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
自己紹介スライド(2023/8)
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Ibuki Nakamura
August 27, 2023
Education
480
1
Share
自己紹介スライド(2023/8)
2023年8月に研究室OB・OG会で使用したLT資料を自己紹介資料として流用しています
Ibuki Nakamura
August 27, 2023
More Decks by Ibuki Nakamura
See All by Ibuki Nakamura
たった1人から始めて深層学習によるニュース要約をプロダクトに実装した方法〜ファーストペンギンでやりきる力〜
inakam
1
3.6k
Other Decks in Education
See All in Education
SSH公開鍵認証 / 02-b-ssh
kaityo256
PRO
0
150
共感から、つくる: 変わり続ける自分と、誰かのための創造
micknerd
1
320
青森県の人口減少について | | 下山学園高等学校
aomori6
0
130
Modelamiento Matematico (Ingresantes UNI 2026)
robintux
0
270
「機械学習と因果推論」入門 ③ 漸近効率な推定量と二重機械学習
masakat0
0
520
Lenguajes de Programacion (Ingresantes UNI 2026)
robintux
0
160
Gluon Recruit Deck
gluon
0
170
✅ レポート採点基準 / How Your Reports Are Assessed
yasslab
PRO
0
320
勾配ブースティングと決定木の話 / gradient boosting and decision trees
kaityo256
PRO
5
1k
バージョン管理とは / 01-a-vcs
kaityo256
PRO
1
310
(2026) Quelle(s) mathématique(s) dans la "grande" culture?
mansuy
1
110
コマンドラインの使い方 / 01-d-cli
kaityo256
PRO
0
140
Featured
See All Featured
The innovator’s Mindset - Leading Through an Era of Exponential Change - McGill University 2025
jdejongh
PRO
1
160
Ten Tips & Tricks for a 🌱 transition
stuffmc
0
99
Technical Leadership for Architectural Decision Making
baasie
3
330
A Guide to Academic Writing Using Generative AI - A Workshop
ks91
PRO
1
270
Lessons Learnt from Crawling 1000+ Websites
charlesmeaden
PRO
1
1.2k
First, design no harm
axbom
PRO
2
1.2k
Marketing to machines
jonoalderson
1
5.2k
Neural Spatial Audio Processing for Sound Field Analysis and Control
skoyamalab
0
260
Large-scale JavaScript Application Architecture
addyosmani
515
110k
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
25
1.9k
Docker and Python
trallard
47
3.8k
Exploring the Power of Turbo Streams & Action Cable | RailsConf2023
kevinliebholz
37
6.4k
Transcript
ػցֶशͷಓΛΓ ։͍ͯػցֶशͷ ࣄΛͯ͠Δ -JHIUOJOH5BML "VH *#6,*/",".63"
ձһγεςϜ(ୈ։ൃνʔϜ χϑςΟχϡʔε։ൃϦʔμʔςοΫϦʔυ /.BDIJOF-FBSOJOH1SPEVDU&OHJOFFS େֶӃͰਂֶशΛ༻͍ͨಈըೝࣝͷݚڀ %FWFMPQFST4VNNJUొஃऀ ϕετεϐʔΧʔҐɺެื உঁࠞνΞϦʔσΟϯάຊද தଜҏਧ
ੈքେձ͕ϑϩϦμͷσΟζχʔϥϯυͰߦΘΕΔΜͰ͕͢ େձޙʹߦͬͨຊͷΞόλʔͷΞτϥΫγϣϯ͕ΊͬͪΌ໘ന͔ͬͨʜ
ֶੜ࣌
ҰਓশߦಈೝࣝͷͨΊͷ ਂϚϧνλεΫֶशͷݚڀ ઍ༿େֶେֶӃ༥߹ཧֶ ֶใՊֶઐ߈ใՊֶίʔε ຊݚڀࣨதଜҏਧ म࢜ൃද
*#6,*/",".63" Ұਓশࢹө૾ w (PPHMF(MBTTͳͲͰࡱӨ͞Εͨରऀ͔Βͷࢹө૾ w ϥΠϑϩάରऀͷߦಈੳͱͯ͠ͷ׆༻ʹظ Damen, Dima, et
al. "Scaling Egocentric Vision: The EPIC-KITCHENS Dataset." arXiv preprint arXiv:1804.02748 (2018).
*#6,*/",".63" ݚڀഎܠ Ұਓশߦಈೝࣝʹ͓͚Δਂֶशͷ ΫϥεͷΈ߹Θͤരൃ σʔληοτͷෆ ϚϧνλεΫֶश
*#6,*/",".63" ຊݚڀͰఏҊ͢ΔΞʔΩςΫνϟ தؒͰͷ༥߹ʹΑΔਂಛྔͷ౷߹ ମͱಈ࡞ͷग़ྗͷϚϧνλεΫֶश छྨͷϞσϧ͕λεΫد༩͢ΔεΩοϓߏ
*#6,*/",".63" ೝࣝ݁Ռ
ݱࡏͷۀ
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE χϑςΟχϡʔε ΞάϦήʔγϣϯܕχϡʔεϝσΟΞͷ։ൃ 1$ εϚϗ ΞϓϦ
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE ৽γεςϜͷҠߦ
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE ϚΠχϑςΟ !OJGUZͷαʔϏεΛͬͱۙʹշదʹɻ ೝূɺϓογϡ௨ɺαʔϏε"1*ͷूͳͲͷόοΫΤϯυશൠΛ୲
ͪΐͬͱ͚ͩมͳ ੜ͖ํΛհ͠·͢
νΞϦʔσΟϯάຊදʹͳͬͨ w େֶੜ͔ΒνΞϦʔσΟϯά ʹؔΘΓ͡Ίͨ w ࣮உࢠͷνΞϦʔσΟϯά ͱ͍͏ͷ͕ଘࡏ͢Δ w
ʹॴଐνʔϜ͕ຊද ʹͳͬͨ͜ͱͰɺͦͷ࣌ʹબख ͩͬͨࣗຊදʹ w ֶදজͯ͠Βͬͨ w ਤॻΧʔυສԁΒͬͨ
୯७ʹνΞϦʔσΟϯά͕Γ͔ͨͬͨ w νΞϦʔσΟϯά͕ଓ͚ΒΕͦ͏ ͱ͍͏ߟ͚͑ͩͰχϑςΟʹೖࣾ w ฏۉۀ͕࣌ؒ࣌ؒ w ී௨࣌ؒ͘Β͍Β͍͠ wݚڀΛੜ͔͢ͱ͍͏ߟ͑
શ͘ͳ͔ͬͨ wকع"*͕͖Ͱݚڀલ͔Βڵຯ͕ ͋ͬͨͷͰɺझຯͰͰ͖ΔͩΖ͏ ͱࢥ͍ͬͯͨ wʮχϑςΟͰຊʹ͍͍ͷ͔ʁʯ ͱຊઌੜʹݺΕͨ
8&#ΤϯδχΞͷษڧΛΊͬͪΌͨ͠ w %PDLFSͲ͜Ζ͔(JUͷ͍ํ͢ΒΒͳ ͔ͬͨͷͰɺΊͪΌΊͪΌษڧͨ͠ w %PDLFS (JU "84 5FSSBGPSNʜ w
εΫϥϜ։ൃษڧ w ࠷ۙɺೝఆεΫϥϜϚελʔʹͳͬͨ w ࠓҰ௨Γѻ͑ΔΑ͏ʹͳͬͨʢͣʣ w ࠷ۙ"[VSFΛ͑ΔΑ͏ʹͳΓ͍ͨͱࢥͬ ͯɺձࣾʹަবத w ΤϯδχΞͷ৽ਓڭҭʹੵۃతʹೖ͍ͬͯ ͘Α͏ʹ͍ͯ͠Δ w ։ൃΠϯλʔϯͰաڈ࠷ߴͷຬΛग़ͨ͠
Ջͳ࣌ؒʹউखʹػցֶशΛ͍ͬͯͨ wχϑςΟ*41ࣄۀͱ͍͏͜ͱ͋ΓɺܦӦج൫͕൫ੴ w Πϯλʔωοτճઢ͕ಥવܹݮ͢Δͱ͍͏͜ͱͳ͍ w རӹιϑτΣΞࣄۀͱ͔ͯ͠ͳΓߴ͍ wۀ࣌ؒͷΛ͖ͳ͜ͱʹ͍͍ͯͯͱ͍͏੍͕ ࣗͷࣄۀ෦ʹ͋ͬͨ w(PPHMFͷϧʔϧΛਅࣅ੍ͨ͠ɻࠓࣗͷνʔϜͰউखʹ࠾༻͍ͯ͠Δɻ
wػցֶशָͦ͠͏ͱࢥͬͯΒ͑ͦ͏ͳ͜ͱΛͬͨ w͖ͦ͏ͳঁ༏ΛϨίϝϯυͯ͘͠ΕΔαʔϏε wΤϯδχΞ͚ػցֶशݚम wڝഅ"*Λ࡞ͬͯɺ࣮ࡍʹ͓ۚΛ͔͚Δʢෛ͚ͨʣ
ػցֶशεϖγϟϦετʹબΕͨ w χϑςΟͷ/ͱ͍͏੍Ͱɺػցֶशε ϖγϟϦετʹબΕͨ w :BIPPͷࠇଳͱಉ੍͡ w ϓϩμΫτʹػցֶशʹಋೖ͢Δ͜ͱΛ ϛογϣϯʹ׆ಈ w
χϡʔεͷ"*ཁ w ࣾφϨοδͷ$IBU(15Ͱͷݕࡧ w ෆਖ਼Ϣʔβʔݕ w ྑ͍ͯ͘͠Δ"84ͷਓʹʮதଜ͞Μ ػցֶशͷ෦ॺͰ͢ΑͶʁʯͱݴΘΕΔ w ࣮୯ಠͰػցֶशΛ͍ͬͯΔ෦ॺͳ͍
%FWFMPQFST4VNNJUͰ͠·ͨ͠ IUUQTDPEF[JOFKQEFWPOMJOFBSDIJWFTFTTJPO %FWFMPQFST4VNNJUϕετεϐʔΧʔҐެื
݁ہԿΛ͍ͯ͠Δͷʁ wνΞϦʔσΟϯάࢦಋΛଓ͚͍ͯΔ🤸 w νʔϜΛࢦಋ͠ͳ͕Βɺ͜ͷલશࠃେձͷ৹ࠪһͨ͠ w ͕ࣗϓϨΠϠʔΛΔͷલʹҾୀ wχϑςΟχϡʔεͷ։ൃϦʔμʔ݉εΫϥϜϚελʔ💻 w ࠷ߴͷ։ൃڥΛٻΊ͍ͯͨΒ͍ͭͷؒʹ্ཱ͔͕͕͍ͬͯͨ wػցֶशʹ݁ہࠓͰؔΘ͍ͬͯΔ🤖
w"*ྲྀߦͰ͋Δ͕࣮Ͱ͖Δਓ͕গͳ͘ɺ͖͕ߴͯࣾ͡ͰϙδγϣϯΛऔΕͨ wࠓ$IBU(15ͰاۀσʔλΛѻ͏ͨΊͷ3"(ύλʔϯΛࢥҊத
͍͖͍͖ͱͨ͠νʔϜΛͭ͘Δ wݩʑମૢڝٕΛ͍ͬͯͯɺࣗڀۃͷݸਓओٛͩͬͨ🤸 w ଞਓʹཔΒͣҰਓͰಥ͖ਐΈɺڠྗཧղΛಘΒΕͳ͍͜ͱଟ͔ͬͨ w ݸਓతͳܦݧͱͯ͠ ࣗউखͳਓपΓʹଟ͘ɺෆຬΛײ͡Δ͜ͱଟʑ͋ͬͨ wνΞϦʔσΟϯάͱ͍͏ڀۃͷूஂεϙʔπʹग़ձͬͨ📣 w ୭͕͚ܽͯԋཱٕ͕͠ͳ͍ͱ͍͏ҙຯͰɺਅͷूஂεϙʔπͩͱߟ͍͑ͯΔ
w ଞऀΛԠԉࣗ͠Ԡԉ͞ΕΔͱ͍͏ʮνΞεϐϦοτʯͷਫ਼ਆΛΊ͍ͨͱࢥ͏Α͏ʹͳͬͨ wʮεΫϥϜʯʹΑΔ͍͖͍͖ͱͨ͠νʔϜΛ࡞Δ🏉 w ͕ࣗେ͖ͳΤϯδχΞϦϯάͷੈքͰྑ͍ूஂɾྑ͍νʔϜΛ࡞Γ͍ͨͱࢥ͏Α͏ʹ w ͠ੲͷࣗͷΑ͏ʹνʔϜϫʔΫ͕ඞཁͳ͍ͱײ͍ͯ͡Δਓ͕͍ͨͱͯ͠ɺ ਓͱڠྗͨ͠ΓԿ͔Λୡ͢Δָ͠͞ັྗΛײͯ͡ཉ͍͠ͱ͍͏ؾ࣋ͪ͋Δ
ͦΜͳੜ͖ํ͔Βͷ ब׆ΞυόΠε
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE ίϛϡχέʔγϣϯΛڪΕͳ͍ اۀͷจԽΛഽͰΔ ษڧձͳͲʹࢀՃͯ͠ɺࣾʹͲ͏͍͏ਓ͕͍Δͷ͔Λͬͯ΄͍͠ ԿΛ͍ͬͯΔ͔ΑΓձࣾͷจԽɾงғؾ͕ࣗʹ߹͍ͬͯΔ͔ͷํ͕େࣄ BXTXBLBSBO5PLZP ϞμϯϑϩϯτΤϯυษڧձ σʔλੳج൫%FWFMPQFST/JHIU
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE ຊʹΓ͍ͨ͜ͱΛߟ͑Δ ͕ࣗେࣄʹ͍ͨ͜͠ͱΛ͑ͯ͘ΕΔձࣾʹߦ͜͏ ࣗࣾձਓʹͳ͔ͬͯΒνΞϦʔσΟϯάʹܞΘΓ͔ͨͬͨ ϥΠϑɾϫʔΫɾόϥϯεΛҰ൪ʹߟ͍͑ͯΔχϑςΟ͕߹͍ͬͯͨ
$PQZSJHIU/*'5:$PSQPSBUJPO"MM3JHIUT3FTFSWFE $POOFDUUIFEPUT ͱ͕ͭͳ͕Δ͜ͱΛ৴͡Δ ࠓͰ͖Δ͜ͱɾࠓ͍ͬͯΔ͜ͱ͖ͬͱʹཱͭ .*36 ݚڀܴࣨձʢ݄ʣ ւಓͷϏΞΨʔσϯͰઌੜՏͱ