README

Description of the repository
-----------------------------

This is the companion code for Piotr Mirowski's tutorial 
on Auto-Encoders, given for the first time at the London Deep Learning
meetup on March 26, 2014.
This tutorial implements Marc'Aurelio Ranzato's Sparse Encoding 
Symmetric Machine (SESM) and tests it on MNIST handwritten digits.

The main reference is the following paper:
M.'A. Ranzato, Y.-L. Boureau, and Y. LeCun.
"Sparse Feature Learning for Deep Belief Networks"
NIPS 2007

The main reference to the MNIST dataset is:
Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner.
"Gradient-based learning applied to document recognition."
Proceedings of the IEEE, 86(11):2278-2324, November 1998

We use the Matlab version of the data, available at Sam Roweis' NYU page:
http://www.cs.nyu.edu/~roweis/data.html


Contents of the repository
--------------------------

* code: contains the code of the SESM and the tutorial.
        Under Matlab/Octave, execute Script_MNIST.
* data: contains the MNIST handwritten digits needed for the script.
* bibliography: PDF versions of several papers related to Auto-Encoders
                and to (deep) dimensionality reduction techniques.