Frank-Wolfe Bayesian Quadrature: Probabilistic Integration with Theoretical Guarantees

Abstract

There is renewed interest in formulating integration as an inference problem, motivated by obtaining a full distribution over numerical error that can be propagated through subsequent computation. Current methods, such as Bayesian Quadrature, demonstrate impressive empirical performance but lack theoretical analysis. An important challenge is to reconcile these probabilistic integrators with rigorous convergence guarantees. In this paper, we present the first probabilistic integrator that admits such theoretical treatment, called Frank-Wolfe Bayesian Quadrature (FWBQ). Under FWBQ, convergence to the true value of the integral is shown to be exponential and posterior contraction rates are proven to be superexponential. In simulations, FWBQ is competitive with state-of-the-art methods and out-performs alternatives based on Frank-Wolfe optimisation. Our approach is applied to successfully quantify numerical error in the solution to a challenging model choice problem in cellular biology.

References in corpus (5)

• On the Equivalence between Herding and Conditional Gradient Algorithms
• Probability Measures for Numerical Solutions of Differential Equations
• Probabilistic ODE Solvers with Runge-Kutta Means
• Sampling for Inference in Probabilistic Models with Fast Bayesian Quadrature
• On the relation between Gaussian process quadratures and sigma-point methods

Cited by in corpus (21)

• Gaussian Processes and Kernel Methods: A Review on Connections and Equivalences
• Variational Bayesian Monte Carlo with Noisy Likelihoods
• Strong convergence rates of probabilistic integrators for ordinary differential equations
• Convergence guarantees for kernel-based quadrature rules in misspecified settings
• On two ways to use determinantal point processes for Monte Carlo integration
• Optimal quantisation of probability measures using maximum mean discrepancy
• Exact Sampling from Determinantal Point Processes
• Kernel interpolation with continuous volume sampling
• Convergence Analysis of Deterministic Kernel-Based Quadrature Rules in Misspecified Settings
• Positively Weighted Kernel Quadrature via Subsampling
• Probabilistic Models for Integration Error in the Assessment of Functional Cardiac Models
• Fully symmetric kernel quadrature
• Monte Carlo Anti-Differentiation for Approximate Weighted Model Integration
• Bayesian posterior approximation via greedy particle optimization
• Neural Conditional Gradients
• Kernel quadrature by applying a point-wise gradient descent method to discrete energies
• Bayesian Probabilistic Numerical Integration with Tree-Based Models
• A Bulirsch-Stoer algorithm using Gaussian processes
• Approximate Frank-Wolfe Algorithms over Graph-structured Support Sets
• Kernel quadrature with DPPs
• Black Box Probabilistic Numerics