.. _example_plot_mnist_generator.py:


=======================================================
Generative networks for random MNIST digits
=======================================================

This demo of an MNIST generator is based on the work of
I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair,
A. Courville, Y.Bengio. *Generative Adversarial Networks*, June 2014.

The generators trained as part of the published experiment have been wrapped in
sklearn-theano, and can easily be used to fetch an arbitrary number of plausible
MNIST digits.

Additionally, this example also shows how to make an automatically updating plot
with the 'TkAgg' backend to matplotlib.




.. image:: images/plot_mnist_generator_001.png
    :align: center


**Script output**::

  Iteration 0
  Iteration 20
  Iteration 40
  Iteration 60
  Iteration 80



**Python source code:** :download:`plot_mnist_generator.py <plot_mnist_generator.py>`

.. literalinclude:: plot_mnist_generator.py
    :lines: 18-

**Total running time of the example:**  19.75 seconds
( 0 minutes  19.75 seconds)