lmcinnes / lmcinnes/umap

transform() speed for online outlier detection

Open
#219 6 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
8.3k
Forks
871
Avg merge
1d 13h
Merged PRs (30d)
5

Description

Hi there!
I am having very good performances with UMAP on my dataset for outlier detection (Isolation forest after UMAP projection).
However, when I am trying to use only `transform()` on 1 sample after a `fit()` on 500 samples, it takes longer on the single sample setting than on the full data (500).

I saw another issue mentioning the fact that UMAP will just take as much time on the full dataset than one sample, but then it shouldn't be longer on 5000 samples, right? Well, the `transform()` on 5000 samples is in fact much slower than 1 or 500 samples, so I do not fully understand how things work between fit and transform for UMAP.

Could anybody help me on this? I am trying to do some online outlier detection.

Best,
Timon

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

Start by reproducing the reported transform timings after fitting on 500 samples, comparing inputs of 1, 500, and 5000 samples. Use the timing results to trace the relevant UMAP transform path; done means explaining or correcting the unexpected scaling behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.