Data Visualization and Unsupervised Learning - STAT 841 Winter 2017

Explore advanced unsupervised learning techniques through this comprehensive lecture series delivered by Professor Ali Ghodsi at the University of Waterloo during Winter 2017. Master fundamental and cutting-edge methods for high-dimensional data visualization, dimensionality reduction, and clustering analysis. Begin with Principal Component Analysis (PCA) in its ordinary, dual, and kernel forms, then progress through multidimensional scaling (MDS), Isomap, and Locally Linear Embedding (LLE). Delve into spectral clustering techniques, Laplacian Eigenmaps, and Maximum Variance Unfolding (MVU) before advancing to supervised dimensionality reduction methods. Learn about Non-negative Matrix Factorization (NMF) algorithms, Sum-Product Networks, and neural network architectures including autoencoders. Discover modern embedding techniques like Word2Vec skip-gram models and conclude with advanced topics in clustering using Mixture of Gaussians, t-SNE visualization, and Variational Autoencoders. Gain practical insights into interactive methods for exploring data structure and dependencies while building expertise in both traditional and contemporary approaches to unsupervised machine learning.