[mlpack-git] [mlpack/mlpack] Universal B tree implementation (#746)

[Ryan Curtin]

Looks great, I think this is ready for merge, after the `PerformSplit()` refactoring and `NeedRearrangeDataset` removal.  If you'd like me to do either of those, let me know and I will.  I think I still have yet to find any bugs in any of your code. :+1:

While going over this, I thought that the ordering of dimensions makes a difference for the performance of the tree, because of how the addresses are calculated.  So I wondered if ordering the dimensions in the address according to the variance of each dimension (or simply scaling the variances to 1) might help performance.  But I think that the UB tree is so much slower than the kd tree that it's not really worth investigating.  I thought it was an interesting idea to write down anyway though. :)

-- 
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/pull/746#issuecomment-242506335
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20160825/3719bc9b/attachment.html>