서강대학교 전자공학과

1. 주제: Logic Compatible Flash Memory Based Non-Volatile Neural Network Accelerator
2. 연사: 송승환 박사 (ANAFLASH, USA)
3. 일시: 2019년 7월 16일 10:30 ~ 12:30
4. 장소: 서강대학교 R관 715호


초록

Power efficient edge computing with some degree of Artificial
Intelligence (AI) is a massive trend. The society is moving to include
AI feature virtually in every edge device such as intelligent home
product, smart mobile, security monitor, wireless sensor, self-driving
car to name a few. A cost-effective deep neural network (DNN)
accelerator is a key building block for such edge devices. Existing
solution for the DNN HW typically requires off-chip access to retrieve
neural network parameters from external memories, incurring huge
communication latency and power consumption. Alternative approach
integrating the DNN engine in a special Non-Volatile Memory (NVM)
process requires additional steps beyond the conventional logic CMOS
process which is not cost-effective for medium density DNN engine in
cost-sensitive edge devices.
In this presentation, single-poly embedded flash memory implemented in a
generic logic process is revisited for moderate density Non-Volatile DNN
applications. Then, the latest test chip result of the logic compatible
flash based neural network accelerator featuring high energy efficiency
and fine-grained analog programming scheme will be discussed.


송승환 박사 약력

Dr. Seung-hwan Song is a CTO and Co-Founder of ANAFLASH
(www.anaflash.com), Inc. that develops logic compatible embedded flash
memory based solution since 2017. Previously, he has held various
research and development positions at SK-HMS America, HGST (acquired by
WD), Seagate, Qualcomm, Broadcom, and Samsung. He received the B.S. and
M.S. degrees in electrical engineering from Seoul National University,
in 2004 and 2006, and the Ph. D. degree in electrical engineering (with
a minor in biomedical engineering) from University of Minnesota, in
2013. Dr Song is a principal investigator of the company’s SBIR awards
received from NSF and DoD Air Force, and has 45+ US patents granted
around SSD, NVM, and Flash memory. He is a recipient of a best paper
award at IEEE ICC 2016 and a low power design contest award at IEEE
ISLPED 2012.


세미나 문의: 전자공학과 최우영 교수 (wchoi@sogang.ac.kr)