Convergence guarantees for kernel-based quadrature rules in misspecified settings

Abstract

Kernel-based quadrature rules are becoming important in machine learning and statistics, as they achieve super-$\sqrt{n}$ convergence rates in numerical integration, and thus provide alternatives to Monte Carlo integration in challenging settings where integrands are expensive to evaluate or where integrands are high dimensional. These rules are based on the assumption that the integrand has a certain degree of smoothness, which is expressed as that the integrand belongs to a certain reproducing kernel Hilbert space (RKHS). However, this assumption can be violated in practice (e.g., when the integrand is a black box function), and no general theory has been established for the convergence of kernel quadratures in such misspecified settings. Our contribution is in proving that kernel quadratures can be consistent even when the integrand does not belong to the assumed RKHS, i.e., when the integrand is less smooth than assumed. Specifically, we derive convergence rates that depend on the (unknown) lesser smoothness of the integrand, where the degree of smoothness is expressed via powers of RKHSs or via Sobolev spaces.

To appear at NIPS2016

References in corpus (2)

• Three discussions of the paper "sequential quasi-Monte Carlo sampling", by M. Gerber and N. Chopin
• On the Equivalence between Herding and Conditional Gradient Algorithms

Cited by in corpus (11)

• Kernel Mean Embedding of Distributions: A Review and Beyond
• Learning Decentralized Controllers for Robot Swarms with Graph Neural Networks
• Sampling based approximation of linear functionals in Reproducing Kernel Hilbert Spaces
• Kernel Recursive ABC: Point Estimation with Intractable Likelihood
• Convergence Analysis of Deterministic Kernel-Based Quadrature Rules in Misspecified Settings
• On the Sampling Problem for Kernel Quadrature
• A Bayes-Sard Cubature Method
• Positively Weighted Kernel Quadrature via Subsampling
• Probabilistic Models for Integration Error in the Assessment of Functional Cardiac Models
• Estimating Rényi's $\alpha$-Cross-Entropies in a Matrix-Based Way
• Rejoinder for "Probabilistic Integration: A Role in Statistical Computation?"