Librería Samer Atenea
Kálamo Books
Librería Elías (Asturias)
Librería Kolima (Madrid)
Librería Proteo (Málaga)
This is a complete, from-first-principles course text on two of the most important tools for protecting individuals in data-driven systems: differential privacy and federated learning. Differential privacy gives a rigorous mathematical definition of what it means for a computation to safeguard the people whose data it uses, along with the mechanisms, accounting methods, and learning algorithms that make the definition usable in practice. Federated learning offers a way to train models on data that never leaves the devices or institutions holding it. Together, they let practitioners learn from sensitive, distributed data while offering provable privacy guarantees.The book unfolds in seven parts: an opening account of how early anonymisation techniques failed and why that failure matters; the mathematical background needed for the rest of the text; the core theory of differential privacy, including composition and accounting; practical applications such as query release, DP-SGD, local privacy, and synthetic data; federated learning and its optimisation; the attacks federated systems invite and the defences against them; and a closing synthesis covering governance and ethics.