Research map / Reinforcement learning / MDPs and planning
Reward-Free Exploration for 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 |
|---|---|---|---|
| Collect data without reward feedback, then plan for rewards using the shared dataset | Reward-Free RL · Reward-free tabular episodic MDP | exploration episodes: c·[H^5S^2Aι/ε^2 + S^4AH^7ι^3/ε] | Helped |
| Require coverage for arbitrary reward functions rather than exploration for one prespecified reward | Reward-free RL · Reward-free tabular episodic MDP | exploration episodes: Ω(S^2AH^2/ε^2); prespecified-reward RL: Õ(SAH^2/ε^2) | Mixed |
| Use ZeroRMax for reward-free exploration instead of the proposed exploration method | Reward-free tabular episodic MDP · Reward-free tabular episodic MDP | exploration episodes: Ω(H^11S^2A/(ε^3p)(log(SAH/(pε))^2) | Hurt |
| Use the MaxEnt exploratory-policy guarantee to establish coverage | Reward-free exploration · Reward-free infinite-horizon discounted MDP | trajectories: Õ(S^2A/(ε^3(1−γ)^2)); the paper's coverage calculation yields S^5 samples | Hurt |
| Explore by maximizing a single prespecified reward rather than covering significant states for later rewards | Reward-free RL · Reward-free tabular episodic MDP | exploration episodes: Ω(S^2AH^2/ε^2); prespecified-reward RL: Õ(SAH^2/ε^2) | Hurt |
Closest papers by meaning
- Near Optimal Reward-Free Reinforcement Learning (ICML 2021 · similarity 0.94)
- On Reward-Free RL with Kernel and Neural Function Approximations: Single-Agent MDP and Markov Game (ICML 2021 · similarity 0.93)
- On Reward-Free Reinforcement Learning with Linear Function Approximation (NeurIPS 2020 · similarity 0.92)
- Reward-Free Kernel-Based Reinforcement Learning (ICML 2024 · similarity 0.91)
- Beyond Optimism: Exploration With Partially Observable Rewards (NeurIPS 2024 · similarity 0.91)
- DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning (NeurIPS 2025 · similarity 0.90)
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.