In Pending

We could be getting closer to computers and other devices known as neuromorphic, i.e. based on the reproduction of the connections between the nerve cells of our brains, according to a study published in the journal Nature Communication by engineers from the University of Aquisgrana (Germany) and the Politecnico di Torino. The researchers, in fact, demonstrated how, thanks to a special nanomaterial and the proper conditions, it is possible to recreate a circuit that, in contrast with that of normal computers, works in a way that is similar to synapses (i.e. the points where neurons communicate with one another).

The material in question is zinc oxide in nanowire form, namely structures with a diameter the size of one ten-thousandth of a millimetre, called memristors (as opposed to transistors, which work in a different way) fixed to two nano-electrodes, one made of platinum and one of silver, which the current is passed through. Due to the physicochemical properties of the memristors, a flow of ions that behaves in a way that is similar, in certain aspects, to the calcium ions released by nerve endings, is created: hence the name neuromorphic computers, meaning similar to our nervous system.

From a practical point of view, this means that due to the memristors it is possible to make the device work, not in a consequential and parallel way, as occurs in normal computers, but rather in a “dual” mode, processing data and storing it at the same time (impossible for classic computers to do, which must instead use a separate memory that inevitably slows them down). Furthermore, if an element loses its function, the neuromorphic system is not affected and, in fact, compensates whatever is missing (like the neural networks of the human brain). In addition to this, each active part can be active at the same time as others, contributing to the creation of a series of very complex information, compared to that of traditional computers, implying that the structure can learn and improve each individual time (neuromorphic computers will enable the capacity of artificial intelligence systems to be significantly expanded).

For now, the materials made by the researchers from the University of Aquisgrana and the Politecnico di Torino cannot be inserted in a true and proper chip inspired by the brain, due to technical problems that have not yet been overcome. However, new research, already planned in the future, will make it possible for the researchers to get closer and closer to achieving this result, by working alongside the studies of many other laboratories worldwide that are trying to produce neuromorphic computers.

artificial intelligenceFoto edgar morin