ABSTRACT
The development of models of brain function remains a complex problem given the difficulty of extracting organizational principles from observations on a variety of morphologically and physiologically different neurons. State of the art results in this modeling research have been obtained by a different route, by leveraging the power of deep learning. However, this approach takes advantage of neuroscientific knowledge only to a limited extent. Here, I adopt a perspective that aims at combining experimental data and optimization algorithms by framing this modeling research as an inverse problem. To illustrate the method, I collected calcium imaging data from the first two regions of the olfactory processing pathway of the fruit fly Drosophila melanogaster, the antennal lobe and the calix of the mushroom bodies. In each case, our method gives accurate predictions for large fractions of recorded glomeruli and neurons, and the inferred networks recover known features of the biological counterpart.
Competing Interest Statement
The authors have declared no competing interest.