HopSkipJumpAttack: A Query-Efficient Decision-Based Attack

The goal of a decision-based adversarial attack on a trained model is to generate adversarial examples based solely on observing output labels returned by the targeted model. We develop HopSkipJumpAttack, a family of algorithms based on a novel estimate of the gradient direction using binary informa...

Ausführliche Beschreibung

Bibliographische Detailangaben

Hauptverfasser:	Chen, Jianbo, Jordan, Michael I., Wainwright, Martin J.
Format:	Tagungsbericht
Sprache:	eng
Schlagworte:	Estimation Iterative methods Measurement Neural networks Optimization Perturbation methods Predictive models
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!