Abstract

Stochastic gradient MCMC methods, such as stochastic gradient Langevin dynamics (SGLD), employ fast but noisy gradient estimates to enable large-scale posterior sampling. Although we can easily extend SGLD to distributed settings, it suffers from two issues when applied to federated non-IID data. First, the variance of these estimates increases significantly. Second, delaying communication causes the Markov chains to diverge from the true posterior even for very simple models. To alleviate both these problems, we propose conducive gradients, a simple mechanism that combines local likelihood approximations to correct gradient updates. Notably, conducive gradients are easy to compute, and since we only calculate the approximations once, they incur negligible overhead. We apply conducive gradients to distributed stochastic gradient Langevin dynamics (DSGLD) and call the resulting method federated stochastic gradient Langevin dynamics (FSGLD). We demonstrate that our approach can handle delayed communication rounds, converging to the target posterior in cases where DSGLD fails. We also show that FSGLD outperforms DSGLD for non-IID federated data with experiments on metric learning and neural networks.
Original languageEnglish
Title of host publicationProceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence
PublisherJMLR
Pages1703-1712
Number of pages10
Publication statusPublished - Dec 2021
MoE publication typeA4 Conference publication
EventConference on Uncertainty in Artificial Intelligence - Virtual, Online
Duration: 27 Jul 202129 Jul 2021
https://auai.org/uai2021/

Publication series

NameProceedings of Machine Learning Research
PublisherPMLR
Number161
ISSN (Electronic)2640-3498

Conference

ConferenceConference on Uncertainty in Artificial Intelligence
Abbreviated titleUAI
CityVirtual, Online
Period27/07/202129/07/2021
Internet address

Fingerprint

Dive into the research topics of 'Federated Stochastic Gradient Langevin Dynamics'. Together they form a unique fingerprint.

Cite this