Loading…
Embedded Technologies Expo & Conference 2020 has ended
Back To Schedule
Tuesday, November 17 • 10:00am - 10:45am
AT1: Low-Cost Embedded AI

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

A main cost in embedded computing is memory. For this reason, embedded processors invariably provide a small amount of on-chip memory that can be extended with external memory. This session presents methods to make best use of small embedded memories in embedded AI inferencing by using binarized and 8-bit quantized networks, and best use of  limited bandwidth to external memories in quantized convolutional neural networks.

Speakers
avatar for Laszlo Kindrat

Laszlo Kindrat

Machine Learning Engineer, XMOS
Laszlo Kindrat is currently in the role of Machine Learning Engineer at XMOS in Hampton, NH. He has aided the design of the XS3 architecture's vector unit and plays a key role in developing inference tools for the xCORE-AI platform. Laszlo also has experience with traditional and... Read More →
avatar for Henk Muller

Henk Muller

Principal Technologist, XMOS
Henk Muller is currently the Principal Technologist at XMOS Ltd. He is the lead architect of the XS3 architecture, has assisted in the implementation of the xCORE200 and xCORE-AI platforms and the design and implementation of software for real time systems. Prior to working at XMOS... Read More →


Tuesday November 17, 2020 10:00am - 10:45am PST
LL21A