OpenAI 官方博客T1·Models
45 pts
This study establishes a benchmark for safe exploration in deep reinforcement learning. It quantifies risk constraints during agent exploration, providing a standardized testing framework for algorithm safety.
安全探索
Strategies to avoid high-risk states during trial-and-error processes
This study establishes a benchmark for safe exploration in deep reinforcement learning. It quantifies risk constraints during agent exploration, providing a standardized testing framework for algorithm safety.