Why do the same BP model get different results on the same test set?

18 Views Asked by At

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.

0

There are 0 best solutions below