We propose an independence criterion based on the eigenspectrum of covariance operators in reproducing kernel Hilbert spaces (RKHSs), consisting of an empirical estimate of the Hilbert-Schmidt norm of the cross-covariance operator (we term this a Hilbert-Schmidt Independence Criterion, or HSIC). This approach has several advantages, compared with previous kernel-based independence criteria. Fir...

In this paper we analyze a greedy procedure to approximate linear functional defined in reproducing kernel Hilbert space by nodal values. This computes quadrature rule which can be applied general functionals. For large class of functionals, that includes integration functionals and other interesting cases, but does not include differentiation, prove convergence results for the approximation me...

We develop an extension of the sliced inverse regression (SIR) framework for dimension reduction using kernel models and Tikhonov regularization. The result is a numerically stable nonlinear dimension reduction method. We prove consistency of the method under weak conditions even when the reproducing kernel Hilbert space induced by the kernel is infinite dimensional. We illustrate the utility o...

A mean function in reproducing kernel Hilbert space, or a kernel mean, is an important part of many applications ranging from kernel principal component analysis to Hilbert-space embedding of distributions. Given finite samples, an empirical average is the standard estimate for the true kernel mean. We show that this estimator can be improved via a well-known phenomenon in statistics called Ste...

A new method for performing a kernel principal component analysis is proposed. By kernelizing the generalized Hebbian algorithm, one can iteratively estimate the principal components in a reproducing kernel Hilbert space with only linear order memory complexity. The derivation of the method and preliminary applications in image hyperresolution are presented. In addition, we discuss the extensio...