Exploring Adversarial Machine Learning - Poisoning Assessment

Hey - I’ve definitely saved the file by running the final code block, however I get an error saying missing file. Unsure why that is?
Step6: missing file; SCORE 0 of 16

Disregard. Mixed up the “Steps” from the assessor, and the labelling of the assessment labs.

Although I am unsure why I’m getting "Step5: bad data; SCORE 0 of 16 " for the poisoning assessment. As opposed to even partial marks as I did for previous labs, when my solution wasn’t 100% correct. Unless my solution for this assessment is completely wrong which might very well be the case?

This has been my worst experience taking an assessment. When graded, Step5: bad data; SCORE 0 of 16 for the poisoning assignment. Can anyone help?

Hello,
I had the same problem at the beginning before taking this step. For my part, I understood that I have a problem when “cifar_display(unmodified_data, 0)” and “cifar_display(poisoned_data, 0)” gave almost the same result.

Anyone to help ? i spent more than 6 hours on this and i can’t get it past the bad model .

I’m a bit lost with this questionnaire. I’m using GradientMatchingAttack and with the build function from the cifar file, it gives me a bad model, but the attack works.

I have questions about index_target = np.where(…)[0][WHAT?], epsilon and learning_rate_schedule as example?.

I’ve increased the poisoning to 0.5, but it still gives me a bad model.

Any clues?