AI Paper Atlas

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

TC-DWA:Text Clustering with Dual Word-Level Augmentation

Bo Cheng, Ximing Li, Yi Chang

AAAI 2023 · 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
add anchor-word augmentation (ablation)TC-DWA · AG NewsACC: 0.810 -> 0.822Improved
add anchor-word augmentation (ablation)TC-DWA · DBPediaACC: 0.780 -> 0.795Improved
add anchor-word augmentation (ablation)TC-DWA · NewsgroupACC: 0.477 -> 0.491Improved
add expectation-word augmentation (ablation)TC-DWA · AG NewsACC: 0.819 -> 0.822Improved
add expectation-word augmentation (ablation)TC-DWA · DBPediaACC: 0.784 -> 0.795Improved
add expectation-word augmentation (ablation)TC-DWA · NewsgroupACC: 0.484 -> 0.491Improved
add both anchor-word and expectation-word augmentation (ablation)BERT self-training text clustering · AG NewsACC: 0.808 -> 0.822Improved
add both anchor-word and expectation-word augmentation (ablation)BERT self-training text clustering · DBPediaACC: 0.776 -> 0.795Improved
add both anchor-word and expectation-word augmentation (ablation)BERT self-training text clustering · NewsgroupACC: 0.470 -> 0.491Improved
use BERT layer 12 rather than layer 7 for norm-based attention weights (ablation)TC-DWA · NewsgroupACC: 0.401 -> 0.491Improved
use BERT layer 12 rather than layer 9 for norm-based attention weights (ablation)TC-DWA · NewsgroupACC: 0.447 -> 0.491Improved
vary the number of anchor words from 0 to 5 (ablation)TC-DWA · AG News, DBPedia, and NewsgroupACCMixed

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