AI another step closer with near human-like memory chip

13 May 20151 Share

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

AI brain, via Shutterstock

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

Robots may soon possess the ability to remember much like a human does after a team of researchers created the world’s first artificial neural circuit capable of image memory.

Led by Dimitri Strukov from the University of California Santa Barbara (UCSB), the team has created a circuit that features 100 different artificial synapses that could one day be expanded to a scale similar to the processing power of the human brain, known commonly as artificial intelligence (AI).

By current scientific understanding, the human brain has somewhat more synapses within its structure, numbering in the region of one quadrillion (1015).

According to the university’s release on the news, Strukov and his team were able to get their circuit to classify the three letters of ‘z’, ‘v’ and ‘n’ by their images and, despite significant ‘noise’ being put in front of the images of these letters, – similar to a human making out a face among a busy crowd – it was able to correctly identify them.

Memristor circuit

The memristor developed by the UCSB team. Image via Sonia Fernandez

Multiplying the scale ten-fold for AI

The key to its success lies in a special electronic component known as a memristor (memory and resistor), the electrical resistance of which is dependent on the direction electricity flows through the circuit, basing itself on ionic movement that is comparable to how a human’s brain generates neural signals.

It’s now a case of creating many, many multiples of these memristors to recreate something similar to our own brains.

In the meantime, the memristor in its current form potentially has a number of uses in medical imaging, online image searching and navigational systems, which would be reliant on fast, correct image recognition.

Memristor illustration

An artist’s conception of a neural network. Image via Peter Allen

One of the team members, Mirko Prezioso, said of the breakthrough: “The exciting thing is that, unlike more exotic solutions, it is not difficult to imagine this technology integrated into common processing units and giving a serious boost to future computers.

“There are so many potential applications — it definitely gives us a whole new way of thinking.”

AI brain image via Shutterstock

Colm Gorey is a journalist with Siliconrepublic.com

editorial@siliconrepublic.com