There are plenty of data sets for which you can apply clustering. Any function that maps for example a real value to let's say 0 or 1 is a classifier that can produce a data set for you. At the end, you compare the two ANNs to see which one made the same output that the function did. However, the data set can help you to discriminate between the two ANNs that you have in mind is what you have to find based on the application. If you need more complicated data set, this link might help you: