Research map / Reinforcement learning / MDPs and planning
#Exploration: A Study of Count-Based Exploration for Deep Reinforcement Learning
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 |
|---|---|---|---|
| Add hashed state-visitation count bonuses to TRPO | TRPO · MountainCar | average return | Helped |
| Add hashed state-visitation count bonuses to TRPO | TRPO · CartPoleSwingup | average return | Helped |
| Add hashed state-visitation count bonuses to TRPO | TRPO · HalfCheetah | average return | Mixed |
| Add hashed state-visitation count bonuses to TRPO | TRPO · SwimmerGather | average return | Helped |
| Use raw-pixel SimHash instead of baseline TRPO exploration | TRPO · Freeway | average total reward: 16.5 -> 31.6 | Helped |
| Use raw-pixel SimHash instead of baseline TRPO exploration | TRPO · Frostbite | average total reward: 2869 -> 4683 | Helped |
| Add BASS preprocessing before SimHash (ablation) | TRPO-pixel-SimHash · Montezuma’s Revenge | average total reward: 0 -> 238 | Helped |
| Add BASS preprocessing before SimHash (ablation) | TRPO-pixel-SimHash · Venture | average total reward: 263 -> 616 | Helped |
| Add BASS preprocessing before SimHash (ablation) | TRPO-pixel-SimHash · Solaris | average total reward: 2897 -> 1201 | Hurt |
| Add autoencoder-learned image features before SimHash (ablation) | TRPO-pixel-SimHash · Freeway | average total reward: 31.6 -> 33.5 | Helped |
| Add autoencoder-learned image features before SimHash (ablation) | TRPO-pixel-SimHash · Solaris | average total reward: 2897 -> 4467 | Helped |
| Add autoencoder-learned image features before SimHash (ablation) | TRPO-pixel-SimHash · Montezuma’s Revenge | average total reward: 0 -> 75 | Helped |
Closest papers by meaning
- Flipping Coins to Estimate Pseudocounts for Exploration in Reinforcement Learning (ICML 2023 · similarity 0.90)
- Count-Based Exploration with the Successor Representation (AAAI 2020 · similarity 0.89)
- Random Latent Exploration for Deep Reinforcement Learning (ICML 2024 · similarity 0.87)
- Counteractive RL: Rethinking Core Principles for Efficient and Scalable Deep Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Count Counts: Motivating Exploration in LLM Reasoning with Count-based Intrinsic Rewards (ICLR 2026 · similarity 0.87)
- Provable and Practical: Efficient Exploration in Reinforcement Learning via Langevin Monte Carlo (ICLR 2024 · 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.