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
『IT Text自然言語処理』第1章
Search
kassy11
April 17, 2024
0
6
『IT Text自然言語処理』第1章
kassy11
April 17, 2024
Tweet
Share
More Decks by kassy11
See All by kassy11
Git入門
kassy11
0
5
BERTでのテキスト分類
kassy11
0
4
過学習と未学習について
kassy11
0
5
LLMアプリケーション開発 ~シンとくぽんTalkと論文要約くんの開発について~
kassy11
0
18
Featured
See All Featured
CoffeeScript is Beautiful & I Never Want to Write Plain JavaScript Again
sstephenson
155
14k
Raft: Consensus for Rubyists
vanstee
133
6.3k
Put a Button on it: Removing Barriers to Going Fast.
kastner
58
3.1k
Let's Do A Bunch of Simple Stuff to Make Websites Faster
chriscoyier
501
140k
[Rails World 2023 - Day 1 Closing Keynote] - The Magic of Rails
eileencodes
9
1.3k
Fantastic passwords and where to find them - at NoRuKo
philnash
39
2.5k
Pencils Down: Stop Designing & Start Developing
hursman
117
11k
Visualization
eitanlees
137
14k
Code Review Best Practice
trishagee
56
15k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
12
1k
Statistics for Hackers
jakevdp
790
220k
A designer walks into a library…
pauljervisheath
201
23k
Transcript
"ݚڀࣨ.ദݪޭଠ ୈষࣗવݴޠॲཧͷ֓ཁ ʰ*55FYUࣗવݴޠॲཧͷجૅʱษڧձ
ࣗવݴޠॲཧͱ • ίϯϐϡʔλ͕ਓؒͷΑ͏ʹݴ༿ΛૢΕΔΑ͏ʹͳΔ͜ͱΛࢦ͢ݚڀ • ίϯϐϡʔλɾαΠΤϯεͷͰɺ • ʢਓʣݴޠɿϓϩάϥϛϯάݴޠʢ1ZUIPOͳͲʣͷ͜ͱ • ࣗવݴޠɿਓؒͷݴ༿ʢຊޠɺӳޠʣͷ͜ͱ
• ˞୯ʹʮݴޠॲཧʯͩͱɺϓϩάϥϛϯάݴޠΛॲཧ͢ΔݚڀʢίϯύΠϥ ʹ͍ͭͯͳͲʣʹͳΔͷͰҙ • ݴޠॲཧֶձʢࣗવݴޠॲཧͷֶձʣྫ֎😅
ࣗવݴޠॲཧͷԠ༻ • ػց༁ • ͔ͳࣈม • ใݕࡧ • ࣭ԠɿใݕࡧͱࣅͯΔ͕ɺϢʔβͷෳࡶɾᐆດͳ͍͔͚ʹରԠ͢Δ
• Α͘Θ͔Βͳ͍͜ͱΛฉ͖ฦ͢ͳͲ • ରγεςϜɿίϯϐϡʔλͱͷձʢࡶஊPSλεΫࢦʣ • ࣭ԠରγεςϜͷҰछͱߟ͑ΒΕΔ • ࣗಈཁ • ࣗવݴޠੜɿݴ༿Λੜ͢Δٕज़શൠʢ࣭ԠରγεςϜͳͲ͜ͷҰछʣ • จ๏ޡΓగਖ਼
ςΩετϚΠχϯάɾςΩετੳ • ςΩετΛੳͯ͠༗༻ͳݟΛಘΔɺࣗવݴޠॲཧͷҰ • ࣗવݴޠॲཧͷͳ͔Ͱσʔλͷੳʹॏ͖Λ͓͍ͨ • Ԡ༻ྫ • Ωʔϫʔυநग़ɿେࣄͳΩʔϫʔυΛൈ͖ग़͢
• ใநग़ • จॻྨɿจॻΛΧςΰϦ͝ͱʹྨ • จॻΫϥελϦϯάɿจষΛΫϥελʹࣗಈͰ͚Δ • จষྨͱҟͳΔͷओʹڭࢣͳ͠Ͱߦ͏ͱ͜Ζ
ίʔύεͱ • ԿΒ͔ͷతʹؔͯ͠ͷςΩετΛେྔʹऩूɾՃͨ͠ͷ • ԿͷͨΊʹ͏͔ʁ • ࣗવݴޠॲཧ"*ʹݴ༿Λֶशͤ͞ΔͨΊ • ςΩετϚΠχϯάΛߦͬͯɺूΊͨςΩετͷΛੳ͢Δ
• ݴޠֶʢจܥʣͷݚڀͷͨΊ • ࣌ͷҠΓมΘΓͰͲ͏ݴ༿͕มΘΔ͔ͳͲ
ίʔύεͷछྨ • ੜίʔύε • ςΩετΛͦͷ··ूΊ͖ͯͯΔ͚ͩ • (15େྔͷੜίʔύε͔Βࣗݾڭࢣ͋Γֶश • ऍ͖ʢλά͖ɺϥϕϧ͖ʣίʔύε
• ςΩετΛूΊͯɺͦΕΒͷࢺͳͲͷऍΛ͚ͭͨίʔύε • จ๏Λཧղ͢Δ"*ͷͨΊʹॏཁ • ଟݴޠɾର༁ʢύϥϨϧɺίϯύϥϒϧʣίʔύε • ಉ͡༰͕ͩෳͷݴޠͰॻ͔ΕͨจΛऩूʢӳͳͲʣ • ػց༁ͷͨΊʹ༻͍ΒΕΔ
Ξϊςʔγϣϯ • ԿΒ͔ͷσʔλʢςΩετʹݶΒͳ͍ʣʹ͍ͭͯऍΛ͚ͭΔ͜ͱ • ࡞ۀ͢ΔਓΛΞϊςʔλͱݺͿ • ऍ͖ίʔύεΛ࡞͢ΔࡍʹɺΞϊςʔλʹΞϊςʔγϣϯΛͯ͠Β͏ • ࡞ۀͷࡍʹɺ࣭Λ্͛ΔͨΊʹΞϊςʔλͷΨΠυϥΠϯΛ࡞͢Δඞཁ
͕͋Δ ࢀߟɿIUUQTUFDICMPHBCFKBBTJBFOUSZNMBOOPUBUJPO
ίʔύεʹର͢Δ౷ܭతੳ • ίʔύεதͷςΩετͰॏཁͳ୯ޠΛ୳͍ͨ͠ʢॏཁΛܭࢉ͍ͨ͠ʣ • ୯७ʹ୯ޠͷग़ݱසΛߟ͑ͯΈΔ • ୯ޠͷग़ݱසΛ୯७ʹूܭ͢Δͱʮʯʮ͔͠͠ʯͳͲͷػೳޠʢॏཁͰͳ͍ʣ্͕Ґ ʹདྷͯ͠·͏🤔 •
ػೳޠɿ୯ޠͷҙຯΑΓจ๏తͳׂΛڧࣔ͘͢୯ޠʢલஔࢺɺଓࢺͳͲʣ • ༰ޠ͚ͩʹݶఆͯ͠ग़ݱසΛूܭͯ͠ɺʮࢲʯʮਓʯʮ͜ͱʯͷΑ͏ͳͲΜͳจষʹ ग़͖ͯͦ͏ʢॏཁͰͳ͍ʣͳ୯ޠ্͕Ґʹདྷͯ͠·͏🤔 • ༰ޠɿҙຯ༰Λڧ͘ද͢୯ޠʢ໊ࢺɺಈࢺͳͲʣ • Ͳ͏ͬͯ୯ޠͷॏཁΛܭࢉ͢Εྑ͍ͷ͔ʁ㱺5'*%'
5'*%' • ʮग़ݱස͕ߴ͍ଞͷςΩετͰ͋·Γग़ݱ͠ͳ͍୯ޠʯ͕ॏཁ͕ߴ͍ • 5'ʢ5FSN'SFRVFODZʣɿ͋Δ୯ޠͷग़ݱස • %'ʢ%PDVNFOU'SFRVFODZʣɿ͋Δ୯ޠ͕ग़ݱ͢ΔςΩετͷ • *%'ʢ*OWFSTF%'ʣɿ%'ͷٯͷMPHΛऔͬͨͷ
• 5'ͱ*%'Λ͔͚ΘͤͨͰ୯ޠͷॏཁΛܭࢉ͢Δ
δοϓͷ๏ଇ • ςΩετͰ͘͝গͷ୯ޠ͕සൟʹ༻͍ΒΕɺͦΕҎ֎ͷ୯ޠ͋·Γ༻͍Β Εͳ͍ • ͋ΒΏΔݴޠʹ͍ͭͯzܦݧతʹzΓཱͭ • ཧతͳূ໌͞Ε͍ͯͳ͍
ࣗવݴޠॲཧͷ͠͞ • ࣗવݴޠͰͷςΩετΛཧղ͢Δͱ͖ɺͦͷݴޠͷจࣈɾ୯ޠɾจ๏ʹ͍ͭͯͷ ͕ࣝඞཁ • ଟ͘ͷ04จࣈίʔυ6OJDPEFʹରԠ͍ͯ͠ΔͷͰจࣈʹ͍ͭͯ౷Ұతʹѻ ͑Δ • ୯ޠจ๏ʹ͍ͭͯͲ͏ѻ͑ྑ͍ͷ͔ʁ
• ओʹͭͷํ๏͕͋Δ㱺ܗଶૉղੳɾߏจղੳɾҙຯղੳ
ܗଶૉղੳ • ͋ΔςΩετΛʢจ๏తʹଥʹʣ୯ޠʢܗଶૉʣʹ͚Δ͜ͱʢ͔ͪॻ͖ʣ • ӳޠεϖʔεͰׂ͢Ε͍͍͕ɺຊޠͰඞཁ • ͚ͨޙͷ୯ޠͷࢺΛਪఆ͢Δ͜ͱؚ·ΕΔ߹͕͋Δ • ҙຯͷऔΓํʹΑͬͯɺ͚ͨ݁Ռ͕ෳଘࡏ͢Δ߹͕͋Δʢᐆດੑʣ
• ྫʣ֎ࠃਓͷࢀݖPS֎ࠃʹ͋Δਓࢀͷݖ
ߏจղੳʢΓड͚ղੳʣ • จઅͱจઅͷम০ؔʢΓड͚ʣΛղੳ͢Δ͜ͱ • ҙຯͷऔΓํʹΑͬͯɺෳͷΓड͚ߏ͕Ͱ͖Δ߹͋Δʢᐆດੑʣ • ྫʣ৽ฉΛͬͯ૭Λআͨ͠PS৽ฉ͕͍ͬͺ͍͍ͯԚΕͨ૭
ҙຯղੳʢड़ޠ߲ߏղੳʣ • ςΩετ͔Βʮ୭͕ɾ͍ͭɾͲ͜ͰɾԿΛͨ͠ʯΛऔΓग़͢ॲཧ • ςΩετதͰ໊ࢺ͕লུ͞ΕΔθϩ໊ࢺ͕͋Δͱҙຯͷཧղ͕͘͠ͳΔ • ྫʣʮ͓͍͡͞ΜർΕ·ͨ͠ɻ͕ͯͬͯ͠·͍·ͨ͠ɻʯ • θϩ໊ࢺΛಛఆ͢Δ͜ͱΛθϩরԠղੳͱݺͿ
• ݻ༗දݱ͕͋Δ߹ҙຯͷཧղ͕͘͠ͳΔ • ྫʣੁट૬ೋਓଘࡏ͢ΔͷͰ • ݻ༗දݱ͕ࢦ࣮͢ଶΛಛఆ͢Δ͜ͱΛ ΤϯςΟςΟϦϯΩϯάͱݺͿ
·ͱΊ ࣗવݴޠॲཧͱͦͷԠ༻ ίʔύεͱͦͷछྨ ςΩετதͷ୯ޠͷॏཁΛܭࢉ͢Δ5'*%' ܗଶૉղੳɾߏจղੳɾҙຯղੳͷ֓ཁ