Please use this identifier to cite or link to this item:
|Scopus||Web of Science®||Altmetric|
|Title:||On the importance of pair-wise feature correlations for image classification|
De Chazal, P.
|Citation:||Proceedings of International Joint Conference on Neural Networks, 2016, vol.2016-October, pp.2290-2297|
|Series/Report no.:||IEEE International Joint Conference on Neural Networks (IJCNN)|
|Conference Name:||International Joint Conference on Neural Networks (IJCNN 2016) (24 Jul 2016 - 29 Jul 2016 : Vancouver, CANADA)|
|Mark D. McDonnell, Robby G. McKilliam, Philip de Chazal|
|Abstract:||We show that simple linear classification of pairwise products of convolutional features achieves near state-of-the-art performance on some standard labelled image databases. Specifically, we found test classification error rates on the MNIST handwritten digits image database of under 0.5%, and achieved under 19% and under 44% error rates on the CIFAR-10 and CIFAR-100 RGB image databases. Since the number of weights in such a classifier grows with the square of the number of features, we discuss how implementation of such a pair-wise products classifier can be achieved in an SLFN architecture where the hidden unit function is the simple quadratic nonlinearity: we can this a Quadratic Neural Network (QNN). We compare this method to setting the input weights in a QNN randomly, and find optimal performance can be achieved provided the hidden layer is sufficiently large. This analysis provides insight on why `extreme-learning machines' can achieve classification performance equal to or better than the use of backpropagation training.|
|Description:||IJCNN 2016 (as part of WCCI 2016)|
|Appears in Collections:||Aurora harvest 3|
Electrical and Electronic Engineering publications
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.