[mlpack-git] [mlpack/mlpack] R+ and R++ trees implementation (#699)

[lozhnikov]

I didn't remove the R+ tree since it is similar to the R++ tree and I think it is better to add a small layer of abstraction and achieve more possibilities. 

No, I add an intermediate node if I failed in enlarging a node.

If a dataset has a lot of equal points, we should increase maxLeafSize. I forgot about it.

And I forgot that I found a tiny error in the R++ paper. The algorithm that splits an intermediate node should try each lower or upper bound of maximum bounding rectangles (instead of minimum bounding rectangles). And there is no reason to try both lower and upper bounds since they are paired. If we use minimum bounding rectangles we can sometimes fail in finding an acceptable partition. We should notify the authors about that:)

---
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/699#issuecomment-229497241
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20160629/c7694c5c/attachment.html>