Boosting for Online Convex Optimization

                         Elad Hazan 1 2 Karan Singh 3

              Abstract                 et al., 2012). Itp is well established that the regret can be
                                  bounded by O( T log |H|), where |H| is the number of
   We consider the decision-making framework of
                                  experts, and that this is tight. However, in many problems
   online convex optimization with a very large num-
                       ...