A study recently published by CCI Associate Professor Edward Kim, PhD and researchers from the Los Alamos National Laboratory (LANL) shows that neural networks benefit from periods of downtime, just like humans need a good night's rest.
Titled “Using Sinusoidally-Modulated Noise as a Surrogate for Slow-Wave Sleep to Accomplish Stable Unsupervised Dictionary Learning in a Spike-Based Sparse Coding Model,” the study was published in June in the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, and was also recently featured in Popular Mechanics, Discover Magazine and Fortune.
Led by LANL computer scientist Yijing Watkins, the research team discovered that network simulations become unstable after continuous periods of unsupervised learning, but regain stability once they are exposed to states that are analogous to slow-wave patterns that human brains experience while sleeping.
According to LANL, the group’s next goal is to implement their algorithm on Intel’s Loihi neuromorphic chip.