Research map / Reinforcement learning / Policy gradient and actor-critic
Safe Exploration in Reinforcement Learning: A Generalized Formulation and Algorithms
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 |
|---|---|---|---|
| Use MASE with GP-based safety estimation and TRPO policy optimization instead of Sauté RL | Safety Gym PointGoal1 and CarGoal1 · Safety Gym | Cumulative reward during training | Helped |
| Use MASE with GP-based safety estimation and TRPO policy optimization instead of baseline algorithms | Safety Gym PointGoal1 and CarGoal1 · Safety Gym | Safety constraint violations | Helped |
| Use MASE with GP-based safety estimation and TRPO policy optimization instead of baseline algorithms | Safety Gym CarGoal1 · Safety Gym, converged performance | Reward | Hurt |
| Use MASE's preemptive emergency stop instead of ET-MDP's termination after a safety violation | Safety Gym PointGoal1 and CarGoal1 · Safety Gym | Safety constraint violations | Helped |
| Use MASE instead of SNO-MDP | Grid-world · Grid-world, time-invariant and time-variant safety thresholds | Reward and safety violations | Mixed |
| Use a safety-agnostic or expected-cost baseline instead of MASE | TRPO, TRPO-Lagrangian, and CPO · Safety Gym | Safety constraint violations | Hurt |
Closest papers by meaning
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · similarity 0.92)
- Safeguarded Progress in Reinforcement Learning: Safe Bayesian Exploration for Control Policy Synthesis (AAAI 2024 · similarity 0.91)
- Balance Reward and Safety Optimization for Safe Reinforcement Learning: A Perspective of Gradient Manipulation (AAAI 2024 · similarity 0.91)
- Safe Reinforcement Learning with Instantaneous Constraints: The Role of Aggressive Exploration (AAAI 2024 · similarity 0.90)
- Long-Term Safe Reinforcement Learning with Binary Feedback (AAAI 2024 · similarity 0.90)
- ActSafe: Active Exploration with Safety Constraints for Reinforcement Learning (ICLR 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.