Ruben Coen-Cagli will discuss a string of publications that demonstrate the use of deep generative networks in closed-loop neurophysiology experiments to drive neural populations towards pre-determined activity states. The method provides new insights into neural coding of visual inputs, and could be generalized to control and explore internal representations of any system whose inputs can be captured by a generative model. The discussion will be based on the following papers:
Bashivan et al. (2019)
Ponce et al. (2019) Neuron 177:999
Walkler et al. (preprint)
It is recommended that you check out those reference(s) in advance. Everyone is welcome to join us and interact!