Comment by cbuq

How is this different from a sparse autoencoder?