Re:How close are we to solving vision?
by oifnwoivnoinvo on Sunday November 13, 2016 @12:09PM
(#12944310)
"Generative models are only vulnerable to adversarial examples that are actually unlikely in the data distribution. They do not have patterns or filters that can be added to ordinary images to cause wild misclassifications."
^ Actually, you are wrong on this. See this recent paper "Universal adversarial perturbations"
Re:How close are we to solving vision?
by eli_gottlieb on Sunday November 13, 2016 @02:38PM
(#12945181)
That paper deals with non-stochastic deep neural networks, and its mathematical analysis deals with discriminative classification. It doesn't deal with generative models, which model the joint probability distribution of classes and data instances rather than just taking a maximum-a-posteriori estimate from the posterior.
Re:How close are we to solving vision?
by Houshalter on Sunday November 13, 2016 @05:31PM
(#12946083)
We don't know that the brain uses generative models. Generative models are pretty inefficient.
Also the original adversarial examples paper found that autoencoders were just as vulnerable. I don't see why generative models wouldn't be vulnerable.
Re:How close are we to solving vision?
by eli_gottlieb on Sunday November 13, 2016 @08:29PM
(#12946867)
>We don't know that the brain uses generative models.
We have some fair evidence, see: http://www.fil.ion.ucl.ac.uk/~karl/A%20free%20energy%20princ...
Generative models are only vulnerable to adversarial examples that are actually unlikely in the data distribution. They do not have patterns or filters that can be added to ordinary images to cause wild misclassifications.
The brain, as far as we know, uses generative modeling.
So yeah.