Research map / Bandits and online learning / RL theory and MDPs
A Near-Optimal Algorithm for Safe Reinforcement Learning Under Instantaneous Hard Constraints
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 |
|---|---|---|---|
| Construct safe subgraphs backward so chosen actions lead only to safe future states | LSVI-NEW · Episodic linear mixture MDP with unsafe states and actions | instantaneous hard-constraint satisfaction | Helped |
| Add bonuses for uncertain transitions and future and past subsubgraphs | LSVI-NEW · Episodic linear mixture MDP with unsafe states and actions | regret: \tilde{O}\left( \frac{dH^3 \sqrt{dK}}{\bar{c}-\bar{c}_1^0-\Delta_{\boldsymbol{\phi}}(c)} \right) | Helped |
| Handle unsafe states and actions under instantaneous hard constraints | LSVI-NEW · Episodic linear mixture MDP | regret: \tilde{O}\left( \frac{dH^3 \sqrt{dK}}{\bar{c}-\bar{c}_1^0-\Delta_{\boldsymbol{\phi}}(c)} \right); \tilde{O}(dH^2\sqrt{K} | Mixed |
| Impose the instantaneous safety requirement | Any safe algorithm · Episodic linear mixture MDP with unsafe states and actions | regret: \max\left\{ \frac{dH\sqrt{K}}{16\sqrt{2}}, \frac{H/24}{(\bar{c}-\bar{c}_1^0-\Delta_{\boldsymbol{\phi}}(c))^2} \right\} | Hurt |
Closest papers by meaning
- Safe Reinforcement Learning with Instantaneous Constraints: The Role of Aggressive Exploration (AAAI 2024 · similarity 0.94)
- Long-Term Safe Reinforcement Learning with Binary Feedback (AAAI 2024 · similarity 0.92)
- Near-Optimal Sample Complexity for Online Constrained MDPs (NeurIPS 2025 · similarity 0.91)
- Probabilistic Shielding for Safe Reinforcement Learning (AAAI 2025 · similarity 0.90)
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · similarity 0.90)
- Pessimism Meets Risk: Risk-Sensitive Offline Reinforcement Learning (ICML 2024 · 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.