search

carlini / nn_robust_attacks

Robust evasion attacks against neural network to find adversarial examples
BSD 2-Clause "Simplified" License
778 stars 229 forks source link

I want to attack my own model training by tensorflow2.0. #40

Closed Madanfeng closed 3 years ago

Madanfeng commented 3 years ago

I want to attack my own model training by tensorflow2.0, how do I modify the code?

carlini commented 3 years ago

You can probably get almost all of this to work by importing tf.compat.v1 if you want to use this implementation.

Alternatively, you can use another implementation of this attack built for TF2 from CleverHans, FoolBox, or ART.

Madanfeng commented 3 years ago

You can probably get almost all of this to work by importing tf.compat.v1 if you want to use this implementation.

Alternatively, you can use another implementation of this attack built for TF2 from CleverHans, FoolBox, or ART.

Thank your reply, I refer to your code rewriting to tf2, the code is being refined. https://github.com/Madanfeng/CW_TF2