Overview
UMAP is a modern dimensionality reduction technique that is often used as an alternative to t-SNE. It is faster and better at preserving the global structure of the data while still revealing local clusters.
How it Works
It is based on theoretical foundations in Riemannian geometry and algebraic topology. It constructs a high-dimensional graph representation of the data and then optimizes a low-dimensional graph to be as structurally similar as possible.
Use Cases
- Visualizing high-dimensional datasets (e.g., single-cell RNA sequencing).
- Pre-processing data for clustering algorithms.