Non-Sparse Multiple Kernel Learning for Fisher Discriminant Analysis
Yan, F, Kittler, J, Mikolajczyk, K and Tahir, A (2009) Non-Sparse Multiple Kernel Learning for Fisher Discriminant Analysis In: ICDM '09, 2009-12-06 - 2009-12-09, Miami, USA.
Available under License : See the attached licence file.
We consider the problem of learning a linear combination of pre-specified kernel matrices in the Fisher discriminant analysis setting. Existing methods for such a task impose an Â¿1 norm regularisation on the kernel weights, which produces sparse solution but may lead to loss of information. In this paper, we propose to use Â¿2 norm regularisation instead. The resulting learning problem is formulated as a semi-infinite program and can be solved efficiently. Through experiments on both synthetic data and a very challenging object recognition benchmark, the relative advantages of the proposed method and its Â¿1 counterpart are demonstrated, and insights are gained as to how the choice of regularisation norm should be made.
|Item Type:||Conference or Workshop Item (Conference Paper)|
|Divisions :||Faculty of Engineering and Physical Sciences > Electronic Engineering > Centre for Vision Speech and Signal Processing|
|Identification Number :||https://doi.org/10.1109/ICDM.2009.84|
|Additional Information :||© 2009 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.|
|Depositing User :||Symplectic Elements|
|Date Deposited :||21 Dec 2012 12:32|
|Last Modified :||23 Sep 2013 19:47|
Actions (login required)
Downloads per month over past year