Open kyleabeauchamp opened 11 years ago
This is a new clustering algorithm I'm working on. I'll share the manuscript with you soon.
Is the key idea to use MDS to convert metrics into features? If so, then we could re-formulate this in terms of the feature vectorizer and combine RMSD with other features.
Not really. MDS is used inside the k-means update step, but this doesn't really permit RMSD to be combined w/ other features. You could do weighted RMSD (with a weight-set over the atoms), but that's already been an option.
You're github stalking me!