Optimised weight programming for analogue memory-based deep neural networks
- MRS Fall Meeting 2022
HsinYu (Sidney) Tsai received her Ph.D. from the Electrical Engineering and Computer Science department at Massachusetts Institute of Technology in 2011. In 2021, Sidney joined the Nanofabrication and Electron Beam Lithography group at the IBM T.J. Watson Research Center as Research Staff Member and developed directed self-assembly (DSA) lithography for finFETs. From 2015-2016, Sidney served as the manager of the Advanced Lithography group in the Microelectronics Research Laboratory (MRL), managing operations of a 200mm research prototyping line.
Sidney currently works in the Alamden Research Center in San Jose, CA, as a Principal Research Staff Memebr and manager of the Analog AI group. Analog AI based on Phase Change Memory (PCM) devices utilizes emerging non-volatile memory embedded in the backend to compute vector-matrix multiplication at the location of of data, potentially achieving high power performance for deep learning workloads in the Cloud and on the edge. The group demonstrates software compatible accuracies for both training and inference of Deep Neural Networks (DNNs). In 2021, the IBM Analog AI teams published two highlight papers at the VLSI conference based on two inference chips with PCM devices fabricated on top of 14nm CMOS transistors. The group is now focusing on developing large scale, configurable hardware for both inference and training applications.