News for hackers. Stuff that matters.

A comment on Understanding UMAP (2019).

Re:Understanding UMAP (2019) by ubutler on Monday May 06, 2024 @07:49PM (#40280779)
Check out PaCMAP, AFAIK it’s the current SoTA in dimensionality reduction: https://github.com/YingfanWang/PaCMAP
[ Parent | on HN ]
Re:Understanding UMAP (2019) by abhgh on Monday May 06, 2024 @10:02PM (#40281532)
Yes, also TriMap does quite well. PacMAP is faster though. This paper [1] (linked at the GitHub repo you have mentioned) goes into a fantastic amount of detail in comparing UMAP, t-sne, PacMAP and TriMAP.

[1] https://jmlr.org/papers/volume22/20-1061/20-1061.pdf

[ Parent | on HN ]
Re:Understanding UMAP (2019) by rsfern on Tuesday May 07, 2024 @08:18AM (#40284595)
Thanks for the link, that is a really nice paper!

I found it really refreshing that they report a bunch of stuff they tried that didn’t work in a way that clarifies the problem and leads to a lot of insight into the strengths and limitations of their final method and the leading alternatives

[ Parent | on HN ]
Re:Understanding UMAP (2019) by abhgh on Tuesday May 07, 2024 @10:44PM (#40293848)
You're welcome! There is a talk too [1], which is how I had learned about the paper.

[1] https://www.youtube.com/watch?v=sD-uDZ8zXkc

[ Parent | on HN ]