[mlpack-git] [mlpack/mlpack] Implementation of Multiprobe LSH (#691)

[Ryan Curtin]

I think that the offsets have a maximum, right?  So every point from, e.g., the lower left cluster will fall not in the lower-leftmost bin, but instead in (I think) the 4 (or 9?) bins closest to the center of the lower-leftmost bin.  (Or something like that.)  If that's the case, we could then put the query point outside of the furthest possible bin, and then set a higher number of probes to ensure that at least one of the 4 (or 9) bins are checked.

My thought there isn't complete, but I hope it is coherent enough to make sense.  Let me know if I can explain further.

The rest of the code itself looks good, so when you decide what to do with `std::vector<bool>` or the previous implementation and figure out this test, I think it's good to go---unless there is something else I overlooked.

I'll go through and point out the style, but it will be a lot of comments, so remember, you asked for it. :)

---
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/691#issuecomment-228853434
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack-git/attachments/20160627/e0ff27ca/attachment.html>