interactive learning: MAB, linear bandits, contextual bandits, linear dynamical systems, MDP (open problem) Linear bandits: pure exploration algorithms -> LinGapE (computationally efficient), RAGE