Manifold Learning#
about#
Manifold Learning is a class of unsupervised learning techniques that seeks to uncover low-dimensional structures hidden in high-dimensional data. It’s particularly useful when traditional linear methods, like PCA, are inadequate because the data structure is inherently nonlinear.
Key Concepts#
High-dimensional space: The original space where the data resides.
Manifold: A lower-dimensional space embedded within the high-dimensional space. A manifold is often a shape that cannot be described linearly but locally resembles a lower-dimensional linear space.
Nonlinear dimensionality reduction: The process of reducing the dimensionality of data while capturing the nonlinear relationships within it.