François Fleuret, professor in the Machine Learning group of the Department of Computer Science at UNIGE, presents an accessible examination of modern artificial neural networks. He explores the fundamental principles that enable these systems to learn autonomously, how they process complex signals such as images and text, and how they can generate meaningful content. The lecture examines the limitations of current methods and discusses the partial understanding we have of their inner workings, with implications for research and applications.
In French.
