Re:End to End Learning for Self-Driving Cars [pdf]
by dkarapetyan on Monday May 09, 2016 @11:58PM
(#11665062)
I don't think it has to do with the combinatorics of the input space. Adversarial inputs are hard to generate until someone figures how to point a set of laser pointers at exactly the right spots on a truck on a highway to get it to swerve out of control.
[ Parent
| on HN ]
Re:End to End Learning for Self-Driving Cars [pdf]
by sgk284 on Tuesday May 10, 2016 @01:22AM
(#11665306)
How is that different than today with human drivers? A laser to the eye will cause lots of swerving.
Re:End to End Learning for Self-Driving Cars [pdf]
by argonaut on Tuesday May 10, 2016 @12:03AM
(#11665077)
> because any random fluctuations can cause the whole thing to malfunction
You made a very specific claim that random fluctuations could have the same effect as adversarial examples. I was addressing that.
Re:End to End Learning for Self-Driving Cars [pdf]
by dkarapetyan on Tuesday May 10, 2016 @12:05AM
(#11665085)
Yes, that makes sense and you're right. I don't have a proper definition of randomness and I wouldn't expect generic noise to cause issues.