Neural Networks as a Guide to Optimization

Subtitle: The Chess Middle Game Explored

Article type: Research Article

Authors: van Tiggelen, A.^{; *}

Affiliations: Amsterdam-Zuid Oost, The Netherlands

Note: [*] The author is a self-employed consultant. For the purposes of the project ALEXS he is also affiliated, as an individual research worker, to the University of Limburg, Department of Computer Science, Maastricht, The Netherlands.

Abstract: Applying genetic learning to parameter optimization in the middle game is so heavily penalized by its excessive demand of resources that a more efficient approach had to be searched for. This was found in a neural-network approach with an efficiency far beyond that of genetic learning. No great penalty seems to be incurred in effectiveness. It is concluded that tuning parameters by algorithm for the chess middle game now is feasible.

DOI: 10.3233/ICG-1991-14302

Journal: ICGA Journal, vol. 14, no. 3, pp. 115-118, 1991