hub-points-radovanovic-2010-jmlr
IN premise — summaries/2026/08/24/wiki-Curse_of_dimensionality-chunk-3.md
Created 2026-08-24T17:11:09+00:00
Radovanović, Nanopoulos & Ivanović (2010, JMLR) demonstrated that in high-dimensional spaces, a small set of 'hub' points appear disproportionately in the k-NN neighbor lists of many other points, distorting classification, clustering, and retrieval.
Summary
When data has many dimensions, a small number of points end up showing up in the nearest-neighbor lists of a huge fraction of all other points, creating a structural bias that isn't visible in low-dimensional data. Any algorithm that relies on nearest-neighbor logic, whether for classifying, grouping, or retrieving, will be quietly skewed toward those dominant points unless the system explicitly corrects for the imbalance.