Resumen

MARTIN, Tonysé de la Rosa. Proposed algorithms for the reduction of sample space on cefalosporin. Universidad y Sociedad [online]. 2020, vol.12, n.4, pp. 316-324.  Epub 02-Ago-2020. ISSN 2218-3620.

This paper presents the implementation and evaluation of various methods of dimensionality reduction based on artificial intelligence techniques, and addresses one of their complex problems, such as identifying and reducing a representative set of attributes to assist in the improvement of the classification and prediction models. The quest for optimal subsets of attributes for the classification of data sets have the disadvantage that its time complexity. Search procedures were implemented by genetic algorithms, simulated cooling, sequential search and a hybrid between this and genetic algorithms, so to achieve greater robustness and efficiency. It also implemented several measures of association between variable subsets, based on concepts borrowed from classical statistical theory of Shannon Information. In all cases tested the sample space is reduced by more than 65%. The best results are achieved through the Simulated Annealing algorithm using support vector machines classifier. All these search procedures present a polynomial time complexity of order, this demonstrates the practical feasibility and cost of each procedure computing resources deployed.

Palabras clave : Artificial intelligence; classification; dimensionality; reduction; prediction; sub-attributes.

        · resumen en Español     · texto en Español     · Español ( pdf )