DOI: 10.22184/1993-8578.2020.13.7-8.466.475

A biomorphic neuroprocessor implements a hardware impulse neural network for traditional information processing tasks, as well as for reproducing the work of the cortical column of the brain. The key nodes of the hardware neural network are the memory and logic matrices, developed on the basis of a new component of nanoelectronics –
a combined memristor-diode crossbar, which have a high integration of elements and energy efficiency compared to known neuroprocessors and separate matrices. The concept of a biomorphic neuroprocessor, a description of a neural network built on the basis of an original software model of a neuron and adapted to the hardware of the neuroprocessor, as well as nanotechnology for manufacturing a memristor-diode crossbar and the results of a study of its electrophysical properties are presented.


Разработка: студия Green Art