A Problem Case for UCT

Cameron Browne

doi:10.1109/TCIAIG.2012.2220138

A Problem Case for UCT

Cameron Browne^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

This paper examines a simple 5 5 Hex position that not only completely defeats flat Monte Carlo search, but also initially defeats plain upper confidence bounds for trees (UCT) search until an excessive number of iterations are performed. The inclusion of domain knowledge during playouts significantly improves UCT performance, but a slight negative effect is shown for the rapid action value estimate (RAVE) heuristic under some circumstances. This example was drawn from an actual game during standard play, and highlights the dangers of relying on flat Monte Carlo and unenhanced UCT search even for rough estimates. A brief comparison is made with RAVE failure in Go.

Original language	English
Pages (from-to)	69-74
Number of pages	6
Journal	IEEE Transactions on Computational Intelligence and AI in Games
Volume	5
Issue number	1
DOIs	https://doi.org/10.1109/TCIAIG.2012.2220138
Publication status	Published - Mar 2013
Externally published	Yes

Access to Document

10.1109/TCIAIG.2012.2220138

https://dblp.org/rec/journals/tciaig/Browne13

Cite this

@article{154adb046d0941c09c194dc1f21f07de,

title = "A Problem Case for UCT",

abstract = "This paper examines a simple 5 5 Hex position that not only completely defeats flat Monte Carlo search, but also initially defeats plain upper confidence bounds for trees (UCT) search until an excessive number of iterations are performed. The inclusion of domain knowledge during playouts significantly improves UCT performance, but a slight negative effect is shown for the rapid action value estimate (RAVE) heuristic under some circumstances. This example was drawn from an actual game during standard play, and highlights the dangers of relying on flat Monte Carlo and unenhanced UCT search even for rough estimates. A brief comparison is made with RAVE failure in Go.",

author = "Cameron Browne",

note = "DBLP's bibliographic metadata records provided through http://dblp.org/search/publ/api are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.",

year = "2013",

month = mar,

doi = "10.1109/TCIAIG.2012.2220138",

language = "English",

volume = "5",

pages = "69--74",

journal = "IEEE Transactions on Computational Intelligence and AI in Games",

issn = "1943-068X",

publisher = "IEEE",

number = "1",

}

TY - JOUR

T1 - A Problem Case for UCT

AU - Browne, Cameron

N1 - DBLP's bibliographic metadata records provided through http://dblp.org/search/publ/api are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.

PY - 2013/3

Y1 - 2013/3

N2 - This paper examines a simple 5 5 Hex position that not only completely defeats flat Monte Carlo search, but also initially defeats plain upper confidence bounds for trees (UCT) search until an excessive number of iterations are performed. The inclusion of domain knowledge during playouts significantly improves UCT performance, but a slight negative effect is shown for the rapid action value estimate (RAVE) heuristic under some circumstances. This example was drawn from an actual game during standard play, and highlights the dangers of relying on flat Monte Carlo and unenhanced UCT search even for rough estimates. A brief comparison is made with RAVE failure in Go.

AB - This paper examines a simple 5 5 Hex position that not only completely defeats flat Monte Carlo search, but also initially defeats plain upper confidence bounds for trees (UCT) search until an excessive number of iterations are performed. The inclusion of domain knowledge during playouts significantly improves UCT performance, but a slight negative effect is shown for the rapid action value estimate (RAVE) heuristic under some circumstances. This example was drawn from an actual game during standard play, and highlights the dangers of relying on flat Monte Carlo and unenhanced UCT search even for rough estimates. A brief comparison is made with RAVE failure in Go.

U2 - 10.1109/TCIAIG.2012.2220138

DO - 10.1109/TCIAIG.2012.2220138

M3 - Article

SN - 1943-068X

VL - 5

SP - 69

EP - 74

JO - IEEE Transactions on Computational Intelligence and AI in Games

JF - IEEE Transactions on Computational Intelligence and AI in Games

IS - 1

ER -

A Problem Case for UCT

Abstract

Access to Document

Fingerprint

Cite this