Research map / Reinforcement learning / Visual and representation RL
Structure Detection for Contextual Reinforcement Learning
Tianyue Zhou, Jung-Hoon Cho, Cathy Wu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| switch between M-MBTL and GP-MBTL via structure detection (vs GP-MBTL alone) | GP-MBTL · CMDP benchmarks aggregated | aggregated normalized score: 0.1681 -> 0.2930 | Improved |
| use M-MBTL clustering instead of GP-MBTL | GP-MBTL · BipedalWalker (K=12) | normalized performance: 0.8073 -> 0.8315 | Improved |
| use M-MBTL clustering instead of GP-MBTL | GP-MBTL · IntersectionZoo (K=50) | normalized performance: 0.5840 -> 0.4878 | Hurt |
| use M-MBTL clustering instead of GP-MBTL | GP-MBTL · CyclesGym (K=50) | normalized performance: 0.2193 -> 0.2205 | No clear effect |
| add structure detection to switch to GP-MBTL when MOUNTAIN is violated (vs M-MBTL alone) | M-MBTL · IntersectionZoo (K=50) | normalized performance: 0.4878 -> 0.5682 | Improved |
| use M/GP-MBTL instead of M-MBTL | M-MBTL · BipedalWalker (K=12) | normalized performance: 0.8315 -> 0.8261 | Hurt |
| use multi-task training instead of strategic task selection | GP-MBTL · CartPole (K=12) | normalized performance: 0.9919 -> 0.9967 | Improved |
| use multi-task training instead of strategic task selection | GP-MBTL · BipedalWalker (K=12) | normalized performance: 0.8073 -> 0.5680 | Hurt |
| M-MBTL vs GP-MBTL under non-MOUNTAIN synthetic (constant f, linear g, non-distance h) | GP-MBTL · Synthetic CMDP | normalized performance: 0.7212 -> 0.6976 | Hurt |
| M/GP-MBTL vs GP-MBTL | GP-MBTL · Synthetic CMDP aggregated | aggregated normalized score: 0.2127 -> 0.3099 | Improved |
Closest papers by meaning
- Model-Based Transfer Learning for Contextual Reinforcement Learning (NeurIPS 2024 · similarity 0.91)
- XLand-100B: A Large-Scale Multi-Task Dataset for In-Context Reinforcement Learning (ICLR 2025 · similarity 0.87)
- Zero-Shot Context Generalization in Reinforcement Learning from Few Training Contexts (NeurIPS 2025 · similarity 0.87)
- Zero-shot Model-based Reinforcement Learning using Large Language Models (ICLR 2025 · similarity 0.87)
- Behavior-agnostic Task Inference for Robust Offline In-context Reinforcement Learning (ICML 2025 · similarity 0.86)
- Towards Large-Scale In-Context Reinforcement Learning by Meta-Training in Randomized Worlds (NeurIPS 2025 · similarity 0.86)
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.
