The course is intended as a broad overview to neural networks, as used today in a number of applicative fields. It provides a strong theoretical and practical understanding of how neural networks and modern deep networks are designed and implemented, highlighting the most common components, ideas, and current limitations. We will review the general paradigm of building differentiable models that can be optimized end-to-end with gradient descent from data, and then overview essential components to design architectures able to work on images (convolutive layers), sequences (recurrent layers), and sets (transformer layers). The last part of the course will then focus on a selection of emerging research topics, including generative models, among others.
Calendar: 26 February - 30 May, 2025.
Class Schedule: Monday 15:00 - 18:00 Classroom 41, SPV, Via Eudossiana, 18
Wednesday 08:30 - 10:00 Classroom 8, SPV, Via Eudossiana, 18
Material: Slides, notebooks, assignments, and news will be available on the course's Google Classroom. Students are invited to register.
Exams
Exams must be booked electronically via the INFOSTUD portal. Scheduled exam sessions for the year 2024/2025:
Session I: 26 June 2025 - Aula A2, Via Ariosto, 10:00
Session II: 21 July 2025 - Aula A2, Via Ariosto, 10:00
Session III: 5 September 2025 - Aula A3, Via Ariosto, 10:00
Extraordinary Session I: October 2025, TBD
Session IV: January 2026, TBD
Session V: February 2026, TBD
Extraordinary Session II: March 2026, TBD