Revisiting Policy Gradients for Restricted Policy Classes: Escaping Myopic Local Optima with $k$-step Policy Gradients

• URL: http://arxiv.org/abs/2605.10909v1
• Date: Mon, 11 May 2026 17:49:09 GMT
• Title: Revisiting Policy Gradients for Restricted Policy Classes: Escaping Myopic Local Optima with $k$-step Policy Gradients
• Abstract summary: This work revisits standard policy gradient methods used on restricted policy classes.<n>We propose a generalized $k$-step policy gradient method that couples the randomness within a $k$-step time window.<n>We show this new method is theoretically guaranteed to converge to a solution that is exponentially close in performance to the optimal deterministic policy.
• Score: 8.64427265159929
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: This work revisits standard policy gradient methods used on restricted policy classes, which are known to get stuck in suboptimal critical points. We identify an important cause for this phenomenon to be that the policy gradient is itself fundamentally myopic, i.e. it only improves the policy based on the one-step $Q$-function. In this work, we propose a generalized $k$-step policy gradient method that couples the randomness within a $k$-step time window and can escape the myopic local optima in MDPs with restricted policy classes. We show this new method is theoretically guaranteed to converge to a solution that is exponentially close in performance to the optimal deterministic policy with respect to $k$. Further, we show projected gradient descent and mirror descent with this $k$-step policy gradient can achieve this exponential guarantee in $O(\frac{1}{T})$ iterations, despite only assuming smoothness and differentiability of the value function. This will provide near optimal solutions to previously elusive applications like state aggregation and partially observable cooperative multi-agent settings. Moreover, our bounds avoid the ubiquitous distribution mismatch factors $||d_μ^{π^*} / d_μ^π||_\infty$ and $||d_μ^{π^*} / μ||_\infty$ enabling the $k$-step policy gradient method to escape suboptimal critical points that emerge from poor exploration in fully observable settings.

Related papers

• Achieve Performatively Optimal Policy for Performative Reinforcement Learning [55.983627302691424]
  This work proposes a zeroth-order FrankWolfe- (0FW) algorithm with a gradient of performative policy in the framework.<n> Experimental results demonstrate that our 0FW is more effective than the existing approximation in finding the desired PO policy.
  arXiv  Detail & Related papers  (2025-10-06T01:56:31Z)
• Convergence and Sample Complexity of First-Order Methods for Agnostic Reinforcement Learning [66.4260157478436]
  We study reinforcement learning in the policy learning setting.<n>The goal is to find a policy whose performance is competitive with the best policy in a given class of interest.
  arXiv  Detail & Related papers  (2025-07-06T14:40:05Z)
• Convergence of Policy Mirror Descent Beyond Compatible Function Approximation [66.4260157478436]
  We develop theoretical PMD general policy classes where we strictly assume a weaker variational dominance and obtain convergence to the best-in-class policy.<n>Our main notion leverages a novel notion induced by the local norm induced by the occupancy- gradient measure.
  arXiv  Detail & Related papers  (2025-02-16T08:05:46Z)
• Near-Optimal Policy Identification in Robust Constrained Markov Decision Processes via Epigraph Form [26.01796404477275]
  This paper presents the first algorithm guaranteed to identify a near-optimal policy in a robust constrained MDP (RCMDP)<n>An optimal policy minimizes cumulative cost while satisfying constraints in the worst-case scenario across a set of environments.
  arXiv  Detail & Related papers  (2024-08-29T06:37:16Z)
• Confident Natural Policy Gradient for Local Planning in $q_π$-realizable Constrained MDPs [44.69257217086967]
  The constrained Markov decision process (CMDP) framework emerges as an important reinforcement learning approach for imposing safety or other critical objectives.<n>In this paper, we address the learning problem given linear function approximation with $q_pi$-realizability.
  arXiv  Detail & Related papers  (2024-06-26T17:57:13Z)
• Last-Iterate Convergent Policy Gradient Primal-Dual Methods for Constrained MDPs [107.28031292946774]
  We study the problem of computing an optimal policy of an infinite-horizon discounted Markov decision process (constrained MDP) We develop two single-time-scale policy-based primal-dual algorithms with non-asymptotic convergence of their policy iterates to an optimal constrained policy. To the best of our knowledge, this work appears to be the first non-asymptotic policy last-iterate convergence result for single-time-scale algorithms in constrained MDPs.
  arXiv  Detail & Related papers  (2023-06-20T17:27:31Z)
• Homotopic Policy Mirror Descent: Policy Convergence, Implicit Regularization, and Improved Sample Complexity [40.2022466644885]
  homotopic policy mirror descent (HPMD) method for solving discounted, infinite horizon MDPs with finite state and action space. We report three properties that seem to be new in the literature of policy gradient methods.
  arXiv  Detail & Related papers  (2022-01-24T04:54:58Z)
• Understanding the Effect of Stochasticity in Policy Optimization [86.7574122154668]
  We show that the preferability of optimization methods depends critically on whether exact gradients are used. Second, to explain these findings we introduce the concept of committal rate for policy optimization. Third, we show that in the absence of external oracle information, there is an inherent trade-off between exploiting geometry to accelerate convergence versus achieving optimality almost surely.
  arXiv  Detail & Related papers  (2021-10-29T06:35:44Z)
• Softmax Policy Gradient Methods Can Take Exponential Time to Converge [60.98700344526674]
  The softmax policy gradient (PG) method is arguably one of the de facto implementations of policy optimization in modern reinforcement learning. We demonstrate that softmax PG methods can take exponential time -- in terms of $mathcalS|$ and $frac11-gamma$ -- to converge.
  arXiv  Detail & Related papers  (2021-02-22T18:56:26Z)
• On the Convergence and Sample Efficiency of Variance-Reduced Policy Gradient Method [38.34416337932712]
  Policy gives rise to a rich class of reinforcement learning (RL) methods, for example the REINFORCE. Yet the best known sample complexity result for such methods to find an $epsilon$-optimal policy is $mathcalO(epsilon-3)$, which is suboptimal. We study the fundamental convergence properties and sample efficiency of first-order policy optimization method.
  arXiv  Detail & Related papers  (2021-02-17T07:06:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.