UMAP author recommends PyNNDescent for much faster kNN-graph construction
leland_mcinnes · x · 2026-11-12
Leland McInnes, author of UMAP, suggests that if you just want kNN-graph construction as fast as possible, or more options in the speed/quality trade space, you should look at PyNNDescent or RINND directly. He adds that PyNNDescent got much faster recently, especially for kNN-graph construction, and the new UMAP tunes its parameters to make better use of it—trading a small amount of accuracy for very significant performance gains.
PyNNDescent is a BSD-2 licensed Python library implementing Nearest Neighbor Descent for k-neighbor-graph construction and approximate nearest neighbor search, with 976 GitHub stars.
More from Research
- 50+ NeurIPS 2025 papers found containing AI hallucinations that slipped past peer review — CSProfKGD · 2026-11-12
- Indie researcher pauses RNN training over environment bugs, shifts to FunSearch-style tasks — cephaloform · 2026-11-12
- Self-supervised visual similarity partially explains word generalization in humans and machines — Patricio M. Vera · 2026-11-12
- P2Echo: one text-prompted model segments multi-view echocardiography with high accuracy and low cost — UBC-V · 2026-11-12
- I Fixed Alamo Drafthouse's Search With Query Logs Alone: +18.4% MRR, No Model Training — CShorten30 · 2026-11-12
- NYU lab open-sources Outerloop: autonomous research agents that open PRs only when benchmarks improve — mengyer · 2026-11-12