Neural network ensembles

Abstract

We propose several means for improving the performance and training of neural networks for classification. We use crossvalidation as a tool for optimizing network parameters and architecture. We show further that the remaining residual "generalization" error can be reduced by invoking ensembles of similar networks.Zndex Terms-Crossvalidation, fault tolerant computing, neural networks, N-version programming.

Artificial neural network ensembles in time series forecasting: an application of rainfall forecasting in Sri Lanka

“…Hansen and Salamon's work [3] shows that the generalisation ability of ANN increase through ensemble. Due to this ANN ensemble techniques have become very popular in ANN applications.…”

“…A collection of a finite number of ANN trained for the same task is called an ensemble neural network. Hansen and Salamon [3] explains that the generalisation ability of an ANN can be significantly improved through an ENN and the advantages of separate ANN can be combined to give a better result. In an ensemble separate ANN are trained individually and then their outputs are combined.…”

Explaining the output of ensembles in medical decision support on a case by case basis

“…In ML research, the most common way to introduce diversity into an ensemble is to train the ensemble members on different subsets of the training data [2,12]. As stated above, this works very well for unstable learners such as neural networks but will not work for a stable learner such as a nearest neighbour classifier.…”

Neural Networks Fusion for Regression Problems

“…It has been shown that the performance of an ensemble method to generalize the results of the training for the rest of datasets is a function of diversity and the accuracy of the individual machines applied in the ensemble [18], [19]. The diversity of the nets is provided through the strategy which was presented in the previous subsection.…”