A new kernel function between two labeled graphs is presented. Feature vectors are defined as the counts of label paths produced by random walks on graphs. The kernel computation ...
Abstract. The present study investigates a geometrical method for optimizing the kernel function of a support vector machine. The method is an improvement of the one proposed in [4...
Ensemble learning algorithms such as boosting can achieve better performance by averaging over the predictions of some base hypotheses. Nevertheless, most existing algorithms are ...
In this paper we define a novel similarity measure between examples of textual entailments and we use it as a kernel function in Support Vector Machines (SVMs). This allows us to ...
Abstract. In this paper a novel framework for brain classification is proposed in the context of mental health research. A learning by example method is introduced by combining loc...
Umberto Castellani, E. Rossato, Vittorio Murino, M...