Calibrated Derivative-Process Sensitivity for Gaussian-Process Variable Selection

DOI10.48550/arXiv.2609.33549MaRDI QIDQ7362014arXivFDO

Authors Jia Cai

Publication date 27 September 2026

Published in arXiv

Tldr We present an adaptive, calibrated method for variable selection via the derivative process of a Gaussian-process model: we use a studentized multiplier bootstrap to transform uncalibrated derivatives into valid p-values and obtain family-wise and false-discovery-rate control.

Full work available at URL https://doi.org/10.48550/arXiv.2609.33549

arXiv classification

cs.LGmath.NAstat.ML

Cites work

• A central limit theorem for generalized quadratic forms
• A significance test for covariates in nonparametric regression
• Active subspaces. Emerging ideas for dimension reduction in parameter studies
• Bayesian approximate kernel regression with variable selection
• Comparing nonparametric versus parametric regression fits
• Controlling the false discovery rate via knockoffs
• Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing
• Efficiency of Weighted Average Derivative Estimators and Index Models
• Efficient estimation of integral functionals of a density
• Estimation and Hypothesis Testing of Derivatives in Smoothing Spline ANOVA Models
• Estimation of the derivative-based global sensitivity measures using a Gaussian process metamodel
• Gaussian and bootstrap approximations for high-dimensional U-statistics and their applications
• Gaussian approximations and multiplier bootstrap for maxima of sums of high-dimensional random vectors
• Gaussian Processes for Machine Learning
• Global sensitivity analysis with dependence measures
• Investigating Smooth Multiple Regression by the Method of Average Derivatives
• Kernel Knockoffs Selection for Nonparametric Additive Models
• Kernel Ridge Regression Inference
• Least angle regression
• Optimal plug-in Gaussian processes for modeling derivatives
• Optimal Plug-in Gaussian Processes for Modeling Derivatives
• Random quadratic forms and the bootstrap for \(U\)-statistics
• Rates of contraction of posterior distributions based on Gaussian process priors
• Sequential Learning of Active Subspaces
• Sequential Learning of Active Subspaces
• Supremum norm posterior contraction and credible sets for nonparametric multivariate regression
• Test comparison for Sobol Indices over nested sets of variables
• The control of the false discovery rate in multiple testing under dependency.
• The role of the range parameter for estimation and prediction in geostatistics
• Variable prioritization in nonlinear black box methods: a genetic association case study
• Variable selection consistency of Gaussian process regression

This page was built for publication: Calibrated Derivative-Process Sensitivity for Gaussian-Process Variable Selection

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7362014)