AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Quantifying Uncertainties in Natural Language Processing Tasks

Yijun Xiao, William Yang Wang

AAAI 2019 · Large language models › Language, safety and interpretability · approach: Faithfulness of LLM explanations

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
Add model uncertainty (ablation)CNN sentiment regression · Yelp 2013, Yelp 2014, Yelp 2015, and IMDBMSE: 0.71 -> 0.57; 0.72 -> 0.55; 0.72 -> 0.55; 3.62 -> 3.20Improved
Add data uncertainty alone (ablation)CNN sentiment regression · Yelp 2013, Yelp 2014, Yelp 2015, and IMDBMSE: 0.71 -> 0.84; 0.72 -> 0.75; 0.72 -> 0.73; 3.62 -> 3.74Hurt
Add both model and data uncertainty (ablation)CNN sentiment regression · Yelp 2013, Yelp 2014, Yelp 2015, and IMDBMSE: 0.71 -> 0.57; 0.72 -> 0.54; 0.72 -> 0.53; 3.62 -> 3.13Improved
Add uncertainty measures (ablation)CNN sentiment classification · Sentiment analysis classificationaccuracyMixed
Add data uncertainty (ablation)Bidirectional LSTM tagger · CoNLL 2003F1 score: 77.5 -> 79.6; 2.7%Improved
Add model uncertainty (ablation)Bidirectional LSTM tagger · CoNLL 2003F1 score: 77.5 -> 76.5Hurt
Add both model and data uncertainty (ablation)Bidirectional LSTM tagger · CoNLL 2003F1 score: 77.5 -> 78.5Improved
Add model uncertainty (ablation)Two-layer LSTM language model · PTBperplexity: 82.7 -> 81.3Improved
Add data uncertainty (ablation)Two-layer LSTM language model · PTBperplexity: 82.7 -> 80.5Improved
Add both model and data uncertainty (ablation)Two-layer LSTM language model · PTBperplexity: 82.7 -> 79.2; 4.2%Improved

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