Machine learning and control theory

We survey in this chapter the connections between Machine Learning and Control Theory. Theory provide useful concepts tools for Learning. Conversely can be used to solve large control problems. In first part of paper, we develop reinforcement learning Markov Decision Processes, which are discrete time second part, review concept supervised relation with static optimization. Deep extends learning, viewed as a problem. third present links stochastic gradient descent mean field theory. Conversely, fourth fifth parts, machine approaches problems, focus on deterministic case, explain, more easily, numerical algorithms.