cs294a_2011-sparseAutoencoders

Solution for Stanford cs294a, sparse autoencoder in MatLab and Python

• This is a solution to the Sparse Autoencoder exercise in the Stanford UFLDL Tutorial(http://ufldl.stanford.edu/wiki/index.php/Exercise:Sparse_Autoencoder)
• The solution is implemented in both MatLab and Python (with numpy, scipy, and Matplotlib)

Running the code:

For Python

• Download the data file 'IMAGES.mat' and the code file 'sparse_autoencoder.py'
• Put them in the same folder, and run the program by typing in 'python sparse_autoencoder.py' in the terminal
• You should get a result similar to the 'weights.first.jpg'
• The code takes about 1 minute to finish on a 2016 computer

For MatLab

• Download the data file 'IMAGES.mat' and the code files '*.m'
• Put them in the same folder, and run the program by typing in 'train' in the MatLab
• You should get a result similar to the 'weights.first.jpg'
• The code takes about 1 minute to finish on a 2016 computer