High performance interface memristor came out

Tech 2023-06-06 09:15:29 15 Source: Network

Science and Technology Daily, Beijing, June 5 (Reporter Liu Xia) Scientists at Los Alamos National Laboratory in the United States tried to replicate the unparalleled computing power of the human brain, and they created a new interface type memristor device. The research results indicate that the device has good programmability and reliability, and can be used as an artificial synapse for next-generation neural morphology calculations

Science and Technology Daily, Beijing, June 5 (Reporter Liu Xia) Scientists at Los Alamos National Laboratory in the United States tried to replicate the unparalleled computing power of the human brain, and they created a new interface type memristor device. The research results indicate that the device has good programmability and reliability, and can be used as an artificial synapse for next-generation neural morphology calculations. The relevant paper was published in the latest issue of the magazine "Advanced Intelligent Systems".

The research team pointed out that unlike the digital computing of von Neumann architecture, neural morphological computing inspired by biological systems can work like the brain. The synapses of the human brain are connected to 100 billion neurons that send and receive chemical information. Synapses store and process information in the same location, saving time and energy for the human brain to process information. In traditional computing, computation and storage are separated. The advantages of neural morphology computing include low energy consumption, high parallelism, and excellent fault tolerance. After all, the operating power of the human brain is only 20 watts, but its learning efficiency is extremely high, making it very suitable for advanced computing tasks such as learning, recognition, and decision-making.

The existing neural morphology calculation relies on memristor and other emerging equipment. Different from current resistors, memristor aims to copy the structure and function of synapses, and has the ability of programming and memory. The existing memristor includes a filament system, but this system is prone to overheating and lacks stability and reliability.

In view of this, the research team has developed a new structure called "interface type memristor", which is simple, reliable and has high performance. They used artificial neural network simulation to test the computational performance of the interface memristor. The results show that the device has good consistency, programmability and reliability, and the recognition accuracy reaches 94.72%.

In addition, the interface type memristor can be reduced to nanometer size, which cannot be achieved even by the current filament memristor technology. In particular, compared with the transistor based neural morphology chip, the new interface type memristor requires much less power, and the transistor is limited by Moore's Law, so it is difficult to continue to miniaturize. Many advantages make these new interface type memristor devices the fundamental hardware for the next generation of neuromorphic computing.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:win-e@163.com)

Mobile advertising space rental

Tag: High performance interface memristor came out