Can I retrieve points within each cluster from spherical k-means clustering?

[KimMinSang96]

Platform

OS: ubuntu 22.04

Faiss version:

Installed from: source

Faiss compilation options:

Running on:

• CPU
• GPU

Interface:

• C++
• Python

I'm considering using spherical k-means clustering for my project. I'm curious if there's a way to obtain a list, preferably in a 2D format, that contains all points within each cluster. For instance, assuming a k value of 2 and my vectors are (1, 1), (1, 2), (1, 3), (-1, -2), and (-3, -4), I'm looking for a method that returns the points in clusters similar to the following structure:

Centroid point a: [(1, 1), (1, 2), (1, 3)]
Centroid point b: [(-1, -2), (-3, -4)]
Is there a straightforward way to achieve this? Any guidance or suggestions would be greatly appreciated.

[mdouze]

you can use kmeans.assign(points)