Due to the NP -hardness of many machine learning problems such as clustering, decision tree, and neural network, one primary belief is that solving ML problems to global optimality is computationally ...