Paper
in
Workshop: Security and Safety in Machine Learning Systems
Fighting Gradients with Gradients: Dynamic Defenses against Adversarial Attacks
Dequan Wang · David Wagner · Trevor Darrell
Abstract: