Axioms for rational reinforcement learning

Authors:
Peter Sunehag;Marcus Hutter
Affiliations:
Research School of Computer Science, Australian National University, Canberra, ACT, Australia;Research School of Computer Science, Australian National University, Canberra, ACT, Australia
Venue:
ALT'11 Proceedings of the 22nd international conference on Algorithmic learning theory
Year:
2011

Citing 3
Cited 2

The uncertain reasoner's companion: a mathematical perspective

The uncertain reasoner's companion: a mathematical perspective
A counterexample to theorems of Cox and fine

Journal of Artificial Intelligence Research
Universal Artificial Intelligence: Sequential Decisions Based on Algorithmic Probability

Universal Artificial Intelligence: Sequential Decisions Based on Algorithmic Probability

Optimistic AIXI

AGI'12 Proceedings of the 5th international conference on Artificial General Intelligence
Learning agents with evolving hypothesis classes

AGI'13 Proceedings of the 6th international conference on Artificial General Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

We provide a formal, simple and intuitive theory of rational decision making including sequential decisions that affect the environment. The theory has a geometric flavor, which makes the arguments easy to visualize and understand. Our theory is for complete decision makers, which means that they have a complete set of preferences. Our main result shows that a complete rational decision maker implicitly has a probabilistic model of the environment. We have a countable version of this result that brings light on the issue of countable vs finite additivity by showing how it depends on the geometry of the space which we have preferences over. This is achieved through fruitfully connecting rationality with the Hahn-Banach Theorem. The theory presented here can be viewed as a formalization and extension of the betting odds approach to probability of Ramsey and De Finetti [Ram31, deF37].