SlideShare a Scribd company logo
1 of 10
Download to read offline
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
論⽂紹介
村上
SI部 第1サービス開発担当
⾃然⾔語処理勉強会(2020/05/22)
※ACL2020に採択された論⽂(Short paper)
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
⽂法誤り訂正(Grammatical Error Correction; GEC)とは
2
• ⽂法誤り訂正(GEC)とは、⽂法的な誤りを含む⽂を, 原⽂の意味を保ちつつ正しい⽂に変換するタスクである
• ⽂法誤り訂正を誤り⽂から訂正⽂を”翻訳”する⽣成タスクとして考える
2
I plays baseball yestarday . I played baseball yesterday .
Transformer
(誤り⽂ → 訂正⽂)
誤り⽂から訂正⽂を⽣成
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
論⽂を3⾏でまとめると
• ⽂法誤り訂正(GEC)モデルの学習において、どのようにMasked Language Model(MLM)を使うと性能向
上に寄与するか検証した論⽂(MLMとしてBERTを採⽤)
• ⽂法誤り検出(GED)タスクでファインチューニングしたBERTの出⼒ベクトルを追加特徴量として⼊⼒する
Encoder-Decoderモデル(BERT-fuse GED)が最も精度向上に寄与した。
• 提案⼿法のうち最も精度が⾼いモデルが2つのベンチマーク(CoNLL2014, BEA2019)でSOTAを達成。
3
⽂法誤り訂正(GEC) ⽂法誤り検出(GED) Masked Language Model
⽂法誤りを訂正するタスク ⽂法誤りを検出するタスク
I go to school by bus.
I go to school at bus.
マスクされた単語を予測する
⾔語モデル。BERTの事前学
習タスクとして⽤いられる。
I go to school at bus .
0 0 0 0 1 0 0
マスクマスク
BERT論⽂より引⽤ https://arxiv.org/pdf/1810.04805.pdf
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
提案⼿法︓事前学習済みMLM(BERT)の導⼊
• BERT-init
• Lample(2019)らに提案された事前学習済みBERTの重みパラメータを⽤いてTransformer-
based Encoder-Decoderモデルを初期化する⽅法。(詳細は割愛)
• BERT-fuse
• Zhu(2020)らに提案されたTransformer-based Encoder-Decoderモデルに事前学習済
みBERTの出⼒ベクトルに対するアテンションを導⼊する⼿法
• こちらの⼿法ほうが性能が良かった。詳細を次ページ以降で説明
4
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
BERT-fused Model
• Zhu et al., “Incorporating BERT into Neural Machine Translation”, ICLR2020
• Neural Machine Translationモデルに効果的にBERTを導⼊する⼿法を提案した論⽂
5
BERT-Enc Attention
(Zhu et al., 2020より引⽤)
Self Attention
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
Drop-Net trick (Zhu et al., 2020)
• モデル学習における正則化(過学習の回避)の⽬的で、Drop-Net Trickを導⼊
• 学習時には「BERT-Enc Attention」、「Self-Attention」、または、両⽅を使うか、をランダムに選択する
6
Drop-Netを⽤いない場合(p=0.0)、過学習に陥る
𝑈!は0〜1の連続値. ランダムに決定することで、Drop-Netを実現.
(Zhu et al., 2020より引⽤)
(Zhu et al., 2020より引⽤)
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
提案⼿法︓導⼊するBERTのチューニング
• BERT-fuse mask
• ⽂法誤りを含むコーパスを⽤いてBERTを追加学習(MLM, NSPタスク)する⽅法
• BERT-fuse GED
• BERTを⽂法誤り検出モデルとしてファインチューニングする⽅法
• ※⽂法誤り検出
• ⼊⼒テキストの⽂法誤り箇所を検出する系列ラベリングタスク
7
BERTは正しい⽂(Wiki, BookCorpus)で
学習されているので、⽂法的誤りを認識で
きるように予めチューニングをしたほうが
良いのでは?
I go to school at bus .
0 0 0 0 1 0 0
⼊⼒テキスト
ラベル
仮説
(Zhu et al., 2020より引⽤)
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
実験結果
• w/o BERT vs BERT → BERTの導⼊により、精度向上に寄与することが分かった
• BERT-init vs BERT-fuse → BERT-fuseの⽅が精度⾼い(Zhu et al, 2020でも同様の結果が出ている)
• BERT-fuse mask vs GED → ⽂法誤り検出(GED)でBERTをファインチューニングすると更に良くなった
8
• 疑似データを⽤いた事前学習(○)
• モデルアンサンブル(○)
• 疑似データを⽤いた事前学習(○)
• モデルアンサンブル(✕)
• 疑似データを⽤いた事前学習(✕)
• モデルアンサンブル(✕)
精度向上テクニック
論⽂より引⽤
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
分析
• GEDでBERTをファインチューニングすることで精度は上がったが、はたしてBERTは⽂法誤りを
認識することができているのか︖
• 検証
• 8つの単語が正しくまたは間違って使われていることを区別できているかを確認
• “the”, “,”, “in”, “to”, “of”, “a”, “for”, “is”
• 結果
• ⽣BERTは正しい⽤法で使われた場合と間違いを区別できていないが、
• ファインチューニング済みBERTは正しいものと誤りのものを区別できている
9論⽂より引⽤
ⓒ2019 NTT DOCOMO, INC. All Rights Reserved.
まとめ
10
• やったこと
• ⽂法誤り訂正(GEC)モデルの学習において、どのようにMasked Language Model(MLM)を使
うと性能向上に寄与するか検証した論⽂(MLMとしてBERTを採⽤)
• 貢献したこと
• ⽂法誤り検出(GED)タスクでファインチューニングしたBERTの出⼒ベクトルを追加特徴量として⼊⼒
するEncoder-Decoderモデル(BERT-fuse GED)が最も精度向上に寄与した。
• 提案⼿法のうち最も精度が⾼いモデルが2つのベンチマーク(CoNLL2014, BEA2019)でSOTA
を達成。
エラータイプごとの精度についても
精度向上していることを確認した
GEDタスクのファインチューニングにより
⽂法誤りの認識性能が向上することが分かった
論⽂より引⽤

More Related Content

Recently uploaded

How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessPixlogix Infotech
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUK Journal
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)Gabriella Davis
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Drew Madelung
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?Igalia
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024Results
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking MenDelhi Call girls
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountPuma Security, LLC
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfEnterprise Knowledge
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
Factors to Consider When Choosing Accounts Payable Services Providers.pptx
Factors to Consider When Choosing Accounts Payable Services Providers.pptxFactors to Consider When Choosing Accounts Payable Services Providers.pptx
Factors to Consider When Choosing Accounts Payable Services Providers.pptxKatpro Technologies
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Enterprise Knowledge
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsJoaquim Jorge
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 

Recently uploaded (20)

How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your Business
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path Mount
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
Factors to Consider When Choosing Accounts Payable Services Providers.pptx
Factors to Consider When Choosing Accounts Payable Services Providers.pptxFactors to Consider When Choosing Accounts Payable Services Providers.pptx
Factors to Consider When Choosing Accounts Payable Services Providers.pptx
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 

Featured

Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsPixeldarts
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthThinkNow
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfmarketingartwork
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024Neil Kimberley
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)contently
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024Albert Qian
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsKurio // The Social Media Age(ncy)
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Search Engine Journal
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summarySpeakerHub
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next Tessa Mero
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentLily Ray
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best PracticesVit Horky
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project managementMindGenius
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...RachelPearson36
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Applitools
 
12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at WorkGetSmarter
 

Featured (20)

Product Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage EngineeringsProduct Design Trends in 2024 | Teenage Engineerings
Product Design Trends in 2024 | Teenage Engineerings
 
How Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental HealthHow Race, Age and Gender Shape Attitudes Towards Mental Health
How Race, Age and Gender Shape Attitudes Towards Mental Health
 
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdfAI Trends in Creative Operations 2024 by Artwork Flow.pdf
AI Trends in Creative Operations 2024 by Artwork Flow.pdf
 
Skeleton Culture Code
Skeleton Culture CodeSkeleton Culture Code
Skeleton Culture Code
 
PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024PEPSICO Presentation to CAGNY Conference Feb 2024
PEPSICO Presentation to CAGNY Conference Feb 2024
 
Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)Content Methodology: A Best Practices Report (Webinar)
Content Methodology: A Best Practices Report (Webinar)
 
How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024How to Prepare For a Successful Job Search for 2024
How to Prepare For a Successful Job Search for 2024
 
Social Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie InsightsSocial Media Marketing Trends 2024 // The Global Indie Insights
Social Media Marketing Trends 2024 // The Global Indie Insights
 
Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024Trends In Paid Search: Navigating The Digital Landscape In 2024
Trends In Paid Search: Navigating The Digital Landscape In 2024
 
5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary5 Public speaking tips from TED - Visualized summary
5 Public speaking tips from TED - Visualized summary
 
ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd ChatGPT and the Future of Work - Clark Boyd
ChatGPT and the Future of Work - Clark Boyd
 
Getting into the tech field. what next
Getting into the tech field. what next Getting into the tech field. what next
Getting into the tech field. what next
 
Google's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search IntentGoogle's Just Not That Into You: Understanding Core Updates & Search Intent
Google's Just Not That Into You: Understanding Core Updates & Search Intent
 
How to have difficult conversations
How to have difficult conversations How to have difficult conversations
How to have difficult conversations
 
Introduction to Data Science
Introduction to Data ScienceIntroduction to Data Science
Introduction to Data Science
 
Time Management & Productivity - Best Practices
Time Management & Productivity -  Best PracticesTime Management & Productivity -  Best Practices
Time Management & Productivity - Best Practices
 
The six step guide to practical project management
The six step guide to practical project managementThe six step guide to practical project management
The six step guide to practical project management
 
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
Beginners Guide to TikTok for Search - Rachel Pearson - We are Tilt __ Bright...
 
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
Unlocking the Power of ChatGPT and AI in Testing - A Real-World Look, present...
 
12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work12 Ways to Increase Your Influence at Work
12 Ways to Increase Your Influence at Work
 

【論文紹介】Encoder-Decoder Models Can Benefit from Pre-trained Masked Language Models in Grammatical Error Correction.

  • 1. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 論⽂紹介 村上 SI部 第1サービス開発担当 ⾃然⾔語処理勉強会(2020/05/22) ※ACL2020に採択された論⽂(Short paper)
  • 2. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. ⽂法誤り訂正(Grammatical Error Correction; GEC)とは 2 • ⽂法誤り訂正(GEC)とは、⽂法的な誤りを含む⽂を, 原⽂の意味を保ちつつ正しい⽂に変換するタスクである • ⽂法誤り訂正を誤り⽂から訂正⽂を”翻訳”する⽣成タスクとして考える 2 I plays baseball yestarday . I played baseball yesterday . Transformer (誤り⽂ → 訂正⽂) 誤り⽂から訂正⽂を⽣成
  • 3. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 論⽂を3⾏でまとめると • ⽂法誤り訂正(GEC)モデルの学習において、どのようにMasked Language Model(MLM)を使うと性能向 上に寄与するか検証した論⽂(MLMとしてBERTを採⽤) • ⽂法誤り検出(GED)タスクでファインチューニングしたBERTの出⼒ベクトルを追加特徴量として⼊⼒する Encoder-Decoderモデル(BERT-fuse GED)が最も精度向上に寄与した。 • 提案⼿法のうち最も精度が⾼いモデルが2つのベンチマーク(CoNLL2014, BEA2019)でSOTAを達成。 3 ⽂法誤り訂正(GEC) ⽂法誤り検出(GED) Masked Language Model ⽂法誤りを訂正するタスク ⽂法誤りを検出するタスク I go to school by bus. I go to school at bus. マスクされた単語を予測する ⾔語モデル。BERTの事前学 習タスクとして⽤いられる。 I go to school at bus . 0 0 0 0 1 0 0 マスクマスク BERT論⽂より引⽤ https://arxiv.org/pdf/1810.04805.pdf
  • 4. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 提案⼿法︓事前学習済みMLM(BERT)の導⼊ • BERT-init • Lample(2019)らに提案された事前学習済みBERTの重みパラメータを⽤いてTransformer- based Encoder-Decoderモデルを初期化する⽅法。(詳細は割愛) • BERT-fuse • Zhu(2020)らに提案されたTransformer-based Encoder-Decoderモデルに事前学習済 みBERTの出⼒ベクトルに対するアテンションを導⼊する⼿法 • こちらの⼿法ほうが性能が良かった。詳細を次ページ以降で説明 4
  • 5. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. BERT-fused Model • Zhu et al., “Incorporating BERT into Neural Machine Translation”, ICLR2020 • Neural Machine Translationモデルに効果的にBERTを導⼊する⼿法を提案した論⽂ 5 BERT-Enc Attention (Zhu et al., 2020より引⽤) Self Attention
  • 6. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. Drop-Net trick (Zhu et al., 2020) • モデル学習における正則化(過学習の回避)の⽬的で、Drop-Net Trickを導⼊ • 学習時には「BERT-Enc Attention」、「Self-Attention」、または、両⽅を使うか、をランダムに選択する 6 Drop-Netを⽤いない場合(p=0.0)、過学習に陥る 𝑈!は0〜1の連続値. ランダムに決定することで、Drop-Netを実現. (Zhu et al., 2020より引⽤) (Zhu et al., 2020より引⽤)
  • 7. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 提案⼿法︓導⼊するBERTのチューニング • BERT-fuse mask • ⽂法誤りを含むコーパスを⽤いてBERTを追加学習(MLM, NSPタスク)する⽅法 • BERT-fuse GED • BERTを⽂法誤り検出モデルとしてファインチューニングする⽅法 • ※⽂法誤り検出 • ⼊⼒テキストの⽂法誤り箇所を検出する系列ラベリングタスク 7 BERTは正しい⽂(Wiki, BookCorpus)で 学習されているので、⽂法的誤りを認識で きるように予めチューニングをしたほうが 良いのでは? I go to school at bus . 0 0 0 0 1 0 0 ⼊⼒テキスト ラベル 仮説 (Zhu et al., 2020より引⽤)
  • 8. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 実験結果 • w/o BERT vs BERT → BERTの導⼊により、精度向上に寄与することが分かった • BERT-init vs BERT-fuse → BERT-fuseの⽅が精度⾼い(Zhu et al, 2020でも同様の結果が出ている) • BERT-fuse mask vs GED → ⽂法誤り検出(GED)でBERTをファインチューニングすると更に良くなった 8 • 疑似データを⽤いた事前学習(○) • モデルアンサンブル(○) • 疑似データを⽤いた事前学習(○) • モデルアンサンブル(✕) • 疑似データを⽤いた事前学習(✕) • モデルアンサンブル(✕) 精度向上テクニック 論⽂より引⽤
  • 9. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. 分析 • GEDでBERTをファインチューニングすることで精度は上がったが、はたしてBERTは⽂法誤りを 認識することができているのか︖ • 検証 • 8つの単語が正しくまたは間違って使われていることを区別できているかを確認 • “the”, “,”, “in”, “to”, “of”, “a”, “for”, “is” • 結果 • ⽣BERTは正しい⽤法で使われた場合と間違いを区別できていないが、 • ファインチューニング済みBERTは正しいものと誤りのものを区別できている 9論⽂より引⽤
  • 10. ⓒ2019 NTT DOCOMO, INC. All Rights Reserved. まとめ 10 • やったこと • ⽂法誤り訂正(GEC)モデルの学習において、どのようにMasked Language Model(MLM)を使 うと性能向上に寄与するか検証した論⽂(MLMとしてBERTを採⽤) • 貢献したこと • ⽂法誤り検出(GED)タスクでファインチューニングしたBERTの出⼒ベクトルを追加特徴量として⼊⼒ するEncoder-Decoderモデル(BERT-fuse GED)が最も精度向上に寄与した。 • 提案⼿法のうち最も精度が⾼いモデルが2つのベンチマーク(CoNLL2014, BEA2019)でSOTA を達成。 エラータイプごとの精度についても 精度向上していることを確認した GEDタスクのファインチューニングにより ⽂法誤りの認識性能が向上することが分かった 論⽂より引⽤