Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Uncertainty Quantification and Exploration for Reinforcement Learning

About

We investigate statistical uncertainty quantification for reinforcement learning (RL) and its implications in exploration policy. Despite ever-growing literature on RL applications, fundamental questions about inference and error quantification, such as large-sample behaviors, appear to remain quite open. In this paper, we fill in the literature gap by studying the central limit theorem behaviors of estimated Q-values and value functions under various RL settings. In particular, we explicitly identify closed-form expressions of the asymptotic variances, which allow us to efficiently construct asymptotically valid confidence regions for key RL quantities. Furthermore, we utilize these asymptotic expressions to design an effective exploration strategy, which we call Q-value-based Optimal Computing Budget Allocation (Q-OCBA). The policy relies on maximizing the relative discrepancies among the Q-value estimates. Numerical experiments show superior performances of our exploration strategy than other benchmark policies.

YI Zhu, Jing Dong, Henry Lam• 2019

Related benchmarks

TaskDatasetResultRank
Empirical Coverage EstimationRiverSwim
Q^π(1, 0)0.948
120
Optimal Policy SelectionStochastic Gridworld (S=16, A=4)
PCS92
28
Action-Value coverage estimationRiverSwim mostly-right target policy T=50
Q-Value Estimate (s=1, a=0)0.503
20
State Value Estimation CoverageRiverSwim
Value Estimate State 10.949
20
State-Action Value Estimation CoverageRiverSwim
Q-Value Estimate (s=1, a=0)0.949
20
State-Value coverage estimationRiverSwim mostly-right target policy T=50
V(s=1)0.503
20
Empirical Coverage EstimationRiverSwim episode length T = 10 (nominal 95% coverage)
Q* (1, 0)38.8
20
Off-policy EvaluationRiverSwim mostly-left policy, T=50
Qπ(1, 0) Coverage52.3
20
Empirical Coverage EstimationRiverSwim T=50 90% nominal coverage
Q* (1, 0)47.8
20
Optimal Policy Recovery (Empirical Coverage)RiverSwim T=50 nominal 95% coverage
Q* Recovery (s=1, a=0)49.7
20
Showing 10 of 14 rows

Other info

Follow for update