Two-Stage Learning Kernel Algorithms

Abstract

This paper examines two-stage techniques for learning kernels based on a notion of alignment. It presents a number of novel theoretical, algorithmic, and empirical results for alignment-based techniques. Our results build on previous work by Cristianini et al. (2001), but we adopt a different definition of kernel alignment and significantly extend that work in several directions: we give a novel and simple concentration bound for alignment between kernel matrices; show the existence of good predictors for kernels with high alignment, both for classification and for regression; give algorithms for learning a maximum alignment kernel by showing that the problem can be reduced to a simple QP; and report the results of extensive experiments with this alignment-based method in classification and regression tasks, which show an improvement both over the uniform combination of kernels and over other state-of-the-art learning kernel methods.