Research map / Reinforcement learning / MDPs and planning
Safe Exploration Incurs Nearly No Additional Sample Complexity for Reward-Free RL
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 safe exploration using a known safe baseline policy | Tabular-SWEET · Tabular MDP reward-free exploration | trajectory sample complexity: \tilde{O}\left(HS^2A/\epsilon^2\right) (constraint-free); safe bound: \tilde{O}\left(\frac{HSA(S+\log(1/\delta))}{\Delta | No clear effect |
| Add safe exploration using a known safe baseline policy | Low-rank-SWEET · Low-rank MDP reward-free exploration | trajectory sample complexity: \tilde{O}\left(H^3d^4A^2/\epsilon^2\right) (constraint-free); safe bound: \tilde{O}\left(\frac{H^3d^4A^2\log(1/\delta)}{ | No clear effect |
| Use truncated value-function clipping rather than conventional clipping of the action-value function | SWEET · Safe reward-free exploration and planning | approximation-error estimation | Helped |
Closest papers by meaning
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · similarity 0.91)
- Risk-Sensitive Reward-Free Reinforcement Learning with CVaR (ICML 2024 · similarity 0.91)
- Safe Reinforcement Learning with Instantaneous Constraints: The Role of Aggressive Exploration (AAAI 2024 · similarity 0.91)
- Balance Reward and Safety Optimization for Safe Reinforcement Learning: A Perspective of Gradient Manipulation (AAAI 2024 · similarity 0.89)
- Safeguarded Progress in Reinforcement Learning: Safe Bayesian Exploration for Control Policy Synthesis (AAAI 2024 · similarity 0.89)
- Uncertainty-Aware Reward-Free Exploration with General Function Approximation (ICML 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.