PAC-Bayesian Lifelong Learning For Multi-Armed Bandits
Abstract
We present a PAC-Bayesian analysis of lifelong learning. In the lifelong learning problem, a sequence of learning tasks is observed one-at-a-time, and the goal is to transfer information acquired from previous tasks to new learning tasks. We consider the case when each learning task is a multi-armed bandit problem. We derive lower bounds on the expected average reward that would be obtained if a given multi-armed bandit algorithm was run in a new task with a particular prior and for a set number of steps. We propose lifelong learning algorithms that use our new bounds as learning objectives. Our proposed algorithms are evaluated in several lifelong multi-armed bandit problems and are found to perform better than a baseline method that does not use generalisation bounds.
Cite
@article{arxiv.2203.03303,
title = {PAC-Bayesian Lifelong Learning For Multi-Armed Bandits},
author = {Hamish Flynn and David Reeb and Melih Kandemir and Jan Peters},
journal= {arXiv preprint arXiv:2203.03303},
year = {2022}
}
Comments
29 pages, 5 figures