Greedy Learning: Using Advantages of Distribution Functions to Improve Generalization of ANNs