Paul Tunison
Paul Tunison
Valid. The only difference that I know of is that it provides another mode that uses histogram intersection kernel. @z-harry-sun provided me this particular version so he would know best...
One of them (this libsvm is effectively required for the currently only implementation of the RelevancyIndex interface required for IQR).
I believe that would be optimal and can already be done. Its just the case that not all plugins provided within SMQTK would be usable without other optional dependencies, some...
The ideal state however is that in the raw form of SMQTK, there be at least one implementation for every algorithm/representation interface, which is currently not the case.
Postponing this feature due to VP-Trees not performing as expected.
Given the recent overhaul of the LSH algorithm modularity, this would now be implemented as a new `HashIndex` implementation
Hi @wphicks! Thanks for your interest. Incoming wall of text. 1) I don't have a specific implementation in mind with regards to back an algorithm wrapper. I have a local...
I should be able to put my tree implementations on a branch sometime today for you to pick apart.
Other implementations I've learned from and could be compared against the one coming on a branch in a few: * http://www.logarithmic.net/pfh/blog/01164790008 * I've been initially comparing against this one. *...
VpTree utilities now on branch dev/vp-tree-utils