PhD Course: Deep Generative Models

The course is aimed at reviewing the foundational aspects as well as the recent advances in generative probabilistic models, which learn distributions from data aimed at generating new data instances from the learned distribution. In recent years, these models have been parameterized using deep neural networks and combined with advanced stochastic optimization methods. This has facilitated their widespread adoption for modeling complex data (such as images, text, and audio) across various application domains, including computer vision, speech and natural language processing, and social media analytics.

The course will focus on the mathematical foundations of generative modeling techniques. We will cover some key concepts from the recent literature, including variational autoencoders, generative adversarial networks, diffusion models, as well as newly developed approaches and the application of the aforementioned models across various domains and tasks.

Orario corso:

Lunedì 12/10/2026 15:00-17:00
Martedì 13/10/2026 9:00-11:00
Mercoledì 14/10/2026 9:00-13:00

Le lezioni si terranno presso l'Aula Tecnica DIMES (Cubo 41C, I° piano)