript Robots from Outer Space
tanoku
273
27k
The #1 spot is gone: here's how to win anyway
tamaranovitovic
2
1.1k
State of Search Keynote: SEO is Dead Long Live SEO
ryanjones
0
210
Gemini Prompt Engineering: Practical Techniques for Tangible AI Outcomes
mfonobong
2
440
Marketing to machines
jonoalderson
1
5.5k
How To Stay Up To Date on Web Technology
chriscoyier
790
250k
Hiding What from Whom? A Critical Review of the History of Programming languages for Music
tomoyanonymous
2
860
The Limits of Empathy - UXLibs8
cassininazir
1
360
Save Time (by Creating Custom Rails Generators)
garrettdimon
PRO
32
3.5k
Raft: Consensus for Rubyists
vanstee
141
7.5k
Six Lessons from altMBA
skipperchong
29
4.3k
Money Talks: Using Revenue to Get Sh*t Done
nikkihalliwell
0
250
Transcript
History & Evolution with F o o d ฉ ຠᏴ
H26
Background - many researches on food and history of food
already - History of Food - focus on food recipes in early stage History with Food ? - not only recipe
ideas on Recipe - Graph - Ranking - among recipes
- within recipe
ideas on Recipe - Ranking - represent change & evolution
process - predict trends - verify the trends - ranking factor could be beyond time
among recipes - trends - total amount of recipes of
one typical food (keyword) - similar features
total amount - to some degree represent the popularity and
traditional / modern
trends (Google trend) - appears around 2011
trends (Google trend) - some relative search keywords
example 2007 2009 2009 2009 2007 2007
similar features ༲͛ ΧϨʔ ຯḩϥʔϝϯ ͡Ό͕ … - similar features
close time period
among recipes - why someone survives but others not -
trends appear year by year - what are the survivors’ common features - what will be the next
within recipe - calculate typicality of recipes - text analysis
- find most common ingredients - similarity
within recipe - intuitive explaination of typicality Jing, Yushi, and
Shumeet Baluja. ! "Visualrank: Applying pagerank to large- scale image search." ! Pattern Analysis and Machine Intelligence, IEEE Transactions on 30.11 (2008): 1877-1890.
Visualrank
Visualrank - begins with local feature vectors being extracted from
images - allow similarity to be considered between images with potential rotation, scale, and perspective transformations
Visualrank
within recipe - Why typicality ʁ - prototype is the
base - an object is more similar to the prototype of a concept, it has a higher typicality degree in the concept.
within recipe - Why typicality ʁ - to find the
prototype (base) - then we have a point to start or stand on
͡Ό͕ ಲ δϟΨΠϞ χϯδϯେ ۄͶ͗େ ͩ͠ͷૉ ਫ৯ࡐ͕͔ͿΔ ࡾԹ ে༉ ञ
ຯᬫ αϥμ༉ ͡Ό͕͍ ۄͶ͗ ਓࢀ ڇόϥബΓ ਫ ͩ͠ͷૉ ࡾԹ(࠭) ञ ΈΓΜ ͠ΐ͏Ώ ϥʔυ ͍Μ͛Μ౾(τοϐϯά༻) δϟΨΠϞ ਓࢀ ಲࡉΕ1 ͑ͷ͖ ͍Μ͛Μ ে༉ ञ ΈΓΜ ࠭ ͡Ό͕͍ ʹΜ͡Μ ͨ·Ͷ͗ ࡉΕ (ಲͰڇͰ) ਫ ͠ΐ͏Ώ ञ ΈΓΜ ࠭ ΄Μͩ͠ ಲόϥ(ബΓ) ͡Ό͕͍(ϝʔΫΠϯ) ۄͶ͗ ਓࢀ খ݁͠Β͖ͨ αϥμ༉(ᖱΊ༻) ͖͢ম͖ͷͨΕʢࢢൢʣ ΊΜͭΏʢࢢൢʣ ਫ
Future - data other than recipe ? - relationship between
history and food - what is history and in which what to focus on - what is food
Future - History of Food History of Food Culture ?
History with Food Food Culture History ? ? -‐ candidates -‐
Fin
「 我 是 分 割 线 」
trends (Google trend)
ຯḩϥʔϝϯ നୌେ1ାʢ180gʣ ΏͰཛ1ݸ Θ͔Ί3g ਫ300cc ຯḩେ1ͱ1/2 ͝·༉খ1/4 ͕ܲΒͷૉখ2 ʹΜʹ͘খ1/4 ౾൘েখ1/4
ञେ1/2 Ԙ͜͠ΐ͏গʑ ͓Έͷத՚໙̍ା ਫ500cc ञେ̎͞͡ ΣΠύʔେ̍͞͡ Ԙίγϣదྔ ຯͷૉదྔ Ͷ͗̎ʙ̏cm ͠1/2ା χϥ1/4ା ಲṢ30ʙ50g ̋ʹΜʹ͘νϡʔϒ1cm ̋͠ΐ͏͕νϡʔϒ1cm ̋౾൘েখ͞͡1/2 శ໙েେ̍͞͡ ຯḩେ̍͞͡ ͝·༉େ̎͞͡ ϥʔ༉దྔ ໙ΛᣐͰΔ͓౬దྔ ͝·༉େ͞͡1 ຯḩେ͞͡5 νϡʔϒ͠ΐ͏͕3cm νϡʔϒχϯχΫ4cm ౾൘েখ͞͡ࢁΓ1 ࠭খ͞͡1 ΦΠελʔιʔεେ͞ ͡1ऑ ਫ100cc ΣΠύʔখ͞͡ࢁ Γ1 ੜத՚ΊΜ(ଠ໙)2ۄ ຯḩϥʔϝϯεʔϓ̎ ା େ伹ম̍ύοΫ നࡊ̏ɺ̐ຕ ۄͶ͗ۄ ΤϦϯΪ̎ຊ χϯχΫ͔͚̍ ΩϜν͓ΈͰ •ΰϚ༉େ͞͡1 •Ԙίγϣগʑ •ຯḩখ͞͡2ͷ̍ ◇ಲͻ͖150g ◇Ͷ͗5cm ◇͠ΐ͏͕1/2͔͚ ◇౾൘েখ͞͡1/2 ◇͝·༉খ͞͡1/2 ◆ΣΠύʔ(·͕ͨܲΒ εʔϓͷૉ)େ͞͡1ɾ1/2 ◆ञ50cc ◆͠ΐ͏Ώେ͞͡1 ◆ਫ1600cc ◆Ԙগʑ ຯḩେ͞͡4 ˑΩϟϕπ·ͨനࡊ2ຕ ˑ͠1ା ˑͶ͗(͋Ε੨͍෦ )10cm τοϐϯά(মಲɺനͶ͗ ͳͲ)దྔ த՚໙(ੜ)4ۄ
ΧϨʔ ܲ180g ۄͶ͗1ʙ2ݸ τϚτͷ؈٧ ʹΜʹ͘(͢Γ͓Ζ͠)1ย ੜᇙ(͢Γ͓Ζ͠)1ย γφϞϯεςΟοΫগྔ ΧϧμϞϯ4ཻ ϩʔϦΤͷ༿1ຕ ਏࢠ1
Ϋϛϯγʔυখ͞͡1 Ϋϛϯύμʔখ͞͡2 ίϦΞϯμʔύμʔখ͞͡2 λʔϝϦοΫύμʔখ͞͡1 νϦύμʔখ͞͡1 αϥμΦΠϧେ͞͡1 Ϥʔάϧτେ͞͡3 ਫ100m ͳ̍͢ຊ τϚτத̎ݸ ͍Μ͛Μ̍̌ຊ ͡Ό͕͍ʢ͔΅ͪΌʣ த̎ݸ ۄͶ͗ ̎ͷ̍ݸ ߹ͼ͖ ̶̍̌̌̍̌ ̶̌ ▪ ʹΜʹ͘ ΨϥϜϚαϥେ̍͞͡ Ϥʔάϧτେ̍͞͡ ΧϨʔϧʔ̎ݸ ڇͻ͖͘͠߹Ṣ ͖150g ۄͶ͗1/2ݸ ࢠ2~3ݸ ϐʔϚϯ2~3ݸ χϯχΫɺੜᇙ֤1/2 ͔͚ αϥμ༉େ͞͡1 ে༉େ͞͡1.5 ञେ͞͡1.5 ελʔιʔεେ͞͡ 1.5 ένϟοϓେ͞͡1.5 ΧϨʔคখ͞͡1 ˒ۄͶ͗(1cm֯ʹΔ)1ݸ 200g ˒ࢠ(1cm֯ʹΔ)2ຊ ˒ύϓϦΧ(1cm֯ʹΔ)1ݸ ˒ύϓϦΧԫ(1cm֯ʹΔ)1ݸ ˒ΦΫϥ(1cm֯ʹΔ)6ຊ Ԙ2ͭ·Έ ʹΜʹ͘(Έ͡ΜΓ)Ұย ߹͍Ṣ͖ϛϯν200g Ԙίγϣదྔ ΧϨʔคେ͞͡3 ে༉େ͞͡2 ◼ਫ600cc ◼νΩϯίϯιϝ(ݻܗ)2ݸ ਫ༹͖ย܀คదྔ ཛ4ݸ ͝൧దྔ νΩϯखӋݩ8ຊ ۄͶ͗2ݸ ਓࢀ1ຊ ΧϨʔคେ4ఔ Ԙɺ͜͠ΐ͏ద ίϯιϝ2ݸ ๘ຶେ2͓ΈͰ ελʔιʔεখ2 ਏࢠ2ຊ χϯχΫɺ͠ΐ͏͕ νϡʔϒখ2ͣͭ ΦϦʔϒ͔ɺαϥμ༉ େ1 ਫ300ccҐ τϚτδϡʔε200ccҐ
༲͛ ܲ3ʙ4ຕ നϫΠϯେ̎͞͡ ίϯιϝ(ԽֶௐຯྉແఴՃ)େ ̍͞͡ ΫϨΠδʔιϧτখ̎͞͡ δϣχʔζΨʔϦοΫখ̍͞͡ ΦʔϧεύΠεখ͞͡1/2 ย܀คେ̏͞͡ ্৽คେ̎͞͡
ܲ2Ωϩ ে༉100cc ΈΓΜখ3 ञখ3 ʹΜʹ͘νϡʔϒ1cm ͠ΐ͏͕νϡʔϒ1cm ཛ1ݸ ย܀ค50cc ย܀คదྔ ༲͛༉গͳ͗͢ͳ͍ྔ ܲ2ຕ ˎ͓ਣখ͞͡2 ˎനͩ͠খ͞͡2 ˎԘখ͞͡1 ˎ͝·༉େ͞͡1ͱ1/2 ˎ͓Ζ͠χϯχΫͻͱ ͔͚ ˎૈṢ͖މᑦগʑ ย܀คେ͞͡3 খഴคେ͞͡3 ܲ580g •͠ΐ͏Ώେ2 •ञେ2 •ͪΈͭେ2 •͝·༉େ1 •ʹΜʹ͘νϡʔϒখ1 •͠ΐ͏͕νϡʔϒখ1 •෩ͩ͠ͷૉখ1 •ίνϡδϟϯখ1/2 ˕ཛ1ݸ ˕ϚϤωʔζେ2 ˕ਫେ2 ˑখഴคେ3 ˑย܀คେ3 ܲڳ2ຕ ে༉߿·ͨԘ߿ˎେ ͞͡1 ञˎখ͞͡1 ਣˎখ͞͡1 ΈΓΜˎʢ͋Εʣখ ͞͡1 Ԙίγϣˎগʑ ʹΜʹ͘νϡʔϒˎখ ͞͡1 ੜᇙνϡʔϒˎখ͞͡ 1 ย܀คదྔ
「 我 是 分 割 线 」