Taming the Interacting Particle Langevin Algorithm: The Superlinear case
Open Access
- 13 June 2025
- journal article
- research article
- Published by Springer Nature in Applied Mathematics & Optimization
- Vol. 91 (3) , 1-34
Abstract
Recent advances in stochastic optimization have yielded the interacting particle Langevin algorithm (IPLA), which leverages the notion of interacting particle systems (IPS) to efficiently sample from approximate posterior densities. This becomes particularly crucial in relation to the framework of Expectation-Maximization (EM), where the E-step is computationally challenging or even intractable. Although prior research has focused on scenarios involving convex cases with gradients of log densities that grow at most linearly, our work extends this framework to include polynomial growth. Taming techniques are employed to produce an explicit discretization scheme that yields a new class of stable, under such non-linearities, algorithms which are called tamed interacting particle Langevin algorithms (tIPLA). We obtain non-asymptotic convergence error estimates in Wasserstein-2 distance for the new class under the best known rate.Keywords
All Related Versions
This publication has 22 references indexed in Scilit:
- Non-asymptotic estimates for TUSLA algorithm for non-convex learning with applications to neural networks with ReLU activation functionIMA Journal of Numerical Analysis, 2023
- Taming Neural Networks with TUSLA: Nonconvex Learning via Adaptive Stochastic Gradient Langevin AlgorithmsSIAM Journal on Mathematics of Data Science, 2023
- Efficient stochastic optimisation by unadjusted Langevin Monte CarloStatistics and Computing, 2021
- The tamed unadjusted Langevin algorithmStochastic Processes and their Applications, 2019
- On explicit order 1.5 approximations with varying coefficients: The case of super-linear diffusion coefficientsJournal of Complexity, 2019
- Strong convergence of an explicit numerical method for SDEs with nonglobally Lipschitz continuous coefficientsThe Annals of Applied Probability, 2012
- Strong and weak divergence in finite time of Euler's method for stochastic differential equations with non-globally Lipschitz continuous coefficientsProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2010
- An Introduction to Infinite-Dimensional Analysis2006
- Maximum Likelihood Algorithms for Generalized Linear Mixed ModelsJournal of the American Statistical Association, 1997
- Maximum Likelihood from Incomplete Data Via the EM AlgorithmJournal of the Royal Statistical Society Series B: Statistical Methodology, 1977