Deep Belief Nets
================

A C++ implementation of the Deep Belief Net training algorithm
described in [1], with more detail available in [2]. There's also a
Google tech talk covering the application [3]. There's a second tech
talk [4] which I haven't had a chance to watch yet.

You'll need to grab the MNIST digit data from [5] and gunzip the files
in a directory called 'data' in the project root.

References
----------

[1] Hinton, G. E., Osindero, S. and Teh, Y., 
A fast learning algorithm for deep belief nets.
Neural Computation 18 (2006), pp 1527-1554.
(http://www.cs.toronto.edu/~hinton/absps/ncfast.pdf)

[2] Bengio, Y.,
Learning deep architectures for AI
Now Publishers (2009)
(http://www.iro.umontreal.ca/~bengioy/papers/ftml_book.pdf)

[3] http://www.youtube.com/watch?v=AyzOUbkUf3M

[4] http://www.youtube.com/watch?v=VdIURAu1-aU

[5] http://yann.lecun.com/exdb/mnist/