In a classification task, I trained a BP network with the same training set several times and evaluated it with the same test set. However, the resulting classification accuracy showed two extremes, either close to 0 or close to 100%. It would be of great help if anyone could explain this problem.
I have tested about five times.