CONF Grandvalet_icml_2008/IDIAP Composite Kernel Learning Szafranski, Marie Grandvalet, Yves Rakotomamonjy, Alain McCallum, A. Ed. Roweis, S. Ed. EXTERNAL http://publications.idiap.ch/attachments/papers/2008/Grandvalet_icml_2008.pdf PUBLIC http://publications.idiap.ch/index.php/publications/showcite/Grandvalet_Idiap-RR-59-2008 Related documents Proceedings of the 25th Annual International Conference on Machine Learning (ICML 2008) 2008 Omnipress 1040-1047 IDIAP-RR 08-59 The Support Vector Machine (SVM) is an acknowledged powerful tool for building classifiers, but it lacks flexibility, in the sense that the kernel is chosen prior to learning. Multiple Kernel Learning (MKL) enables to learn the kernel, from an ensemble of basis kernels, whose combination is optimized in the learning process. Here, we propose Composite Kernel Learning to address the situation where distinct components give rise to a group structure among kernels. Our formulation of the learning problem encompasses several setups, putting more or less emphasis on the group structure. We characterize the convexity of the learning problem, and provide a general wrapper algorithm for computing solutions. Finally, we illustrate the behavior of our method on multi-channel data where groups correpond to channels. REPORT Grandvalet_Idiap-RR-59-2008/IDIAP Composite Kernel Learning Szafranski, Marie Grandvalet, Yves Rakotomamonjy, Alain EXTERNAL http://publications.idiap.ch/attachments/reports/2008/Grandvalet_Idiap-RR-59-2008.pdf PUBLIC Idiap-RR-59-2008 2008 IDIAP Published in A. McCallum and S. Roweis (Eds.,',','), Proceedings of the 25th Annual International Conference on Machine Learning (ICML 2008,',','), (pp. 1040--1047). Omnipress, 2008. The Support Vector Machine (SVM) is an acknowledged powerful tool for building classifiers, but it lacks flexibility, in the sense that the kernel is chosen prior to learning. Multiple Kernel Learning (MKL) enables to learn the kernel, from an ensemble of basis kernels, whose combination is optimized in the learning process. Here, we propose Composite Kernel Learning to address the situation where distinct components give rise to a group structure among kernels. Our formulation of the learning problem encompasses several setups, putting more or less emphasis on the group structure. We characterize the convexity of the learning problem, and provide a general wrapper algorithm for computing solutions. Finally, we illustrate the behavior of our method on multi-channel data where groups correpond to channels.