Skip navigation
Skip navigation

Axioms for rational reinforcement learning

Sunehag, Peter; Hutter, Marcus

Description

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...[Show more]

dc.contributor.authorSunehag, Peter
dc.contributor.authorHutter, Marcus
dc.date.accessioned2015-08-20T04:35:56Z
dc.date.available2015-08-20T04:35:56Z
dc.identifier.isbn978-3-642-24411-7
dc.identifier.issn0302-9743
dc.identifier.urihttp://hdl.handle.net/1885/14813
dc.description.abstractWe 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].
dc.publisherSpringer Verlag
dc.relation.ispartofAlgorithmic learning theory : 22nd international conference, ALT 2011, Espoo, Finland, October 5-7, 2011 : proceedings
dc.rights© Springer-Verlag Berlin Heidelberg 2011. http://www.sherpa.ac.uk/romeo/issn/0302-9743/..."Author's post-print on any open access repository after 12 months after publication" from SHERPA/RoMEO site (as at 20/08/15)
dc.subjectRationality
dc.subjectProbability
dc.subjectUtility
dc.subjectBanach Space
dc.subjectLinear Functional
dc.titleAxioms for rational reinforcement learning
dc.typeConference paper
local.identifier.citationvolume6925
dc.date.issued2011-10
local.type.statusAccepted Version
local.contributor.affiliationSunehag, P., Research School of Computer Science, The Australian National University
local.contributor.affiliationHutter, M., Research School of Computer Science, The Australian National University
dc.relationhttp://purl.org/au-research/grants/arc/DP0988049
local.bibliographicCitation.startpage338
local.bibliographicCitation.lastpage352
local.identifier.doi10.1007/978-3-642-24412-4_27
CollectionsANU Research Publications

Download

File Description SizeFormat Image
Sunehag and Hutter Axioms for Rational Reinforcement Learning 2011.pdf125.2 kBAdobe PDFThumbnail


Items in Open Research are protected by copyright, with all rights reserved, unless otherwise indicated.

Updated:  19 May 2020/ Responsible Officer:  University Librarian/ Page Contact:  Library Systems & Web Coordinator