When more than a single classifier has been trained for the same recognition problem the question arises how this set of classifiers may be combined into a final decision rule. Several fixed combining rules are used that depend on the output values of the base classifiers only. They are almost always suboptimal.

The generalization from two-class classification to multiclass classification is not straightforward for discriminants which are not based on density estimation. Simple combining methods use voting, but this has the drawback of inconsequent labelings and ties. More advanced methods map the discriminant outputs to approximate posterior probability estimates and combine these, while other methods use error-correcting output codes. In this paper we want to show the possibilities of simple generalizations of the twoclass classification, using voting and combinations of approximate posterior probabilities.

In this paper, a cost-sensitive learning method for support vector machine (SVM) classifiers is proposed. We focus on a particular case of cost-sensitive problems, namely, classification with reject option. Standard learning algorithms, the one for SVMs included, are not cost-sensitive. In particular, they can not handle the reject option. However, we show that, under the framework of the structural risk minimisation induction principle, on which standard SVMs are based, the rejection region should be determined during the training phase of a classifier, by the learning algorithm. We apply this approach to develop a cost-sensitive SVM classifier, by following Vapnik's maximum margin method to the derivation of standard SVMs.

For learning purposes, representations of real world objects can be built by using the concept of dissimilarity (distance). In such a case, an object is characterized in a relative way, i.e. by its dissimilarities to a set of the selected prototypes. Such dissimilarity representations are found to be more practical for some pattern recognition problems. When experts cannot decide for a single dissimilarity measure, a number of them may be studied in parallel. We investigate two possibilities of combining either dissimilarity representations themselves or classifiers built on each of them separately. Our experiments conducted on a handwritten digit set demonstrate that when the dissimilarity representations are of different nature, a much better performance can be obtained by their combination than on individual representations.

Segmentation methods, combining spectral and spatial information, are essential for analysis of multi-spectral images. In this article, we propose such a method based on statistical pattern recognition algorithms and a combined classifier approach. A set of experiments is presented with multi-spectral images of detergent laundry powders acquired by imaging cross-sections with scanning electron microscopy using energy-dispersive X-ray microanalysis (SEM/EDX). The algorithm stability and the segmentation quality are investigated. The use of apriori information for the segmentation of images with similar spectral properties is studied as well. Finally, a comparison with probabilistic relaxation method for multi-spectral image segmentation is made.

Abstract not found

Abstract—Patch-based face recognition is a recent method which uses the idea of analyzing face images locally, in order to reduce the effects of illumination changes and partial occlusions. Feature fusion and decision fusion are two distinct ways to make use of the extracted local features. Apart from the well-known decision fusion methods, a novel approach for calculating weights for the weighted sum rule is proposed in this paper. Improvements in recognition accuracies are shown and superiority of decision fusion over feature fusion is advocated. In the challenging AR database, we obtain significantly better results using decision fusion as compared to conventional methods and feature fusion methods by using validation accuracy weighting scheme and nearest-neighbor discriminant analysis dimension reduction method. Keywords-face recognition, patch-based face recognition, decision fusion, linear combiner training. I.

For learning purposes, representations of real world objects can be built by using the concept of dissimilarity. In such a case, an object is characterized in a relative way, i.e. by its dissimilarities to a set of the selected prototypes. Such dissimilarity representations are found to be more practical for some pattern recognition problems.

For combining classifiers at measurement level, the diverse outputs of classifiers should be transformed to uniform measures that represent the confidence of decision, hopefully, the class probability or likelihood. This paper presents our experimental results of classifier combination using confidence evaluation. We test three types of confidences: log-likelihood, exponential and sigmoid. For re-scaling the classifier outputs, we use three scaling functions based on global normalization and Gaussian density estimation. Experimental results in handwritten digit recognition show that via confidence evaluation, superior classification performance can be obtained using simple combination rules. 1.

Proceedings