Provably Efficient Model-Free Constrained RL with Linear Function Approximation. (arXiv:2206.11889v1 [cs.LG])

Web: http://arxiv.org/abs/2206.11889

We study the constrained reinforcement learning problem, in which an agent
aims to maximize the expected cumulative reward subject to a constraint on the
expected total value of a utility function. In contrast to existing model-based
approaches or model-free methods accompanied with a `simulator', we aim to
develop the first model-free, simulator-free algorithm that achieves a
sublinear regret and a sublinear constraint violation even in large-scale
systems. To this end, we consider the episodic constrained Markov decision
processes with linear …

approximation arxiv free function lg linear model rl