MID / iSTAC Receptive Field Models for LNP Neurons

This notebook demonstrates nonparametric model generation with an information theoretic framework. Specifically using maximally informative dimension (MID) and iSTAC models. Compares model output with a (whitened) spike-triggered average (STA) model. It uses the output of synthetic, Linear-Nonlinear Poisson (LNP) neurons stimulated with simply correlated noise as a data source.

Inspiration/code from:

1. Pillow Lab code (https://github.com/pillowlab/MIDestimator)
2. Sharpee Lab code (http://cnl-t.salk.edu/Code/, https://github.com/sharpee/mid)

TODO: add more inline annotation and consistent labeling

SLH 2017