Stochastic gradient descent is a widely used method to find locally-optimal models in machine learning and data mining. However, it is naturally a sequential algorithm, and parallelization involves severe compromises because the cost of synchronizing across a cluster is much larger than the time required to compute an optimal-sized gradient step. Here we explore butterfly mixing, where gradient steps are interleaved with the k stages of a butterfly network on 2^k nodes. Udp based butterfly mix steps should be extremely fast and failure-tolerant, and convergence is almost as fast as a full mix (AllReduce) on every step.
Title
Communication-Efficient Distributed Stochastic Gradient Descent with Butterfly Mixing
Published
2012-05-11
Full Collection Name
Electrical Engineering & Computer Sciences Technical Reports
Other Identifiers
EECS-2012-96
Type
Text
Extent
15 p
Archive
The Engineering Library
Usage Statement
Researchers may make free and open use of the UC Berkeley Library’s digitized public domain materials. However, some materials in our online collections may be protected by U.S. copyright law (Title 17, U.S.C.). Use or reproduction of materials protected by copyright beyond that allowed by fair use (Title 17, U.S.C. § 107) requires permission from the copyright owners. The use or reproduction of some materials may also be restricted by terms of University of California gift or purchase agreements, privacy and publicity rights, or trademark law. Responsibility for determining rights status and permissibility of any use or reproduction rests exclusively with the researcher. To learn more or make inquiries, please see our permissions policies (https://www.lib.berkeley.edu/about/permissions-policies).
