A way of describing data as a blend of several overlapping groups, where each point belongs partly to each rather than being assigned to one. It suits situations where the boundaries are genuinely fuzzy, such as customer types who do not fall neatly into categories.
Technical view
Models the density as a weighted sum of Gaussians, fitted by expectation maximisation, giving soft cluster assignments and a full generative model of the data. K-means is a limiting special case with equal spherical components and hard assignment. Component count is chosen by information criteria or cross-validation; fitting is sensitive to initialisation and can converge to degenerate solutions.
Related
Part of AI Terminology Index, section: Classical Machine Learning.