We are searching data for your request:
Upon completion, a link will appear to access the found materials.
We're not quite at the stage where we know what an artificial brain is dreaming about, or if it dreams at all.
However, a new study carried out by Los Alamos National Laboratory scientists points towards androids requiring rest time in order to perform at their best, much like our human brains need sleep.
SEE ALSO: SLEEPING WASHES OUR BRAINS CLEAN OF UNWANTED TOXINS, NEW STUDY FINDS
All brains need sleep
Sleep is incredibly useful, and very much needed, in order for humans and animals to function adequately. So, it was only a matter of time until artificial brains needed similar requirements.
Los Alamos National Laboratory computer scientist, Yijing Watkins, explained the research team's motivation for the study: "We were fascinated by the prospect of training a neuromorphic processor in a manner analogous to how humans and other biological systems learn from their environment during childhood development."
Much like our brains, Watkins and her team observed that neural simulations became unstable after a long period of self-learning without rest. And when the team placed these simulations under sleep-like states, stability was restored.
"It was as though we were giving the neural networks the equivalent of a good night’s rest," said Watkins.
The trickiest part of the research, as per Garrett Kenyon, study coauthor and computer scientist at Los Alamos, was finding a way to stop the neural networks from going unstable. "The issue of how to keep learning systems from becoming unstable really only arises when attempting to utilize biologically realistic, spiking neuromorphic processors or when trying to understand biology itself," he said.
"The vast majority of machine learning, deep learning, and AI researchers never encounter this issue because in the very artificial systems they study they have the luxury of performing global mathematical operations that have the effect of regulating the overall dynamical gain of the system."
The team's last resort to try and keep the networks stable was looking into how to simulate a sleep-like state for the artificial brains. Noise was the answer. Creating a noise similar to the static you hear when tuning a radio station did the trick. The best option as something called Gaussian noise, which involves a large range of frequencies and amplitudes.
As per the research, this type of noise helps stabilize the neural networks and not to hallucinate as it provides some much-needed resting time.