Some Studies in Machine Learning Using the Game of Checkers

  title={Some Studies in Machine Learning Using the Game of Checkers},
  author={Arthur L. Samuel},
  journal={IBM J. Res. Dev.},
  • A. Samuel
  • Published 1 November 1967
  • Computer Science
  • IBM J. Res. Dev.
Abstract A new signature-table technique is described together with an improved book-learning procedure which is thought to be much superior to the linear polynomial method. Full use is made of the so-called “alpha-beta” pruning and several forms of forward pruning to restrict the spread of the move tree and to permit the program to look ahead to a much greater depth than it otherwise could do. While still unable to outplay checker masters, the program's playing ability has been greatly… 

Figures and Tables from this paper


  • R. Bellman
  • Economics, Computer Science
    Proceedings of the National Academy of Sciences of the United States of America
  • 1965
How the theory of dynamic programming can be used to determine optimal play in the great majority of Pawn-King end-games in Chess, with computers currently available, and in all probability, to determined optimal play for the entire game of checkers is indicated.

Deep Learning and the Game of Checkers

An approach which given only a set of rules is able to learn to play the game of Checkers, using neural networks and reinforced learning combined with Monte Carlo Tree Search and alpha-beta pruning is presented.

Classical Checkers

A computer program created to play the game of Classical Checkers, Turska, is described and, for the first time, the complexity of Classical checkers is estimated and analysed.


Chinook is the strongest 8 × 8 checkers program around today. Its strength is largely a result of brute-force methods. The program is capable of searching to depths that make it a feared tactician.

Applying the Experience of Building a High-Performance Search Engine for One Domain to Another

The experience of transferring the technology used to develop a chess program to the creation of a highperformance checkers program, used in CHINOOK, the World Man-Machine Checkers Champion is described.

TD-GAC: Machine Learning Experiment with Give-Away Checkers

Results of applying the temporal difference learning methods in the game of give-away checkers show the success of Temporal Difference methods in improving the quality of computer player’s policy.

Solving the Game of Checkers

This paper discusses achieving three different levels of solving the game of checkers: publicly (as evidenced by Samuel's results), practically (by the checkers program Chinook, the best player in the world) and provably (by considering the 5×1020 positions in the search space).

The importance of look-ahead depth in evolutionary checkers

The experiments show that increasing the depth of a look-ahead has significant improvements to the performance of the checkers program and has a significant effect on its learning abilities.

Use of Evolutionary Algorithms to Play the Game of Checkers: Historical Developments, Challenges and Future Prospects

The question is can a programme be developed which can beat humans with complete success, and it may appears that some challenges may be formed which may substantiate the argument of the paper.



Chess-Playing Programs and the Problem of Complexity

The Los Alamos program is an almost perfect example of the type of system specified by Shannon, and the static evaluation function consisted of a sum of material and mobility measures.

Experiments in Chess

Some experiments performed on a fast computing machine (M~.nIAC I -Los Alamos) on the coding of computers to play the game of chess are reported on.

Programming a computer for playing chess

This paper is concerned with the problem of constructing a computing routine or “program” for a modern general purpose computer which will enable it to play chess. Although perhaps of no practical

Logical or non-mathematical programmes

It is generally true to say, I think, that the non-mathematical parts of the programme use far more instructions than one would at first sight expect, and that a relatively large part of effort in preparing the programme is spent dealing with these non-Mathematical operations.

Tests on a cell assembly theory of the action of the brain, using a large digital computer

Theories by D.O. Hebb and P.M. Milner on how the brain works were tested by simulating neuron nets on the IBM Type 704 Electronic Calculator. The formation of cell assemblies from an unorganized net

The brain computing machine

The author, a medical doctor, employs electrical engineering terminology to show how the brain may be likened to a digital computing machine consisting of ten billion relays called neurons. To carry

Faster than thought

The perceptron: a probabilistic model for information storage and organization in the brain.

This article will be concerned primarily with the second and third questions, which are still subject to a vast amount of speculation, and where the few relevant facts currently supplied by neurophysiology have not yet been integrated into an acceptable theory.