This repo contains basic code to train multi-region data-constrained recurrent neural network (RNN) models and perform current-based decomposition (CURBD).

For details, see: Perich MG, Arit C, Soares S, Young ME, Mosher CP, Minxha J, Carter E, Rutishauser U, Rudebeck PH, Harvey CD, Rajan K. Inferring brain-wide interactions using data-constrained recurrent neural network models. bioRxiv. https://doi.org/10.1101/2020.12.18.423348

We provide both Matlab and Python implementations. Run CURBD_example in the language of choice to see a demo of RNN fitting and CURBD using simulated data.

All software distributed under GNU GPL (see LICENSE.md).