An Iterative Bayesian Robbins--Monro Sequence

TL;DR

This paper presents an iterative Bayesian Robbins--Monro sequence that enhances classical root-finding methods by incorporating Bayesian updates, leading to faster convergence and improved robustness in noisy, real-world applications like brain stimulation threshold estimation.

Contribution

It introduces a Bayesian variant of the Robbins--Monro sequence that adaptively updates priors for improved convergence speed and robustness in noisy settings.

Findings

Almost sure convergence of the sequence.

Superior performance in noisy threshold estimation.

Reduced error margins and outlier frequency in simulations.

Abstract

This study introduces an iterative Bayesian Robbins--Monro (IBRM) sequence, which unites the classical Robbins--Monro sequence with statistical estimation for faster root-finding under noisy observations. Although the standard Robbins--Monro method iteratively approaches solutions, its convergence speed is limited by noisy measurements and naivety to any prior information about the objective function. The proposed Bayesian sequence dynamically updates the prior distribution with newly obtained observations to accelerate convergence rates and robustness. The paper demonstrates almost sure convergence of the sequence and analyses its convergence rates for both one-dimensional and multi-dimensional problems. We evaluate the method in a practical application that suffers from large variability and allows only a few function evaluations, specifically estimating thresholds in noninvasive brain stimulation, where the method is more robust and accurate than conventional alternatives. Simulations involving 25,000 virtual subjects illustrate reduced error margins and decreased outlier frequency with direct impact on clinical use.