Scientists at MIT have come up with a way of creating an ‘artificial brain’ using memristors, silicon components that mimic how the synapses in our brains function. While memristors have existed for some time, MIT has used alloys of silver and copper with silicon to create this new type, which seems to be better at ‘remembering’ images as compared to traditional memristors.
According to Jeehwan Kim, Associate Professor of Mechanical Engineering at MIT, these new memristors could potentially be used to build ‘real’ neural networks for portable AI applications. “Imagine connecting a neuromorphic device to a camera on your car, and having it recognize lights and objects and make a decision immediately, without having to connect to the internet. We hope to use energy-efficient memristors to do those tasks on-site, in real-time,” he added.
Memristors are considered vital to AI and other fields of technology because of their ability to generate signals that vary depending on the input signal, unlike transistors, which switch between two values – 0 and 1, or on and off. According to some experts, memristors could lead to far more powerful computing hardware, which as of now is based on transistors. Other computing industry titans are also working on this: Intel has the Loihi chip, which uses 130,000 artificial ‘neurons’, and IBM has TrueNorth, which has a million.
As part of their tests, the MIT researchers used a chip, packed with ‘tens of thousands’ of memristors, to reproduce and alter images, which it managed to do so more reliably than existing designs.
According to Prof Kim, advances in memristor tech could lead to breakthroughs in image recognition tasks: “We would like to develop this technology further to have larger-scale arrays to do image recognition tasks. And some day, you might be able to carry around artificial brains to do these kinds of tasks, without connecting to supercomputers, the internet, or the cloud.”