Developing a robust prediction interval based criterion for neural network model selection

Authors:
Abbas Khosravi;Saeid Nahavandi;Doug Creighton
Affiliations:
Centre for Intelligent Systems Research, Deakin University, Geelong, Australia;Centre for Intelligent Systems Research, Deakin University, Geelong, Australia;Centre for Intelligent Systems Research, Deakin University, Geelong, Australia
Venue:
ICONIP'10 Proceedings of the 17th international conference on Neural information processing: models and applications - Volume Part II
Year:
2010

Citing 5
Cited 0

Prediction intervals for neural networks via nonlinear regression

Technometrics
Modeling with constructive backpropagation

Neural Networks
Neural Networks for Pattern Recognition

Neural Networks for Pattern Recognition
A prediction interval-based approach to determine optimal structures of neural network metamodels

Expert Systems with Applications: An International Journal
COVNET: a cooperative coevolutionary model for evolving artificial neural networks

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper studies how an optimal Neural Network (NN) can be selected that is later used for constructing the highest quality delta-based Prediction Intervals (PIs). It is argued that traditional assessment criteria, including RMSE, MAPE, BIC, and AIC, are not the most appropriate tools for selecting NNs from a PI-based perspective. A new NN model selection criterion is proposed using the specific features of the delta method. Using two synthetic and real case studies, it is demonstrated that this criterion outperforms all traditional model selection criteria in terms of picking the most appropriate NN. NNs selected using this criterion generate high quality PIs evaluated by their length and coverage probability.