Probabilistic Graphical Models 3: Learning

Introduction to learning tasks for probabilistic graphical models, providing an overview of the key concepts and challenges to be covered in the course.

Review of fundamental machine learning concepts including regularization, overfitting, bias-variance tradeoff, and model evaluation techniques.

Comprehensive coverage of parameter estimation in Bayesian networks, including maximum likelihood estimation, Bayesian estimation, and prediction techniques.

Advanced study of parameter estimation for Markov networks and undirected graphical models, covering maximum likelihood and MAP estimation for log-linear models and conditional random fields.

Comprehensive coverage of Bayesian network structure learning, including likelihood scores, BIC, Bayesian scores, and algorithms for learning tree-structured and general networks.

Advanced techniques for learning models with incomplete data, focusing on the Expectation Maximization (EM) algorithm and its applications to various problems.

Summary of key concepts in learning probabilistic graphical models and final examination.

Overview of PGM methods and practical considerations when applying them in real-world scenarios.