Reinforcement Learning for Online Optimization of Banner Format and Delivery Figure 10. Evolution of priority indexes for each arm of the Bandit problem Comparison is measured and has an effect in the subsequently