Research map / Label-efficient and robust learning / Recognition and long-tail
Adaptive Region Embedding for Text Classification
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 |
|---|---|---|---|
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Yelp Review Polarity | accuracy: 96.4 -> 96.6 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Yelp Review Full | accuracy: 64.9 -> 65.9 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Amazon Review Polarity | accuracy: 95.3 -> 95.9 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Amazon Review Full | accuracy: 60.9 -> 62.6 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · AG's News | accuracy: 92.8 -> 93.1 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Sogou News | accuracy: 97.6 -> 97.5 | Hurt |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · Yahoo! Answers | accuracy: 73.7 -> 74.9 | Helped |
| Replace word-indexed LCU with instance-adaptive ACU generated by a meta-network | ARE · DBPedia | accuracy: 98.9 -> 99.1 | Helped |
| Use a smaller output context filter shared across region positions (ablation) | ARE CNN meta-network · AG's News, DBPedia, Yahoo! Answers | accuracy: 93.1 -> 93.0; 99.1 -> 98.9; 74.9 -> 74.9 | Mixed |
| Factorize generated filters with u=32 (ablation) | ARE CNN meta-network · AG's News, DBPedia, Yahoo! Answers | accuracy: 93.1 -> 92.7; 99.1 -> 98.9; 74.9 -> 74.3 | Hurt |
| Replace CNN meta-network with LSTM (ablation) | ARE CNN meta-network · AG's News, DBPedia, Yahoo! Answers | accuracy: 93.1 -> 92.5; 99.1 -> 98.6; 74.9 -> 73.5 | Hurt |
| Replace CNN meta-network with GRU (ablation) | ARE CNN meta-network · AG's News, DBPedia, Yahoo! Answers | accuracy: 93.1 -> 92.5; 99.1 -> 98.6; 74.9 -> 73.5 | Hurt |
Closest papers by meaning
- A New Method of Region Embedding for Text Classification (ICLR 2018 · similarity 0.91)
- Active Discriminative Text Representation Learning (AAAI 2017 · similarity 0.88)
- Supervised and Semi-Supervised Text Categorization using LSTM for Region Embeddings (ICML 2016 · similarity 0.88)
- Making Text Embedders Few-Shot Learners (ICLR 2025 · similarity 0.85)
- Contextual Document Embeddings (ICLR 2025 · similarity 0.85)
- 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.