“Universal adversarial perturbations” seems most dramatic ML result in years; if so, not getting deserved attention https://arxiv.org/pdf/1610.08401v1.pdf …
-
-
And, as
@mattskala says, some of the perturbation maps work pretty well across many different architectures:pic.twitter.com/rZKnQu3egf
-
it's worth noting that all those networks were trained on the same training set, i.e learn same distribution
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.