Plastic Prototype Customization,Vacuum Casting Prototype,Reaction Injection Molding Plastic,Rubber Silicone Machining Guangdong Shunde Teamwork Model Technology Co., Ltd , https://www.gdtwmx.com
Institute of Microelectronics and others make progress in the field of brain-like perception computing
[ Instrument Network Instrument R & D ] At present, human society is developing from information to intelligence. The construction of an intelligent society requires information technology systems to be able to obtain external environment information in real time, process it efficiently, and make decisions in a timely manner. It is an important trend to develop a low-power intelligent information processing system of "integration of storage and computing". As a next-generation neuromorphic computing technology, impulsive neural network is an ideal choice for building an energy-efficient data processing center with integrated computing. In order to realize the sensory-storage-computing integrated intelligent processing system of the pulse mechanism, an efficient sensory information interface (biologically called afferent nerve) needs to be constructed to establish a real-time connection between the pulse data processing center and the sensor.
Recently, Liu Ming's team at the Key Laboratory of the Institute of Microelectronics of the Chinese Academy of Sciences and Yang Jianhua's team at the University of Massachusetts, USA, based on the NbOx-Mott memristor developed by Microelectronics. Interface to the pulse data processing center. As a new principle device, NbOx-Mott memristor has simple structure, good scalability, easy to integrate on a large scale, high reliability and rich dynamic transition (insulation-metal transition) characteristics. The interface circuit constructed successfully converted the analog electrical signal perceived by the sensor into a frequency-dependent pulse signal, simplifying the information interface between the sensor and the pulse data processing center. Its frequency response exhibits a quasi-linear relationship with the input voltage under non-noxious stimulus intensity, and tends to stop issuing pulses under noxious input intensity, similar to biological afferent nerves. In order to verify the application of the interface, the researchers further connected the afferent neural circuit with the piezoelectric sensor, and constructed an artificial pulsed mechanical sensing system without any external power supply. The system is able to respond to pressure signals and convert the pressure intensity into the corresponding pulse frequency for use by the pulse data processing center.
In addition, the interface circuit can be easily extended to process signals from various sensors, such as smell, taste, vision, hearing, temperature, magnetic field, and humidity. In addition to the afferent neural circuit (interface circuit), which can be used to construct a variety of sensing systems, it also has the integral emission characteristics of neurons, which is suitable for implementing impulse neural networks. The neural circuit can be further used to construct a complex impulsive neural network to process information, and it is expected to realize an efficient intelligent terminal system that integrates sensing and calculation.
The above research results were published in the journal Nature Communications on January 2 (Nature Communications, DOI: 10.1038 / s41467-019-13827-6). Zhang Xumeng, PhD student of the Institute of Microelectronics, is the first author of the article, and Liu Ming, Liu Qi, a researcher at the Institute of Microelectronics, and co-corresponding author Yang Jianhua, a professor at the University of Massachusetts. The above-mentioned work was partially funded by the National Funding Committee, the Ministry of Science and Technology, and the Chinese Academy of Sciences.