Compressing gradients by exploiting temporal correlation in momentum-SGD

Abstract

An increasing bottleneck in decentralized optimization is communication. Bigger models and growing datasets mean that decentralization of computation is important and that the amount of information exchanged is quickly growing. While compression techniques have been introduced to cope with the latter, none has considered leveraging the temporal correlations that exist in consecutive vector updates. An important example is distributed momentum-SGD where temporal correlation is enhanced by the low-pass-filtering effect of applying momentum. In this paper we design and analyze compression methods that exploit temporal correlation in systems both with and without error-feedback. Experiments with the ImageNet dataset demonstrate that our proposed methods offer significant reduction in the rate of communication at only a negligible increase in computation complexity. We further analyze the convergence of SGD when compression is applied with error-feedback. In the literature, convergence guarantees are developed only for compressors that provide error-bounds point-wise, i.e., for each input to the compressor. In contrast, many important codes (e.g. rate-distortion codes) provide error-bounds only in expectation and thus provide a more general guarantee. In this paper we prove the convergence of SGD under an expected error assumption by establishing a bound for the minimum gradient norm.

arXiv commentThis paper was presented in part at the 11th International Symposium on Topics in Coding (ISTC), Montreal, QC, Canada, August 2021, and the paper has been accepted for publication in the IEEE Journal on Selected Areas in Information Theory (JSAIT) Volume 2, Issue 3 (2021), https://ieeexplore.ieee.org/document/9511618

References in corpus (13)

• Wide Residual Networks
• Deep Gradient Compression: Reducing the Communication Bandwidth for Distributed Training
• QSGD: Communication-Efficient SGD via Gradient Quantization and Encoding
• TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning
• Sparse Communication for Distributed Gradient Descent
• Horovod: fast and easy distributed deep learning in TensorFlow
• A Downsampled Variant of ImageNet as an Alternative to the CIFAR datasets
• Sparsified SGD with Memory
• Gradient Sparsification for Communication-Efficient Distributed Optimization
• ATOMO: Communication-efficient Learning via Atomic Sparsification
• Error Feedback Fixes SignSGD and other Gradient Compression Schemes
• Communication-Efficient Distributed Blockwise Momentum SGD with Error-Feedback
• Understanding Top-k Sparsification in Distributed Deep Learning