Randomized Gradient Adversarial Training

Randomized Gradient Adversarial Training

Randomized Gradient Step scales sign gradients pixel-wise in order to improve state of the art Projected Gradient Descent adversarial training to have higher diversity in adversarial generation thereby improve robustness and generalization of deep neural networks.

Poster: https://drive.google.com/file/d/19SWgF03UShhbXrAgwLtveuxYEQA-6y_5/view?usp=sharing

Video: https://drive.google.com/file/d/1I1hzO2J_NsIXxt3wLvZ_-YVILS0wbrz9/view?usp=sharing

Project Members: 

Oğuz Kaan Yüksel

Project Advisor: 

İnci Meliha Baytaş

Project Status: 

Project Year: 

2020
  • Spring

Contact us

Department of Computer Engineering, Boğaziçi University,
34342 Bebek, Istanbul, Turkey

  • Phone: +90 212 359 45 23/24
  • Fax: +90 212 2872461
 

Connect with us

We're on Social Networks. Follow us & get in touch.