Communication-Efficient Federated Learning via Sparse Training with Regularized Error Correction

Federated Learning (FL) is an emerging paradigm that allows for decentralized machine learning (ML), where multiple models are collaboratively trained in a privacy-preserving manner. However, since communications and computation resources are limited, training DNN models in FL systems face challenges such as elevated computational and communication costs in complex tasks. Sparse training schemes gain increasing attention in order to scale down the dimensionality of each client (i.e., node) transmission. Specifically, sparsification with error correction methods is a promising technique, where only important updates are sent to the parameter server (PS) and the rest are accumulated locally. While error correction methods have shown to achieve a significant sparsification level of the client-to-PS message without harming convergence, pushing sparsity further remains unresolved due to the staleness effect. In this paper, we propose a novel algorithm, dubbed Federated Learning with Accumulated Regularized Embeddings (FLARE), to overcome this challenge. FLARE presents a novel sparse training approach via accumulated pulling of the updated models with regularization on the embeddings in the FL process, providing a powerful solution to the staleness effect, and pushing sparsity to an exceptional level. Our theoretical analysis demonstrates that FLARE's regularized error feedback achieves significant improvements in scalability with sparsity parameter. The performance of FLARE is validated through experiments on diverse and complex models, achieving a remarkable sparsity level (10 times and more beyond the current state-of-the-art) along with significantly improved accuracy.