Outlier detection using classifier instability
When a classifier is used to classify objects, it is important to know if these objects resemble the training objects the classifier is trained with. Several methods to detect novel objects exist. In this paper a new method is presented which is based on the instability of the output of simple classifiers on new objects. The performances of the outlier detection methods is shown in a handwritten digit recognition problem.
KeywordsGaussian Mixture Model Outlier Detection Simple Classifier Training Object Probability Density Estimation
- [BDT97]Breukelen van M., Duin R.P.W, and Tax D.M.J. Combining classifiers for the recognition of handwritten digits. In Pudil P., Novovicova J, and Grim J, editors, 1st international workshop on statistical techniques in pattern recognition, pages 13–18. Institute of Information Theory and Automation, June 1997.Google Scholar
- [Bis95]Bishop C.M. Neural Networks for Pattern Recognition. Oxford University Press, Walton Street, Oxford OX2 6DP, 1995.Google Scholar
- [BL78]Barnett V. and Lewis T. Outliers in statistical data. Wiley series in probability and mathematical statistics. John Wiley & Sons Ltd., 2nd edition, 1978.Google Scholar
- [MKH93]Moya, MR., Koch, M.W., and Hostetler, L.D. One-class classifier networks for target recognition applications. In Proceedings world congress on neural networks, pages 797–801, Portland, OR, 1993. International Neural Network Society, INNS.Google Scholar
- [Rip96]Ripley B.D. Pattern Recognition and Neural Networks. Cambridge University Press, Cambridge, 1996.Google Scholar