ATML: Deep generative modelling
This track is an introduction to deep generative modelling algorithms, where we understand that term as follows:
- A generative model is a probability distribution, or generative process, that is derived from data so as to approximate the distribution that produced the data.
- A deep generative model is one that uses deep neural networks to represent the generative process or its components.
- A deep generative modelling algorithm consists of: a choice of generative process, a family of distributions parametrised by neural networks to represent that process, and a learning algorithm to fit those networks' parameters to data.
This track will survey the main classes of deep generative modelling algorithms, particularly for high-dimensional data, and their applications.
Course information
Instructor: Nikolay Malkin; assistants: Kirill Tamogashev and Rajit Rajpal.
Lectures: Tuesdays 17:10-18:00, Anatomy Lecture Theatre.
Tutorials: Mondays 13:10-14:00 (NM present) and 14:10-15:00 (KT present), Appleton Tower M2, starting in Week 3. Theory exercises and code demonstrations will be posted the preceding week and linked to in the table below.
Recommended references: Probabilistic Machine Learning: An Introduction and Advanced Topics by Kevin P. Murphy. Chapter IV of the latter book concerns deep generative models.
Sample exam: see here.
Course schedule
Slides will be posted the evening before each lecture (possibly in draft form). In full pdfs, embedded videos work only in Adobe Reader and possibly some other programs. The compressed pdfs have images in lower quality and no embedded videos.
| Week | Date | Topic | Slides | Exercises |
|---|---|---|---|---|
| 1 | 13.01 | Introduction and overview | full (39MB), compressed (1.6MB) | |
| 2 | 20.01 | Preliminaries: probability review, distribution approximation and divergences | full (6.5MB), compressed (0.9MB) | theory, code |
| 3 | 27.01 | Latent variable models as generative models: autoencoders, variational autoencoders, hierarchical models, and their evaluation | full (7.4MB), compressed (0.7MB) | theory, code |
| 4 | 03.02 | Generative models with exact density evaluation: normalising flows, autoregressive models | ||
| 5 | 10.02 | Adversarial objectives for generative models: generative adversarial networks, density-free evaluation of generative models | ||
| Break | ||||
| 6 | 24.02 | Downstream uses of generative models: representation learning, conditioning and control, improvement with human feedback | ||
| 7 | 03.03 | Guest lecture by Zee Talat, “Evaluating generative AI” | ||
| 8 | 10.03 | Diffusion and continuous-time models I: diffusion models as hierarchical VAEs, denoising objective | ||
| 9 | 17.03 | Diffusion and continuous-time models II: score matching, stochastic differential equations | ||
| 10 | 24.03 | Diffusion and continuous-time models III: flow matching, advanced topics | ||
| 11 | 31.03 | Revision and advanced topics depending on interest: implicit (energy-based) models, discrete latent variable models and neurosymbolic methods, various applications, etc. | ||