Talk by Prof. Jonathan Kadmon
We hereby announce the next talk in the 'Computational and Systems Neuroscience Virtual Seminar' in short: 'CSN Virtual Seminar'
Order from chaos: computational and learning in cortical circuits
Neural activity in the living brain is noisy and unreliable. Nevertheless, the brain can produce accurate and reliable behavior. In this talk, I will suggest that chaotic activity is favorable for learning dynamic patterns and show how chaotic cortical circuits can learn and output reliable signals. Using dynamic mean-field theory, I will show how a neural network can suppress chaos in a low-dimensional subspace of its activity. In addition, I will present a plausible learning mechanism for low-dimensional dynamics and argue that the cerebellum—an evolutionary-conserved subcortical region—plays a crucial role in cortical learning.
Prof. Jonathan Kadmon
The Hebrew University of Jerusalem