Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
adversarial_features [2018/05/15 10:53]
admin 		adversarial_features [2018/09/29 13:51] (current)
admin
Line 251: Line 251:
 flexibility and blackbox approach of deep learning models while providing a viable flexibility and blackbox approach of deep learning models while providing a viable
 alternative to other state-of-the-art methods. alternative to other state-of-the-art methods.
 +
 +https://​arxiv.org/​abs/​1805.12152v1 There Is No Free Lunch In Adversarial Robustness (But There Are Unexpected Benefits)
 +
 +
 + ​Robust models turn out to have interpretable gradients and feature representations that align unusually well with salient data characteristics. In fact, they yield striking feature interpolations that have thus far been possible to obtain only using generative models such as GANs.
 +
 +https://​arxiv.org/​abs/​1806.06108v1 Non-Negative Networks Against Adversarial Attacks
 +
 +https://​arxiv.org/​pdf/​1805.12177v1.pdf Why do deep convolutional networks generalize so
 +poorly to small image transformations?​
 +
 +https://​arxiv.org/​abs/​1806.11146 Adversarial Reprogramming of Neural Networks
 +
 +https://​github.com/​anishathalye/​obfuscated-gradients
 +
 +https://​arxiv.org/​pdf/​1805.12152.pdf There Is No Free Lunch In Adversarial Robustness
 +(But There Are Unexpected Benefits)
 +An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution
 +
 +https://​arxiv.org/​abs/​1807.03247 An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution
 +
 +
 +https://​arxiv.org/​abs/​1808.03305The Elephant in the Room
 +
 +https://​openreview.net/​pdf?​id=S1xoy3CcYX ADVERSARIAL EXAMPLES ARE A NATURAL CONSEQUENCE
 +OF TEST ERROR IN NOISE
 +