Adversarial Learning for Good: On Deep Learning Blindspots

Re:Adversarial Learning for Good: On Deep Learning Blindspots by daenz on Friday December 29, 2017 @01:15PM (#16030881)

The video example of a turtle being misclassified as a rifle is pretty scary https://www.youtube.com/watch?v=piYnd_wYlT8

We're approaching a world where AI will be more and more relied upon in dangerous situations. Imagine someone getting killed for something ridiculous like inadvertently holding an adversarial example. Public trust would have a hard time recovering.

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by oh_sigh on Friday December 29, 2017 @02:28PM (#16031378)

Would adding random noise to an image before running it through the classifier mitigate these kinds of attacks?

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by yorwba on Friday December 29, 2017 @10:29PM (#16033687)

No, ironically because classifiers are trained to be robust to noise. Adversarial examples are generated by altering the original in one specific dimension of a high-dimensional vector space. Random noise is vanishingly unlikely to undo that transformation, so the classification will stay wrong.

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by doener on Friday December 29, 2017 @01:42PM (#16031068)

>Imagine someone getting killed for something ridiculous like inadvertently holding an adversarial example.

This would never be the case with humans, right? No.

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by daenz on Saturday December 30, 2017 @12:01AM (#16033925)

With humans there is accountability (in theory). Who is accountable for the model?

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by trhway on Saturday December 30, 2017 @03:38AM (#16034440)

model is adjustable/correctable/trainable/improvable. Humans are much less so.

[ Parent | on HN ]