Multi-objective evolutionary algorithm based optimization of neural network ensemble classifier

摘要

The purpose of this paper is to investigate a Multi-Objective Evolutionary Algorithm (MOEA) for optimizing neural ensemble classifiers. This paper provides an automatic procedure based on MOEA to identify the best accuracy and diversity. A MOEA is used to search for the combination of layers and clusters in ensemble classifiers to obtain the non-dominated set of accuracy and diversity. The experiments were conducted on UCI machine learning benchmark datasets using the MOEA and also single objective evolutionary algorithms. The detailed results and analysis show that MOEA has improved the performance of ensemble classifier and obtained better accuracy compared to recently published approaches.