Research map / Reinforcement learning / Policy gradient and actor-critic
Learning to Sample with Local and Global Contexts in Experience Replay Buffer
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 NERS instead of the strongest reported baseline sampler | TD3 and SAC · MuJoCo Ant-v3, Walker2D-v3, and Hopper-v3 | average cumulative reward: 4193.69 (+677.81); 3938.59 (+378.62); 3062.32 (+451.92); 2913.54 (+678.20); 3720.11 (+323.12); 2763.69 (+354.18) | Helped |
| Use NERS instead of the strongest reported baseline sampler | Rainbow · Atari environments in Table 2 | average cumulative reward: 1125.18 (+281.16); 167.50 (+18.85); 516.52 (+26.46); 679.00 (+80.90); 18584.99 (+1500.19); 2.51 (+4.83); 878.73 (+125.87 | Helped |
| Use NERS with the full feature set rather than only reward, TD-error, and timestep (ablation) | NERS · Continuous-control tasks shown in the feature ablation | performance / sample efficiency | Helped |
| Include global context to model relative importance among sampled transitions (ablation) | NERS · Continuous-control tasks shown in the global-context ablation | sample efficiency | Helped |
| Use training-episode reward differences instead of evaluation-policy reward differences (ablation) | NERS* · BipedalWalker-v3 and LunarLanderContinuous-v2 | performance | No clear effect |
Closest papers by meaning
- Model-augmented Prioritized Experience Replay (ICLR 2022 · similarity 0.90)
- Variance Reduction via Resampling and Experience Replay (AAAI 2026 · similarity 0.90)
- Remember and Forget for Experience Replay (ICML 2019 · similarity 0.90)
- Actor Prioritized Experience Replay (Abstract Reprint) (AAAI 2024 · similarity 0.89)
- Prioritized Generative Replay (ICLR 2025 · similarity 0.88)
- When Does Non-Uniform Replay Matter in Reinforcement Learning? (NeurIPS 2026 · 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.