This talk has moved to Spring 2017 — Some theoretical results on Thompson sampling for Multi-armed Bandits.