Variance Reduction Methods for Sublinear Reinforcement Learning

Abstract

There is a technical issue in the analysis that is not easily fixable. We, therefore, withdraw the submission. Sorry for the inconvenience.

arXiv commentThere is a technical issue in the analysis that is not easily fixable

References in corpus (2)

• On Lower Bounds for Regret in Reinforcement Learning
• Near-Optimal Time and Sample Complexities for Solving Discounted Markov Decision Process with a Generative Model

Cited by in corpus (17)

• Algorithmic Framework for Model-based Deep Reinforcement Learning with Theoretical Guarantees
• Tighter Problem-Dependent Regret Bounds in Reinforcement Learning without Domain Knowledge using Value Function Bounds
• Almost Optimal Model-Free Reinforcement Learning via Reference-Advantage Decomposition
• Learning Linear Dynamical Systems with Semi-Parametric Least Squares
• Agnostic Q-learning with Function Approximation in Deterministic Systems: Tight Bounds on Approximation Error and Sample Complexity
• Is Q-learning Provably Efficient?
• MADE: Exploration via Maximizing Deviation from Explored Regions
• Regret Minimization for Reinforcement Learning by Evaluating the Optimal Bias Function
• Max-Plus Matching Pursuit for Deterministic Markov Decision Processes
• Towards Minimax Optimal Reinforcement Learning in Factored Markov Decision Processes
• Exploration Bonus for Regret Minimization in Undiscounted Discrete and Continuous Markov Decision Processes
• Provably Efficient Model-Free Algorithm for MDPs with Peak Constraints
• Posterior sampling for reinforcement learning: worst-case regret bounds
• Provably Efficient Exploration for Reinforcement Learning Using Unsupervised Learning
• Is Q-Learning Provably Efficient? An Extended Analysis
• Model-Free Algorithm and Regret Analysis for MDPs with Long-Term Constraints
• Planning with Submodular Objective Functions