MSc Machine Learning Theory

A mathematically demanding master's on why learning algorithms work. Students prove generalisation bounds, analyse the convergence of optimisation methods, and study the theory of deep networks, online learning, reinforcement learning and causal inference. Most assessment is through proofs, derivations and small computational experiments that test theoretical predictions. The programme suits graduates with strong backgrounds in mathematics, statistics, theoretical computer science or physics who want to pursue doctoral research or rigorous algorithm design, and it ends with a theoretical or methodological thesis.