AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Have We Solved The Hard Problem? It’s Not Easy! Contextual Lexical Contrast as a Means to Probe Neural Coherence

Wenqiang Lei, Yisong Miao, Runpeng Xie, Bonnie Webber, Meichun Liu, Tat-Seng Chua, Nancy F. Chen

AAAI 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 12 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use contextual BERT embeddings instead of FastText embeddingsCLC recognition model · Cont2Lexaccuracy: 66.3 -> 70.0Improved
Use contextual embeddings rather than static embeddingsOut-of-context lexical contrast recognition · Out-of-context lexical contrastaccuracy: 84.1 -> 79.5Hurt
Add a BiLSTM encoder to contextual BERT embeddingsCLC recognition model · Cont2Lexaccuracy: 70.0 -> 69.2; 70.0 -> 69.1No clear effect
Use BERT rather than BERT.Lex, which lacks surrounding-context inputCLC recognition model · Cont2Lexaccuracy: 66.4 -> 70.0Improved
Use BERT rather than FastTextCLC recognition model · Same-sentence pairsaccuracy: 60.4 -> 67.4Improved
Use BERT rather than FastTextCLC recognition model · Adjacent-sentence pairsaccuracy: 69.8 -> 71.4Improved
Use full BERT instead of BERT.LexCLC recognition model · Same-sentence pairsaccuracy: 60.7 -> 67.4Improved
Use full BERT instead of BERT.LexCLC recognition model · Adjacent-sentence pairsaccuracy: 69.8 -> 71.4Improved
Use BERT rather than FastTextCLC recognition model · Instances with word repetitionaccuracy: 61.1 -> 68.7Improved
Use BERT rather than FastTextCLC recognition model · Instances without word repetitionaccuracy: 68.8 -> 70.7Improved
Use full BERT instead of BERT.LexCLC recognition model · Instances with word repetitionaccuracy: 58.7 -> 68.7Improved
Use full BERT instead of BERT.LexCLC recognition model · Instances without word repetitionaccuracy: 69.9 -> 70.7Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy