Research map / Reinforcement learning / Policy gradient and actor-critic
Attentive Experience Replay
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 |
|---|---|---|---|
| replace uniform sampling with attentive (state-similarity) sampling | SAC · MuJoCo tasks (6) | sample efficiency / steps to match vanilla-ER final performance: less than 65% of vanilla-ER's total steps | Helped |
| replace PER with AER | TD3, DDPG · Hopper, HalfCheetah | cumulative reward | Helped |
| add PER (TD-error prioritization) | SAC, TD3, DDPG · MuJoCo tasks | cumulative reward | No clear effect |
| add PER (TD-error prioritization) | DQN · Atari Pong | cumulative reward | Helped |
| add learning rate annealing to uniform sampling (annealing-ER) (ablation) | SAC with vanilla-ER · Hopper, HalfCheetah, Walker2d | cumulative reward | No clear effect |
| use AER instead of PER in terms of compute | SAC · six MuJoCo tasks | running time (10^3 s): Hopper 8 -> 5.4 | Helped |
Closest papers by meaning
- Remember and Forget for Experience Replay (ICML 2019 · similarity 0.92)
- Revisiting Fundamentals of Experience Replay (ICML 2020 · similarity 0.92)
- Actor Prioritized Experience Replay (Abstract Reprint) (AAAI 2024 · similarity 0.90)
- Learning to Sample with Local and Global Contexts in Experience Replay Buffer (ICLR 2021 · similarity 0.89)
- Synthetic Experience Replay (NeurIPS 2023 · similarity 0.89)
- Turning Sand to Gold: Recycling Data to Bridge On-Policy and Off-Policy Learning via Causal Bound (NeurIPS 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.