Argyriou, Andreas and Hauser, Raphael and Micchelli, Charles A. and Pontil, Massimiliano (2006) A DC-programming algorithm for kernel selection. Technical Report. Unspecified. (Submitted)
We address the problem of learning a kernel for a given supervised learning task. Our approach consists in searching within the convex hull of a prescribed set of basic kernels for one which minimizes a convex regularization functional. A unique feature of this approach compared to others in the literature is that the number of basic kernels can be infinite. We only require that they are continuously parameterized. For example, the basic kernels could be isotropic Gaussians with variance in a prescribed interval or even Gaussians parameterized by multiple continuous parameters. Our work builds upon a formulation involving a minimax optimization problem and a recently proposed greedy algorithm for learning the kernel. Although this optimization problem is not convex, it belongs to the larger class of DC (difference of convex functions) programs. Therefore, we apply recent results from DC optimization theory to create a new algorithm for learning the kernel. Our experimental results on benchmark data sets show that this algorithm outperforms a previously proposed method.
Raphael Hauser was supported through grant GR/M30975 from the Engineering and Physical Sciences Research Council of the UK.
|Item Type:||Technical Report (Technical Report)|
|Subjects:||H - N > Numerical analysis|
|Research Groups:||Numerical Analysis Group|
|Deposited By:||Lotti Ekert|
|Deposited On:||11 May 2011 09:56|
|Last Modified:||29 May 2015 18:48|
Repository Staff Only: item control page