UID:
almahu_9949420054202882
Format:
XIII, 127 p. 40 illus., 38 illus. in color.
,
online resource.
Edition:
1st ed. 2023.
ISBN:
9783031190674
Series Statement:
Synthesis Lectures on Learning, Networks, and Algorithms,
Content:
This book discusses state-of-the-art stochastic optimization algorithms for distributed machine learning and analyzes their convergence speed. The book first introduces stochastic gradient descent (SGD) and its distributed version, synchronous SGD, where the task of computing gradients is divided across several worker nodes. The author discusses several algorithms that improve the scalability and communication efficiency of synchronous SGD, such as asynchronous SGD, local-update SGD, quantized and sparsified SGD, and decentralized SGD. For each of these algorithms, the book analyzes its error versus iterations convergence, and the runtime spent per iteration. The author shows that each of these strategies to reduce communication or synchronization delays encounters a fundamental trade-off between error and runtime.
Note:
Distributed Optimization in Machine Learning -- Calculus, Probability and Order Statistics Review -- Convergence of SGD and Variance-Reduced Variants -- Synchronous SGD and Straggler-Resilient Variants -- Asynchronous SGD and Staleness-Reduced Variants -- Local-update and Overlap SGD -- Quantized and Sparsified Distributed SGD -- Decentralized SGD and its Variants.
In:
Springer Nature eBook
Additional Edition:
Printed edition: ISBN 9783031190667
Additional Edition:
Printed edition: ISBN 9783031190681
Additional Edition:
Printed edition: ISBN 9783031190698
Language:
English
DOI:
10.1007/978-3-031-19067-4
URL:
https://doi.org/10.1007/978-3-031-19067-4
