Gradient Sparification for Asynchronous Distributed Training

Abstract

Modern large scale machine learning applications require stochastic optimization algorithms to be implemented on distributed computational architectures. A key bottleneck is the communication overhead for exchanging information, such as stochastic gradients, among different nodes. Recently, gradient sparsification techniques have been proposed to reduce communications cost and thus alleviate the network overhead. However, most of gradient sparsification techniques consider only synchronous parallelism and cannot be applied in asynchronous scenarios, such as asynchronous distributed training for federated learning at mobile devices. In this paper, we present a dual-way gradient sparsification approach (DGS) that is suitable for asynchronous distributed training. We let workers download model difference, instead of the global model, from the server, and the model difference information is also sparsified so that the information exchanged overhead is reduced by sparsifying the dual-way communication between the server and workers. To preserve accuracy under dual-way sparsification, we design a sparsification aware momentum (SAMomentum) to turn sparsification into adaptive batch size between each parameter. We conduct experiments at a cluster of 32 workers, and the results show that, with the same compression ratio but much lower communication cost, our approach can achieve better scalability and generalization ability.

arXiv comment8 pages

References in corpus (11)

• Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
• HOGWILD!: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent
• One weird trick for parallelizing convolutional neural networks
• TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning
• Large Batch Training of Convolutional Networks
• Train longer, generalize better: closing the generalization gap in large batch training of neural networks
• Highly Scalable Deep Learning Training System with Mixed-Precision: Training ImageNet in Four Minutes
• AdaComp : Adaptive Residual Gradient Compression for Data-Parallel Distributed Training
• Stochastic Nonconvex Optimization with Large Minibatches
• SparCML: High-Performance Sparse Communication for Machine Learning
• Asynchronous Parallel Stochastic Gradient Descent - A Numeric Core for Scalable Distributed Machine Learning Algorithms