AI Paper Atlas

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

Effective Neural Topic Modeling with Embedding Clustering Regularization

Xiaobao Wu, Xinshuai Dong, Thong Thanh Nguyen, Anh Tuan Luu

ICML 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 9 citations (OpenAlex)

Read the paper

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 embedding clustering regularization (ECR) to topic model (ablation)ECRTM · 20NG, IMDB, Yahoo, AG News K=50topic diversityImproved
add ECR to topic model (ablation)ECRTM · document clustering (ablation)Purity/NMIImproved
replace Deep KMeans clustering regularization with ECR (optimal transport, sparse, size-constrained) (ablation)ECRTM · topic diversityTDImproved
add entropy minimization to DKM for sparse assignments (DKM+Entropy) (ablation)DKM regularization · IMDB topicsTDNo clear effect
use ECRTM instead of NSTMNeural topic model · Yahoo Answer K=50C_V: 0.390 -> 0.405Improved
use ECRTM instead of NSTMNeural topic model · Yahoo Answer K=50TD: 0.658 -> 0.985Improved
use ECRTM instead of NSTMNeural topic model · AG News K=100C_V: 0.421 -> 0.416Hurt
use ECRTM instead of runner-up for document clusteringTopic model · 20NG K=50Purity: 0.367 -> 0.560Improved
remove high-frequency words by lowering max-df for baselinesLDA, ETM, NSTM · 20NGTDImproved
use ECRTM regardless of preprocessing (max-df)ECRTM · 20NG max-df 0.5 to 0.1topic quality and clusteringImproved
use ECRTM instead of LDA, DVAE, ETMTopic model · 20NGperplexity (lower better): 2044.6 -> 1896.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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