Multi-Class Discriminant Analysis Based on Support Vector Machine Ensembles
The areas of pattern recognition and analysis of image databases require the managing of datasets originally represented in high dimensional spaces. Besides, the original data representation implies, in general, in redundancy and noisy. Thus, we must compute a more suitable feature space, reducing both the dimension and redundancy of representation. Once a feature space has been defined there is the necessity of determining the most important discriminant features for pattern recognition tasks, like classification. Discriminant analysis techniques, address this problem. Thus, the goal of the proposed thesis is to develop discriminant analysis methods for multi-class problems. The key idea is to combine N classifiers to form a global discriminant function, which allows to rank the components of the space according to the importance of each feature to the classification problem. To achieve this goal, we use separate hyperplane computed by traditional support vector machine (SVM) or defined by a Kernel SVM (KSVM) decision boundary, and use ensemble methodologies known as AdaBoost to combine linear classifier. In this work, principal components analysis (PCA), Convolutional neural networks (CNNs) and texture descriptors, are used to create feature space that serve as input to discriminant analysis techniques. In terms of application for validation of the proposed techniques our focus are human face and texture images obtained from granite tiles. Further works will be undertaken by exploring color images, tensor subspaces as well as to improve performance.
Artur Ziviani - Laboratório Nacional de Computação Científica - firstname.lastname@example.org
Fabio André Machado Porto - Laboratório Nacional de Computação Científica
Gilson Antônio Giraldi - Laboratório Nacional de Computação Científica - email@example.com
Helio José Corrêa Barbosa - Laboratório Nacional de Computação Científica - firstname.lastname@example.org
Tiene Andre Filisbino - LNCC