Stanford UFLDL教程 MATLAB Modules

MATLAB Modules

MATLAB Modules

Sparse autoencoder |sparseae_exercise.zip

• checkNumericalGradient.m - Makes sure that computeNumericalGradient is implmented correctly
• computeNumericalGradient.m - Computes numerical gradient of a function (to be filled in)
• display_network.m - Visualizes images or filters for autoencoders as a grid
• initializeParameters.m - Initializes parameters for sparse autoencoder randomly
• sampleIMAGES.m - Samples 8x8 patches from an image matrix (to be filled in)
• sparseAutoencoderCost.m - Calculates cost and gradient of cost function of sparse autoencoder
• train.m - Framework for training and testing sparse autoencoder

Using the MNIST Dataset |mnistHelper.zip

• loadMNISTImages.m - Returns a matrix containing raw MNIST images
• loadMNISTLabels.m - Returns a matrix containing MNIST labels

PCA and Whitening |pca_exercise.zip

• display_network.m - Visualizes images or filters for autoencoders as a grid
• pca_gen.m - Framework for whitening exercise
• sampleIMAGESRAW.m - Returns 8x8 raw unwhitened patches

Softmax Regression |softmax_exercise.zip

• checkNumericalGradient.m - Makes sure that computeNumericalGradient is implmented correctly
• display_network.m - Visualizes images or filters for autoencoders as a grid
• loadMNISTImages.m - Returns a matrix containing raw MNIST images
• loadMNISTLabels.m - Returns a matrix containing MNIST labels
• softmaxCost.m - Computes cost and gradient of cost function of softmax
• softmaxTrain.m - Trains a softmax model with the given parameters
• train.m - Framework for this exercise

from: http://ufldl.stanford.edu/wiki/index.php/MATLAB_Modules