How to find the wrong prediction cases in a test case (CNN using Keras)

Question

I am using the MNIST example with 60,000 training images and 10,000 test images. How to find out which of the 10,000 test images has the wrong classification / forecast?

Answer 1

Just use model.predict_classes() and compare the output with true labes. i.e:

 incorrects = np.nonzero(model.predict_class(X_test).reshape((-1,)) != y_test) 

to get indices of incorrect forecasts