Instance-Dependent Regret for CMDPs with Step-Wise Constraints
We study online learning in episodic tabular constrained Markov decision processes with step-wise safety constraints. In such a setting, the constraints induce a safe subgraph that shapes the variance of cumulative rewards under feasible policies and, consequently, the difficulty of learning. Exploiting this structure,...
Qiang Zuo, F. Stradi, Le-Yang Xue et al.
· 0 citations