Mean-Field Langevin Dynamics for Signed Measures via a Bilevel Approach

TL;DR

This paper extends Mean-Field Langevin Dynamics to signed measures using a bilevel approach, providing stronger convergence guarantees and faster rates for convex optimization problems on manifolds, including neural network training.

Contribution

It introduces a bilevel reduction method for applying MLFD to signed measures, demonstrating improved convergence rates and applicability to neural network learning.

Findings

Bilevel reduction yields stronger guarantees than lifting.

Annealing schedule improves convergence to fixed accuracy.

Local exponential convergence for single neuron learning.

Abstract

Mean-field Langevin dynamics (MLFD) is a class of interacting particle methods that tackle convex optimization over probability measures on a manifold, which are scalable, versatile, and enjoy computational guarantees. However, some important problems -- such as risk minimization for infinite width two-layer neural networks, or sparse deconvolution -- are originally defined over the set of signed, rather than probability, measures. In this paper, we investigate how to extend the MFLD framework to convex optimization problems over signed measures. Among two known reductions from signed to probability measures -- the lifting and the bilevel approaches -- we show that the bilevel reduction leads to stronger guarantees and faster rates (at the price of a higher per-iteration complexity). In particular, we investigate the convergence rate of MFLD applied to the bilevel reduction in the low-noise regime and obtain two results. First, this dynamics is amenable to an annealing schedule, adapted from Suzuki et al. (2023), that results in improved convergence rates to a fixed multiplicative accuracy. Second, we investigate the problem of learning a single neuron with the bilevel approach and obtain local exponential convergence rates that depend polynomially on the dimension and noise level (to compare with the exponential dependence that would result from prior analyses).