University of Pennsylvania researchers have developed an approach that enables a neural network to run a code like a standard computer, in an effort to speed up certain calculations.
The researchers determined the effects of individual artificial neurons and used that information to assemble a neural network capable of addition and other basic tasks.
To mimic the behavior of the logic gates in computer chips, the researchers connected several of these networks in chains so they could perform more complex operations, then combined the chains into a network that could run a virtual neural network and play the game Pong.
Francesco Martinuzzi at Germany's Leipzig University said, "There will definitely be specific applications where these computers are going to be outperforming standard computers. And by far, I believe."
From New Scientist
View Full Article - May Require Paid Subscription
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA
No entries found