I'm trying to implement Normalized Cuts. Input is a image and the algorithm needs view each pixel of image as a node.
Suppose the image size is 100*200, i have to create a numpy matrix of size (100*200,100*200) during the execution of the algorithm.
I'v tried to use sparse matrix (csr_matrix), but it didn't work.
How can i store this huge matrix?
For Normalized Cuts, is there any method which i don't need to store huge matrix and can still figure out the eigenvector of it?