Preventing Gradient Attenuation in Lipschitz Constrained Convolutional Networks

Qiyang Li
Qiyang Li
Saminul Haque
Saminul Haque
Cem Anil
Cem Anil
Joern-Henrik Jacobsen
Joern-Henrik Jacobsen

ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 32 (NIPS 2019), pp. 15364-15376, 2019.

Cited by: 3|Views38
EI
Keywords:

Abstract:

Lipschitz constraints under L-2 norm on deep neural networks are useful for provable adversarial robustness bounds, stable training, and Wasserstein distance estimation. While heuristic approaches such as the gradient penalty have seen much practical success, it is challenging to achieve similar practical performance while provably enforc...More

Code:

Data:

Your rating :
0

 

Tags
Comments