README

-> This is a solution to the Stacked Autoencoder exercise in the Stanford UFLDL Tutorial(http://ufldl.stanford.edu/wiki/index.php/Exercise:_Implement_deep_networks_for_digit_classification)
-> The code has been written in Python using Scipy and Numpy
-> The code is bound by The MIT License (MIT)

Running the code:

-> Download the gunzip data files and the code file 'stackedAutoencoder.py'
-> Put them in the same folder, extract the gunzips and run the program by typing in 'python stackedAutoencoder.py' in the command line
-> You should get two text outputs as follows
-> The first one should say 'Accuracy after greedy training : 0.87', which signifies an accuracy of 87%
-> The second one should say 'Accuracy after finetuning : 0.97', which signifies an accuracy of 97%
-> The code takes about 150 minutes to execute on an i3 processor