Niche Particle Swarm Optimization for Neural Network Ensembles

摘要

This research investigates a swarm intelligence based multi-objective optimization algorithm for optimizing the behavior of a group of Artificial Neural Networks (ANNs), where each ANN specializes to solving a specific part of a task, such that the group as a whole achieves an effective solution. Niche Particle Swarm Optimization (NichePSO) is a speciation technique that has proven effective at locating multiple solutions in complex multivariate tasks. This research evaluates the efficacy of the NichePSO method for training a group of ANNs that form a neural network ensemble (NNE) for the purpose of solving a set of multivariate tasks. NichePSO is compared with a gradient descent method for training a set of individual ANNs to solve different parts of a multivariate task, and then combining the outputs of each ANN into a single solution. To date, there has been little research that has compared the effectiveness of applying NichePSO versus more traditional supervised learning methods for the training of neural network ensembles.