LGOCMLFeb 18, 2021

Finite-Sample Analysis of Off-Policy Natural Actor-Critic Algorithm

arXiv:2102.09318v234 citations
Originality Incremental advance
AI Analysis

This addresses the challenge of sample efficiency and variance control in reinforcement learning for practitioners, though it is incremental as it builds on existing methods like V-trace.

The paper tackles the problem of providing finite-sample convergence guarantees for an off-policy natural actor-critic algorithm, achieving a sample complexity of O(ε^{-3} log^2(1/ε)) to converge to a global optimal policy.

In this paper, we provide finite-sample convergence guarantees for an off-policy variant of the natural actor-critic (NAC) algorithm based on Importance Sampling. In particular, we show that the algorithm converges to a global optimal policy with a sample complexity of $\mathcal{O}(ε^{-3}\log^2(1/ε))$ under an appropriate choice of stepsizes. In order to overcome the issue of large variance due to Importance Sampling, we propose the $Q$-trace algorithm for the critic, which is inspired by the V-trace algorithm \cite{espeholt2018impala}. This enables us to explicitly control the bias and variance, and characterize the trade-off between them. As an advantage of off-policy sampling, a major feature of our result is that we do not need any additional assumptions, beyond the ergodicity of the Markov chain induced by the behavior policy.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes