Generating adversarial data from cleverhans attack models

Question

I want a code example to how to generate train data from clever hans' adversarial attacks.

adv_x = fgsm.generate_np(X_test, **fgsm_params)

This generates adversarial x data but how can I get y?

adv_pred = model.predict_classes(adv_x)

And this will give the "fooled" results right?

What I want is to correctly show generated x, y, fooled y (by which I mean results of models predictions that may be false because of the attack). I'm using Mnist btw, if it helps.

@NicolasPapernot May I have your wisdom?:D – Jeredriq Demas Dec 18 '18 at 06:30 — Jeredriq Demas, Dec 18 '18 at 06:30

score 1 · Accepted Answer · answered Dec 18 '18 at 16:23

1

Based on the code snippets you shared, I would make two suggestions:

It is generally not a good idea to train the model on test data (if you are going to use that test data to evaluate its performance afterwards) so I would replace X_test by X_train in your first line.
To get the label for your adversarial examples, you can use the original labels of the training data or the predictions of the model on the original training data model.predict_classes(X_train) (this assumes that the adversarial example is not perturbed enough to change the label of the input).

answered Dec 18 '18 at 16:23

Nicolas Papernot

351
1
2

So `adv_x = fgsm.generate_np(X_train, **fgsm_params)` is my x #### y_train is already my y that belongs to perturbed adv_x ##### `adv_pred = model.predict_classes(adv_x)` is my fooled results right? – Jeredriq Demas Dec 19 '18 at 05:35
I believe that's correct yes (not sure what the #### stands for though). – Nicolas Papernot Dec 21 '18 at 10:15
Thanks nicolas, you're a great guy:D – Jeredriq Demas Dec 21 '18 at 10:31

Generating adversarial data from cleverhans attack models

1 Answers1

Linked