PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A feature selection algorithm based on the global minimization of a generalization error bound
Dori Peleg and Ron Meir
In: NIPS 2004, 13-16 December 2004, Vancouver, Canada.


A novel feature selection algorithm is presented based on the global minimization of a data-dependent generalization error bound. Feature selection and scaling algorithms often lead to non-convex optimization problems, which in many previous approaches were addressed through gradient descent procedures that can only guarantee convergence to a local minimum. We propose an alternative approach, whereby the global solution of the non-convex optimization problem is derived via an equivalent optimization problem. Moreover, the convex optimization task is reduced to a conic quadratic programming problem for which efficient solvers are available. Highly competitive numerical results on both artificial and real-world data sets are reported.

EPrint Type:Conference or Workshop Item (Poster)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
Theory & Algorithms
ID Code:911
Deposited By:Ron Meir
Deposited On:06 January 2005