Impact of PCM Flicker Noise and Weight Drift on Analog Hardware Inference for state-of-the-art Deep Learning Networks

Jin-Ping Han

doi:https://doi.org/10.7567/SSDM.2020.B-7-01

09:00 〜 09:15

[B-7-01] Impact of PCM Flicker Noise and Weight Drift on Analog Hardware Inference for state-of-the-art Deep Learning Networks

〇Jin-Ping Han¹, Malte Rasch¹, Zuoguang Liu², Paul Solomon¹, Kevin Brew², Kangguo Cheng², Injo OK², Victor Chan², Michael Longstreet³, Wanki Kim¹, Robert L Bruce¹, Cheng-Wei Cheng¹, Nanbo Gong¹, Praneet Adusumilli², Shawn Pinkett², Hongmei Li³, Nicole Saulnier², Tenko Yamashita², Matthew J BrightSky¹, Vijay Narayanan¹ (1. IBM T. J. Watson Research Center(United States of America), 2. IBM Research - Albany Nano Tech(United States of America), 3. IBM System(United States of America))

https://doi.org/10.7567/SSDM.2020.B-7-01

2020 International Conference on Solid State Devices and Materials

[B-7-01] Impact of PCM Flicker Noise and Weight Drift on Analog Hardware Inference for state-of-the-art Deep Learning Networks