For the formulas of the generalized cross entropy is calculated as follows:
0-1 but for the cross-entropy distribution calculation formula can be simplified as follows:
Why just the beginning of such a change does not understand, that contains what deep-seated concepts and techniques, and is actually very simple, 0-1 distribution is binary, can be seen as a binary classification problem is there are two categories of classification, know an event probability p, the probability of another event is 1-p, the probability of a category is forecast to q, then the probability of another event be 1-q. Thus changes to the formula above is a matter of course.