Acting irrationally to improve performance in stochastic worlds

Belavkin, Roman V. (2005) Acting irrationally to improve performance in stochastic worlds. In: 25th SGAI International Conference on Innovative Techniques and Applications of Artificial Intelligence, 2005, Cambridge, UK.

[img]
Preview
PDF (paper) - Final accepted version (with author's formatting)
Download (165kB)

Abstract

Despite many theories and alogorithms for decision-making, after estimating the utility function the choice is usually made by maximising its expected value (the max EU principle). This traditional and 'rational' conclusion of the decision-making process is compared in this paper with several 'irrational' techniques that make choice in Monte-Carlo fashion. The comparison is made by evaluating the performance of simple decision-theoretic agents in stochastic environments. It is shown that not only the random choice strategies can achieve performance comparable to the max EU method, but under certain conditions the Monte-Carlo choice methods perform almost two times better than the max EU. The paper concludes by quoting evidence from recent cognitive modelling works as well as the famous decision-making paradoxes.

Item Type: Conference or Workshop Item (Paper)
Research Areas: A. > School of Science and Technology > Computer Science
A. > School of Science and Technology > Computer Science > Artificial Intelligence group
ISI Impact: 0
Item ID: 6
Useful Links:
Depositing User: Repository team
Date Deposited: 08 Sep 2008 11:13
Last Modified: 01 May 2015 12:32
URI: http://eprints.mdx.ac.uk/id/eprint/6

Actions (login required)

Edit Item Edit Item

Full text downloads (NB count will be zero if no full text documents are attached to the record)

Downloads per month over the past year