Bio-inspired algorithms for optimal feature subset selection

Abstract

Feature subset selection is one of the most important tasks for the success of pattern classification, data mining or machine learning applications. The basic objective of feature subset selection is to reduce the dimensionality of the problem while retaining the most discriminatory information necessary for accurate classification. Thus it is necessary to evaluate feature subsets for their ability to discriminate different classes of pattern. Now the fact that “two best features do not comprise the best feature subset of two features” demands evaluation of all possible subset of features to find out the best feature subset. If the number of features increases, the number of possible feature subsets grows exponentially leading to a combinatorial optimization problem. Biologically inspired evolutionary algorithms are known to be well suited for optimization problems and recently research in this direction is gaining momentum. In this talk, I am going to present an overview of proposed feature subset selection algorithms based on biologically inspired approaches, Artificial Neural Network (ANN), Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and their hybrids.