Dimensionality Reduction in Action
Saturday, April 18, 2026 · 9:00 AM – 10:30 AM · Classroom 205
As humans, we naturally perceive low-dimensional data: one-dimensional sound waves, two-dimensional images and three-dimensional spatial objects. As data scientists, however, we often work with high-dimensional datasets, whose dimensions far exceed what we can easily imagine. In these spaces, intuition breaks down: distances behave strangely, and the distance to the nearest neighbor becomes almost equal to the distance to the farthest point. This is known as the curse of dimensionality, and it makes tasks like clustering or pattern detection extremely challenging. One solution is to reduce dimensionality while preserving the important structure of the data.
In this tutorial, we will:
1) Apply dimensionality reduction techniques that are especially good for:
• interpretation and preprocessing: PCA, Factor Analysis, ICA • visualization: t-SNE, UMAP • scalability with theoretical guarantees: Random Projections (plus Johnson-Lindenstrauss condition) • complex, high-dimensional data: Autoencoders and Variational Autoencoders (VAEs)
2) Use lower-dimensional representations of datasets to:
• visualize • cluster • identify statistically independent components
This tutorial enables participants to apply, compare and interpret different dimensionality reduction methods on real high-dimensional datasets.
About the Speaker
Evzenie Coupkova
Purdue Math PhD
Evzenie Coupkova holds a PhD in Mathematics from Purdue University and specializes in statistical learning theory, machine learning, and high-dimensional data analysis. Her research explores the dimensionality reduction, proportion of high-accuracy models and generalizability in connection with randomness and dataset labels. She has collaborated with industry partners and presented her work at conferences including IPAM-UCLA and SIAM MDS, blending mathematical rigor with practical data science applications.