MaLGa Summer PhD School in Genoa, Italy | Theoretical Foundations of Machine Learning (TFML)

[Malga Unige]

We apologize for multiple postings and kindly request that this message be circulated among students and researchers who might be interested in participating.

We are pleased to announce the Summer PhD course: Theoretical Foundations of Machine Learning (TFML): https://malga.unige.it/education/schools/tfml/.

The course is organized by MaLGa - Machine Learning Genoa Center, as part of the ELLIS Genoa unit activities, and is taught by Ernesto De Vito, Lorenzo Rosasco, and Silvia Villa.

The objective of the course is to introduce the mathematical framework and key foundational results in Machine Learning Theory. In particular, the course will present the statistical learning theory framework with empirical risk minimization and regularization as guiding principles for algorithm design. Linear and nonlinear models will be discussed, including neural networks and kernel methods (reproducing kernel Hilbert spaces). The course will cover optimization aspects, discussing and analyzing stochastic gradient methods and backpropagation. Additionally, it will cover the error analysis of different algorithms, considering both statistical and approximation theory aspects. Further advanced topics will be covered in a half-day workshop by Nicolo’ Cesa Bianchi (UniMi) and Andrea Agazzi (UniBe).

--------------------------------------------------------------------------------------------------

Format:

The course will consist of 28 hours, held from June 23 to June 27, 2025, at the DIBRIS/DIMA@UNIGE building, Via Dodecaneso 35, 16146 Genoa, Italy.

Lessons will be in-person only.

Participants will receive a certificate of attendance upon completing the course. Students will have the chance to take a final exam if they wish so.

--------------------------------------------------------------------------------------------------

Apply:

Submit your application here: https://forms.gle/FPMz4bcwzfExhud69

Application deadline: March 30, 2025.

The maximum number of participants is 120. Selection will be based on CV evaluation.

We encourage early applications as spaces are limited.

Best regards,

The organizers: Marco Rando, Hippolyte Labarrière, Giulia Casu,

on behalf of the instructors:

Ernesto De Vito, Lorenzo Rosasco, and Silvia Villa