Reviews: Model-based Bayesian inference of neural activity and connectivity from all-optical interrogation of a neural circuit

This papers proposes an inference method of (biological) neural connectivity from fluorescence (calcium) traces. The model includes the spiking model (GLM low-rank factor) with an external input (optical stimulation) and a fluorescence model. The inference methods is based on variational Bayes, where the approximate posterior is modeled using a neural network. Novelty and originality: The methods in this paper are adequately novel and original, nicely combining various elements from previous work. Technical issues: My main problem with this paper is that I can't really be sure that the proposed method is actually working well. It is very good that the authors tested their method on real data, but since there is no ground truth, I it is hard to estimate the quality of the inferred weights (see footnote (1) below).