A research group at the University of Southern California has unveiled a novel type of artificial neuron capable of mimicking the electrochemical behavior of biological brain cells. Their findings have been published in the prestigious journal Nature Electronics. These artificial neurons process information through chemical reactions, essentially 'emulating' the brain's mechanisms via physical processes. The team utilized 'diffusive memristors' to achieve a highly accurate simulation of how the brain transmits information at the physical level. The ion movements within these artificial neurons closely mirror the natural activities of real neurons, allowing for learning capabilities to be implemented at the hardware level.

These artificial neurons boast remarkable energy efficiency and a streamlined structure. In fact, a single diffusive memristor-based artificial neuron occupies a space equivalent to just one transistor. Looking ahead, they hold the potential to significantly reduce the size and energy consumption of future chips, thereby paving the way for the development of highly efficient artificial intelligence systems.