An Edge Computing Marketplace for Distributed Machine Learning

Published In

SIGCOMM Posters and Demos '19: Proceedings of the ACM SIGCOMM 2019 Conference Posters and Demos

Document Type

Citation

Publication Date

8-2019

Abstract

There is an increasing demand among machine learning researchers for powerful computational resources to train their machine learning models. In order to train these models in a reasonable amount of time, the training is often distributed among multiple machines; yet paying for such machines is costly. DeepMarket attempts to reduce these costs by creating a marketplace that integrates multiple computational resources over a distributed tensorflow framework. Instead of requiring users to rent expensive resources from a third party cloud provider, DeepMarket will allow users to lend their computing resources to each other when they are available. Such a marketplace, however, requires a credit mechanism that ensures users receive resources in proportion to the resources they lend to others. Moreover, DeepMarket must respect users' needs to use their own resources and the resulting limits on when resources can be lent to others. This Demo will introduce the audience to PLUTO: DeepMarket's intuitive graphical user interface. The audience will be able to see how PLUTO in coordination with DeepMarket servers tracks the performance of each user's training jobs, matches jobs to resources made available by other users, and tracks the resulting credits that regulate the exchange of resources.

Description

© 2019 Association for Computing Machinery.

DOI

10.1145/3342280.3342299

Persistent Identifier

https://archives.pdx.edu/ds/psu/32845

Share

COinS