The 2022 course by Stéphane Mallat, Professor at the Collège de France, is devoted to Information Theory. It focuses on two seminal articles: R. A. Fisher's in 1922, 'On the mathematical foundations of theoretical statistics,' and Cl. Shannon's in 1948, 'A Mathematical Theory of Communication.' Fisher develops concepts such as the consistency of an estimator, particularly maximum likelihood and inference, the notion of information and sufficient statistics. Meanwhile, Shannon extracts entropy from its framework in physics to make it a universal tool, with which he addresses, among other things, coding (compression), through the concentration of probabilities on typical sets. These concepts are at the heart of the implementation of neural networks and the understanding of their success