[mlpack-git] [mlpack] GMM::Classify() is slow (#215)

[Ryan Curtin]

Okay, so this took more than a couple days.  But here it is...
http://www.ratml.org/pub/pdf/2015single.pdf

The algorithm detailed there is general, and it should not be (too) difficult to use the existing mlpack framework for dual-tree and single-tree algorithms to figure out how to adapt the `BaseCase()` and `Score()` function given there.  An example of another single-tree algorithm in mlpack is the `NeighborSearchRules` class, which implements `BaseCase()` and `Score()`; the version of `Score()` for the single-tree algorithm is `Score(size_t, TreeType&)`.

For this GMM algorithm, the first parameter (`size_t`) isn't really needed so it could be ignored during implementation.

Anyway, if there are any errors or difficulties in the document, let me know and I'll happily revise it...

---
Reply to this email directly or view it on GitHub:
https://github.com/mlpack/mlpack/issues/215#issuecomment-106095571
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20150527/26179565/attachment.html>