PhD Course - Introduction to Probabilistic Machine Learning (2026)

Pre-requisites

📌 Probability & Machine Learning: We expect basic knowledge of probability theory and machine learning. You can review Chapters 1–4 of the PML book for a solid foundation.

Murphy, Kevin P. Probabilistic Machine Learning: An Introduction. MIT Press, 2022. A free online version is available here: 🔗 PML Book

📌 Python & Numpy: Some familiarity with Python is also expected. You can use following two Google Colab notebooks to help you get comfortable with Python and NumPy.

Day 1

Block 0: Introduction & Assignment

• Slides

Block 1: Probabilistic Regression

• Slides
• Notebook: 01_probabilistic_linear_regression
• Notebook: 02_nonlinear_regression

Block 2: Probabilistic Classification

• Slides
• Notebook: 03_probabilistic_classification
• Notebook: 04_neural_classifier

Block 3: Clustering

• Slides
• Notebook: 05_gmm_clustering

Day 2 - Before Lunch

• Slides
• Notebook: students_simple_model
• Notebook: solutions_simple_model
• Notebook: CAVI-linreg

Day 2 - After Lunch

• Slides
• Notebook: BayesianNeuralNetworks
• Notebook: students_BBVI
• Notebook: solutions_BBVI
• Notebook: student_simple_gaussian_model_pyro
• Notebook: solution_simple_gaussian_model_pyro
• Notebook: Bayesian_linear_regression
• Notebook: students_bayesian_logistic_regression
• Notebook: solutions_bayesian_logistic_regression
• Notebook: students_VAE
• Notebook: solutions_VAE

Assignment

Students are required to apply probabilistic machine learning to a real-world dataset of their choice, designing and implementing a probabilistic model in Pyro and evaluating the results. Deadline: 30 April 2026.

📄 Assignment description