$\beta$-Stochastic Sign SGD: A Byzantine Resilient and Differentially Private Gradient Compressor for Federated Learning

Ming Xiang, Lili Su

Keywords: distributed systems, differential privacy, communication efficiency, convergence rate analysis, robust optimization

TL;DR: Clients send stochastic sign-bits gradient estimates to a server, which aggregates updates based on majority vote. We show that this algorithm is provable differentially private, convergent, communication efficient, and Byzantine fault tolerant.

Abstract: Federated Learning (FL) is a nascent privacy-preserving learning framework under which the local data of participating clients is kept locally throughout model training. Scarce communication resources and data heterogeneity are two defining characteristics of FL. Besides, a FL system is often implemented in a harsh environment -- leaving the clients vulnerable to Byzantine attacks. To the best of our knowledge, no gradient compressors simultaneously achieve quantitative Byzantine resilience and privacy preservation. In this paper, we fill this gap via revisiting the stochastic sign SGD \cite{Jin2020}. We propose $\beta$-stochastic sign SGD, which contains a gradient compressor that encodes a client's gradient information in sign bits subject to the privacy budget $\beta>0$. We show that as long as $\beta>0$, $\beta$-stochastic sign SGD converges in the presence of partial client participation and mobile Byzantine faults, showing that it achieves quantifiable Byzantine-resilience and differential privacy simultaneously. In sharp contrast, when $\beta=0$, the compressor is not differentially private. Notably, for the special case when each of the stochastic gradients involved is bounded with known bounds, our gradient compressor with $\beta=0$ coincides with the compressor proposed in \cite{Jin2020}. As a byproduct, we show that when the clients report sign messages, the popular information aggregation rules simple mean, trimmed mean, median and majority vote are identical in terms of the output signs. Our theories are corroborated by experiments on MNIST and CIFAR-10 datasets.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Theory (eg, control theory, learning theory, algorithmic game theory)