arXiv：cs.LG· Samuel Tetteh, Cody Fleming·· 4 小时前AI 评分35
平均安全、尾部不安全：情景成本尾部何时可控？
Safe on Average, Unsafe in the Tail: When Is the Episodic-Cost Tail Controllable?
AI 导读
安全强化学习通常只约束期望情景成本，而忽略成本在情景间的分布，导致策略可能平均安全却仍存在不安全的最坏情景。该研究用 CVaR_0.1（最差 10% 情景的平均成本）衡量情景成本尾部，并在三个 Safety-Gymnasium 导航任务上评估五种标准算法，识别出平均安全但尾部不安全的策略。随后在密集危险导航中考察四类约束，并在四个导航与四个运动任务上评估尾部控制能否在保持回报的同时实现。
正文
Abstract:Safe reinforcement learning seeks policies that maximize return while satisfying constraints on cumulative cost. Most methods impose these constraints on expected episodic cost. Consequently, standard evaluations report mean episodic cost without characterizing how cost is distributed across episodes. A policy that satisfies the mean-cost criterion may therefore remain unsafe in its worst episodes. Mean-cost reporting neither identifies this tail violation nor shows whether it can be brought within budget while preserving return. In this work, we measure the episodic-cost tail using $\mathrm{CVaR}_{0.1}$, the average cost of the worst $10\%$ of episodes. We classify a policy as tail-safe when $\mathrm{CVaR}_{0.1}$ is within the safety budget. This allows us first to identify policies that are safe on average but unsafe in the tail and then to study whether their tail violations can be controlled while preserving return. To identify tail-unsafe policies, we evaluate five standard algorithms on three Safety-Gymnasium navigation tasks. We then examine four constraint families on dense-hazard navigation and assess tail control across four navigation and four locomotion tasks.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.09508 [cs.LG] |
| (or arXiv:2610.09508v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09508 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Samuel Tetteh [view email]
[v1]
Wed, 7 Oct 2026 06:07:06 UTC (282 KB)
来源：arXiv：cs.LG · arxiv.org