Section outline
-
We close the gap between neural networks and probabilistic learning by discussing generative deep learning models. We introduce a taxonomy of the existing generative deep learning approaches and study in-depth relevant families of models for each element of the taxonomy, including: autoregressive generation, variational autoencoders, generative adversarial networks, diffusion models, flow-based methods and score matching-
Date Topic References Additional Material 23 16/04/2026
(11-13)Neural Autoencoders
Introduction to the generative deep learning module; generative models taxonomy; undercomplete neural autoencoders; deep autoencoders.[SD] Coverage of the Prince book on this lecture is inadequate but you can use the lecture slides and complement with the additional material if necessary. (e.g. chapter 14 of the deep learning book). Additional Readings
[24] DBN: the paper that started deep learning
[25] Deep Boltzmann machines paper
[26] Review paper on deep generative models
[27] Long review paper on autoencoders from the perspective of representation learning24 21/04/2026
(11-13)Variational Autoencoders
explicit distribution models; score learning in DAE; neural ELBO; variational approximation; reparameterization trick; latent space properties
[SD] Chapter 14
[SD] Chapter 17
Additional Readings
[28] Tutorial on VAE
Sofware- A tutorial on VAE with code
25 22/04/2026
(16-18)Generative Adversarial Networks
learning a sampling process; adversarial learning principles; wasserstein GANs; conditional generation; notable GANs; adversarial autoencoders
[SD] Chapter 15 Additional readings
[29] Tutorial on GAN (here another online resource with GAN tips)
[30] Wasserstein GAN
[31] Tutorial on sampling neural networks
[32] Progressive GAN
[33] Cycle Gan
[34] Seminal paper on Adversarial AEs
Sofware- Official Wasserstein GAN code
- A (long) list of GAN models with (often) associated implementation
26 23/04/2026
(11-13)Coding practice III - Lecture by Riccardo Massidda
27 28/04/2026
(11-13)Normalizing flow models I
tractable explicit likelihood; autoregressive generative learning; probabilistic change of variable, forward/normalization pass; from 1D to multidimensional flows.
[SD] Chapter 16 29/04/2026
(16-18)LECTURE CANCELLED DUE TO STUDENTS' ASSEMBLY
28 30/04/2026
(11-13)Normalizing flow models II
coupling flows; masking & squeezing; invertible convolutions; autoregressive flows; residual & continous normalizing flows
[SD] Chapter 16 Additional Readings
[35] Survey paper on normalizing flows
[36] RealNVP paper
[37] GLOW paper
[38] MADE autoregressive flow
Sofware- Normalizing flows are implemented natively in Tensorflow Probability
- Two PyTorch-based packages for Normalizing Flows: Normflows (pure PyTorch) - Flowtorch (PyRo)
05/05/2026
(11-13)LECTURE CANCELLED DUE TO LECTURER UNAVAILABILITY
29 06/05/2026
(16-18)Diffusion models
noising-denoising processes; kernelized diffusion; latent space diffusion; conditional diffusion models
[SD] Chapter 18 Additional Readings
[39] Introductory and survey paper on diffusion models
[40] Seminal paper introducing diffusion models
[41] Paper introducing the diffusion model reparameterization
[42] Diffusion beats GAN paper30 07/05/2026
(11-13)Causal representation learning - Lecture by Riccardo Massidda
31 12/05/2026
(11-13)Matching approaches
alternative views of diffusion approaches; score matching; continous score matching as stochastic DE; flow matching; generative DL module wrap-up
The content of this lecture is too new to be covered by SD. You can however find it in Chapter 9 of Tomczac's new book, available for free download from UNIPI's network
(Slide + handouts is also ok)
Additional Readings
[43] Foundational work on learning by score matching
[44] The flow matching paper
[45] Rectified flow matching