A hybrid multiobjective particle swarm optimization approach for non-redundant gene marker selection

Abstract

The gene markers or biological markers indicate change in expression or state of protein that correlates with the risk or progression of a disease, or with the susceptibility of the disease to a given treatment. There are many approaches for detecting these informative genes from high dimensional microarray data. But in practice, for most of the cases a set of redundant marker genes are identified. Motivated by this fact a hybrid multiobjective optimization method has been proposed which can find small set of non-redundant disease related genes. In this article the optimization problem has been modeled as multiobjective problem which is based on the framework of particle swarm optimization. As the wrapper approaches depend on a specific classifier evaluation, hence artificial neural network classifier is used as evaluation criteria. Using the real life datasets, performance of proposed algorithm has been compared with other different techniques. © 2013 Springer.