Sign In

Communications of the ACM

ACM TechNews

Neural Networks That Replicate Biological Brains

View as: Print Mobile App Share:

Imperial College London professor of cognitive robotics Murray Shanahan is using graphics processing technology originally developed for the gaming industry to advance artificial intelligence (AI) research. Shanahan says that gamers' demands for increasingly realistic graphics have driven a corresponding increase in the processing capabilities of graphics processing units (GPUs). He says researchers in other fields are capitalizing on the potential these GPUs contain, and manufacturers are now producing specialized units for applications outside the gaming industry.

Shanahan is developing large-scale neural networks that replicate how biological brains provide intelligence. "We're interested in simulating large numbers of neurons," he says. "If we ever wanted to make a robot move around using something that was a simulation of a brain, we would need to make millions of neurons work in real time."

The human brain has about 100 billion neurons, and Shanahan's team is  at around 100,000, which is roughly equivalent to an ant. He says biologically inspired artificial intelligence will be needed to advance traditional AI research. "Using traditional AI techniques, we've pretty much reached a plateau of intelligence and it's hard to see how we're going to be able to move beyond that," he says. "That's the motivation for trying to go back to the way nature has done it and try to replicate the way brains do things."

From The Engineer (United Kingdom)
View Full Article


Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA


No entries found