Abstrakti

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.
AlkuperäiskieliEnglanti
OtsikkoProceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence
Sivut1703-1712
Sivumäärä10
TilaJulkaistu - joulukuuta 2021
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisuussa
TapahtumaConference on Uncertainty in Artificial Intelligence - Virtual, Online
Kesto: 27 heinäkuuta 202129 heinäkuuta 2021
https://auai.org/uai2021/

Julkaisusarja

NimiProceedings of Machine Learning Research
KustantajaPMLR
Numero161
ISSN (elektroninen)2640-3498

Conference

ConferenceConference on Uncertainty in Artificial Intelligence
LyhennettäUAI
KaupunkiVirtual, Online
Ajanjakso27/07/202129/07/2021
www-osoite

Sormenjälki

Sukella tutkimusaiheisiin 'Federated Stochastic Gradient Langevin Dynamics'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä