Information Theoretic Kernel Integration
Published on Jan 19, 20103578 Views
In this paper we consider a novel information-theoretic approach to multiple kernel learning based on minimising a Kullback-Leibler (KL) divergence between the output kernel matrix and the input ker