Argonne's insect brain-inspired AI chip design to be unveiled in August webinar

A team at Argonne National Lab used physics, computer science and materials science to design and test a chip that can perform on less than a watt of power. Left to right: Anil Mane, Prasanna Balaprakash, Angel Ynaguas-Gil, Sandeep Madireddy and Jeff Elam. (Argonne)

Argonne National Lab is teasing an upcoming webinar that promises to unveil a new neuromorphic computing chip design that can drop power by 10 times or more without sacrificing accuracy. Insect brains are the inspiration for the work.

The webinar, being held August 12 for only 15 minutes, will feature Angel Yanguas-Gil, principal materials scientist, discussing his team’s discoveries to allow AI to perform on a chip design using less than 1 watt. Registration is required.

The lab promoted the event with a vague reference that the new design relies on “new materials, designs and hardware,” with few details. “Can we help AI adapt to new and extreme environments—in space, inside nuclear power plants, or anywhere temperatures exceed 500 degrees Fahrenheit?” the lab said in its promotion. "Yanguas-Gil will show how a newly designed neuromorphic computer chip can reduce power by one order of magnitude without sacrificing accuracy. To do this, his team drew on new materials, designs, and hardware."

Despite the suggestion, an Argonne official clarified that the webinar will not unveil a physical chip. "The lab doesn't and won't have expertise to make the chips," the spokesman explained. "The scientist and others are exploring potential designs and required materials, which is the webinar's topic."  It isn't clear how close the design is to the point of fabrication. *

Yangues-Gil and other researchers published a scientific paper in 2019 that offers insights at where their AI work comes from. Entitled “The Insect Brain as a model system for low power electronics and edge processing applications,” the paper notes, “insects carry out multisensory integration and are able to change the way they process information, learn and adapt to changes in their environment with a very limited power budget.”

The team implemented algorithms in three ways:  in a neuromorphic chip, a customized FPGA and a hybrid analog/digital implementation based on cross-bar arrays

Also in 2019, Argonne separately described research into low-power neuromorphic computing, using the abilities of ants, bees and fruit flies, among others. Researchers simulated a neuromorphic chip based on two discoveries. One discovery was that dynamic filters and weights change the strength of neural connections depending on what the system finds important in real time. The other was that tungsten-aluminum oxide created by Argonne chemists Jeff Elam and Anil Mane would allow the chip to operate at below 1 watt, compared to conventional chips consuming 100 watts or more.

* This story was corrected to say Argonne researchers are only showing off a chip design and not a physical chip as described in an earlier version. 

RELATED: Chip designers must rethink low power for AI