Automatic Design of Multiple Classifier Systems by Unsupervised Learning
In the field of pattern recognition, multiple classifier systems based on the combination of the outputs of a set of different classifiers have been proposed as a method for the development of high performance classification systems. Previous work clearly showed that multiple classifier systems are effective only if the classifiers forming them make independent errors. This achievement pointed out the fundamental need for methods aimed to design ensembles of “independent” classifiers. However, the most of the recent work focused on the development of combination methods. In this paper, an approach to the automatic design of multiple classifier systems based on unsupervised learning is proposed. Given an initial set of classifiers, such approach is aimed to identify the largest subset of “independent” classifiers. A proof of the optimality of the proposed approach is given. Reported results on the classification of remote sensing images show that this approach allows one to design effective multiple classifier systems.
KeywordsError Probability Weight Seed Automatic Design Unsupervised Learn Probabilistic Neural Network
Unable to display preview. Download preview PDF.
- 4.K. Tumer and J. Ghosh, “Error correlation and error reduction in ensemble classifiers”, Connection Science 8, December 1996, pp. 385–404Google Scholar
- 5.A. J. C. Sharkey (Ed.), Special Issue: Combining Artificial Neural Nets: Ensemble Approaches. Connection Science Vol. 8, No. 3 & 4, Dec. 1996Google Scholar
- 8.G. Giacinto and F. Roli, “Ensembles of Neural Networks for Soft Classification of Remote Sensing Images”, Proc. of the European Symposium on Intelligent Techniques, Bari, Italy, pp. 166–170Google Scholar
- 16.A.K. Jain and R.C. Dubes, Algorithms for clustering data, Prentice Hall, 1988Google Scholar
- 18.S. B. Serpico, and F. Roli, “Classification of multi-sensor remote-sensing images by structured neural networks”, IEEE Trans. Geoscience Remote Sensing 33, 1995, pp. 562–578.Google Scholar