Nacionalni portal odprte znanosti

Consider the online convex optimization problem, in which a player has to choose actions iteratively and suffers corresponding losses according to some convex loss functions, and the goal is to minimize the regret. In the full-information setting, the player after choosing her action can observe the ...