On Policy Evaluation Algorithms in Distributional Reinforcement Learning

• URL: http://arxiv.org/abs/2407.14175v1
• Date: Fri, 19 Jul 2024 10:06:01 GMT
• Title: On Policy Evaluation Algorithms in Distributional Reinforcement Learning
• Authors: Julian Gerstenberg, Ralph Neininger, Denis Spiegel,
• Abstract summary: We introduce a novel class of algorithms to efficiently approximate the unknown return distributions in policy evaluation problems from distributional reinforcement learning (DRL) For a plain instance of our proposed class of algorithms we prove error bounds, both within Wasserstein and Kolmogorov--Smirnov distances. For return distributions having probability density functions the algorithms yield approximations for these densities; error bounds are given within supremum norm.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We introduce a novel class of algorithms to efficiently approximate the unknown return distributions in policy evaluation problems from distributional reinforcement learning (DRL). The proposed distributional dynamic programming algorithms are suitable for underlying Markov decision processes (MDPs) having an arbitrary probabilistic reward mechanism, including continuous reward distributions with unbounded support being potentially heavy-tailed. For a plain instance of our proposed class of algorithms we prove error bounds, both within Wasserstein and Kolmogorov--Smirnov distances. Furthermore, for return distributions having probability density functions the algorithms yield approximations for these densities; error bounds are given within supremum norm. We introduce the concept of quantile-spline discretizations to come up with algorithms showing promising results in simulation experiments. While the performance of our algorithms can rigorously be analysed they can be seen as universal black box algorithms applicable to a large class of MDPs. We also derive new properties of probability metrics commonly used in DRL on which our quantitative analysis is based.

Related papers

• Distributed Markov Chain Monte Carlo Sampling based on the Alternating Direction Method of Multipliers [143.6249073384419]
  In this paper, we propose a distributed sampling scheme based on the alternating direction method of multipliers. We provide both theoretical guarantees of our algorithm's convergence and experimental evidence of its superiority to the state-of-the-art. In simulation, we deploy our algorithm on linear and logistic regression tasks and illustrate its fast convergence compared to existing gradient-based methods.
  arXiv  Detail & Related papers  (2024-01-29T02:08:40Z)
• A unified consensus-based parallel ADMM algorithm for high-dimensional regression with combined regularizations [3.280169909938912]
  parallel alternating multipliers (ADMM) is widely recognized for its effectiveness in handling large-scale distributed datasets. The proposed algorithms serve to demonstrate the reliability, stability, and scalability of a financial example.
  arXiv  Detail & Related papers  (2023-11-21T03:30:38Z)
• Provably Efficient UCB-type Algorithms For Learning Predictive State Representations [55.00359893021461]
  The sequential decision-making problem is statistically learnable if it admits a low-rank structure modeled by predictive state representations (PSRs) This paper proposes the first known UCB-type approach for PSRs, featuring a novel bonus term that upper bounds the total variation distance between the estimated and true models. In contrast to existing approaches for PSRs, our UCB-type algorithms enjoy computational tractability, last-iterate guaranteed near-optimal policy, and guaranteed model accuracy.
  arXiv  Detail & Related papers  (2023-07-01T18:35:21Z)
• Multivariate Systemic Risk Measures and Computation by Deep Learning Algorithms [63.03966552670014]
  We discuss the key related theoretical aspects, with a particular focus on the fairness properties of primal optima and associated risk allocations. The algorithms we provide allow for learning primals, optima for the dual representation and corresponding fair risk allocations.
  arXiv  Detail & Related papers  (2023-02-02T22:16:49Z)
• Asynchronous Distributed Reinforcement Learning for LQR Control via Zeroth-Order Block Coordinate Descent [7.6860514640178]
  We propose a novel zeroth-order optimization algorithm for distributed reinforcement learning. It allows each agent to estimate its local gradient by cost evaluation independently, without use of any consensus protocol.
  arXiv  Detail & Related papers  (2021-07-26T18:11:07Z)
• Efficient semidefinite-programming-based inference for binary and multi-class MRFs [83.09715052229782]
  We propose an efficient method for computing the partition function or MAP estimate in a pairwise MRF. We extend semidefinite relaxations from the typical binary MRF to the full multi-class setting, and develop a compact semidefinite relaxation that can again be solved efficiently using the solver.
  arXiv  Detail & Related papers  (2020-12-04T15:36:29Z)
• Implicit Distributional Reinforcement Learning [61.166030238490634]
  implicit distributional actor-critic (IDAC) built on two deep generator networks (DGNs) Semi-implicit actor (SIA) powered by a flexible policy distribution. We observe IDAC outperforms state-of-the-art algorithms on representative OpenAI Gym environments.
  arXiv  Detail & Related papers  (2020-07-13T02:52:18Z)
• Study of Diffusion Normalized Least Mean M-estimate Algorithms [0.8749675983608171]
  This work proposes diffusion normalized least mean M-estimate algorithm based on the modified Huber function. We analyze the transient, steady-state and stability behaviors of the algorithms in a unified framework. Simulations in various impulsive noise scenarios show that the proposed algorithms are superior to some existing diffusion algorithms.
  arXiv  Detail & Related papers  (2020-04-20T00:28:41Z)
• A Distributional Analysis of Sampling-Based Reinforcement Learning Algorithms [67.67377846416106]
  We present a distributional approach to theoretical analyses of reinforcement learning algorithms for constant step-sizes. We show that value-based methods such as TD($lambda$) and $Q$-Learning have update rules which are contractive in the space of distributions of functions.
  arXiv  Detail & Related papers  (2020-03-27T05:13:29Z)

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.