Brain-inspired recurrent neural network

Abstract

Brain-inspired recurrent neural networks are remarkably efficient at performing tasks, however, their performance is challenged by tasks involving temporal multiplexing, defined as the ability to manipulate information over multiple timescales simultaneously. Here, we will test the hypothesis that recurrent circuits endowed with multiple timescales, comprising both fast and slow neural populations, can robustly perform temporal multiplexing.

  • In collaboration with prof. Luca Mazzucato, University of Oregon, Eugene
Avatar
LIPh
Laboratory of Interdisciplinary Physics

Related