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
Using millions of emoji occurrences to learn an...
Search
Yuto Kamiwaki
December 16, 2018
Research
130
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Using millions of emoji occurrences to learn any-domain representations for detecting sentiment, emotion and sarcasm
2018/12/17 文献紹介の発表内容
Yuto Kamiwaki
December 16, 2018
More Decks by Yuto Kamiwaki
See All by Yuto Kamiwaki
Emo2Vec: Learning Generalized Emotion Representation by Multi-task Training
yuto_kamiwaki
0
120
Modeling Naive Psychology of Characters in Simple Commonsense Stories
yuto_kamiwaki
1
220
Epita at SemEval-2018 Task 1: Sentiment Analysis Using Transfer Learning Approach
yuto_kamiwaki
0
140
Tensor Fusion Network for Multimodal Sentiment Analysis
yuto_kamiwaki
0
290
Sentiment Analysis: It’s Complicated!
yuto_kamiwaki
0
94
ADAPT at IJCNLP-2017 Task 4: A Multinomial Naive Bayes Classification Approach for Customer Feedback Analysis task
yuto_kamiwaki
0
180
EmoWordNet: Automatic Expansion of Emotion Lexicon Using English WordNet
yuto_kamiwaki
0
120
ATTENTION-BASED LSTM FOR PSYCHOLOGICAL STRESS DETECTION FROM SPOKEN LANGUAGE USING DISTANT SUPERVISION
yuto_kamiwaki
0
160
BB_twtr at SemEval-2017 Task 4: Twitter Sentiment Analysis with CNNs and LSTMs
yuto_kamiwaki
0
260
Other Decks in Research
See All in Research
[CV勉強会@関東 CVPR2026] PSDesigner: Automated Graphic Design with a Human-Like Creative Workflow / kantocv 67th CVPR 2026
shunk031
0
290
MM-OVSeg: Multimodal Optical–SAR Fusion for Open-Vocabulary Segmentation in Remote Sensing
satai
3
120
Claude Code × autoresearch 実践
mathbullet
0
270
PHTalks Bengaluru - SSRF When All Else Fails
dk999
0
1.1k
Karkada さんの論文 × 2 の紹介: (1) Closed-Form Training Dynamics Reveal Learned Features and Linear Structure in Word2Vec-like Models, (2) Symmetry in language statistics shapes the geometry of model representations
eumesy
PRO
1
660
20260624 NLP colloquium: 単一のhubテキストがCLIPを壊す:hubnessによる埋め込みの脆弱性特定
de9uch1
2
230
Language and AI
ayaniwa
0
220
SOTAのさらに先へ:厳しい推論制約下での高性能モデルのPost-Training
analokmaus
0
1.5k
SAKURAONE:An Open Ethernet-based AI HPC System And Its Observed Workload Dynamicsin a Single-Tenant LLM Development Environment
yuukit
1
580
typst の使い方:言語学を研究する学生のために
gitomochang
0
570
LLM の Attention 機構まとめ — 数式・計算量・メモリ
puwaer
8
2.6k
量子サマースクール2026「量子計算機アーキテクチャ分野の概観」
youten622
1
860
Featured
See All Featured
Bash Introduction
62gerente
615
220k
Gemini Prompt Engineering: Practical Techniques for Tangible AI Outcomes
mfonobong
2
510
Improving Core Web Vitals using Speculation Rules API
sergeychernyshev
21
1.6k
The Pragmatic Product Professional
lauravandoore
37
7.4k
How Software Deployment tools have changed in the past 20 years
geshan
1
34k
Designing for Performance
lara
611
70k
Put a Button on it: Removing Barriers to Going Fast.
kastner
60
4.6k
The SEO Collaboration Effect
kristinabergwall1
1
550
How to optimise 3,500 product descriptions for ecommerce in one day using ChatGPT
katarinadahlin
PRO
2
3.8k
Leveraging Curiosity to Care for An Aging Population
cassininazir
1
490
Typedesign – Prime Four
hannesfritz
42
3.2k
Designing for Timeless Needs
cassininazir
1
470
Transcript
Using millions of emoji occurrences to learn any-domain representations for
detecting sentiment, emotion and sarcasm Nagaoka University of Technology Yuto Kamiwaki Literature Review
Literature • Using millions of emoji occurrences to learn any-domain
representations for detecting sentiment, emotion and sarcasm • Bjarke Felbo, Alan Mislove, Anders Søgaard, Iyad Rahwan, Sune Lehmann • EMNLP 2017 2
Abstract • sentiment analysis, emotion analysis and sarcasm classificationにおける8つのbenchmarkでSoTA達成 •
感情ラベルの多様性が以前のdistant supervisonのアプ ローチよりもパフォーマンスの向上をもたらすことを確認 3
Introduction • NLPのタスクでは,アノテーション済み(感情が付与された)の データは少ない. • Distant supervisionを用いてSoTAを達成している研究があ る. Distant supervision
: (http://web.stanford.edu/~jurafsky/mintz.pdf) ラベル付きデータの情報を手がかりに全く別のラベルなしデータからラベル付きの学 習データを生成し、モデルを学習する手法 4
Related work • Ekman, Plutchikなどの感情の理論を用いて手作業によって 分類 ◦ 感情の理解が難しく,時間がかかる. • official
emoji tables (Eisner et al., 2016)からembeddingす る手法 ◦ emojiの使われ方を考慮しない. • マルチタスク学習 ◦ データストレージの観点から問題あり. 5
Pretraining • 2013年1月から2017年6月までのTweet data(emojiあり) • Only English tweets without URL’s
are used for the pretraining dataset. • All tweets are tokenized on a word-by-word basis. 6
Model 7
Transfer Learning(ChainThaw) 8
Emoji Prediction 9
Benchmarking 10 8 Benchmarks(3tasks,5domains)
Benchmarking 11
Importance of emoji diversity 12 Pos/Neg Emoji:8 types DeepMoji:64 types
感情ラベルの多様性が重要 64種類のemojiの細かい ニュアンスを学習できている. (次ページの図を参照)
Importance of emoji diversity 13
Model architecture 14 Pretraining時点では,差がない benchmark時点では,Attention ありの方が精度が高い 低層の特徴へのアクセスが簡単 勾配消失がなく,学習可能
Analyzing the effect of pretraining 15 Pretraining+chainthawで語彙が 増加 ->word coverageが改善
Comparing with human-level agreement 16 Human:76.1% Deepmoji:82.4% Deepmojiの方が,精度 が高い (実験内容については,論文
を参照)
Conclusion • sentiment analysis, emotion analysis and sarcasm classificationにおける8つのbenchmarkでSoTA達成 •
感情ラベルの多様性が以前のdistant supervisonのアプ ローチよりもパフォーマンスの向上をもたらすことを確認 • Pretraining済みモデルを公開 ◦ (Demo : https://deepmoji.mit.edu/) 17