About the importance of choosing the right priors in your machine learning models

Dr. Fabio Anselmi
Università degli Studi di Trieste

Machine learning models are widely used in various elds for prediction, classi cation, and decision-making tasks. However, their performance strongly depends on the specific choice of the model together with the optimization method. In this talk I will discuss about the importance of understanding the information processing operated by a neural network and how incorporating good data priors (such as symmetries) is crucial for applications in elds such as physics or chemistry.