[mlpack-git] [mlpack/mlpack] Neighbor Search - Incorrect bound. (#642)

[MarcosPividori]

Hi @rcurtin,
Thanks for your comments. I agree with the proposal.
I have implemented B_aux, in the commit https://github.com/MarcosPividori/mlpack/commit/558a4fd15536eb6d71464a8c519ab709f0b163d0 , and tested it with some datasets. It doesn't show any difference in the number of base cases for Cover Trees.
I modified the benchmarking system to add "BaseCases" as a new metric, and I have uploaded the results to: http://marcospividori.github.io/mlpack-aux/
+ mlpack is the original library using b_2.
+ mlpack-aux includes the modification to use b_aux for all trees. It doesn't differentiate between ball trees and the rest.

As you can see there, ALLKNN in mlpack and mlpack-aux shows the same number of base cases for all datasets.

Maybe this happens because of the characteristics of Cover Trees. I have not an indepth understanding of them...

So, I was wondering if we could simply use the b_aux modification without taking into account the tighter bound...

---
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/642#issuecomment-223000550
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20160601/88561087/attachment.html>