"Adversarial perturbations" reliably trick AIs about what kind of road-sign they're seeing

FGD135 August 7, 2017, 9:38pm 4

That’s already possible without “AI”.
They are usually called “trains”.

9 Likes

Topic		Replies	Views
The "universal adversarial preturbation" undetectably alters images so AI can't recognize them boing	28	3477	April 3, 2017
Adversarial patches: colorful circles that convince machine-learning vision system to ignore everything else boing	22	3199	January 13, 2018
Techniques for reliably fooling AI machine-vision classifiers boing	20	2806	July 23, 2017
Machine learning models keep getting spoofed by adversarial attacks and it's not clear if this can ever be fixed boing	31	3476	March 14, 2018
Tiny alterations in training data can introduce "backdoors" into machine learning models boing	11	995	November 30, 2019