Machine Learning Pipelines: Training, Deployment and Opportunities for Reconfigurable Hardware
Markus Weimer - Principal Scientist Manager at Microsoft
Monday, May 21, 2018

Abstract

This talk will introduce the machine learning workflow at Microsoft. This approach centers on the concept of pipelines of operators. Those are used at training time to produce the prediction pipeline. The latter is used in various deployment systems across IoT, mobile devices, servers and the cloud. The talk will discuss challenges and current solutions in that transfer from training to deployment, deployment itself. Throughout this discussion, Markus Weimer will highlight existing use of reconfigurable hardware as well as opportunities for future work.

About the speaker

Markus Weimer leads the machine learning algorithms and toolkit group in Microsoft’s Cloud & AI division. The group develops Microsoft's major internal machine learning toolkit. He his also a member of the Apache Software Foundation and was the inaugural PMC chair (VP) of Apache REEF. His work focuses on machine learning techniques, systems therefore and applications thereof. Prior, He led the machine learning research group of the Cloud Information Services Laboratory (CISL) at Microsoft and prior to that, was a researcher at Yahoo! Research.