Leland McInnes
Leland McInnes
There is definitely some overhead that is going to be bad for the small cases, but largely irrelevant for the large cases. As to actually being slower for very small...
I think you will want to look into ParametricUMAP (included in this package) by Tim Sainburg (@timsainb). The documentation has some basic mentions of it, but you way also want...
There is a strange issue in the spectral initialization when multiple connected components of the graph occur. I did sped some time on this and may have tracked down the...
It's been a while since I looked at that code, but I seem to think it should be as parallel as the regular path. It should be fixed by the...
Going through the code, it *should* be parallel, or, at least as parallel as the regular UMAP. And yes, the random state should fix the results. If you can confirm...
In principle you can build a custom metric (say, similar to jaccard) over the multi-labels, and use that are the ``target_metric`` parameters for a surpervised UMAP. You could also try...
I'll see what I can manage.
Thanks, I suspect it is indeed related to #798. It must be an odd interaction. I'll see if I can track it down.
Easy enough in the end -- let me know if that commit fixes the issue for you. It certainly worked locally in my own testing.
Hi @edwindj, thanks for the link. I had been thinking about approaches to do something similar to this for UMAP specifically and had some plans, but this looks like they've...