세미나
- 2021.07.07
- 613
1. Title: Neuromorphic Processor Featuring
Nonvolatile Compute-in-Memory Architecture
2. 일시:
2021년
7월 12일 월요일 오전
10시~11시
3. 장소:
Zoom 을 이용한 online 접속(https://us02web.zoom.us/j/7616590262pwd=K0hQNWVVbzZidTYrdGNpMlJkSVk0UT09,
4. 회의 ID: 761 659 0262, 암호: TIDL)
5. Abstract:
High speed vision processing based on recent
artificial intelligence technology is finding growing interest, since the
processing performance and accuracy have improved beyond traditional signal
processing methods significantly. The anticipated edge applications, however,
require minimal power consumption to perform such intelligence within the
resource constrained smart edge devices. Traditional Von Neumann architecture has
a separate processor and memory, consuming high power and having access
overhead. Differently, non-volatile processor combining processor and memory
may pave the way to enhance the performance under given power budget in the
smart edge devices.
In
this talk, logic compatible embedded Flash memory based 8-bit neural network
core adopting such new architecture will be discussed first. Bit-serial and
bit-parallel operations have been deployed to input and weights, respectively,
resulting 8-bit input and weight operations of the neural network with less
reliability concern on the analog circuitry and memory. As a result, the
silicon test chip result showed improved an improved accuracy and reliability
result compared to the prior generation test chip. Then, we will briefly
introduce our recent research on neuromorphic processor featuring Resistive
Random-Access Memory (ReRAM) for radiation-tolerant applications.
[Bio] Dr.
Seung-hwan Song is a CEO and Co-Founder of ANAFLASH (www.anaflash.com).
that develops brain inspired processor for smart edge devices since 2017.
He is also a CEO and Founder of SEMIBRAIN (www.semibraintech.com) that
develops logic compatible embedded Flash based smart semiconductor in Korea
since 2020. He has held various research and development positions at
SK-HMS America, HGST (acquired by WD), Seagate, Qualcomm, Broadcom, and
Samsung. He received the Ph. D. degree in electrical engineering (with a
minor in biomedical engineering) from University of Minnesota, in 2013 and the
M.S. and B.S. degrees in electrical engineering from Seoul National
University, in 2006 and 2004.
Dr Song has
conducted several SBIR projects from NASA, NSF, and Air Force in US. He has 50+
granted patents and a recipient of a best paper award at IEEE ICC 2016
and a low power design contest award at IEEE ISLPED 2012.
문의: 최우영 교수 (wchoi@sogang.ac.kr, 02-705-8467)