Distributed Non-Convex Optimization with One-Bit Compressors on Heterogeneous Data: Efficient and Resilient Algorithms

TL;DR

This paper introduces two communication-efficient, resilient algorithms for federated learning that use one-bit gradient compression, adapt to unbounded gradients, and outperform existing methods in convergence and robustness.

Contribution

The paper proposes Ada-StoSign and $eta$-StoSign algorithms that enable efficient, resilient federated learning with one-bit compressors and adaptive gradient norm estimation.

Findings

Ada-StoSign converges at rate O(log T/√T + 1/√M)

Ada-StoSign outperforms state-of-the-art when M is large

β-StoSign provides Byzantine resilience and privacy guarantees

Abstract

Federated Learning (FL) is a nascent decentralized learning framework under which a massive collection of heterogeneous clients collaboratively train a model without revealing their local data. Scarce communication, privacy leakage, and Byzantine attacks are the key bottlenecks of system scalability. In this paper, we focus on communication-efficient distributed (stochastic) gradient descent for non-convex optimization, a driving force of FL. We propose two algorithms, named {\em Adaptive Stochastic Sign SGD (Ada-StoSign)} and {\em $\beta$-Stochastic Sign SGD ($\beta$-StoSign)}, each of which compresses the local gradients into bit vectors. To handle unbounded gradients, Ada-StoSign uses a novel norm tracking function that adaptively adjusts a coarse estimation on the $\ell_{\infty}$ of the local gradients - a key parameter used in gradient compression. We show that Ada-StoSign converges in expectation with a rate $O(\log T/\sqrt{T} + 1/\sqrt{M})$, where $M$ is the number of clients. To the best of our knowledge, when $M$ is sufficiently large, Ada-StoSign outperforms the state-of-the-art sign-based method whose convergence rate is $O(T^{-1/4})$. Under bounded gradient assumption, $\beta$-StoSign achieves quantifiable Byzantine resilience and privacy assurances, and works with partial client participation and mini-batch gradients which could be unbounded. We corroborate and complement our theories by experiments on MNIST and CIFAR-10 datasets.