Analysis and algorithms

Pattern-stimulation and large-scale activity imaging both require methods for synthetically generating and analyzing multiple spike trains, taking into account features like their correlation structure.  We have introduced a strategy based on correlation distortions in the Linear-Nonlinear-Poisson (LNP) model for generating multiple spike trains with an exactly controlled correlation structure, and for identifying neural encoding models and directed information flow in networks of neurons.

Selected publications:

%d bloggers like this: