PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Semi-Supervised Bipartite Ranking with the Normalized Rayleigh Coefficient
Liva Ralaivola
In: ESANN 09, 23-25 Apr. 2009, Bruges.

Abstract

We propose a new algorithm for semi-supervised learning in the bipartite ranking framework. It is based on the maximization of a so-called normalized Rayleigh coefficient, which differs from the usual Rayleigh coefficient of Fisher's linear discriminant in that the actual covariance matrices are used instead of the scatter matrices. We show that if the class conditional distributions are Gaussian, then the ranking function produced by our algorithm is the optimal linear ranking function. A kernelized version of the proposed algorithm and a semi-supervised formulation are provided. Preliminary numerical results are promising.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Theory & Algorithms
ID Code:5036
Deposited By:Liva Ralaivola
Deposited On:24 March 2009