Engineers Put Thousands of Artificial Brain Synapses on a Single Chip
Imagine a future where you can carry around artificial brains that can work without supercomputers or the internet. Doesn't that sound like someone from the 1920s dreaming of mobile phones? Well, MIT engineers have designed a "brain-on-a-chip," smaller than a piece of confetti, that could advance the development of small, portable AI devices.
Their "brain-on-a-chip" is made from tens of thousands of artificial brain synapses, namely "memristors," which are silicon-based components that mimic the information-transmitting synapses in the human brain.
Once their promising new memristor design for neuromorphic devices is built into portable devices, they could do complex computational tasks that only today’s supercomputers can carry out.
The chip was able to 'remember' and 'recreate' images
The researchers used the principles of metallurgy to build the memristors, using alloys of silver and copper, in addition to silicon. This twist enabled them to build a chip that can effectively "remember" images in very high detail.
The chip was able to remember and recreate a gray-scale image of Captain America's shield and reliably alter an image of MIT’s Killian Court by sharpening and blurring it. This was done repeatedly, and the images the chip "remembered" were much crisper and more detailed than its predecessors.
Recreating large, complex, artificial neural networks
While tests might seem not that important, this could be the cornerstone of advancing the small, portable AI devices. The team's goal is to recreate large, complex artificial neural networks.
Jeehwan Kim, associate professor of mechanical engineering at MIT, says, "So far, artificial synapse networks exist as software. We’re trying to build real neural network hardware for portable artificial intelligence systems.
Here is the future told by Kim, "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."
This design could be a cornerstone
While we have a lot to get there, the team suggests that their design could pave the way for portable artificial brain computers that can perform tasks with minimal power and without any network connection.
"We’re using artificial synapses to do real inference tests," Kim said. "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."
The paper was published in Nature Nanotechnology.