Improved classification of genomic data by Gram-Schmidt feature selection

Abstract

This work explains some important aspects in the world of the neural networks, as the classification methods and the procedures of feature selection. Moreover, there is a practical part that consists in creating a program that provides us the useful information to do the classification. It is important to consider that in this thesis we have touched some biochemical aspects because the program has been designed for bioinformatics applications. Therefore the first part of the work consists in an introduction to genomics, namely, relations of enzymes and amino-acids. Finally all the results obtained in the work have been reported and discussed.