A Novel Multi-class Classification Approach Based on Fruit Fly Optimization Algorithm and Relevance Vector Machine

Abstract

A novel version of multi-class classification method based on fruit fly optimization algorithm (FOA) and relevance vector machine (RVM) is proposed. The one-against-one-against-rest (OAOAR) classification model based on the traditional one-against-one (OAO) and one-against-rest (OAR) algorithm is aimed at combining the advantages of them and translates the multi-class classification problem into multiple three-classification problems to accelerate the running speed with high classification precision. With RVM applied as the binary classifier, the optimal parameter values of RVM kernel function are determined by FOA. Theoretical analysis and computational comparisons on publicly available datasets both indicate that the proposed approach outperforms in terms of diagnosis accuracy and running time with more model sparsity and higher diagnosis efficiency.