Tabular Markov Decision Process

Tabular Markov Decision Processes (MDPs) are a fundamental framework for sequential decision-making problems with a finite number of states and actions, aiming to find optimal policies that maximize cumulative rewards. Current research emphasizes developing efficient algorithms for policy evaluation and optimization, focusing on areas like safe data collection, principled policy gradient methods, and federated learning approaches for collaborative multi-agent settings. These advancements improve sample efficiency, address safety constraints, and enable scalable solutions for complex real-world applications, impacting fields such as robotics, personalized medicine, and resource management.

Papers

June 24, 2022

Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems
Masatoshi Uehara, Ayush Sekhari, Jason D. Lee, Nathan Kallus, Wen Sun
Reinforcement Learning Tabular Markov Decision Process Thermodynamic Observables Partially Observed

June 9, 2022

Regret Bounds for Information-Directed Reinforcement Learning
Botao Hao, Tor Lattimore
Reinforcement Learning Regret Bound Tabular Markov Decision Process Information Directed Sampling Prior Dependent

June 5, 2022

Formally Verified Solution Methods for Infinite-Horizon Markov Decision Processes
Maximilian Schäfeller, Mohammad Abdulaziz
Markov Decision Process Optimal Policy Dynamic Programming Infinite Horizon Tabular Markov Decision Process Interactive Theorem

May 31, 2022

Nearly Minimax Optimal Offline Reinforcement Learning with Linear Function Approximation: Single-Agent MDP and Markov Game
Wei Xiong, Han Zhong, Chengshuai Shi, Cong Shen, Liwei Wang, Tong Zhang
Offline Reinforcement Learning Markov Game Single Agent Linear Function Approximation Offline Multi Agent Reinforcement Learning Tabular Markov Decision Process Minimax Optimal Algorithm

April 26, 2022

BATS: Best Action Trajectory Stitching
Ian Char, Viraj Mehta, Adam Villaflor, John M. Dolan, Jeff Schneider
Offline Reinforcement Learning Optimal Policy Online Reinforcement Learning Learned Dynamic Tabular Markov Decision Process Captive Egyptian Fruit Bat Trajectory Stitching

March 24, 2022

Horizon-Free Reinforcement Learning in Polynomial Time: the Power of Stationary Policies
Zihan Zhang, Xiangyang Ji, Simon S. Du
Real Power Markov Chain Polynomial Time Tabular Markov Decision Process Stationary Policy

January 21, 2022

Optimal variance-reduced stochastic approximation in Banach spaces
Wenlong Mou, Koulik Khamaru, Martin J. Wainwright, Peter L. Bartlett, Michael I. Jordan
Worst Case Zero Sum Markov Game Tabular Markov Decision Process Banach Space Compression Operator Variance Reduced Stochastic

December 20, 2021

Differentially Private Regret Minimization in Episodic Markov Decision Processes
Sayak Ray Chowdhury, Xingyu Zhou
Markov Decision Process Regret Bound Regret Minimization Sublinear Regret Episodic Markov Decision Process Tabular Markov Decision Process

Tabular Markov Decision Process

Papers

Provably Efficient Reinforcement Learning in Partially Observable Dynamical Systems

Regret Bounds for Information-Directed Reinforcement Learning

Formally Verified Solution Methods for Infinite-Horizon Markov Decision Processes

Nearly Minimax Optimal Offline Reinforcement Learning with Linear Function Approximation: Single-Agent MDP and Markov Game

BATS: Best Action Trajectory Stitching

Horizon-Free Reinforcement Learning in Polynomial Time: the Power of Stationary Policies

Optimal variance-reduced stochastic approximation in Banach spaces

Differentially Private Regret Minimization in Episodic Markov Decision Processes