Change graph laplacian to unnormalized

quinngroup / fergus-ssl

Python implementation of the Rob Fergus semi-supervised learning algorithm.

1 stars 0 forks source link

Change graph laplacian to unnormalized #6

Closed magsol closed 8 years ago

magsol commented 8 years ago

Re-implement _normalized_graph_laplacian to instead compute the unnormalized version, or L = D - A, where A is the affinity / kernel matrix, and D is the diagonal matrix.

Feel free to rename the method so it makes sense.

mgadgil09 commented 8 years ago

Created a new method unnormalized_graph_laplacian to compute unnormalized Laplacian L = D - W. Also, computing generalized eigenvectors and eigenvalues now.