Reviews: Approximation and Convergence Properties of Generative Adversarial Learning