Foundations Machine Learning

Catalog description

This is an introductory course in machine learning (ML) that covers basic theory, algorithms, and applications. The class will develop a principled understanding of the various facets of ML and encompass fundamental (supervised and unsupervised) ML primitives that underpin modern technologies. Specifically, the learning theory content will cover the statistical learning paradigm, empirical risk minimization, generalization, bias-variance tradeoff, regularization, and validation. The supervised learning chapter will cover regression, the maximum likelihood principle, generalized linear models, support vector machines, and naïve Bayes. Unsupervised learning methods will include clustering, k-means, EM algorithm, factor analysis, and other dimensionality reduction techniques. The final few lectures will be devoted to large language models and the generative pre-trained transformer (GPT) architecture, as well as topics in ethics and fairness in machine learning. Our treatment of the material will start from theoretical principles, and build up towards implementation and applications dealing with text data, handwriting, music, images, etc. To that end, the course will incorporate a programming.