An Improved Ensemble of Extreme Learning Machine Based on Attractive and Repulsive Particle Swarm Optimization

Although ensemble neural networks perform more effectively than individual one in many cases, it is easy to fall into the situation of comparatively low overall convergence accuracy while the individual members obtain high classification accuracies. To improve the convergence performance of the ensemble system, a modified ensemble of extreme learning machine (ELM) based on attractive and repulsive particle swarm optimization (ARPSO) is proposed in this paper. In the proposed method, ARPSO is applied to select the ELMs to build the ensemble system by optimizing the ensemble weights of the candidate ELMs. Experiment results on two data verify that the proposed method could obtain better convergence performance than some classical ELMs and ensemble ELMs.