Multiobjective Optimization of Indexes Obtained by Clustering for Feature Selection Methods Evaluation in Genes Expression Microarrays

摘要

The selection of relevant genes in microarray is an important task, since that in a single experiment expressions of thousands of genes are extracted. One way to evaluate feature selection methods in a dataset is by clustering the instances that have similar behaviors. The aim of this paper is to use a set of indexes that measure the quality of a clustering and, through the multiobjective optimization of this set, to show how it is possible to find the best feature selection methods in genes expression datasets obtained by microarray technique.