Why Policy Gradient Algorithms Work for Undiscounted Total-Reward MDPs

Jongmin Lee, Ernest K. Ryu

Keywords: undiscoutned expected total-reward infinite horizon MDPs, visitation measure, recurrent and transient states, policy gradient theorem, natural policy gradient, projected policy gradient, convergence analysis

Abstract: The classical policy gradient method is the theoretical and conceptual foundation of modern policy-based reinforcement learning (RL) algorithms. Most rigorous analyses of such methods, particularly those establishing convergence guarantees, assume a discount factor $\gamma < 1$. In contrast, however, a recent line of work on policy-based RL for large language models uses the undiscounted total-reward setting with $\gamma = 1$, rendering much of the existing theory inapplicable. In this paper, we provide analyses of the policy gradient method for undiscounted expected total-reward infinite-horizon MDPs based on two key insights: (i) the classification of the MDP states into recurrent and transient states is invariant over the set of policies that assign strictly positive probability to every action (as is typical in deep RL models employing a softmax output layer) and (ii) the classical state visitation measure (which may be ill-defined when $\gamma = 1$) can be replaced with a new object that we call the transient visitation measure.

Primary Area: reinforcement learning

Submission Number: 9947