Adversarial Learning for Good: On Deep Learning Blindspots

Re:Adversarial Learning for Good: On Deep Learning Blindspots by beefield on Friday December 29, 2017 @01:30PM (#16030977)

I wonder how sensitive these results are. I mean, if you run one more learning round of the network, does the rifle turn to turtle in the eyes of the network immediately so that you would need to generate a new turtle for every single network? Or does that turtle look like rifle for all current neural networks? Or, most likely, somewhere in between?

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by marcosdumay on Friday December 29, 2017 @03:21PM (#16031732)

If you don't change the data, why would the network change its results with just some more learning iterations? It's already converging.

[ Parent | on HN ]

Re:Adversarial Learning for Good: On Deep Learning Blindspots by beefield on Saturday December 30, 2017 @03:25AM (#16034404)

Because if I understtod anything correctly, the methods tries to find the smallest possible changes that causes the network to make the incorrect classofication. And these smallest possible changes just might be very sensitive to whatever small random ripple weight changes a network has. But I am far from neural network expert, so I can't really answer.

[ Parent | on HN ]