NVIDIA / NVIDIA/cuvs

[FEA] Balanced k-means to expose a max points per cluster

Open
#1,714 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

feature request
Dominant language
Cuda
Stars
854
Forks
236
Avg merge
3d 3h
Merged PRs (30d)
62

Description

We have users asking us to expose a hyper-parameter to hard bound the maximum points per cluster. The idea here is that we want to be able to cap the max cluster to force a more even distribution of points across clusters.

Currently, the balanced k-means seems to do a fairly Gaussian spread of points across clusters, forming a nealry perfect bell curve with the average tending towards the center. We would like to make the distribution more uniform, even at the potential cost of performance.

cc @singhmanas1 for any more details you might have gathered.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The issue names no files, tests, or entry points. Start by locating the balanced k-means implementation and its public API, then determine where a maximum-points-per-cluster parameter belongs and how cluster-size limits should be validated. Done means the parameter is exposed, produces the requested distribution, and has coverage for the new behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
machine-learning
Domain
machine-learning, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.