Mercer’s Theorem for Quaternionic Kernels
نویسنده
چکیده
the series being uniformly and absolutely convergent in (x,y). A number of generalisations to Mercer’s theorem may be found in the literature, in particular dealing with kernels K : Y × Y → C for various choices of Y . However there would appear to have been (to the best of the author’s knowledge) no attempts made to extend Mercer’s theorem to cover non-complex valued kernels. In the present paper we show how Mercer’s theorem may be extended to cover one such family of kernels, namely the continuous quaternionic valued kernels K : R × R → H. As quaternions provide a powerful tool for describing geometric problems [6] we anticipate that this extension will find applications in geometrical learning problems. Throughout this paper the quaternionic division algebra [7], [3] will be denoted H, the field of complex numbers C and the completely ordered field of reals R. The conjugate and norm of x ∈ H will be denoted x̄ and |x| respectively, where |x| = x̄x ∈ R\R. We define R = (0,∞) to be the set of positive reals, R = (−∞, 0) the set of negative reals, N the set of natural numbers including 0, Zp = {0, 1, . . . , p− 1} the set of integers modulo p (with the extensions Z∞ = N, Z0 = ∅), N ∞ = N ∪ {∞}, and L the set of square integrable quaternionic functions on R: L = {
منابع مشابه
Mercer's Theorem, Feature Maps, and Smoothing
We study Mercer’s theorem and feature maps for several positive definite kernels that are widely used in practice. The smoothing properties of these kernels will also be explored.
متن کاملMercer’s Theorem on General Domains: on the Interaction between Measures, Kernels, and RKHSs
Given a compact metric space X and a strictly positive Borel measure ν on X , Mercer’s classical theorem states that the spectral decomposition of a positive self-adjoint integral operator Tk : L2(ν) → L2(ν) of a continuous k yields a series representation of k in terms of the eigenvalues and -functions of Tk. An immediate consequence of this representation is that k is a (reproducing) kernel a...
متن کاملReproducing Kernel Hilbert Spaces in Learning Theory: the Sphere and the Hypercube
We analyze the regularized least square algorithm in learning theory with Reproducing Kernel Hilbert Spaces (RKHS). Explicit convergence rates for the regression and binary classification problems are obtained in particular for the polynomial and Gaussian kernels on the n-dimensional sphere and the hypercube. There are two major ingredients in our approach: (i) a law of large numbers for Hilber...
متن کاملImproving the Performance of Text Categorization using N-gram Kernels
Kernel Methods are known for their robustness in handling large feature space and are widely used as an alternative to external feature extraction based methods in tasks such as classification and regression. This work follows the approach of using different string kernels such as n-gram kernels and gappy-n-gram kernels on text classification. It studies how kernel concatenation and feature com...
متن کاملBimodules over Cartan MASAs in von Neumann algebras, norming algebras, and Mercer’s Theorem
In a 1991 paper, R. Mercer asserted that a Cartan bimodule isomorphism between Cartan bimodule algebras A1 and A2 extends uniquely to a normal ∗-isomorphism of the von Neumann algebras generated by A1 and A2 (Corollary 4.3 of Mercer, 1991). Mercer’s argument relied upon the Spectral Theorem for Bimodules of Muhly, Saito and Solel, 1988 (Theorem 2.5, there). Unfortunately, the arguments in the l...
متن کامل