Abstract
How could we solve the machine learning and the artificial intelligence problem if we had infinite computation? Solomonoff induction and the reinforcement learning agent AIXI are proposed answers to this question. Both are known to be incomputable. In this paper, we quantify this using the arithmetical hierarchy, and prove upper and corresponding lower bounds for incomputability. We show that AIXI is not limit computable, thus it cannot be approximated using finite computation. Our main result is a limitcomputable ε-optimal version of AIXI with infinite horizon that maximizes expected rewards.
| Original language | English |
|---|---|
| Title of host publication | Uncertainty in Artificial Intelligence - Proceedings of the 31st Conference, UAI 2015 |
| Editors | Heskes T.Meila M. |
| Place of Publication | TBC |
| Publisher | AUAI Press |
| Pages | 464-473 |
| Edition | peer reviewed |
| Publication status | Published - 2015 |
| Event | Conference on Uncertainty in Artificial Intelligence, UAI 2015 - Amsterdam, Netherland Duration: 1 Jan 2015 → … |
Conference
| Conference | Conference on Uncertainty in Artificial Intelligence, UAI 2015 |
|---|---|
| Period | 1/01/15 → … |
| Other | July 12-16 2015 |