Learning middle-game patterns in chess: A case study

Miroslav Kubat, Jan Žižka

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Despite the undisputed strength of today’s chess-playing programs, the fact that they have to evaluate millions, or even billions, of different positions per move is unsatisfactory. The amount of “computation” carried out by human players is smaller by orders of magnitudes because they employ specific patterns that help them narrow the search tree. Similar approachs hould in principle be feasible also in computer programs. To draw attenion to this issue, we report our experiments with a program that learns to classify chessboard positions that permit the well-known bishop sacrifice at h7. We discuss some problems pertaining to the collection of training examples, their representation, and pre-classification. Classification accuracies achieved with a decision-tree based classifier are encouraging.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages426-433
Number of pages8
Volume1821
ISBN (Print)3540676899, 9783540450498
DOIs
StatePublished - 2000
Externally publishedYes
Event13th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, IEA/AIE 2000 - New Orleans, United States
Duration: Jun 19 2000Jun 22 2000

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1821
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other13th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, IEA/AIE 2000
CountryUnited States
CityNew Orleans
Period6/19/006/22/00

Fingerprint

Game
Search Trees
Decision trees
Decision tree
Computer program listings
Classifiers
Classify
Classifier
Evaluate
Experiment
Experiments
Learning
Human
Training

Keywords

  • Computer chess
  • Concept learning
  • Pattern recognition

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Kubat, M., & Žižka, J. (2000). Learning middle-game patterns in chess: A case study. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 1821, pp. 426-433). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1821). Springer Verlag. https://doi.org/10.1007/3-540-45049-1_52

Learning middle-game patterns in chess : A case study. / Kubat, Miroslav; Žižka, Jan.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1821 Springer Verlag, 2000. p. 426-433 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1821).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kubat, M & Žižka, J 2000, Learning middle-game patterns in chess: A case study. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 1821, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1821, Springer Verlag, pp. 426-433, 13th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, IEA/AIE 2000, New Orleans, United States, 6/19/00. https://doi.org/10.1007/3-540-45049-1_52
Kubat M, Žižka J. Learning middle-game patterns in chess: A case study. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1821. Springer Verlag. 2000. p. 426-433. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/3-540-45049-1_52
Kubat, Miroslav ; Žižka, Jan. / Learning middle-game patterns in chess : A case study. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 1821 Springer Verlag, 2000. pp. 426-433 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{effcfda4bf0c4927bd92c202f03584b8,
title = "Learning middle-game patterns in chess: A case study",
abstract = "Despite the undisputed strength of today’s chess-playing programs, the fact that they have to evaluate millions, or even billions, of different positions per move is unsatisfactory. The amount of “computation” carried out by human players is smaller by orders of magnitudes because they employ specific patterns that help them narrow the search tree. Similar approachs hould in principle be feasible also in computer programs. To draw attenion to this issue, we report our experiments with a program that learns to classify chessboard positions that permit the well-known bishop sacrifice at h7. We discuss some problems pertaining to the collection of training examples, their representation, and pre-classification. Classification accuracies achieved with a decision-tree based classifier are encouraging.",
keywords = "Computer chess, Concept learning, Pattern recognition",
author = "Miroslav Kubat and Jan Žižka",
year = "2000",
doi = "10.1007/3-540-45049-1_52",
language = "English (US)",
isbn = "3540676899",
volume = "1821",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "426--433",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Learning middle-game patterns in chess

T2 - A case study

AU - Kubat, Miroslav

AU - Žižka, Jan

PY - 2000

Y1 - 2000

N2 - Despite the undisputed strength of today’s chess-playing programs, the fact that they have to evaluate millions, or even billions, of different positions per move is unsatisfactory. The amount of “computation” carried out by human players is smaller by orders of magnitudes because they employ specific patterns that help them narrow the search tree. Similar approachs hould in principle be feasible also in computer programs. To draw attenion to this issue, we report our experiments with a program that learns to classify chessboard positions that permit the well-known bishop sacrifice at h7. We discuss some problems pertaining to the collection of training examples, their representation, and pre-classification. Classification accuracies achieved with a decision-tree based classifier are encouraging.

AB - Despite the undisputed strength of today’s chess-playing programs, the fact that they have to evaluate millions, or even billions, of different positions per move is unsatisfactory. The amount of “computation” carried out by human players is smaller by orders of magnitudes because they employ specific patterns that help them narrow the search tree. Similar approachs hould in principle be feasible also in computer programs. To draw attenion to this issue, we report our experiments with a program that learns to classify chessboard positions that permit the well-known bishop sacrifice at h7. We discuss some problems pertaining to the collection of training examples, their representation, and pre-classification. Classification accuracies achieved with a decision-tree based classifier are encouraging.

KW - Computer chess

KW - Concept learning

KW - Pattern recognition

UR - http://www.scopus.com/inward/record.url?scp=84957899807&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84957899807&partnerID=8YFLogxK

U2 - 10.1007/3-540-45049-1_52

DO - 10.1007/3-540-45049-1_52

M3 - Conference contribution

AN - SCOPUS:84957899807

SN - 3540676899

SN - 9783540450498

VL - 1821

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 426

EP - 433

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -