I read this fascinating article recently: Researchers Made an AI Whose Performance Increases if They Let It Sleep And Dream
"the team worked out a way to mathematically implement human sleep patterns - rapid-eye movement sleep and slow-wave sleep, the former of which is thought to remove unnecessary memories, and the latter of which is thought to consolidate important ones.
“So this is what the [artificial neural network’s] ‘sleep’ state does too, cycling through and unlearning unnecessary information, and then consolidating what’s left, the important stuff.”
The result was that the neural network had a much higher storage capacity.
You can read the entire (short) article here. I was particularly interested in this because I had just been reading about those mechanisms for REM and NREM sleep in Matthew Walker’s excellent Why We Sleep (reviewed by Andrew Holecek here).