The aim of this study is to apply a state-of-the-art speech emotion recognition engine on the detection of microsleep endangered sleepiness states. Current approaches in speech em...
Existing cost-sensitive learning methods work with unequal misclassification cost that is given by domain knowledge and appears as precise values. In many real-world applications,...
The problem of identifying approximately duplicate objects in databases is an essential step for the information integration process. Most existing approaches have relied on gener...
—A new formulation for multiway spectral clustering is proposed. This method corresponds to a weighted kernel principal component analysis (PCA) approach based on primal-dual lea...
Abstract. Standard Support Vector Machines (SVM) text classification relies on bag-of-words kernel to express the similarity between documents. We show that a document lattice can ...