Research map / Reinforcement learning / Visual and representation RL
Competitive experience replay
Hao Liu, Alexander Trott, Richard Socher, Caiming Xiong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 CER to DDPG | DDPG · U-shaped and S-shaped AntMaze | success rate | Improved |
| Combine CER with HER | DDPG with HER · U-shaped and S-shaped AntMaze | final success rate | Improved |
| Use interact-CER instead of independent-CER (ablation) | CER · U-shaped and S-shaped AntMaze | success rate | Improved |
| Add independent-CER to HER | HER · 12 multi-goal sparse-reward tasks | success rate | Improved |
| Add CER rather than ICM to HER | HER · Robotic control and maze tasks | task success rate | Improved |
| Use CER rather than ICM as the standalone exploration addition | DDPG · Four robotic control and maze tasks | task performance | Improved |
| Use CER rather than ICM | DDPG · Four robotic control and maze tasks | learning speed | Improved |
| Add CER to DDPG | DDPG · U-shaped and S-shaped AntMaze | sample efficiency | Improved |
| Add CER to HER | HER · U-shaped AntMaze | goal-directed state visitation | Improved |
| Periodically reset agent B's parameters early in training (ablation) | CER · Robotic control tasks | agent A performance | Improved |
| Use imbalanced batch sizes for agents A and B (ablation) | CER · FetchSlide and HandManipulatePenFull | performance | Hurt |
| Use balanced batch sizes for agents A and B (ablation) | CER · FetchSlide and HandManipulatePenFull | performance | Improved |
Closest papers by meaning
- Hindsight Experience Replay (NeurIPS 2017 · similarity 0.90)
- DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Experience Selection in Deep Reinforcement Learning for Control (JMLR 2018 · similarity 0.88)
- Guided Exploration with Proximal Policy Optimization using a Single Demonstration (ICML 2021 · similarity 0.88)
- Cluster-Based Sampling in Hindsight Experience Replay for Robotic Tasks (Student Abstract) (AAAI 2024 · similarity 0.87)
- CIG: Exploration via Conditional Information Gain (NeurIPS 2026 · similarity 0.86)
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.
