Home People Publications Resources Contact

Code Software Datasets Workshops & Tutorials

A joint maximum-entropy model for binary neural population patterns and continuous signals

---

Files

MaxEntFit.zip (11 KB)

Description

This is a collection of python functions to reproduce experiments/figures from the following paper:

@inproceedings{Gerwinn2009a,
  author = "S. Gerwinn and P. Berens and M. Bethge",
  title = "A joint maximum-entropy model for binary neural population patterns and continuous signals",
  year = 2009,
  booktitle = "Advances in Neural Information Processing Systems 22",
  keywords = "maximum entropy, population coding"
}

maxentfit.py:
Contains the algorithm for fitting a maximum entropy model to observed moments. This is done by gradient ascent on the log-likelihood. The moments needed for the gradient are computed by brute force enumeration of all (binary) states.

example.py:
Contains a one-dimensional example and should illustrate how to use the code in maxentfit. It produces figure 1 of the above mentioned publication.

minimize_carl.py:
Conjugate gradient minimizer by Carl Rasmussen. Can probably be replaced by any other gradient based optimizer.

References