A Prescriptive Dirichlet Power Allocation Policy with Deep Reinforcement Learning