Keras reimplementation of "One pixel attack for fooling deep neural networks"

Re:Keras reimplementation of "One pixel attack for fooling d... by xvedejas on Friday February 23, 2018 @05:35PM (#16450218)

I wonder how well capsule networks could deal with this, considering they're more robust than traditional CNNs towards other sorts of adversarial attacks. My guess is that one-pixel changes are going to do very little to alter pose information (though they will still alter the confidence of the feature existing), and so caps nets should be more robust here as well. Does anyone know if my intuition is correct?

Link to the capsule network paper, for those who haven't heard of it: https://arxiv.org/abs/1710.09829

[ Parent | on HN ]

Re:Keras reimplementation of "One pixel attack for fooling d... by eanzenberg on Saturday February 24, 2018 @01:12AM (#16452014)

I’m not sure capsules will change the security much. We’re talking about pixel level attacks while capsules are trying to generalize at larger length scales (rotational/translational invariance).

[ Parent | on HN ]

Re:Keras reimplementation of "One pixel attack for fooling d... by dahart on Friday February 23, 2018 @06:09PM (#16450434)

I suspect the same, that this kind of attack won't work on capnets. Note that "test on capnets" is on the list of not-yet-complete milestones.

[ Parent | on HN ]

Re:Keras reimplementation of "One pixel attack for fooling d... by hyperparticle on Saturday February 24, 2018 @06:25AM (#16452809)

Author of the repo here. I'm training a capsule network right now, will report back on results in the README in 24-48 hours. My hypothesis is that capsule networks are vulnerable as well, but we'll see by how much.

[ Parent | on HN ]

Re:Keras reimplementation of "One pixel attack for fooling d... by hyperparticle on Sunday February 25, 2018 @12:00PM (#16459834)

So I did some preliminary testing with CapsNet, and it seems that while it was harder than all the other CNNs to find adversarial pixels to fool the network, it was still vulnerable to attack. See the README for quantitative results.

[ Parent | on HN ]

Re:Keras reimplementation of "One pixel attack for fooling d... by tomrod on Friday February 23, 2018 @06:45PM (#16450620)

... We should do this. I learn capnets, you get feather in cap?

[ Parent | on HN ]