NTR organizes and hosts scientific webinars on neural networks and invites speakers from all over the world to present their recent work at the webinars.
On June 22 Maksim Bazhenov, University of California, San Diego, USA, led a technical Zoom webinar on Sleep, Memory and Artificial Intelligence.
About the webinar:
Previously encoded memories can be damaged by the encoding of new memories, especially when they are relevant to new data and therefore can be disrupted by new training – a phenomenon called “catastrophic forgetting.”
Human and animal brains are capable of continual learning, allowing them to learn from past experience and to integrate newly acquired information with previously stored memories. A range of empirical data suggest the important role of sleep in consolidation of recent memories and protection of past knowledge from catastrophic forgetting.
I started by describing the main features and mechanisms of sleep in the brain. Then I described some of our recent results on application sleep related ideas in computational models and artificial intelligence.
Materials:
https://ucsdnews.ucsd.edu/pressrelease/can-sleep-protect-us-from-forgetting-old-memories
https://elifesciences.org/articles/51005
https://arxiv.org/abs/1908.02240
Articles that were mentioned during the webinar:
https://arxiv.org/abs/2106.11908
https://arxiv.org/abs/2106.08446
https://arxiv.org/abs/2104.04132
https://arxiv.org/abs/2103.04780
https://arxiv.org/abs/2011.09526
Moderator and contact: NTR CEO Nick Mikhailovsky: nickm@ntrlab.com.