Back to Course
Machine Learning: The Engineering Approach
Module 12 of 13
12. Dimensionality Reduction (t-SNE/UMAP)
1. The Curse of Dimensionality
Points in high-dimensional space are far apart. Distances lose meaning.
2. Manifold Learning
t-SNE and UMAP try to preserve the local structure of data. They allow you to visualize 784 dimensions (MNIST) in 2D.