02/02/2021

Adversarial Training and Provable Robustness: A Tale of Two Objectives

Jiameng Fan, Wenchao Li

Keywords:

Abstract Paper Similar Papers

Abstract: We propose a principled framework that combines adversarial training and provable robustness verification for training certifiably robust neural networks. We formulate the training problem as a joint optimization problem with both empirical and provable robustness objectives and develop a novel gradient-descent technique that can eliminate bias in stochastic multi-gradients. We perform both theoretical analysis on the convergence of the proposed technique and experimental comparison with state-of-the-arts. Results on MNIST and CIFAR-10 show that our method can consistently match or outperform prior approaches for provable l∞ robustness. Notably, we achieve 6.60% verified test error on MNIST at ε = 0.3, and 66.57% on CIFAR-10 with ε = 8/255.

The video of this talk cannot be embedded. You can watch it here:
https://slideslive.com/38949055
(Link will open in new window)
 0
 0
 0
 0
  Share    
This is an embedded video. Talk and the respective paper are published at AAAI 2021 virtual conference. If you are one of the authors of the paper and want to manage your upload, see the question "My papertalk has been externally embedded..." in the FAQ section.

Comments

Post Comment
no comments yet
code of conduct: tbd Characters remaining: 140

Similar Papers

 0
 0
 0
 0
 3:23 
 0
 0
 0
 0
 8:13 
 0
 0
 0
 0
 5:14 
 0
 0
 0
 0
 12:41 
 0
 0
 0
 0
 3:38 
 0
 0
 0
 0
 3:03 
 0
 0
 0
 0
 5:19 
 0
 2
 0
 0
 5:52 
 0
 0
 0
 0
 17:41 
 0
 0
 0
 0
 10:37 
 0
 0
 0
 0
 10:24 
 0
 0
 0
 0
 16:11 
 0
 0
 0
 0
 1:01 
 0
 0
 0
 0
 5:16 
 0
 0
 0
 0
 3:20 
 0
 0
 0
 0
 3:22 
 0
 0
 0
 0
 2:58 
 0
 0
 0
 0
 10:29 
 0
 0
 0
 0
 7:16 
 0
 0
 0
 0
 10:52 
 0
 0
 0
 1
 7:48 
 0
 0
 0
 0
 3:06 
 0
 0
 0
 0
 15:18 
 0
 0
 0
 0
 4:59 
 0
 0
 0
 1
 5:06 
 0
 0
 0
 0
 5:23 
 0
 0
 0
 0
 4:52 
 0
 0
 0
 0
 7:36 
 0
 0
 0
 0
 1:00 
 0
 0
 0
 0
 8:31 
 0
 0
 0
 0
 3:22 
 0
 0
 0
 0
 5:13 
 0
 0
 0
 0
 2:58 
 0
 0
 0
 0
 5:05 
 0
 0
 0
 0
 5:18 
 0
 0
 0
 0
 0:59 
 0
 0
 0
 0
 6:00 
 0
 0
 0
 0
 15:06 
 0
 0
 0
 0
 5:51 
 0
 0
 0
 0
 20:21