Communications of the ACM
New Hardware Could Curb AI's Energy Appetite
Engineers at Purdue University have created hardware that, they say, can learn skills using a type of artificial intelligence that currently runs on software platforms.
Credit: Qi Wang/Purdue University
Purdue University engineers have developed hardware that learns using artificial intelligence (AI) that currently runs on software platforms.
They said sharing intelligence features between hardware and software would offset the massive power requirements for using AI in more advanced applications.
The team is the first to demonstrate artificial "tree-like" memory in hardware at room temperature.
The hardware is fashioned from a quantum material, to which the researchers added a proton and applied an electric pulse that caused the proton to move. Each new proton position generated a different resistance state, creating an information storage site or memory state—and multiple pulses created a branch comprised of memory states.
Purdue's Shriram Ramanathan said, "This discovery opens up new frontiers for AI that have been largely ignored because implementing this kind of intelligence into electronic hardware didn't exist."
From Purdue University News
View Full Article
Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA
No entries found