Artificial neural networks are computer systems that are inspired by biological neural networks found in animal brains. They, like biological models, can learn (be trained) by processing examples and forming probability associations, which they can then apply to other tasks.
Humans require 7 to 13 hours of sleep per 24 hours, depending on age. A lot happens during this time: heart rate, breathing, and metabolism ebb and flow; hormone levels adjust; and the body relaxes. In the brain, not so much.
“When we sleep, the brain is very busy, repeating what we have learned during the day,” said Maxim Bazhenov, PhD, professor of medicine and sleep researcher at the University of California, San Diego School of Medicine. “Sleep assists in reorganising memories and presenting them in the most efficient manner.”
Bazhenov and colleagues have previously reported how sleep improves rational memory, or the ability to remember arbitrary or indirect associations between objects, people, or events, and protects against forgetting old memories.
Artificial neural networks use the architecture of the human brain to improve a wide range of technologies and systems, from basic science and medicine to finance and social media. They have achieved superhuman performance in some ways, such as computational speed, but they fall short in one crucial area: when artificial neural networks learn sequentially, new information overwrites previous information, a phenomenon known as catastrophic forgetting.
“In other projects, we use computer models to develop optimal strategies for applying sleep stimulation, such as auditory tones, to improve sleep rhythms and learning.” This may be especially important when memory is suboptimal, such as when memory declines with age or in conditions such as Alzheimer’s disease.
Co-authors include UC San Diego’s Ryan Golden and Jean Erik Delanois, as well as Pavel Sanda of the Czech Academy of Sciences’ Institute of Computer Science.
