umap-adds-repulsive-term-over-tsne
IN premise — summaries/2026/08/24/wiki-Nonlinear_dimensionality_reduction-chunk-2.md
Created 2026-08-24T17:11:20+00:00
UMAP differs from t-SNE by adding a repulsive term for far points, which improves preservation of global structure, whereas t-SNE focuses primarily on local neighborhood preservation
Summary
When you squash high-dimensional data down to a 2D or 3D plot, t-SNE only guarantees that nearby points stay nearby, so the overall shape of the dataset can get distorted. UMAP fixes this by also pushing distant points apart, meaning the big-picture layout of clusters and gaps is more faithfully preserved, which matters when you need to compare regions of the data rather than just inspect one small neighborhood.