The optimal solution of multi-kernel regularization learning

DOI10.1007/S10114-013-0675-9MaRDI QIDQ2392006zbMATH OpenOpenAlexFDO

Authors Hongwei Sun, Ping Liu

Publication date 6 August 2013

Published in Acta Mathematica Sinica, English Series (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1007/s10114-013-0675-9

zbMATH Keywords

existence of optimal solutionGaussian kernelslearning theorymulti-kernel regularization

Mathematics Subject Classification ID

General nonlinear regression (62J02) Computational learning theory (68Q32) Learning and adaptive systems in artificial intelligence (68T05)

Review text

Multi-kernel regularization schemes provide flexibility and better learning ability in some applications. They are also crucial in the problem of learning the kernel. It was shown by \textit{Y. Ying} and \textit{D.-X. Zhou} [J. Mach. Learn. Res. 8, 249--276 (2007; Zbl 1222.68339)] that the union of the unit balls of reproducing kernel Hilbert spaces generated by Gaussian kernels with flexible variances is learnable. The optimization problem when the variance runs over a compact set was also discussed. In this paper the authors discuss the optimization problem when the variance runs over the set of positive numbers (which is not compact). They present some sufficient conditions for the existence of an optimal solution for the least squares regularized regression associated with Gaussians with flexible variances.

Recommendations

• Learning the kernel function via regularization
• Learnability of Gaussians with flexible variances
• Regularized least square algorithm with two kernels
• Learning Theory
• Multi-kernel regularized classifiers

Cites work

• 10.1162/153244302760200704
• Choosing multiple parameters for support vector machines
• scientific article; zbMATH DE number 1183919 (Why is no real title available?)
• Learnability of Gaussians with flexible variances
• Learning rates of least-square regularized regression
• Learning the kernel function via regularization
• Learning the kernel matrix with semidefinite programming
• Learning theory estimates via integral operators and their approximations
• On the mathematical foundations of learning
• ONLINE LEARNING WITH MARKOV SAMPLING
• Online regression with varying Gaussians and non-identical distributions
• Regularization networks and support vector machines
• Risk bounds for mixture density estimation
• Theory of Reproducing Kernels
• VECTOR VALUED REPRODUCING KERNEL HILBERT SPACES AND UNIVERSALITY

Cited in

(10)

• Symmetry and antisymmetry properties of optimal solutions to regression problems
• Learning rates of kernel-based robust classification
• Feature space perspectives for learning the kernel
• Learning the kernel function via regularization
• Improvement of multiple kernel learning using adaptively weighted regularization
• Learnability of Gaussians with flexible variances
• The kernel regularized learning algorithm for solving Laplace equation with Dirichlet boundary
• Learning Theory
• Regularization in kernel learning
• Ideal regularization for learning kernels from labels

This page was built for publication: The optimal solution of multi-kernel regularization learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2392006)