Advanced Semisupervised SVM Approaches to Classification of Hyperspectral Data

In this paper, we propose two novel semisupervised algorithms based on Support Vector Machine (SVM) for classification of hyperspectral data. In greater detail, we present: i) a Progressive Semisupervised SVM (PS 3VM) developed in the dual formulation of the optimization problem related to the learning of the classifier; ii) a Low Density Separation algorithm developed in the primal formulation of the optimization problem. Experimental results carried out on real hyperspectral data confirm the effectiveness and the reliability of both the proposed approaches.