Engineers develop high-performance, high-reliability synthetic semiconductor device

High performance and reliability synthetic semiconductor device for the next generation of brain-simulating computers

Credit: Korea Institute of Science and Technology (KIST)

A neural computing system technology that mimics the human brain must overcome the limitations of excessive energy consumption, which is a feature of the current von Neumann computing method. A high-performance analog synthetic synapse device capable of expressing the strength of the synapse connection is required to implement a semiconductor device using the brain information transmission method. This method uses the signals transmitted between neurons when the neuron generates a spike signal.

However, with the widely used conventional rheostat memory devices such as Synthetic Clasps, as the filament grows with varying resistance, the electric field increases, causing the feedback phenomenon, resulting in rapid filament growth. Therefore, it is difficult to implement plasticity while maintaining the variance of the analog (gradual) resistance with respect to the type of filament.

The Korea Institute of Science and Technology, led by Dr. YeonJoo Jeong’s team at the Center for Neuroengineering, has solved the limitations of analog synaptic properties, plasticity and information preservation, which are chronic hurdles related to memory, and neural semiconductor devices. Announced the development of an artificial synapse semiconductor device Capable of highly reliable neural computing.

The KIST research team tuned the redox properties of active electrode ions to solve the small synaptic plasticity problems that hinder the performance of current neuron semiconductor devices. In addition to, transition metals Anesthetic and used in the synaptic apparatus, controlling the potential to reduce active electrode ions. The engineers discovered that the high reduction potential of ions is a critical variable in the development of high-performance synthetic crosslinked devices.

High performance and reliability synthetic semiconductor device for the next generation of brain-simulating computers

An example of a visual information processing technology using an artificial interlocking device, which confirms that the error rate is reduced by more than 60% by improving the performance of the device. Credit: Korea Institute of Science and Technology (KIST)

Therefore, the research team introduced a titanium transition metal, which has a high ion-reduction potential, into an existing synthetic crosslinker. This preserves the analog properties of the synapse and the plasticity of the device at the synapse in the biological brain, approximately five times the difference between high and low impedance. Furthermore, they have developed high-performance neuron semiconductors that are about 50 times more efficient.

In addition, due to the high reactivity of the alloy formation shown by the titanium-doped transition metal, the information retention was increased up to 63 times compared to the existing synthetic crosslinker. Moreover, brain functions, including long-term potentiation and long-term depression, can be more accurately simulated.

The team applied an artificial neural network learning pattern using a developed artificial synaptic device and attempted to learn image recognition with artificial intelligence. The mistake percentage It has been reduced by more than 60% compared to the existing artificial interlocking device; In addition, the accuracy of handwriting image pattern recognition (MNIST) increased by more than 69%. The research team confirmed the feasibility of a high-performance neural computing system through this improved synthetic synaptic device.

High performance and reliability synthetic semiconductor device for the next generation of brain-simulating computers

Photographs of (a) solar collector, (b) membrane distillation system. Credit: Korea Institute of Science and Technology (KIST)

Dr. Jeong from KIST said, “This study greatly improved the synaptic range of motion and information preservation, which were the greatest technical barriers to the current synapse simulation. In the developed synthetic synapse device, the analog operating region of the device to express the diverse connectivity of the synapse strengths was maximized, so the strengths will be maximized, Improving the performance of AI computing based on brain simulation.

“In the follow-up research, we will manufacture a neuron semiconductor chip based on the developed artificial synapse device to achieve a high-performance artificial intelligence system, thus enhancing the competitiveness of the domestic system and artificial intelligence. Semiconductors area.”

The search was published in Nature Connections.

The neural memory system simulates neurons and synapses

more information:
Jaehyun Kang et al, Cluster-type analogue memristor by engineering redox dynamics for high-performance neural computing, Nature Connections (2022). DOI: 10.1038 / s41467-022-31804-4

Provided by the National Research Council for Science and Technology

the quote: Engineers develop high-performance, high-reliability synthetic semiconductor device (2022, Sep 20) Retrieved on Sep 20, 2022 from semiconductor.html

This document is subject to copyright. Notwithstanding any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.

Leave a Comment