Roman Treutlein

I have a cpu implementation but I have to go through some company bureaucracy to be allowed to upload it.

@linas I am working on understanding your method. The matrix library seems quite useful but I can't quite figure out how to use it as a GDTV. Specifically, I am...

The more I think about this the more I am agreeing with @linas that having such an API to access the data is clearly superior to having a fixed representation....

@linas Maybe this example will help a bit. Generally we want PLN to use DVs for more robust/flexible inferences instead of the current STVs. So assuming the system has learned...

@bgoertzel In that paper, they store all the samples in a kd-tree which I don't think we want to do. We want to have some Intervals with an associated count...

But it might be possible to build a kd-tree that only contains the count of the elements in each section. This would be similar to my third option with the...

Well, the problem with cover trees is they don't help us with merging of intervals which as I said gets very messy the more dimensions you have. For that, I...

Is anybody working on this? If not I might do it as I could use a FloatValue that's just a single float.

@ngeiswei I am using single values for the distributional values but I can just use a [0] when accessing the value. It just seems less clean then if I knew...

There will be a DistributionalValue Type and the way it is right now also a ConditionalDistributionalValue Type. That's they way it currently works in my dev branch.