GitHub - dgboy2000/sparseAutoencoder

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 52 Commits
minFunc		minFunc
IMAGES.log		IMAGES.log
IMAGES.mat		IMAGES.mat
README		README
checkNumericalGradient.m		checkNumericalGradient.m
computeNumericalGradient.m		computeNumericalGradient.m
display_network.m		display_network.m
initializeParameters.m		initializeParameters.m
octave-workspace		octave-workspace
sampleIMAGES.m		sampleIMAGES.m
sparseAutoencoderCost.m		sparseAutoencoderCost.m
train.m		train.m
web.stanford.edu_class_cs294a_sparseAutoencoder.pdf		web.stanford.edu_class_cs294a_sparseAutoencoder.pdf
weights.jpg		weights.jpg

Repository files navigation

Solution of the sparse autoencoder problem: http://nlp.stanford.edu/~socherr/sparseAutoencoder_2011new.pdf

Note that the formula for the backpropagation update step of the sparsity term in the objective function is wrong in the pdf. The correct formula is implemented here in code.