AI Paper Atlas

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

DeepChannel: Salience Estimation by Contrastive Learning for Extractive Document Summarization

Jiaxin Shi, Chen Liang, Lei Hou, Juanzi Li, Zhiyuan Liu, Hanwang Zhang

AAAI 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining

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 DeepChannel instead of the strongest listed baseline, SWAP-NETCNN/Daily Mail summarization · CNN/Daily Mail testROUGE-1 F-1: 41.60 -> 41.50No clear effect
Use DeepChannel instead of rnn-ext + RLCNN/Daily Mail summarization · CNN/Daily Mail testROUGE-2 F-1: 18.72 -> 17.77Hurt
Use DeepChannel instead of rnn-ext + RLCNN/Daily Mail summarization · CNN/Daily Mail testROUGE-L F-1: 37.76 -> 37.62No clear effect
Train on CNN/Daily Mail and evaluate out of domain with DeepChannel rather than RefreshExtractive summarization · DUC 2007, 75 bytesROUGE-1 recall: 18.39 -> 19.53Improved
Train on CNN/Daily Mail and evaluate out of domain with DeepChannel rather than RefreshExtractive summarization · DUC 2007, 275 bytesROUGE-1 recall: 26.80 -> 28.85Improved
Train DeepChannel with 1/100 of the CNN/Daily Mail training data rather than RefreshExtractive summarization · CNN/Daily Mail test, 1/100 trainingROUGE-1 F-1: 36.05 -> 39.41Improved
Add the attention penalization term (α=0.001) instead of removing it (ablation)DeepChannel · CNN/Daily Mail testROUGE-1 F-1: 40.89 -> 41.50Improved
Add the attention penalization term (α=0.001) instead of removing it (ablation)DeepChannel · CNN/Daily Mail testROUGE-2 F-1: 17.21 -> 17.77Improved
Add the attention penalization term (α=0.001) instead of removing it (ablation)DeepChannel · CNN/Daily Mail testROUGE-L F-1: 37.08 -> 37.62Improved
Increase attention penalization weight from α=0.001 to α=0.01 (ablation)DeepChannel · CNN/Daily Mail testROUGE-1 F-1: 41.50 -> 41.30Hurt
Increase attention penalization weight from α=0.001 to α=0.1 (ablation)DeepChannel · CNN/Daily Mail testROUGE-1 F-1: 41.50 -> 40.49Hurt

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