Sparked by the success of deep neural networks in many practical applications, there has been an increasing interest in analyzing these methods from a statistical perspective. In recent years, advances in the theory of neural networks and kernels have brought both fields closer together.