Scientists at Tyumen State University (Tyumen State University) have developed a biomorphic neuroprocessor based on a new component of nanoelectronics – a combined memristor-diode crossbar. Information on development is published in the Microelectronic journal.
According to the press service of the university, the existing neuroprocessors are designed to accelerate hardware calculations in artificial neural networks on simple neurons and provide computer vision, machine learning, and other systems with weak artificial intelligence (AI). Information processing and decision making in such processors occur by choosing the most plausible decision based on previously established associations.
Unlike existing neuroprocessors with simple neurons, the biomorphic neuroprocessor that we have presented makes it possible to make decisions not only on the basis of pre-established associations but also on the basis of new associations formed during signal processing in dynamically changing conditions.
Sergey Udovichenko, Professor, Department of Applied and Technical Physics, Tyumen State University
The new neuroprocessor, according to the creators, is able to generate new associations, or new knowledge, using a biologically similar mechanism. This means that AI is becoming stronger: it will already be able to comprehend new information.
The new device will help increase the speed and energy efficiency of calculations compared to today’s PCs and supercomputers.
At the moment, the research team continues to test the new system. In 2025, it is planned to launch small-scale production.