No abstract

Abstract. When solving a pattern classification problem, it is common to apply a feature extraction method as a pre-processing step, not only to reduce the computation complexity but also to obtain better classification performance by reducing the amount of irrelevant and redundant information in the data. In this study, we investigate a novel schema for linear feature extraction in classification problems. The method we have proposed is based on clustering technique to realize feature extraction. It focuses in identifying and transforming redundant information in the data. A new similarity measure-based trend analysis is devised to identify those features. The simulation results on face recognition show that the proposed method gives better or competitive results when compared to conventional unsupervised methods like PCA and ICA.

In this paper we suggest an approach to select features for the Support Vector Machines (SVM). Feature selection is efficient in searching the most descriptive features which would contribute in increasing the effectiveness of the classifier algorithm. The process described here consists in backward elimination strategy based on the criterion of the rate of misclassification. We used the tabu algorithm to guide the search of the optimal set of features; each set of features is assessed according to its goodness of fit. This procedure is exploited in the regulation of urban transport network systems. It was first applied in a binary case and then it was extended to the multiclass case thanks to the MSVM technique: Binary Tree.

Weld defect detection is an important application in the field of Non-Destructive Testing (NDT). These defects are mainly due to manufacturing errors or welding processes. In this context, image processing especially segmentation is proposed to detect and localize efficiently different types of defects. It is a challenging task since radiographic images have deficient contrast, poor quality and uneven illumination caused by the inspection techniques. The usual segmentation technique uses a region of interest ROI from the original image. In this article, a robust and automatic method is presented to detect linear defect from the original image without selection of ROI based on canny detector and a modified 'Hough Transform' technique. This task can be subdivided into the following steps: firstly, preprocessing step with Gaussian filter and contrast stretching; secondly, segmentation technique is used to isolate weld region from background and non-weld using Adaptative Thresholding and to extract edges; thirdly, detection, location of linear defect and limiting the welding area by Hough Transform. The experimental results show that our proposed method gives good performance for industrial radiographic images.

This paper describes a new technique for clustering data based on their trend characteristics. The technique that we propose proceed by incorporating a new distance based on qualitative trend analysis into Mean shift clustering algorithm. Mean shift clustering is a powerful non-parametric technique that does not require prior knowledge of the number of clusters and does not constrain the shape of the clusters. Trend analysis is a data-driven semi-quantitative technique that has been used for process monitoring and fault detection and diagnosis. The performances of our approach are assesed through synthetic banana shaped data. Unsupervised clustering is then applied for intelligent decision-making process specifically for fault diagnosis on Tennessee Easteman Process (TEP) challenge.