Research map / Reinforcement learning / Visual and representation RL
Learning to Compose Words into Sentences with Reinforcement Learning
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 |
|---|---|---|---|
| Learn latent task-specific tree structures instead of using fixed composition orders | 100D Tree LSTM · SST binary classification | accuracy: 85.3 -> 86.5 | Helped |
| Use latent task-specific structures instead of supervised syntax | 100D Tree LSTM · SICK | MSE: 0.381 -> 0.359 | Helped |
| Use semi-supervised structures instead of supervised syntax | 100D Tree LSTM · SICK | MSE: 0.381 -> 0.320 | Helped |
| Use latent task-specific structures instead of supervised syntax | 100D Tree LSTM · SNLI | accuracy: 78.5 -> 80.5 | Helped |
| Add initial parse-tree guidance, then optimize the downstream task | 100D Tree LSTM · SNLI | accuracy: 78.5 -> 80.2 | Helped |
| Add initial parse-tree guidance, then optimize the downstream task | 100D Tree LSTM · IMDB sentence generation | perplexity: 100.8 -> 98.4 | Helped |
| Learn task-specific tree structures rather than using fixed composition orders | 100D Tree LSTM · IMDB sentence generation | perplexity: 100.2 -> 98.4 | Helped |
Closest papers by meaning
- Learning Structured Representation for Text Classification via Reinforcement Learning (AAAI 2018 · similarity 0.89)
- Characterizing intrinsic compositionality in transformers with Tree Projections (ICLR 2023 · similarity 0.88)
- Modelling Sentence Pairs via Reinforcement Learning: An Actor-Critic Approach to Learn the Irrelevant Words (AAAI 2020 · similarity 0.87)
- W2S-AlignTree: Weak-to-Strong Inference-Time Alignment for Large Language Models via Monte Carlo Tree Search (AAAI 2026 · similarity 0.85)
- Compositional Reasoning in Language Models under Reinforcement Learning Post-Training (NeurIPS 2026 · similarity 0.85)
- Subwords as Skills: Tokenization for Sparse-Reward Reinforcement Learning (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.