Tom (Tomáš) Pecher

Unsupervised Learning: Course Introduction

From my experience of tutoring and discussing AI with non-experts, I've observed that unsupervised learning is the most misunderstood and underappreciated subfield of ML. The goal of unsupervised learning is certainly more abstract than that of the other subfields, which I believe is the reason for the confusion. Put very simply, unsupervised learning tasks itself with understanding, manipulating and replicating data. In that sense, unsupervised learning is an algorithmic approach to data analytics. At first glance, it is not entirely clear what it means to "understand" data. Just know that everything from clustering, anomaly detection, dimensionality reduction and generative AI (GenAI) are all examples of unsupervised learning. Ironically, this makes unsupervised learning one of the richest and most diverse subfields of ML, and so this course covers a wide variety of concepts.

Course Structure:

• Section 1: Unsupervised Learning Paradigms
• Section 2: Clustering Methods
• Section 3: Hierarchical Clustering
• Section 4: Dimensionality Reduction and Manifold Learning
• Section 5: Density Estimation
• Section 6: Variational Autoencoders
• Section 7: Generative Adversarial Networks
• Section 8: Frequent Pattern Discovery
• Section 9: Anomaly and Outlier Detection
• Section 10: Matrix Factorization
• Section 11: Topic Modelling and Text Analysis
• Section 12: Network Analysis
• Section 13: Temporal Pattern Mining
• Section 14: Self-Supervised Learning
• Section 15: Causal Discovery
• Section 16: Online Unsupervised Learning
• Section 17: Evaluation and Validation
• Section 18: Special Applications

Back to AI Courses...