Research map / Reinforcement learning / Policy gradient and actor-critic
Replacing Rewards with Examples: Example-Based Policy Search via Recursive 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 |
|---|---|---|---|
| Use RCE instead of reward-learning baselines | Example-based control · Five Meta-World Sawyer tasks and two manipulation tasks | evaluation return | Helped |
| Use RCE instead of reward-learning baselines | Example-based control · Hammering and bin picking | task progress | Helped |
| Use RCE instead of the comparison methods | Image-based example-based control · reach_random_position and reach_random_size | task performance | Helped |
| Use RCE instead of the comparison methods | Image-based example-based control · sawyer_clear_image | task progress | Helped |
| Use n-step returns (ablation) | RCE · Manipulation tasks | performance | Helped |
| Apply RCE to full expert trajectories instead of success examples (ablation) | RCE · Full expert trajectories | task performance | Hurt |
| Use an untrained random image encoder instead of training the image encoder (ablation) | RCE and baselines · Image-based tasks | performance | No clear effect |
| Use an untrained random image encoder instead of training the image encoder (ablation) | RCE and baselines · Image-based tasks | training speed | Helped |
| Take the maximum rather than the minimum over the two classifier predictions (ablation) | RCE · sawyer_bin_picking | performance | Helped |
Closest papers by meaning
- To the Max: Reinventing Reward in Reinforcement Learning (ICML 2024 · similarity 0.90)
- STARC: A General Framework For Quantifying Differences Between Reward Functions (ICLR 2024 · similarity 0.88)
- Going Beyond Heuristics by Imposing Policy Improvement as a Constraint (NeurIPS 2024 · similarity 0.87)
- Prompted Policy Search: Reinforcement Learning through Linguistic and Numerical Reasoning in LLMs (NeurIPS 2025 · similarity 0.87)
- Exploration-Driven Policy Optimization in RLHF: Theoretical Insights on Efficient Data Utilization (ICML 2024 · similarity 0.87)
- The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret (ICML 2025 · similarity 0.87)
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.