Lightning Talk: Co-Design and Characterization of AI/ML Embedded Software for Ultra-Low Power Platforms
DescriptionThe design of low-power and resource-constrained platforms has received much attention in the last decades, as technological advancements in chip manufacturing have allowed complex AI/ML algorithms to be executed within the μW range. To ensure continued progress in this domain, new co-design views that optimize hardware and software simultaneously are necessary. This talk summarizes the latest trends in complete embedded software applications designed for the real-time operation of complex AI/ML algorithms. It examines the requirements of typical signal acquisition and processing phases used in the latest AI/deep learning embedded software, including varying computational workloads and relations between active and idle times. Finally, it presents the different trends in benchmark suites for ultra-low power (ULP) AI/ML platform designs under varying requirements.
Event Type
Research Manuscript
TimeThursday, July 13th3:30pm - 3:40pm PDT
Location3006, 3rd Floor
Embedded Systems
Embedded Software