Research map / Reinforcement learning / Visual and representation RL
Exploration by random network distillation
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 random-network distillation rather than PPO without an exploration bonus | PPO with RND · Montezuma's Revenge | mean episodic return: 2,497 -> 8,152 | Helped |
| Use random-network distillation rather than forward-dynamics prediction error (ablation) | RND exploration · Montezuma's Revenge | mean episodic return: 400 -> 8,152 | Helped |
| Use random-network distillation rather than PPO without an exploration bonus | PPO with RND · Venture | mean episodic return: 0 -> 1,859 | Helped |
| Use random-network distillation rather than forward-dynamics prediction error (ablation) | RND exploration · PrivateEye | mean episodic return: 33 -> 8,666 | Helped |
| Use random-network distillation rather than forward-dynamics prediction error (ablation) | RND exploration · Solaris | mean episodic return: 3,246 -> 3,282 | Helped |
| Use random-network distillation rather than forward-dynamics prediction error (ablation) | RND exploration · Venture | mean episodic return: 1,712 -> 1,859 | No clear effect |
| Use random-network distillation rather than forward-dynamics prediction error (ablation) | RND exploration · Gravitar | mean episodic return: 3,371 -> 3,906 | No clear effect |
| Use non-episodic intrinsic returns instead of episodic intrinsic returns (ablation) | RND agent · Montezuma's Revenge | rooms explored | Helped |
| Increase the extrinsic discount factor from 0.99 to 0.999 (ablation) | RND agent · Montezuma's Revenge | mean episodic return | Helped |
| Increase the intrinsic discount factor from 0.99 to 0.999 (ablation) | RND agent · Montezuma's Revenge | performance | Hurt |
| Use two value heads instead of one for episodic intrinsic and extrinsic rewards (ablation) | RND agent · Montezuma's Revenge | performance | No clear effect |
| Increase the number of parallel environments used to collect experience (ablation) | RND agent · Atari hard-exploration games | mean episodic return | Helped |
Closest papers by meaning
- Exploration and Anti-Exploration with Distributional Random Network Distillation (ICML 2024 · similarity 0.92)
- Catching Two Birds with One Stone: Reward Shaping with Dual Random Networks for Balancing Exploration and Exploitation (ICML 2025 · similarity 0.90)
- Random Latent Exploration for Deep Reinforcement Learning (ICML 2024 · similarity 0.89)
- Anti-Exploration by Random Network Distillation (ICML 2023 · similarity 0.88)
- DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning (NeurIPS 2025 · similarity 0.85)
- MaxInfoRL: Boosting exploration in reinforcement learning through information gain maximization (ICLR 2025 · similarity 0.85)
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.