Implementation of https://arxiv.org/pdf/1404.3606.pdf in Tensorflow
download mnist, cifar10, etc... and make sure the name matches the names in the code. Run the appropriate convert scripts to converte the dataset into TFRecords. For example:
# with mnist folder containing the 4 mnist files
./mnist_to_record.py
The run PCA net. Again, make sure you set the line that picks the dataset. It looks like load('mnist')
.
./pcanet.py --temp