I have a few abstract and high dimensional point clouds in the form of distance matrices. I want to do unsupervised learning on this dataset.

The problem is, I am not using one distance matrix, but treating each distance matrix as an object, and trying to differentiate between those objects. On one distance matrix it's very simple: we can do k-mean, hierarchical clustering, etc.

An exemplary problem in 3d would be to differentiate a group of 3d point clouds of cats from dogs. In 2d would be to differentiate the pictures from cats and dogs. But what are the state-of-the-art methods to apply it on distance matrices?

Many thanks!

Although that was a general question to which I'm looking for an answer, here are more details on how I obtained the distance matrices: I am actually looking into a bunch of weighted networks. The entries in the distance matrices are approximately indicating the distances from one node to another in those networks. I used Dijkstra algorithm for calculating the distances between nodes.

  • $\begingroup$ Can you tell us in which setting this dataset arises? $\endgroup$ – kjetil b halvorsen May 18 '15 at 15:08
  • $\begingroup$ From some weighted graphs (networks). I mapped the nodes of weighted networks to a space of metric $\endgroup$ – user3572889 May 18 '15 at 16:25
  • $\begingroup$ Can you add that information to the original question, as an edit? preferably with more details about how you did it? $\endgroup$ – kjetil b halvorsen May 18 '15 at 16:39
  • $\begingroup$ Done! Please let me know if you have more questions. $\endgroup$ – user3572889 May 18 '15 at 23:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.