ZC or CZ? about deep-subspace-clustering-networks HOT 2 CLOSED

It depends on how data is organized. If the data is stored column-wise, we use ZC; if row-wise, we use CZ. For python, the default matrix data format is row-wise, so we use CZ.

from deep-subspace-clustering-networks.

@panji1990 I do not know if I am right, for the summary architecture of the neural network, the input to the algorithm is a matrix in the form [feature, sample], and in the first layer the neuron nodes number is #sample, and then the second and later layer the neuron nodes number is ${some ratio} * #sample if ${some ratio} exist, the neuron nodes number in the last layer of the encoder goes back to #sample and then comes the self-expression layer, which also has #sample neurons. As a result, the parameters between the last encoder layer to the self-expression layer are #sample * #sample due to the fully connected way?

from deep-subspace-clustering-networks.