Robust In-Memory Computing in Light of Model Stability

Department of Electrical Engineering Seminar

Kevin Cao
Arizona State University

“Robust In-Memory Computing in Light of Model Stability”

Tuesday, April 13, 2021
3:00 PM
Zoom Link: Contact dept. for instructions

Abstract: With the ever-increasing model size in machine learning, contemporary AI accelerators are facing the tremendous challenge in off-chip memory access, i.e., the von Neumann bottleneck. Leveraging the advances in device technology and design techniques, in-memory computing (IMC) embeds analog deep-learning operations in the memory array, achieving massively parallel computing with high storage density. On the other side, its performance is still limited by device non-idealities, circuit precision, on-chip interconnection, and algorithm properties.

Based on statistical data from a fully-integrated 65nm CMOS/RRAM test chip and a cross-layer simulation framework, we illustrate that the real bottleneck of current IMC system is not the RRAM cross-bar, but the stability of machine learning models, peripheral circuits and interconnection. They interact with device variations, limiting the inference accuracy and system energy-delay product (EDP). Furthermore, we demonstrate that training for model stability is an effective method to recover the accuracy loss and improve IMC robustness and efficiency. The results are summarized into a roofline model and applied to various datasets, helping shed light on future IMC research focus.

Bio: Yu Cao received the B.S. degree in physics from Peking University in 1996. He received the M.A. degree in biophysics and the Ph.D. degree in electrical engineering from University of California, Berkeley, in 1999 and 2002, respectively. He is now a Professor of Electrical Engineering at Arizona State University, Tempe, Arizona. Dr. Cao has published numerous articles and two books on nano-CMOS modeling and physical design. His research interests include neural-inspired computing, hardware design for on-chip learning, and reliable integration of nanoelectronics. Dr. Cao was a recipient of the 2020 Intel Outstanding Researcher Award, the 2009 ACM SIGDA Outstanding New Faculty Award, the 2006 NSF CAREER Award, the 2006 and 2007 IBM Faculty Award, and four Best Paper awards. He is a Fellow of the IEEE.

Hosted by: Prof. Priya Panda