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
コミック内の発話への読者手書き文字融合による共感度向上手法の提案(JSAI2017)
Search
Junki Saito
May 24, 2017
Research
49
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
コミック内の発話への読者手書き文字融合 による共感度向上手法の提案(JSAI2017)
Junki Saito
May 24, 2017
Other Decks in Research
See All in Research
SAM3を用いたコマ・吹き出しの 領域検出と分割構造からの読み順推定
kzmssk
0
130
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
790
HackSick vol.7 LT資料【LLMアーキテクチャ入門・事前学習時の躓き所解説】 スパースなAttention・状態空間モデル
rikkabotan7
0
180
最先端NLP勉強会2026 論文紹介:Reasoning with Sampling: Your Base Model is Smarter Than You Think (ICLR 2026 paper)
kogoro
4
610
ハードウェア研究で国際トップ会議を目指す!IROS 2027での論文採択を目指して
ayatokanada
6
3.7k
Sleuthcon Keynote - How Cybercriminals (ab)use AI
fr0gger
0
320
[SNLP2026] Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
wataruuuuu
0
310
Claude Code × autoresearch 実践
mathbullet
0
270
IA for theory
gpeyre
1
410
「AIとWhyを深堀る」をAIと深堀る
iflection
0
610
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
21
24k
高性能計算機クラスタを用いた大規模点群処理による森林の単木抽出と構造解析
kentaitakura
1
130
Featured
See All Featured
Navigating Weather and Climate Data
rabernat
0
510
Building Applications with DynamoDB
mza
96
7.2k
10 Git Anti Patterns You Should be Aware of
lemiorhan
PRO
659
62k
Music & Morning Musume
bryan
47
7.4k
Noah Learner - AI + Me: how we built a GSC Bulk Export data pipeline
techseoconnect
PRO
0
430
Why Your Marketing Sucks and What You Can Do About It - Sophie Logan
marketingsoph
0
410
Build your cross-platform service in a week with App Engine
jlugia
234
19k
Building Adaptive Systems
keathley
44
3.2k
State of Search Keynote: SEO is Dead Long Live SEO
ryanjones
0
270
Mobile First: as difficult as doing things right
swwweet
225
10k
How Software Deployment tools have changed in the past 20 years
geshan
1
34k
Building Better People: How to give real-time feedback that sticks.
wjessup
370
20k
Transcript
ίϛοΫͷൃͷಡऀखॻ͖จࣈ༥߹ ʹΑΔڞײ্ख๏ͷఏҊ ੪౻ Ҁج தଜ ૱࢙ ླ ਖ਼໌ ໌࣏େֶ ૯߹ཧֶ෦
ઌϝσΟΞαΠΤϯεֶՊ
ʢ ̍ ʣ ಡऀͷ ͱϚϯΨͷʮϑΥϯτʯΛ༥߹͢Δ͜ͱͰ ϢʔβͷڞײΛߴΊΔ͜ͱ͕Ͱ͖Δిࢠॻ੶ !
ৄ खॻ͖จࣈσʔλϕʔε !
ίϛοΫͷిࢠԽ ϢʔβͦΕͧΕͷσόΠεͰӾཡՄೳʹ ಈతͳදݱ͕Մೳʹ
ిࢠίϛοΫͷදݱΛ๛͔ʹ͢Δ खॻ͖จࣈΞχϝʔγϣϯੜख๏ <ࠤ౻ > खॻ͖จࣈΞχϝʔγϣϯΛࣜͰදݱ ϚϯΨͷηϦϑɾΦϊϚτϖʹ༩
σΟδλϧίϛοΫ੍࡞ͷͨΊͷಈతͳԻᄻදݱੜγεςϜ <দԼ > ίϛοΫͷΦϊϚτϖʹରͯ͠खܰʹಈ͖Λ༩ Ϣʔβ͕ύϥϝʔλΛௐઅ͢Δ͜ͱ͕Մೳ
ిࢠίϛοΫͷݱঢ় ΄ͱΜͲࢴഔମΛը૾σʔλͱͯ͠εΩϟϯ͚ͨͩ͠ͷͷ
ిࢠίϛοΫͷݱঢ় ΄ͱΜͲࢴഔମΛը૾σʔλͱͯ͠εΩϟϯ͚ͨͩ͠ͷͷ ిࢠίϛοΫͷརΛ׆͔͖͠Ε͍ͯͳ͍
ిࢠίϛοΫΛͱΜ͔ͭʹஔ͖͑Δͱ
ిࢠίϛοΫΛͱΜ͔ͭʹஔ͖͑Δͱ ίϛοΫʹௐຯྉΛՃ͑ͯࣗΈͷͷʹʂʂ
ిࢠίϛοΫʹ͓͚ΔௐຯྉΛՃ͑ΒΕͦ͏ͳ෦ ֆ ΦϊϚτϖɾޮՌԻ ίϚׂΓ ηϦϑ
• ஶऀͷͩ͜ΘΓϙΠϯτ • ઌߦݚڀ͋Γ
ిࢠίϛοΫʹ͓͚ΔௐຯྉΛՃ͑ΒΕͦ͏ͳ෦ ֆ ΦϊϚτϖɾޮՌԻ ίϚׂΓ ηϦϑ
• ஶऀͷͩ͜ΘΓϙΠϯτ • ઌߦݚڀ͋Γ
ίϛοΫͷηϦϑͰ༻͞ΕΔϑΥϯτ ଟ͘ՄಡੑΛॏࢹͨ͠ʮʯ ηϦϑͷϑΥϯτ֤ग़൛͕ܾࣾఆ͍ͯ͠ΔͨΊɼ ϚϯΨݪ࡞ऀ΄ͱΜͲհೖ͍ͯ͠ͳ͍ ूӳࣾ ߨஊࣾ BI F A
D G
ूӳࣾ ߨஊࣾ BI F A D G ίϛοΫͷηϦϑͰ༻͞ΕΔϑΥϯτ ଟ͘ՄಡੑΛॏࢹͨ͠ʮʯ ηϦϑͷϑΥϯτ֤ग़൛͕ܾࣾఆ͍ͯ͠ΔͨΊɼ
ϚϯΨݪ࡞ऀ΄ͱΜͲհೖ͍ͯ͠ͳ͍ ηϦϑʹ༻͞ΕΔϑΥϯτʹ ࣗΈͷຯ͚Λʂ
ฏۉจࣈඒ͍͠ <தଜ > ਓࣗࣗͷࣈΛ࠷ΉʢଞऀධՁͷഒҎ্ʣ
खॻ͖จࣈͷࣗଞࣝผೳྗͱײௐࠪ <੪౻ > ࣗͷखॻ͖จࣈݟ͚Δ͜ͱ͕Ͱ͖Δ
खॻ͖จࣈͷࣗଞࣝผೳྗͱײௐࠪ <੪౻ > ࣗͷखॻ͖จࣈݟ͚Δ͜ͱ͕Ͱ͖Δ ࣗͷखॻ͖จࣈʹରͯ͠ײΛ๊͖ɼ ଞऀͷखॻ͖จࣈͱ༥߹͢Δ͜ͱͰΑΓҹʹͳΔʂ
खॻ͖จࣈͷࣗଞࣝผೳྗͱײௐࠪ <੪౻ > ࣗͷखॻ͖จࣈݟ͚Δ͜ͱ͕Ͱ͖Δ ࣗͷखॻ͖จࣈʹରͯ͠ײΛ๊͕͘ɼ ଞऀͷखॻ͖จࣈͱ༥߹͢Δ͜ͱͰΑΓҹʹͳΔʂ
ϑΥϯτͱ༥߹ͯ͠ಉ༷ͷޮՌ͕ಘΒΕΔͷͰʁ
खॻ͖จࣈΛεύΠεͱͨ͠ಡΈखʹ͋ͬͨిࢠίϛοΫ
ྫ̍ ಡऀͷखॻ͖จࣈΛ༥߹ உੑಡऀ
ྫ̎ ঁੑಡऀ ಡऀͷखॻ͖จࣈΛ༥߹
ྫ̏ ঁੑͷײͷߴ͍ खॻ͖จࣈΛ༥߹ உੑಡऀ
ྫ̐ ײͷ͍ खॻ͖จࣈΛ༥߹ ಡऀ
• ϚϯΨ͔ΒηϦϑ෦ͷநग़ <ాத ><3JHBVE > • ηϦϑͷจࣈೝࣝʢ0$3ٕज़ʣ • ऀࣝผ •
ಡऀͷײҠೖਪఆ ˠ ࢹઢ͔Βڵຯ߹͍ਪఆ <Լ > • ϑΥϯτͱखॻ͖จࣈͷ༥߹ख๏
࣮ݱͷୈҰาͱͯ͠ ᶃ ʮϑΥϯτʯͱʮ ͷ༥߹ख๏ͷఏҊ ᶄ ༥߹ख๏Λ༻͍ͨϓϩτλΠϓγεςϜͷ࣮
ϑΥϯτͱखॻ͖จࣈͷ༥߹ख๏
8 96 5 7༥ ϑʔϦΤڃల։ʹΑΔखॻ͖จࣈͷࣜԽ <தଜ >ΛԠ༻ http://satoken.nkmr.io/2015/Char2Fourier/convert.html ! =
# = + .12 3 2 0 3 .12 3 - 2 0 3 - .12 3 2 0 3 - .12 3 ) 2 0 3 + + .12 (3 2 0 (3 - .12 )3 ) 2 0 )3 + .12 3 2 0 3 + .12 +3 2 0 +3 .12 3 2 0 3 4 - -) .12 3 2 0 3 ( .12 3 2 0 3 ( +) .12 3 ) 2 0 3 (( ( .12 3 + 2 0 3 - .12 (3 2 0 (3 ) .12 )3 + 2 0 )3 .12 3 2 0 3 + .12 +3 2 0 +3 .12 3 2 0 3 4 $ %
༥ จࣈͷଠ͞ใΛͭΊͷมͱͯ͠ॲཧ ! " → " $ →
$ z → (", $)
༥ ϑΥϯτͷσʔλͷ͔࣋ͪͨ ࣍ϕδΣۂઢɾ࣍ϕδΣۂઢͰදݱ ϑΥϯτදݱͷͨΊͷ੍ޚͷใ͔͍͍ͬͯ͠ͳ͍ ϑΥϯτͷਊઢͷ ྻ࠲ඪɾଠ͞Λࣄલʹࢉग़
จࣈͷଠ͞ใΛͭΊͷมͱͯ͠ॲཧ ༥ ! " → " $ →
$ z → (", $) ! " = *+ , $ = -+ , z = .+ , ! " = */ , $ = -/ , 0 = ./ , ˞खॻ͖จࣈͷଠ͞Ұఆͱ͢Δ ϑΥϯτ
ಘΒΕͨࣜಉ࢜ΛՃॏฏۉԽ͢Δ͜ͱͰ༥߹ ༥ ! " = $%& ' +
(1 − $)%- ' . = $/& ' + (1 − $)/- ' 0 = $1& ' + (1 − $)1- ' 0 ≤ $ ≤ 1 ʁ खॻ͖ ϑΥϯτ खॻ͖ ϑΥϯτ खॻ͖ ϑΥϯτ खॻ͖ ϑΥϯτ खॻ͖ ϑΥϯτ
ϓϩτλΠϓγεςϜ
ిࢠίϛοΫͷద༻ྫ
ʢЋʣ ʢЋʣ
( ( ( )
( ) ( ) ) ) ஶऀ " #
ΞΠυϧঁ༏ͷखॻ͖จࣈΛ༥߹ͨ͠ϚϯΨΛൢച
Ø ϚϯΨݪ࡞ऀͷωʔϜʹॻ͖ࠐ·Εͨखॻ͖จࣈΛ༥߹ Ø Ξχϝ൛ͷ༏ͷखॻ͖จࣈΛ༥߹͢Δ͜ͱͰ ҰछͷΞςϨίͷΑ͏ͳָ͠Έํ͕Ͱ͖Δ
Ø ϚϯΨݪ࡞ऀͷωʔϜʹॻ͖ࠐ·Εͨखॻ͖จࣈΛ༥߹ Ø Ξχϝ൛ͷ༏ͷखॻ͖จࣈΛ༥߹͢Δ͜ͱͰ ҰछͷΞςϨίͷΑ͏ͳָ͠Έํ͕Ͱ͖Δ
ήʔϜͷΩϟϥΫλͷൃݴɾΧϥΦέͷՎࢺɾ 8FCࠂ FUD
σδλϧίϯςϯπʹϑΥϯτ͕ ͖ͭͷͳͷͰԠ༻ઌͨ͘͞Μ
Ø ηϦϑͷϑΥϯτ͕खॻ͖จࣈͱ༥߹ͯ͠ఏࣔ͞ΕΔ ϚϯΨϏϡʔϫʔͷఏҊ Ø ϑΥϯτͱखॻ͖จࣈͷ༥߹ΞϧΰϦζϜͷఏҊ Ø ͞·͟·ͳγνϡΤʔγϣϯʹ͓͚Δద༻ྫͷհ ࠓޙͷల։ ü
ࣗͷखॻ͖จࣈͱϑΥϯτͷ༥߹จࣈͷҹௐࠪ खॻ͖จࣈͷ༥߹ׂ߹Ͳ͜·ͰΛڐ༰͢Δ͔ ü ༥߹จࣈ͕ΘΕͨϚϯΨΛಡΜͩͱ͖ɼڞײ্͕͢Δ͔ௐࠪ