A Weighted Kernel Method for Approximation that Adapts to Learned Multivariable Structure

DOI10.48550/arXiv.2609.16606MaRDI QIDQ7353389arXivFDO

Authors John E. Darges, Laura Weidensager

Publication date 15 September 2026

Published in arXiv

Tldr We develop weighted kernels (TSKs) for approximating multivariable functions that learn from data and adapt to their structure, improving accuracy compared to standard methods. We show TSKs are consistent under mild conditions and provide a novel characterization of input sensitivity via total Sobol indices in the kernel space.

Full work available at URL https://doi.org/10.48550/arXiv.2609.16606

arXiv classification

cs.LGmath.NA

Cites work

• A Class of Statistics with Asymptotically Normal Distribution
• A Multi-Batch L-BFGS Method for Machine Learning
• Adaptive ANOVA decomposition of stochastic incompressible and compressible flows
• An efficient methodology for modeling complex computer codes with Gaussian processes
• Analysis of variance on function spaces
• ANOVA kernels and RKHS of zero mean functions for model-based sensitivity analysis
• ANOVA-boosting for random Fourier features
• Black Box Approximation in the Tensor Train Format Initialized by ANOVA Decomposition
• Effective dimension of some weighted pre-Sobolev spaces with dominating mixed partial derivatives
• Efficient input-output model representations
• Estimating Mean Dimensionality of Analysis of Variance Decompositions
• Extreme learning machines for variance-based global sensitivity analysis
• Fast approximation by periodic kernel-based lattice-point interpolation with application in uncertainty quantification
• Flexible variable selection for recovering sparsity in nonadditive nonparametric models
• Generalized descent for global optimization
• Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates
• Kernel-based ANOVA decomposition and Shapley effects -- Application to global sensitivity analysis
• Kernel-based measures of association between inputs and outputs using ANOVA
• Learning multivariate functions with low-dimensional structures using polynomial bases
• Learning theory estimates via integral operators and their approximations
• Multiple kernel learning algorithms
• Nonadaptive quasi-optimal points selection for least squares linear regression
• On decompositions of multivariate functions
• On quasi-Monte Carlo methods in weighted ANOVA spaces
• On the equivalence between kernel quadrature rules and random feature expansions
• On the limited memory BFGS method for large scale optimization
• On the mathematical foundations of learning
• On the tractability of the Brownian bridge algorithm
• Optimal rates for the regularized least-squares algorithm
• Quasi-Monte Carlo finite element methods for a class of elliptic partial differential equations with random coefficients
• Sensitivity Analysis on the Sphere and a Spherical ANOVA Decomposition
• Sparse polynomial chaos expansions: literature survey and benchmark
• Support Vector Machines
• Variance based sensitivity analysis of model output. Design and estimator for the total sensitivity index
• When are quasi-Monte Carlo algorithms efficient for high dimensional integrals?

This page was built for publication: A Weighted Kernel Method for Approximation that Adapts to Learned Multivariable Structure

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7353389)