Magnitude Matters: Fixing signSGD Through Magnitude-Aware Sparsification and Error Feedback in the Presence of Data Heterogeneity
- 20 September 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 72 (1053587X) , 5140-5155
Abstract
Communication overhead has become one of the major bottlenecks in the distributed training of deep neural networks. To alleviate the concern, various gradient compression methods have been proposed, and sign-based algorithms are of surging interest. However, SIGNSGD fails to converge in the presence of data heterogeneity, which is commonly observed in the emerging federated learning (FL) paradigm. Error feedback has been proposed to address the non-convergence issue. Nonetheless, it requires the workers to locally keep track of the compression errors, which renders it not suitable for FL since the workers may not participate in the training throughout the learning process. In this paper, we propose a magnitude-driven sparsification scheme, which addresses the non-convergence issue of SIGNSGD while further improving communication efficiency. Moreover, the local update and the error feedback schemes are further incorporated to improve the learning performance (i.e., test accuracy and communication efficiency), and the convergence of the proposed method is established. The effectiveness of the proposed scheme is validated through extensive experiments on Fashion-MNIST, CIFAR-10, CIFAR-100, Tiny-ImageNet, and Mini-ImageNet datasets.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62301487)
- Zhejiang Provincial Natural Science Foundation of China (LQ23F010021)
- National Science Foundation (ECCS-2203214)
- National Research Foundation Singapore
- Infocomm Media Development Authority under its Future Communications Research & Development Programme
This publication has 18 references indexed in Scilit:
- Algorithm Unrolling: Interpretable, Efficient Deep Learning for Signal and Image ProcessingIEEE Signal Processing Magazine, 2021
- UVeQFed: Universal Vector Quantization for Federated LearningIEEE Transactions on Signal Processing, 2020
- Deep Learning for Wireless Communications: An Emerging Interdisciplinary ParadigmIEEE Wireless Communications, 2020
- Federated Learning: Challenges, Methods, and Future DirectionsIEEE Signal Processing Magazine, 2020
- High-Dimensional Stochastic Gradient Quantization for Communication-Efficient Edge LearningIEEE Transactions on Signal Processing, 2020
- Machine Learning at the Wireless Edge: Distributed Stochastic Gradient Descent Over-the-AirIEEE Transactions on Signal Processing, 2020
- Robust and Communication-Efficient Federated Learning From Non-i.i.d. DataIEEE Transactions on Neural Networks and Learning Systems, 2019
- Sparse Binary Compression: Towards Distributed Deep Learning with minimal Communication2019
- Deep Residual Learning for Image Recognition2016
- Stochastic Spectral Descent for Discrete Graphical ModelsIEEE Journal of Selected Topics in Signal Processing, 2015