What is UMAP?

UMAP (Uniform Manifold Approximation and Projection) is a dimension reduction technique that can be used for visualization similarly to t-SNE, but can also be used for general non-linear dimension reduction. UMAP is constructed from a fuzzy simplicial representation of the data. It is often used to reduce the dimensionality of large datasets for visualization purposes, enabling easier exploration and understanding of complex data structures, and can be used to improve the performance of machine learning algorithms.