As energy efficient as the human brain, artificial ion neurons could be used for future electronic memory

As energy efficient as the human brain

Artificial ion neurons could be used for future electronic memories

Science and Technology Daily BEIJING, August 6 (Reporter Zhang Mengran) Getting inspiration from the operation of the human brain to develop electronic products is an area of concern for engineers around the world. This time, the researchers conducted a theoretical analysis that uses ions as nerve cells to carry information to develop artificial neurons. The device uses graphene to make nanoslits, which are filled with a single layer of water to transport ions, with the same transport capacity as neurons. The paper was published in the latest issue of the journal Science.

Arguably, no artificial device is more efficient than the human brain— the human brain can perform many complex tasks and consume as much energy as two bananas a day. Its high efficiency depends on its basic unit, the neuron. Neurons have a membrane with nanopores, called ion channels, that open and close according to the received stimulus, and the resulting stream of ions produces an electric current that is responsible for emitting the action potential (a process of potential change produced on the basis of a resting potential when excitable cells are stimulated), which is a signal that allows neurons to communicate with each other.

Of course, artificial intelligence (AI) can do all of these tasks, but at the cost of energy consumption – AI consumes tens of thousands of times more energy than the human brain.

In view of this, scientists from institutions such as the French National Center for Scientific Research and the Ecole Normale Supérieure de Paris have focused their entire research on designing an electronic system that is as energy efficient as the human brain. For example, carrying information by using ions instead of electrons. Scientists believe that nanofluidics (which studies how fluids behave in channels smaller than 100 nanometers wide) provide many perspectives.

In the new experiment, the research team constructed a prototype of artificial neurons made up of extremely thin graphene slits containing monolayers of water molecules. The researchers demonstrated that under the action of an electric field, ions from this layer of water can aggregate into elongated clusters and produce a property known as the memristor effect: these clusters retain some of the stimuli received in the past, and in comparison with the brain, graphene slits reproduce ion channels, clusters, and ion streams. Using theoretical and digital tools, the researchers demonstrated how these clusters could be assembled to simulate the physical mechanisms of action potential emission, thereby simulating the transmission of information.

The researchers say their goal now is to prove experimentally that such systems can implement simple learning algorithms that could pave the way for future electronic memory.

Editor-in-chief dots

Over the years, scientists have hoped to create electronic systems similar to those of the human brain, but the entry points for their research are different. For example, there have been a lot of studies that have designed artificial neural networks that are larger and larger in scale and have more and more layers by simulating the mechanism of the human brain processing information. In the above study, the scientists focused on making human neurons as energy efficient as human brain neurons. It is believed that these research results from different angles will eventually be integrated, bringing more efficient and powerful artificial intelligence.

Source: Science and Technology Daily