Date/heure
Date(s) - 8 avril 2014

Catégories Pas de Catégories

$Efficient optimization of the multiple kernel learning problem\n\nBy Alain Rakotomamonjy, Univ. Rouen.\n\nKernel methods are widely used to address a variety of learning tasks including classification, regression, ranking, clustering, and dimensionality reduction. The choice of a kernel is often left to the user. But poor selections may lead to sub-optimal performance. Instead, the kernel selection process should be included as part of the overall learning problem. In this way, better performance guarantees can be given and the kernel selection process can be made automatic. Algorithms that are able to address these issues are denoted as Multiple Kernel Learning (MKL) algorithm and they have been now widely used for learning with multiple views of the same data or for selecting/fusing different features and kernels.\n\nIn this talk, I will review the basics of MKL and discuss two issues\na how can we solve the MKL optimization problem when the number of kernels is very large (or infinite) ?\nb how can we solve the MKL optimization problem when only low-rank approximation of the different kernels are available ?[$

Date/heure
Date(s) - 8 avril 2014

Catégories Pas de Catégories

$Efficient optimization of the multiple kernel learning problem\n\nBy Alain Rakotomamonjy, Univ. Rouen.\n\nKernel methods are widely used to address a variety of learning tasks including classification, regression, ranking, clustering, and dimensionality reduction. The choice of a kernel is often left to the user. But poor selections may lead to sub-optimal performance. Instead, the kernel selection process should be included as part of the overall learning problem. In this way, better performance guarantees can be given and the kernel selection process can be made automatic. Algorithms that are able to address these issues are denoted as Multiple Kernel Learning (MKL) algorithm and they have been now widely used for learning with multiple views of the same data or for selecting/fusing different features and kernels.\n\nIn this talk, I will review the basics of MKL and discuss two issues\na how can we solve the MKL optimization problem when the number of kernels is very large (or infinite) ?\nb how can we solve the MKL optimization problem when only low-rank approximation of the different kernels are available ?[$