Fast Global Convergence of Natural Policy Gradient Methods with Entropy Regularization - BizPub.ai