Toward Understanding the Impact of Staleness in Distributed Machine Learning

Abstract

Many distributed machine learning (ML) systems adopt the non-synchronous execution in order to alleviate the network communication bottleneck, resulting in stale parameters that do not reflect the latest updates. Despite much development in large-scale ML, the effects of staleness on learning are inconclusive as it is challenging to directly monitor or control staleness in complex distributed environments. In this work, we study the convergence behaviors of a wide array of ML models and algorithms under delayed updates. Our extensive experiments reveal the rich diversity of the effects of staleness on the convergence of ML algorithms and offer insights into seemingly contradictory reports in the literature. The empirical findings also inspire a new convergence analysis of stochastic gradient descent in non-convex optimization under staleness, matching the best-known convergence rate of O(1/\sqrt{T}).

arXiv comment19 pages, 12 figures

Cited by in corpus (8)

• SpreadGNN: Serverless Multi-task Federated Learning for Graph Neural Networks
• Taming Momentum in a Distributed Asynchronous Environment
• Straggler-Resilient Distributed Machine Learning with Dynamic Backup Workers
• At Stability's Edge: How to Adjust Hyperparameters to Preserve Minima Selection in Asynchronous Training of Neural Networks?
• AdaptCL: Efficient Collaborative Learning with Dynamic and Adaptive Pruning
• Distributed Machine Learning for Wireless Communication Networks: Techniques, Architectures, and Applications
• Scheduling Optimization Techniques for Neural Network Training
• Sparsification as a Remedy for Staleness in Distributed Asynchronous SGD