🗓️ Week 12 - Introduction to Generative Models

Author

Affiliation

Dr. Simone Rossi

EURECOM

This week we introduce generative models and contrast them with the discriminative models covered so far. We will explore how generative models learn the joint distribution of data and labels, enabling data generation, density estimation, and conditional reasoning. We start with Naive Bayes as a simple yet illustrative generative classifier, then move to mixture models and the Expectation-Maximization algorithm for Gaussian mixture models. Finally, we cover probabilistic PCA as a linear latent variable model for dimensionality reduction.

🖥️ Lecture Slides

• Introduction to Generative Models
• Latent Variable Models

📑 Lecture Notes (PDF)

• Introduction to Generative Models
• Latent Variable Models

📝 Exercises

• Generative models and latent variable models exercises

📚 Recommended Reading

• Murphy (2023): Chapter 20 (Generative Models)
• Murphy (2023): Chapter 28 (Mixture Models and Probabilistic PCA)

📙 References

Murphy, K. P. (2023). Probabilistic machine learning: Advanced topics. MIT Press. http://probml.github.io/book2