ORAUE: Optimal Rational Agents in Unknown Environments
Goal of this project is to extend and deepen a recent theory of theoretically optimal universal agents interacting with unknown environments. We build on Solomonoff's celebrated universal theory of induction to derive an optimal reinforcement learning agent, called AIXI, embedded in a world whose responses to the agent's actions are sampled from a computable probability distribution -- this is the only very weak assumption.
From an algorithmic complexity perspective, the AIXI model generalizes optimal passive universal induction to the case of active agents. From a decision theoretic perspective, the AIXI model is a suggestion of a new (implicit) "learning" algorithm, which may overcome all (except computational) problems of previous reinforcement learning algorithms. If the optimality theorems of universal induction and decision theory generalize to the unified AIXI model, we would have, for the first time, a universal (parameterless) model of an optimal rational agent in any computable but unknown environment with reinforcement feedback.

Keywords: algorithmic complexity, sequential decision theory, induction, time series prediction, reinforcement learning, strategic games, function minimization, supervised learning.
People Jan Poland
Shane Legg
Marcus Hutter



Copyright © 2003 - 2012 IDSIA. All Rights Reserved.