Artificial Brains May Need Sleep To Function
Garrett Kenyon and his team of neural network researchers at Los Alamos National Laboratory and the New Mexico Consortium recently were featured in an Inside Science article Why Artificial Brains Need Sleep.
What are artificial neural networks used for these days? They are used in everything from identifying a pedestrian crossing the street to diagnosing cancer. They work by discovering patterns and changes in a pattern and are constantly processing info, adapting, learning, and coming up with solutions, similar to a human brain. And it turns out artificial brains, or artificial neural networks, may need deep sleep in order to keep stable much like a real brain does!
Yijing Watkins and Garrett Kenyon at Los Alamos National Laboratory are both doing work programming neuromorphic processors to learn how to reconstruct images and video data. According to Kenyon all of their attempts to have the processors learn eventually become unstable.
However, conventional techniques used to rapidly train standard artificial neural networks do not work on spiking neural networks. “We are still learning how to train spiking neural networks to perform useful tasks,” said study lead author Yijing Watkins, a computer scientist at Los Alamos National Laboratory in New Mexico.
After trying have the simulation going into a deep sleep with oscillating slow waves of white noise they found that this restored the simulation back to stability, much like what happens to a real brain after going into deep sleep.
All animals need slow-wave sleep, Kenyon said. “Even aquatic mammals — whales, dolphins and so on — require periods of slow-wave sleep, despite the obvious evolutionary pressure to find some alternative. Instead, dolphins and whales sleep with half their brain at a time.
“Why is slow-wave sleep so indispensable?” Kenyon said. “Our results make the surprising prediction that slow-wave sleep may be essential for any spiking neural network, or indeed any organism with a nervous system, to be able to learn from its environment.”
The scientists doing this research presented their findings virtually June 14 at the Women in Computer Vision Workshop.