UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction

About

UMAP (Uniform Manifold Approximation and Projection) is a novel manifold learning technique for dimension reduction. UMAP is constructed from a theoretical framework based in Riemannian geometry and algebraic topology. The result is a practical scalable algorithm that applies to real world data. The UMAP algorithm is competitive with t-SNE for visualization quality, and arguably preserves more of the global structure with superior run time performance. Furthermore, UMAP has no computational restrictions on embedding dimension, making it viable as a general purpose dimension reduction technique for machine learning.

Leland McInnes, John Healy, James Melville · 2018

Related benchmarks

Task	Dataset	Result
Image Classification	Fashion MNIST (test)	Accuracy—	633
Image Classification	MNIST	Accuracy96.6	417
Image Classification	DTD (test)	Accuracy51	316
Image Classification	Oxford Flowers-102 (test)	Top-1 Accuracy55	221
Image Classification	Caltech101 (test)	Accuracy87	204
Clustering	MNIST	NMI0.8375	113
Classification	COIL-20	Accuracy0.921	96
Image Classification	EMNIST	Accuracy74.7	90
Classification	MNIST	Accuracy94.5	89
KNN Classification	MNIST	Accuracy97.4	82