Improving Feature Subset Selection Using a Genetic Algorithm for Microarray Gene Expression Data

摘要: Microarray data usually contains a huge number of genes (features) and comparatively small samples, which make accurate classification or prediction diseases challenging. Feature selection techniques can help us identify important irrelevant (unimportant) features by applying certain criteria. However, different feature algorithms based on various theoretical arguments often produce results when applied to the same set. This makes selecting an optimal near subset for set difficult. In this paper, we propose using genetic algorithm improve combining valuable outcomes from multiple methods. The goal our is achieve balance between accuracy size subsets selected. advantages approach include ability accommodate criteria find that perform well particular inductive learning interest build classifier. experimental demonstrate with higher and/or smaller compared each individual algorithm.