Standard learning procedures are better fitted to estimation than to classification problems, and focusing the training on appropriate samples provides performance advantages in classification tasks. In this paper, we combine these ideas creating smooth targets for classification by means of a convex combination of the original target and the output of an auxiliary classifier, the combination p...

The study of multiple classifier systems has become recently an area of intensive research in pattern recognition in order to improve the results of single classifiers. In this work, two types of features combination for handwritten Arabic literal words amount recognition, using neural network classifiers are discussed. Different parallel combination schemes are presented and their results comp...

We propose a new classifier combination scheme for the ensemble of classifiers. The Pairwise Fusion Matrix (PFM) constructs confusion matrices based on classifier pairs and thus offers the estimated probability of each class based on each classifier pair. These probability outputs can then be combined and the final outputs of the ensemble of classifiers is reached using various fusion functions...

The use of artificial outputs generated by a classifier simulator has recently emerged as a new trend to provide an underlying evaluation of classifier combination methods. In this paper, we propose a new method for the artificial generation of classifier outputs based on additional parameters which provide sufficient diversity to simulate, for a problem of any number of classes and any type of...

Plurality voting is widely used in pattern recognition practice. However, there is little theoretical analysis of plurality voting. In this paper, we attempt to explore the rationales behind plurality voting. The recognition/error/rejection rates of plurality voting are compared with those of majority voting under different conditions. It is demonstrated that plurality voting is more efficient ...

Data fusion under the belief function framework has attracted the interest of many researchers over the past few years. Until now, many combination rules have been proposed in order to aggregate beliefs induced form dependent or independent information sources. Although the choice of the most appropriate rule among several alternatives is crucial, it still requires non-trivial effort. In this i...

Traditional character recognition systems use a single classifier to determine the class of a given character. However, by using different types of classifiers simultaneously, the accuracy of classification could be improved. In this paper, we propose a classifier elimination approach based on correlation analysis and the derived heuristic rules to eliminate the redundant classifiers such that ...

This paper presents a study on the combination of different classifiers for toxicity prediction. Two combination operators for the Multiple-Classifier System definition are also proposed. The classification methods used to generate classifiers for combination are chosen in terms of their representability and diversity and include the Instance-based Learning algorithm (IBL), Decision Tree learni...

Although many decision combination methods have been proposed, most of them did not focus on dependency relationship among classiiers in combining multiple decisions. That makes classiication performance of combining multiple decisions be degraded and biased, in case of adding highly dependent inferior classiiers. To overcome such weaknesses and obtain robust classiication performance, the pres...