MadryLab/mnist_challenge

Issues

• Why epsilon can be larger than 1?

  #13 opened 5 years ago by sjyjytu
  6

• problems between paper(Generating Adversarial Examples with Adversarial Networks) and code

  #18 opened 6 months ago by bieyl
  0

• about version

  #17 opened 3 years ago by lsy-sunny
  1

• Are there any Pytorch version of this challenge? Cause tensorflow usually has conflict with my Pytorch.

  #16 opened 3 years ago by Asber777
  2

• Requirements to reproduce the results

  #15 opened 4 years ago by FAKEBOB-adversarial-attack
  0

• Adversarial performance for MNIST models vary widely with different random seed initializations

  #14 opened 4 years ago by hangletn
  1

• What does this line do?

  #12 opened 5 years ago by ovshake
  1

• Any adversarial attack that sustains after resize attack?

  #11 opened 5 years ago by BalaMallikarjuna-G
  3

• wrong implementation of CW attack

  #10 opened 5 years ago by CNOCycle
  3

• tf.tf.Variable() seems cannot be replaced with tf.get_variable().

  #9 opened 6 years ago by lepangdan
  8

• parameters for train robust adv_trained/secret networks?

  #8 opened 6 years ago by lepangdan
  1

• About random restart

  #6 opened 6 years ago by jinghuichen
  2

• For the reported results with 100 iterations, is the eps_iter/"a" value still 0.01?

  #5 opened 6 years ago by gwding
  1

• Question about reproducing your results

  #4 opened 6 years ago by liuchihuang
  2

• Does PGD not need to perform random restart in every iterative ? Is it enough to start with random noise at FGSM?

  #3 opened 6 years ago by lepangdan
  3

• (Not an issue) Request to keep the challenge open

  #2 opened 7 years ago by akarshzingade
  2

• Release Model

  #1 opened 7 years ago by huanzhang12
  3