Stein's method meets computational statistics: A review of some recent developments

A Anastasiou, A Barp, FX Briol, B Ebner… - Statistical …, 2023 - projecteuclid.org
Stein's method compares probability distributions through the study of a class of linear
operators called Stein operators. While mainly studied in probability and used to underpin …

Postprocessing of MCMC

LF South, M Riabiz, O Teymur… - Annual Review of …, 2022 - annualreviews.org
Markov chain Monte Carlo is the engine of modern Bayesian statistics, being used to
approximate the posterior and derived quantities of interest. Despite this, the issue of how …

Multiplicative noise and heavy tails in stochastic optimization

L Hodgkinson, M Mahoney - International Conference on …, 2021 - proceedings.mlr.press
Although stochastic optimization is central to modern machine learning, the precise
mechanisms underlying its success, and in particular, the precise role of the stochasticity …

Kernel stein discrepancy descent

A Korba, PC Aubin-Frankowski… - International …, 2021 - proceedings.mlr.press
Among dissimilarities between probability distributions, the Kernel Stein Discrepancy (KSD)
has received much interest recently. We investigate the properties of its Wasserstein …

Optimal thinning of MCMC output

M Riabiz, WY Chen, J Cockayne… - Journal of the Royal …, 2022 - academic.oup.com
The use of heuristics to assess the convergence and compress the output of Markov chain
Monte Carlo can be sub-optimal in terms of the empirical approximations that are produced …

Gradient estimation with discrete stein operators

J Shi, Y Zhou, J Hwang, M Titsias… - Advances in neural …, 2022 - proceedings.neurips.cc
Gradient estimation---approximating the gradient of an expectation with respect to the
parameters of a distribution---is central to the solution of many machine learning problems …

Stochastic stein discrepancies

J Gorham, A Raj, L Mackey - Advances in Neural …, 2020 - proceedings.neurips.cc
Stein discrepancies (SDs) monitor convergence and non-convergence in approximate
inference when exact integration and sampling are intractable. However, the computation of …

Metrizing weak convergence with maximum mean discrepancies

CJ Simon-Gabriel, A Barp, B Schölkopf… - Journal of Machine …, 2023 - jmlr.org
This paper characterizes the maximum mean discrepancies (MMD) that metrize the weak
convergence of probability measures for a wide class of kernels. More precisely, we prove …

A Riemann–Stein kernel method

A Barp, CJ Oates, E Porcu, M Girolami - Bernoulli, 2022 - projecteuclid.org
This paper proposes and studies a numerical method for approximation of posterior
expectations based on interpolation with a Stein reproducing kernel. Finite-sample-size …

A kernel stein test of goodness of fit for sequential models

J Baum, H Kanagawa, A Gretton - … Conference on Machine …, 2023 - proceedings.mlr.press
We propose a goodness-of-fit measure for probability densities modeling observations with
varying dimensionality, such as text documents of differing lengths or variable-length …