October issue cover sheet Advanced Practical Materials. Credit Dead Korea Institute of Materials Science (KMS)
For the first time in the history of the country, researchers in Korea have developed a semiconductor for the next generation of neuromorphic (neural network). This is the result of a joint team led by Dr. Jung-de Quun and a study led by Yong-Hun Kim of the Department of Energy and Electronics of the Korean Institute of Material Sciences. KIMS is a government-sponsored research institute under the Ministry of Science and ICT.
This new concept memtransistor uses two-dimensional nano material with multiple nanometers thickness. The researchers copied the electrical plastics of nerve synapses to more than 1,000 electrical stimuli and achieved a high pattern detection speed of 94.2% (98% simulation-based pattern detection).
Molybdenum Sulfur (MS2Widely used as a semiconductor material, it works on the principle that defects in a single crystal move through an external electric field, making it difficult to accurately control the focus or shape of the defect. To solve the problem, the research team sequentially applied a layer of niobium oxide (Nb) oxide.2Oh5) And Molybdenum Sulfur Material They also showed that by changing the thickness of the nebulium oxide coating, the protective properties of the immune system could be controlled freely, and memory information related to memory and forgetfulness could be processed at a low power of 10 PJ (picojoule).
Artificial intelligence hardware currently consumes a significant amount of energy and expenditure on GPUs, FPGAs and ASICs, and is expected to create explosive demand in the future as the industry grows. Wearable AI Market It is expected to reach $ 42.4 billion by 2023, up from $ 29.5 billion by 2018 at a CAGR of 29.75.
A research team led by Dr. Jung-de Quun and Yong-hoon Kim at KIMS said, “The use of high-reliability, new-concept memory-based AI semiconductor is expected to significantly reduce circuit density and driving power. And wearable AI systems apply.
This study was published in the October 1 issue as a cover sheet Advanced Practical Materials.
Development of dendritic-network-functional artificial neurofiber transistors
Jae Hyeon Nam et al, Low Power MoS 2 / Nb 2 O 5 Memtransistor Device Highly reliable heterospinal plastic (advertising function. Mater. 40/2021), Advanced Practical Materials (2021) DOI: 10.1002 / adfm.202170294
Presented by the National Science and Technology Research Council
QuoteMaterial development accelerates AI technology realization (2021, November 8) November 22, 2021 Retrieved from https://phys.org/news/2021-11-materials-advancement-ai-technology.html
This document is subject to copyright. No part of this site may be reproduced without our written permission. Content is for informational purposes only.