The tiny Changes that can cause AI to fail

BBC has published an interesting article on how easily machine learning algorithms can be mistaken by small changes in data. See http://www.bbc.com/future/story/20170410-how-to-fool-artificial-intelligence for more information. The article also refers to recent research at EPFL by Dez Seyed-Mohsen Moosavi-Dezfooli, Alhussein Fawzi, Omar Fawziz, and Pascal Frossardy on universal adversarial perturbations.