Manifold Learning: A technique used for dimensionality reduction that is based on the assumption that high-dimensional data lies on or near a low-dimensional manifold.