Paul van Gerven
20 October

Can you make a brain-like disordered electronic network perform a specific task? Yes, you can. It just needs a little help from a deep-learning network, researchers at the University of Twente (UT) have demonstrated. They recently published their results in Nature Nanotechnology.

Credit: University of Twente

Previously, the UT researchers had shown that an electronic circuit is capable of recognizing basic patterns, even if it isn’t made up of conventional building blocks like logic gates or amplifiers that make it work in a predictable way. This works by steering the output signal in the right direction by changing the voltages on the control electrodes – an iterative process called artificial evolution.

Artificial evolution is quite time consuming, however. This is why the researchers of UT’s Center for Brain-Inspired Nano Systems (Brains) tried speeding it up by training their disordered network with deep-learning neural networks. These are made for learning, but as it turns out, they can be used for teaching as well: once the teaching process has finished, the brain-inspired network can function by itself.