[mlpack-git] [mlpack/mlpack] Density Estimation Tree made sparse-enabled (#802)

[Ivan Georgiev]

I can't see here your sparse matrix sorting comment. But I see it in my email :-) So here is the answer:
- I did already implemented SpMat sort - it is in another branch, but I haven't made tests, so I didn't make pull request.
- The code in ExtractSplits is pretty close to that sorting, but with removed parts that we're not interested in.

Generally, sparse matrices will have large number of zeroes, which we're not interested in for our splits - we just need to one whether there is none or any. Also arma's sort is not in-place, which is a performance penalty (I've checked the arma's dense matrix sort - it is still based on STD's one :-). And it is perfectly fine that row_col_iterator() jumps over the zeroes, because (1) we don't specifically need them, and (2) even in a sorted version of a sparse row, they will be all together.

-- 
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/802#issuecomment-254952619
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20161019/baecb4bc/attachment.html>