Deep learning processor architecture and its application

 일시 : 2019년 6월 7일(금), 16:00 ~ 17:00

 장소 : 서울대학교반도체공동연구소설계관(104-1동) 205호

 연사 : Seokbum Ko, Ph. D., PEng

Abstract

In the first part of the talk, deep learning processor architecture will be discussed mainly from computer arithmetic point of view. Multiple/mixed precision and flexible precision techniques are newly introduced and some benefits for deep learning neural networks will be demonstrated. Additionally, approximate computing and some acceleration techniques will be further discussed.

In the second part of the talk, deep learning applications will be introduced including CT imaging, ultrasound imaging, fast detection of explosives, license plate localization/recognition, and FPGA based advanced hearing aid using deep learning.

Biography

-        Professor and Graduate Chair, Dept of ECE

-        Professor, Div. of Biomedical Engineering

-        2016: Intel Hardware Accelerator Research Program awardee

-        2002-현재: Intel, Xilinx, Microchip, IRD 등 다수의 industry 프로젝트 수행

-        2002-현재: NSERC SPG, CRD, DG 등 다수의 연방정부 프로젝트 수행

-        2002: Ph.D., University of Rhode Island, USA

• 담당교수 : 서울대학교 전기·정보공학부 이 혁 재  / 문의 : 박진우(jwpark@capp.snu.ac.kr)