New paper: a novel set of methods to efficiently fit spiking circuit models to single-trial spike trains. Congratulations to Josef Ladenbauer for outstanding work! Exciting collaboration with @samamckenzie , Daniel English and Olivier Hagens.
nature.com/articles/s4146…
nature.com/articles/s4146…
We derived maximum likelihood-representations for mechanistic (integrate-and-fire) neuron models which can be solved highly efficiently to infer model parameters from data. In particular, our methods allow us to infer connectivity between neurons even for highly sub-sampled data. 
To validate our approach, we systematically used ground truth data from in-vitro and in-vivo electrophysiological recordings, and benchmarked our methods against several existing model-based and model-free techniques.
