Research map / Label-efficient and robust learning / Recognition and long-tail
Inter and Intra Topic Structure Learning with Word Embeddings
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Using topic embeddings (top words ranked by dot product with word embeddings) instead of the local topic distribution | WEDTM topics · WS | NPMI coherence: 0.143 to 0.224 on example topic 5 vs local interpretation | Helped |
| Word embedding informed priors (WEDTM) vs deep GBN baseline | WEDTM local topics · WS | NPMI coherence: local topics more coherent than GBN in example topics | Helped |
| Adding word-embedding-informed sub-topics and deep structure | WEDTM vs GBN · 20NG, WS, TMN, Twitter (perplexity) | relative per-heldout-word perplexity | Helped |
| Increasing network depth (ablation) | WEDTM · WS with 20% training words | perplexity | Hurt |
| Increasing network depth (ablation) | WEDTM · 20NG, TMN, WS (perplexity, general) | perplexity | Helped |
| Adding word embeddings (WEDTM) vs GBN for document classification | WEDTM · TMN, 20NG | document classification accuracy | Helped |
| Adding sub-topics (intra-topic structure) to local topics (ablation) | WEDTM · WS, TMN, 20NG | classification accuracy / perplexity | No clear effect |
Closest papers by meaning
- TopicNet: Semantic Graph-Guided Topic Discovery (NeurIPS 2021 · similarity 0.90)
- Dirichlet belief networks for topic structure learning (NeurIPS 2018 · similarity 0.89)
- Knowledge-Aware Bayesian Deep Topic Model (NeurIPS 2022 · similarity 0.88)
- On the Affinity, Rationality, and Diversity of Hierarchical Topic Modeling (AAAI 2024 · similarity 0.86)
- FASTopic: Pretrained Transformer is a Fast, Adaptive, Stable, and Transferable Topic Model (NeurIPS 2024 · similarity 0.84)
- Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe (NeurIPS 2024 · similarity 0.84)
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.