Research map / Reinforcement learning / Visual and representation RL
A State-Distribution Matching Approach to Non-Episodic 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 |
|---|---|---|---|
| Match the backward policy's state distribution to demonstrations rather than the initial-state distribution (ablation) | MEDAL · Tabletop Organization and Sawyer Door | deployed policy performance | Helped |
| Match the backward policy's state distribution to demonstrations rather than the initial-state distribution | MEDAL · Sawyer Peg | average return: 0.00 (0.00) -> 0.40 (0.16) | Helped |
| Use log(C(s)) instead of -log(1-C(s)) as the backward-policy reward (ablation) | MEDAL · EARL continuous-control tasks | deployed policy performance | Hurt |
| Initialize from states sampled from the optimal policy's stationary distribution instead of the standard initial-state distribution | episodic RL on Tabletop Organization · Tabletop Organization | policy performance: nearly 18% | Helped |
| Use MEDAL rather than FBRL | EARL methods · Sawyer Peg | average return: 0.00 (0.00) -> 0.40 (0.16) | Helped |
| Use MEDAL rather than VaPRL | EARL methods · Tabletop Organization | average return: 0.98 (0.02) -> 0.98 (0.02) | No clear effect |
Closest papers by meaning
- Autonomous Reinforcement Learning: Formalism and Benchmarking (ICLR 2022 · similarity 0.92)
- Leave no Trace: Learning to Reset for Safe and Autonomous Reinforcement Learning (ICLR 2018 · similarity 0.89)
- Enhanced Meta Reinforcement Learning via Demonstrations in Sparse Reward Environments (NeurIPS 2022 · similarity 0.89)
- You Only Live Once: Single-Life Reinforcement Learning (NeurIPS 2022 · similarity 0.89)
- DynaMITE-RL: A Dynamic Model for Improved Temporal Meta-Reinforcement Learning (NeurIPS 2024 · similarity 0.89)
- Dual RL: Unification and New Methods for Reinforcement and Imitation Learning (ICLR 2024 · similarity 0.88)
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.