We use optimism to introduce generic asymptotically optimal reinforcement learning agents. They achieve, with an arbitrary finite or compact class of environments, asymptotically optimal behavior. Furthermore, in the finite deterministic case we provide finite error bounds.
@article{arxiv.1210.0077,
title = {Optimistic Agents are Asymptotically Optimal},
author = {Peter Sunehag and Marcus Hutter},
journal= {arXiv preprint arXiv:1210.0077},
year = {2013}
}