2024/02/07 by Isaac Grosof, Siva Theja Maguluri, Grosof, Isaac +3 · 1 citation
Computer Science · Decision Sciences · Economics, Econometrics and Finance · #Economic Policies and Impacts #FOS: Computer and information sciences #Machine Learning (cs.LG) #Optimization and Search Problems #Risk and Portfolio Optimization
paper · pdf · doi:10.48550/arxiv.2402.05274
openalex publication_date 2024/02/07 · openalex created_date 2024/02/10 · openalex updated_date 2026/07/28
A wide variety of queueing systems can be naturally modeled as infinite-state Markov Decision Processes (MDPs). In the reinforcement learning (RL) context, a variety of algorithms have been developed to learn and optimize these MDPs. At the heart of many popular policy-gradient based learning algorithms, such as natural actor-critic, TRPO, and PPO, lies the Natural Policy Gradient (NPG) policy optimization algorithm. Convergence results for these RL algorithms rest on convergence results for the NPG algorithm. However, all existing results on the convergence of the NPG algorithm are limited to finite-state settings. We study a general class of queueing MDPs, and prove a O(1/√(T)) convergence rate for the NPG algorithm, if the NPG algorithm is initialized with the MaxWeight policy. This is the first convergence rate bound for the NPG algorithm for a general class of infinite-state average-reward MDPs. Moreover, our result applies to a beyond the queueing setting to any countably-infinite MDP satisfying certain mild structural assumptions, given a sufficiently good initial policy. Key to our result are state-dependent bounds on the relative value function achieved by the iterate policies of the NPG algorithm.