Oleg Zabluda's blog
Tuesday, July 25, 2017
 
Robust Adversarial Examples | OpenAI | Anish Athalye
Robust Adversarial Examples | OpenAI | Anish Athalye
"""
We’ve created images that reliably fool neural network classifiers when viewed from varied scales and perspectives.
[...]
Out-of-the-box adversarial examples do fail under image transformations. Below, we show the same cat picture, adversarially perturbed to be incorrectly classified as a desktop computer by Inception v3 trained on ImageNet. A zoom of as little as 1.002 causes the classification probability for the correct label tabby cat to override the adversarial label desktop computer.
[...]
By adding random rotations, translations, scales, noise, and mean shifts to our training perturbations, the same technique produces a single input that remains adversarial under any of these transformations.
"""
https://blog.openai.com/robust-adversarial-inputs/
https://blog.openai.com/robust-adversarial-inputs/

Labels:


- posted by Oleg Zabluda @ 5:00 PM | permalink | | | technorati | bloglines | feedster
| |

Home

Powered by Blogger