Efficient reinforcement learning in deterministic systems with value function generalization
From MaRDI portal
Abstract: We consider the problem of reinforcement learning over episodes of a finite-horizon deterministic system and as a solution propose optimistic constraint propagation (OCP), an algorithm designed to synthesize efficient exploration and value function generalization. We establish that when the true value function lies within a given hypothesis class, OCP selects optimal actions over all but at most K episodes, where K is the eluder dimension of the given hypothesis class. We establish further efficiency and asymptotic performance guarantees that apply even if the true value function does not lie in the given hypothesis class, for the special case where the hypothesis class is the span of pre-specified indicator functions over disjoint sets. We also discuss the computational complexity of OCP and present computational results involving two illustrative examples.
Recommendations
- Adaptive-resolution reinforcement learning with polynomial exploration in deterministic domains
- Deep exploration via randomized value functions
- Towards min max generalization in reinforcement learning
- The effect of representation and knowledge on goal-directed exploration with reinforcement-learning algorithms
- scientific article; zbMATH DE number 1804129
Cited in
(12)- Min max generalization for deterministic batch mode reinforcement learning: relaxation schemes
- Towards min max generalization in reinforcement learning
- scientific article; zbMATH DE number 5957431 (Why is no real title available?)
- scientific article; zbMATH DE number 5957492 (Why is no real title available?)
- scientific article; zbMATH DE number 5525811 (Why is no real title available?)
- Deep exploration via randomized value functions
- Rationality, optimism and guarantees in general reinforcement learning
- General value function networks
- Provably Efficient Reinforcement Learning with Linear Function Approximation
- Faithful and Effective Reward Schemes for Model-Free Reinforcement Learning of Omega-Regular Objectives
- Sample-based planning and learning with function approximation
- Risk-Sensitive Deep RL: Variance-Constrained Actor-Critic Provably Finds Globally Optimal Policy
This page was built for publication: Efficient reinforcement learning in deterministic systems with value function generalization
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5359119)
