search

lmcinnes / umap

Uniform Manifold Approximation and Projection
BSD 3-Clause "New" or "Revised" License
7.41k stars 805 forks source link

UMAP alignment #490

Open caseybasichis opened 4 years ago

caseybasichis commented 4 years ago

Given a set of words I have: embedding vectors one-hot CSV with X:words Y:categories

Does UMAP have functions to align the two by token? I could swear UMAP had acquired alignment functions but Google is looking at me funny.

Can the "Y:categories" word vec embeddings have an influence as well?

lmcinnes commented 4 years ago

Unless I am not understanding the problem properly (which is certainly possible) I don't believe UMAP can do this for you.

caseybasichis commented 4 years ago

Thanks for clarifying. I must have misread. Last night was my first hands on with UMAP -- going through the docs tonight.

I was thinking of something along these lines, but I'm not entirely sure it's even appropriate:

https://www.samtalksml.net/aligning-vector-representations/