Introduction to Dimensionality Reduction: A Hands-On Tutorial on PCA and t-SNE

• Huan-Hsin Tseng (Brookhaven National Laboratory)

**Abstract**: This tutorial will introduce two powerful dimensionality reduction techniques widely used in data science and machine learning. One is Principal Component Analysis (PCA); the other is call t-Distributed Stochastic Neighbor Embedding (t-SNE). Both are essential in the unsupervised learning. PCA reduces data dimensions by a linear projection onto lowerdimensional space vectors. On the other hand, t-SNE is a non-linear technique that excels at visualizing high-dimensional data by preserving local structure in 2D or 3D spaces. We will demonstrate their applications using datasets like MNIST and the Olivetti faces, providing practical insights into their usage and interpretation. **Speaker Biography**: Huan Hsin Tseng is a research scientist at the Computational Science Initiative of the Brookhaven National Lab. He received a Ph.D. and a B.S. degree in Physics and Mathematics from National Tsing Hua University. His background was in High Energy Physics, General Relativity, and Quantum Fields in Spacetime. His current research area is mainly Deep Learning and Quantum Machine Learning, particularly on deep learning speech analysis and deep speech background denoise. His interests include theoretical physics, mathematics, and their intersections.