A Literature Review on Feature Selection using Evolutionary Algorithms

Abstract

Feature Selection is an optimization problem, where a subset of relevant features are derived from a set of features. It's a pre-processing technique performed before training an algorithm. Features/Attributes provide information about the labels/targets, so we may think that more attributes means more information about the target, but this is not the case always. Initially, the algorithm's performance may go up, but gradually it may come down; this is because of the irrelevant and redundant attributes present in the dataset. This phenomenon is called a Curse of Dimensionality. Feature Selection problem can be optimized using Evolutionary algorithms. This paper emphasizes on use of Evolutionary algorithms in optimizing the feature selection problem. This is a review paper where all the works relating to the application of Evolutionary algorithms in the field of Feature Selection are reviewed and presented.