Session 9 Overview: ML Processors from Cloud to Edge Machine Learning Subcommittee

Abstract

Significant progress has been made in machine learning processor design in two different but important topic areas. The first addresses flexible accelerators for inference and training in the most advanced CMOS technology nodes (e.g. 5nm and 7nm) for mobile and the cloud. The second topic area covers application-specific acceleration engines for ultra-low-power applications, including wearable devices. This session comprises nine papers, covering a diverse set of neural networks targeted at a wide range of applications, including gesture recognition, smart cameras, speech-to-text and keyword spotting. © 2021 IEEE