Theoretical topics in computer science and machine learning, with emphasis on regret minimization in reinforcement learning, online learning and multi-arm bandits.