Variance-Dependent Regret Lower Bounds for Contextual Bandits

Open in new window