I'm looking for a lower dimensional projection of data such that the k nearest neighbors (in Euclidean space) in high dimensions remain the k nearest neighbors in low dimensions. I found that Isomap takes k-neighborhood into account, but it looks at gedesic distance rather than Eucledian distance, right? What about Multidimensional scaling (MDS)? Or should I use t-SNE or LLE?

  • 2
    $\begingroup$ As far as I know t-SNE doesn't guarantee conserving neighbourhood, since it uses a "spring"-type distance model, i.e. distances can be compressed/extended a lot depending on context. $\endgroup$ – Denwid Mar 29 '18 at 7:02
  • $\begingroup$ Can you give a dummy problem? Can you motivate the problem? How about providing links to references in the "jargon" that was used? Is this a dead question? $\endgroup$ – EngrStudent - Reinstate Monica Oct 15 '18 at 17:17

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.