Researchers get $28 million to create software that can learn like the brain
Source: Nidhi
Neuroscientists from Harvard are teaming up with coders who make computers smarter on a massive project to understand and mimic how the mammalian brain learns.
The five-year effort kicked off this week with $28 million from the Intelligence Advanced Research Projects Agency, a government organization that makes big bets on artificial intelligence research.
Researchers behind the project come from 13 different labs affiliated with Harvard, MIT, New York University, Rockefeller University, Notre Dame, and the University of Chicago.
Eventually, they hope to create software that could be used in any application that demands spotting patterns from large amounts of data, ranging from self-driving cars to medical diagnostics.
“The goal is to move the needle on machine intelligence,” said David Cox, a researcher at Harvard who will lead the project. “We really want some paradigm shift — something that’s much closer to how neurons work.”
Cox acknowledged that the size of the grant is unique in the field: “It’s a ton of cash,” he said.
But it’s also the first time that neuroscientists, who study brains, and computer scientists, who try to build artificial learning systems, have united in such a big way.
The collaboration a timely one. In the last decade, tools that can create images of the brain have matured in parallel with research on learning algorithms. “We’re at a much better position to do this massive science expedition,” Cox said.
Cox said one of his collaborators at Harvard has begun mapping how neurons fire in “a tiny crumb” of the mouse brain. The new project will expand the observations to about the size of a mustard seed.
By the end of the project, Cox hopes the team will have created more powerful tools that other neuroscientists can use to watch and understand the brain.
“I think we will see a transformation in how every corner of our civilization works in the next 50 years. And the thing that will drive this is intelligent machines,” Cox said.
Nidhi Subbaraman writes about science and research. Email her at nidhi.subbaraman@globe.com.
| }
|