The Role of Coverage in Online Reinforcement Learning

10/09/2022
by   Tengyang Xie, et al.
0

Coverage conditions – which assert that the data logging distribution adequately covers the state space – play a fundamental role in determining the sample complexity of offline reinforcement learning. While such conditions might seem irrelevant to online reinforcement learning at first glance, we establish a new connection by showing – somewhat surprisingly – that the mere existence of a data distribution with good coverage can enable sample-efficient online RL. Concretely, we show that coverability – that is, existence of a data distribution that satisfies a ubiquitous coverage condition called concentrability – can be viewed as a structural property of the underlying MDP, and can be exploited by standard algorithms for sample-efficient exploration, even when the agent does not know said distribution. We complement this result by proving that several weaker notions of coverage, despite being sufficient for offline RL, are insufficient for online RL. We also show that existing complexity measures for online RL, including Bellman rank and Bellman-Eluder dimension, fail to optimally capture coverability, and propose a new complexity measure, the sequential extrapolation coefficient, to provide a unification.

READ FULL TEXT
research
04/25/2023

Provable benefits of general coverage conditions in efficient online RL with function approximation

In online reinforcement learning (RL), instead of employing standard str...
research
11/21/2021

Offline Reinforcement Learning: Fundamental Barriers for Value Function Approximation

We consider the offline reinforcement learning problem, where the aim is...
research
10/22/2020

What are the Statistical Limits of Offline RL with Linear Function Approximation?

Offline reinforcement learning seeks to utilize offline (observational) ...
research
06/23/2023

Active Coverage for PAC Reinforcement Learning

Collecting and leveraging data with good coverage properties plays a cru...
research
06/05/2023

Survival Instinct in Offline Reinforcement Learning

We present a novel observation about the behavior of offline reinforceme...
research
11/23/2021

Understanding the Impact of Data Distribution on Q-learning with Function Approximation

In this work, we focus our attention on the study of the interplay betwe...
research
10/29/2019

Overcoming Catastrophic Interference in Online Reinforcement Learning with Dynamic Self-Organizing Maps

Using neural networks in the reinforcement learning (RL) framework has a...

Please sign up or login with your details

Forgot password? Click here to reset