Approximate Dynamic Programming By Minimizing Distributionally Robust\n Bounds

2012/05/08 by Marek Petrik, Petrik, Marek
Computer Science · Decision Sciences · Engineering · #Adaptive Dynamic Programming Control #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Reinforcement Learning in Robotics #Risk and Portfolio Optimization #Smart Grid Energy Management

paper · pdf · doi:10.48550/arxiv.1205.1782

openalex publication_date 2012/05/08 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Approximate dynamic programming is a popular method for solving large Markov\ndecision processes. This paper describes a new class of approximate dynamic\nprogramming (ADP) methods- distributionally robust ADP-that address the curse\nof dimensionality by minimizing a pessimistic bound on the policy loss. This\napproach turns ADP into an optimization problem, for which we derive new\nmathematical program formulations and analyze its properties. DRADP improves on\nthe theoretical guarantees of existing ADP methods-it guarantees convergence\nand L1 norm based error bounds. The empirical evaluation of DRADP shows that\nthe theoretical guarantees translate well into good performance on benchmark\nproblems.\n

Related

• Scaling Up Robust MDPs by Reinforcement Learning
• A New Optimal Stepsize For Approximate Dynamic Programming
• An Alternating Approach to Approximate Dynamic Programming
• A Generalized Reduced Linear Program for Markov Decision Processes
• Optimistic Planning by Regularized Dynamic Programming