Information technology abroad: The prototype of a self-learning photon computer is created

Spanish physicists have created a working prototype of a photonic computer, which is able to educate itself and in the future is able to process information at a rate beyond the reach of classical silicon electronics, according to an article, published in the Nature Communications journal.

A group of physicists led by Daniel Brunner at the University of the Balearic Islands in Palma de Mallorca (Spain) took a major step towards the creation of photonic computer, having created a working prototype of such a computing device.

A photon computer of Brunner and his colleagues is constructed on the principle of so-called computational tanks. Such computers are similar in structure to the natural or artificial neural networks, and consist of single-site synapses. Each “neuron” is connected to the adjacent nodes at random and can remember its previous state.

Unlike traditional neural networks, “Computers tank” can solve various problems and can be easily reprogrammed without disturbing the physical structure of the system. By manipulating the properties of synapses, this machine can be adapted to solve complex computing problems, including speech recognition, or pictures.

In addition, the computer model of the photon has another advantage over conventional computers – the low power consumption. As noted by Brunner and his colleagues, their prototype uses to recognize a word 200 times less energy.