Kernel machines have been shown as the state-of-the-art learning techniques for classification. In this paper, we propose a novel general framework of learning the Unified Kernel ...
When given a small sample, we show that classification with SVM can be considerably enhanced by using a kernel function learned from the training data prior to discrimination. Thi...