Research map / Reinforcement learning / MDPs and planning
Task-agnostic Exploration in 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 |
|---|---|---|---|
| Explore without reward guidance using UCBZero, then optimize policies from sampled rewards | Tabular episodic MDPs with N tasks · Task-agnostic RL | exploration episodes: O((\log N+\iota)H^5SA/\epsilon^2) | Helped |
| Use task-agnostic exploration rather than separately applying UCB-H to each task | Learning N tasks · Task-agnostic RL | exploration episodes: N times the original sample complexity | Helped |
| Use UCBZero rather than RFE in the N-dependent reward-free setting | Reward-free RL · N tasks, when N is not extremely large | exploration episodes: \tilde O(\log (N)H^5SA/\epsilon^2) versus \tilde{O}(H^5S^2A/\epsilon^2) | Helped |
| Use UCBZero with task-agnostic sampled rewards rather than assuming true rewards are known | RL with multiple tasks · Task-agnostic RL | exploration episodes: \Omega\left(\log (N)H^2SA/\epsilon^2\right) | Hurt |
| Use UCBZero in reward-free RL with reward discretization for an N-independent guarantee | Reward-free RL · Any number of tasks | exploration episodes: O\left(H^6S^2A^2(\log \frac{3H}{\epsilon}+\iota)/\epsilon^2\right) | Mixed |
| Use full exploration without task-specific rewards rather than balancing exploration and exploitation | Task-specific RL · N=1 | exploration episodes: \tilde O(H^2SA/\epsilon^2) | No clear effect |
Closest papers by meaning
- Reward-Free Exploration for Reinforcement Learning (ICML 2020 · similarity 0.92)
- Improved Bounds for Reward-Agnostic and Reward-Free Exploration (ICML 2026 · similarity 0.92)
- Maximize to Explore: One Objective Function Fusing Estimation, Planning, and Exploration (NeurIPS 2023 · similarity 0.91)
- Reward-Free Kernel-Based Reinforcement Learning (ICML 2024 · similarity 0.90)
- DISCOVER: Automated Curricula for Sparse-Reward Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Sample Efficient Myopic Exploration Through Multitask Reinforcement Learning with Diverse Tasks (ICLR 2024 · similarity 0.89)
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.