Adapting the Exploration Rate for Value-of-Information-Based Reinforcement Learning