Publications:Selecting features for neural network committees

From ISLAB/CAISR

Do not edit this section

Keep all hand-made modifications below

Title Selecting features for neural network committees
Author Antanas Verikas and Marija Bacauskiene and Kerstin Malmqvist
Year 2002
PublicationType Conference Paper
Journal
HostPublication Proceedings of the International Joint Conference on Neural Networks
Conference International Joint Conference on Neural Networks (IJCNN '02), Honolulu, HI, United States, 12-17 May, 2002
DOI http://dx.doi.org/10.1109/IJCNN.2002.1005472
Diva url http://hh.diva-portal.org/smash/record.jsf?searchId=1&pid=diva2:1254348
Abstract We present a neural network based approach for identifying salient features for classification in neural network committees. Our approach involves neural network training with an augmented cross-entropy error function. The augmented error function forces the neural network to keep low derivatives of the transfer functions of neurons of the network when learning a classification task. Such an approach reduces output sensitivity to the input changes. Feature selection is based on the reaction of the cross-validation data set classification error due to the removal of the individual features. We compared the approach with two other neural network based feature selection methods. The algorithm developed outperformed the methods by achieving a higher classification accuracy on three real world problems tested. ©2002 IEEE