If the method is indeed found in the training set it's not particularly important to de-anonymize it. You have the proof you need.