DeepChess: End-to-end deep neural network for automatic learning in chess

Omid E. David; Nathan S. Netanyahu; Lior Wolf

doi:10.1007/978-3-319-44781-0_11

DeepChess: End-to-end deep neural network for automatic learning in chess

Omid E. David, Nathan S. Netanyahu, Lior Wolf

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

51 Scopus citations

Abstract

We present an end-to-end learning method for chess, relying on deep neural networks. Without any a priori knowledge, in particular without any knowledge regarding the rules of chess, a deep neural network is trained using a combination of unsupervised pretraining and supervised training. The unsupervised training extracts high level features from a given position, and the supervised training learns to compare two chess positions and select the more favorable one. The training relies entirely on datasets of several million chess games, and no further domain specific knowledge is incorporated. The experiments show that the resulting neural network (referred to as DeepChess) is on a par with state-of-the-art chess playing programs, which have been developed through many years of manual feature selection and tuning. DeepChess is the first end-to-end machine learningbased method that results in a grandmaster-level chess playing performance.

Original language	English
Title of host publication	Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings
Editors	Alessandro E.P. Villa, Paolo Masulli, Antonio Javier Pons Rivero
Publisher	Springer Verlag
Pages	88-96
Number of pages	9
ISBN (Print)	9783319447803
DOIs	https://doi.org/10.1007/978-3-319-44781-0_11
State	Published - 2016
Event	25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016 - Barcelona, Spain Duration: 6 Sep 2016 → 9 Sep 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9887 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016
Country/Territory	Spain
City	Barcelona
Period	6/09/16 → 9/09/16

Bibliographical note

Publisher Copyright:
© Springer International Publishing Switzerland 2016.

Access to Document

10.1007/978-3-319-44781-0_11

Cite this

David, O. E., Netanyahu, N. S., & Wolf, L. (2016). DeepChess: End-to-end deep neural network for automatic learning in chess. In A. E. P. Villa, P. Masulli, & A. J. P. Rivero (Eds.), Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings (pp. 88-96). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9887 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-44781-0_11

David, Omid E. ; Netanyahu, Nathan S. ; Wolf, Lior. / DeepChess : End-to-end deep neural network for automatic learning in chess. Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. editor / Alessandro E.P. Villa ; Paolo Masulli ; Antonio Javier Pons Rivero. Springer Verlag, 2016. pp. 88-96 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a51fe628738e4a8ea45ca90010379b97,

title = "DeepChess: End-to-end deep neural network for automatic learning in chess",

abstract = "We present an end-to-end learning method for chess, relying on deep neural networks. Without any a priori knowledge, in particular without any knowledge regarding the rules of chess, a deep neural network is trained using a combination of unsupervised pretraining and supervised training. The unsupervised training extracts high level features from a given position, and the supervised training learns to compare two chess positions and select the more favorable one. The training relies entirely on datasets of several million chess games, and no further domain specific knowledge is incorporated. The experiments show that the resulting neural network (referred to as DeepChess) is on a par with state-of-the-art chess playing programs, which have been developed through many years of manual feature selection and tuning. DeepChess is the first end-to-end machine learningbased method that results in a grandmaster-level chess playing performance.",

author = "David, {Omid E.} and Netanyahu, {Nathan S.} and Lior Wolf",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2016.; 25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016 ; Conference date: 06-09-2016 Through 09-09-2016",

year = "2016",

doi = "10.1007/978-3-319-44781-0_11",

language = "אנגלית",

isbn = "9783319447803",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "88--96",

editor = "Villa, {Alessandro E.P.} and Paolo Masulli and Rivero, {Antonio Javier Pons}",

booktitle = "Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings",

address = "גרמניה",

}

David, OE, Netanyahu, NS & Wolf, L 2016, DeepChess: End-to-end deep neural network for automatic learning in chess. in AEP Villa, P Masulli & AJP Rivero (eds), Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9887 LNCS, Springer Verlag, pp. 88-96, 25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016, Barcelona, Spain, 6/09/16. https://doi.org/10.1007/978-3-319-44781-0_11

DeepChess: End-to-end deep neural network for automatic learning in chess. / David, Omid E.; Netanyahu, Nathan S.; Wolf, Lior.
Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. ed. / Alessandro E.P. Villa; Paolo Masulli; Antonio Javier Pons Rivero. Springer Verlag, 2016. p. 88-96 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9887 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - DeepChess

T2 - 25th International Conference on Artificial Neural Networks and Machine Learning, ICANN 2016

AU - David, Omid E.

AU - Netanyahu, Nathan S.

AU - Wolf, Lior

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2016.

PY - 2016

Y1 - 2016

N2 - We present an end-to-end learning method for chess, relying on deep neural networks. Without any a priori knowledge, in particular without any knowledge regarding the rules of chess, a deep neural network is trained using a combination of unsupervised pretraining and supervised training. The unsupervised training extracts high level features from a given position, and the supervised training learns to compare two chess positions and select the more favorable one. The training relies entirely on datasets of several million chess games, and no further domain specific knowledge is incorporated. The experiments show that the resulting neural network (referred to as DeepChess) is on a par with state-of-the-art chess playing programs, which have been developed through many years of manual feature selection and tuning. DeepChess is the first end-to-end machine learningbased method that results in a grandmaster-level chess playing performance.

AB - We present an end-to-end learning method for chess, relying on deep neural networks. Without any a priori knowledge, in particular without any knowledge regarding the rules of chess, a deep neural network is trained using a combination of unsupervised pretraining and supervised training. The unsupervised training extracts high level features from a given position, and the supervised training learns to compare two chess positions and select the more favorable one. The training relies entirely on datasets of several million chess games, and no further domain specific knowledge is incorporated. The experiments show that the resulting neural network (referred to as DeepChess) is on a par with state-of-the-art chess playing programs, which have been developed through many years of manual feature selection and tuning. DeepChess is the first end-to-end machine learningbased method that results in a grandmaster-level chess playing performance.

UR - http://www.scopus.com/inward/record.url?scp=84988418847&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-44781-0_11

DO - 10.1007/978-3-319-44781-0_11

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:84988418847

SN - 9783319447803

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

SP - 88

EP - 96

BT - Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings

A2 - Villa, Alessandro E.P.

A2 - Masulli, Paolo

A2 - Rivero, Antonio Javier Pons

PB - Springer Verlag

Y2 - 6 September 2016 through 9 September 2016

ER -

David OE, Netanyahu NS, Wolf L. DeepChess: End-to-end deep neural network for automatic learning in chess. In Villa AEP, Masulli P, Rivero AJP, editors, Artificial Neural Networks and Machine Learning - 25th International Conference on Artificial Neural Networks, ICANN 2016, Proceedings. Springer Verlag. 2016. p. 88-96. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-44781-0_11

DeepChess: End-to-end deep neural network for automatic learning in chess

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this