Classification by Ensembles of Neural Networks

TL;DR

This paper presents a novel ensemble-based training method for neural networks that approximates the objective function through averaging multiple randomly generated networks, offering an alternative to traditional optimization-based training for classification tasks.

Contribution

It introduces a new ensemble training procedure for neural networks that does not rely on standard optimization, expanding the methods for neural network training.

Findings

Ensemble averaging can effectively train neural networks for classification.

The method does not require each network to approximate the objective function.

The approach offers an alternative to traditional optimization-based training.

Abstract

We introduce a new procedure for training of artificial neural networks by using the approximation of an objective function by arithmetic mean of an ensemble of selected randomly generated neural networks, and apply this procedure to the classification (or pattern recognition) problem. This approach differs from the standard one based on the optimization theory. In particular, any neural network from the mentioned ensemble may not be an approximation of the objective function.