An algorithm with nearly optimal pseudo-regret for both stochastic and adversarial bandits.

Peter Auer, Chao-Kai Chiang

Browse the full COLT paper archive.