인공지능 서비스가 구체화 되면서 이의 핵심이 되는 neural network 의 hardware 화가 진행되고 있다. IBM, Google 등은 big data 처리를 위한 server 용 neural processor 에 집중하고 Qualcomm 은 mobile 기기용으로, 그리고 Intel 은 자율주행용 processor 로 방향을 잡았다. 이에 따라 국내 반도체 기업을 고려했을 때 바람직한 neural processor 를 개발 방향을 검토하고 이를 대비하기 위하여 연구기관에서 준비해야 할 반도체소자를 제안한다. 한편, neural network 기법은 이미 bioinformatics 에서 활용해 왔던 tool 이었기 때문에 소재 개발에도 적용할 수 있도록 materials informatics 관점에서 그 용도를 제안하고자 한다. 특히 data 가 없거나 빈약한 경우 신소재 발굴 시간을 단축할 수 있는 방안과 계획을 사례를 이용하여 소개한다. 끝으로 인공지능을 가장 절실하게 요구하고 있는 금융계에 필요한 neural network 이 어떠한 형태인가를 review 하고 이의 현상을 hardware 에 어떻게 적용할 것인지를 언급한다.
In Kyeong Yoo received his BS degree in Metallurgical Engineering from Hanyang University in Korea in 1975 and his MS and Ph.D. degrees in Materials Science and Engineering from VA Tech in USA in 1986 and 1990, respectively. Dr. Yoo worked as a member of Nondestructive Testing Engineers from 1978 through 1984 at Hyundai Heavy Industries. He joined the Materials Engineering Department at VA Tech as a Research Scientist from 1991 through 1993 and joined Samsung Advanced Institute of Technology (SAIT) in 1993 as a team leader. Dr. Yoo described several electrical failure mechanisms in ferroelectrics including breakdown mechanisms and fatigue mechanisms when he joined VA Tech from 1991 through 1993. He is an expert in electrical characterization, modeling, and reliability analysis in oxides, dielectrics, and ferroelectrics. He developed the world’s first 1T-1C 64K PZT FRAM(Ferroelectric Random Access Memory) in 1996. He pioneered oxide TFT for LCD display application and developed the world’s first 1T-1R NiO resistance change memory (OxRRAM) in 2005. Dr. Yoo has published more than 130 papers and filed more than 145 Patents in fields including FRAM and OxRRAM. His most recent paper describes the source of data scatter of the asymmetric double switching in NiOx interpreted on the basis of percolation. He has served MRS as a Symposium Organizer for Ferroelectric Thin Films V, VI, and VI. He also served MRS as a Meeting Chair for 2014 Fall Meeting. He was appointed as Samsung Fellow in 2002, Samsung’s highest technical honor. He has consulted Materials Research Center at SAIT by December 2015 in finding new opportunities in the field of materials informatics and analytics device for IoT in relation to strategies for impending social infrastructure. He is currently working at POSTECH as a research professor in the field of Synaptic devices and Reconfigurable logics Dr. Yoo has received the FRAM development Award in 1996 and four Patent Awards as well as the Paper Award at SAIT.