Adversarial Dueling Bandits

                  Aadirupa Saha 1 Tomer Koren 2 Yishay Mansour 2

             Abstract               regret with respect to the best item in hindsight, according
    We introduce the problem of regret minimization     to a certain score function.
    in Adversarial Dueling Bandits. As in classic      Numerous real-world applications are naturally modelled as
    Dueling Bandits, the learner has to repeatedly      dueling bandit problems, including movie recommendati ...