Research map / Reinforcement learning / Policy gradient and actor-critic
Transfer Reinforcement Learning with Shared Dynamics
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 |
|---|---|---|---|
| add optimism (reward upper confidence bound, kappa>0) to reward-proxy transition reuse (ablation) | Transition reuse with Fitted-Q · 25-task maze transfer | average reward | Helped |
| add optimism (ablation) | Transition reuse with Fitted-Q · 25-task maze transfer, very early transitions | average reward | Hurt |
| reuse transition samples across tasks rather than learning without transfer (ablation) | Fitted-Q learner · 25-task maze transfer | average reward | Helped |
| use continuous Gaussian-similarity confidence intervals instead of discrete state knowledge (ablation) | optimistic transfer · 25-task maze transfer | average reward: 80 vs. 40 updates | Helped |
| inject strong expert knowledge (rewards only at corners) (ablation) | optimistic transfer · 4 corner tasks | early convergence | Helped |
| lower policy update parameter (1.05 instead of 2) (ablation) | discrete corner setting · 4 corner tasks | convergence: less than 10 -> 50 updates | Helped |
| vary kappa vs sigma (ablation) | continuous setting · 25-task maze transfer | cumulative reward | Mixed |
| use optimism vs non-optimistic in Multi-Task RL (ablation) | continuous setting · Multi-task maze | average reward | Helped |
Closest papers by meaning
- Transfer Q-Learning with Composite MDP Structures (ICML 2025 · similarity 0.89)
- To the Max: Reinventing Reward in Reinforcement Learning (ICML 2024 · similarity 0.88)
- Towards the Transferability of Rewards Recovered via Regularized Inverse Reinforcement Learning (NeurIPS 2024 · similarity 0.88)
- Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage Perspective (ICML 2025 · similarity 0.88)
- DUO: Diverse, Uncertain, On-Policy Query Generation and Selection for Reinforcement Learning from Human Feedback (AAAI 2025 · similarity 0.88)
- Reward-Consistent Dynamics Models are Strongly Generalizable for Offline Reinforcement 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.