“Universal adversarial perturbations” seems most dramatic ML result in years; if so, not getting deserved attention https://arxiv.org/pdf/1610.08401v1.pdf …
-
-
Thanks to Jeff Shrager (https://explorecourses.stanford.edu/instructor/jshrager …) for drawing the paper to my attention.
- 2 more replies
New conversation -
-
-
Haven't read the paper yet, but very intriguing! your analysis seems astute and measured, which is rare in the "how DL systems fail" genre.
-
possibly relevant: https://arxiv.org/abs/1703.06857
- 1 more reply
New conversation -
-
-
You're on the money. Since perception is built up from fine-grained detail (texture) to coarse (contour)...
-
...it gives adjacent cells in the lowest layers just enough information to disagree with each other.
- 1 more reply
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.