Targeted DeepFool

Targeted DeepFool attacks are a class of adversarial attacks designed to fool deep neural networks (DNNs) into misclassifying inputs into a specific, pre-selected target class, using minimal perturbations. Current research focuses on improving the efficiency and effectiveness of these attacks, particularly within the context of various DNN architectures like AlexNet and Vision Transformers, while also considering the impact on image quality and the confidence of the misclassification. This research is crucial for understanding and improving the robustness of DNNs, particularly in applications where security and reliability are paramount, such as image recognition and autonomous systems.

Papers

October 18, 2023

Tailoring Adversarial Attacks on Deep Neural Networks for Targeted Class Manipulation Using DeepFool Algorithm
S. M. Fazle Rabby Labib, Joyanta Jyoti Mondal, Meem Arafat Manab, Sarfaraz Newaz, Xi Xiao
Deep Neural Network Class Specific Targeted DeepFool

March 22, 2023

Revisiting DeepFool: generalization and improvement
Alireza Abdollahpourrostam, Mahed Abroshan, Seyed-Mohsen Moosavi-Dezfooli
Adversarial Attack Strong Generalization Adversarial Example Adversarial Training Adversarial Perturbation Large Relevance Improvement State of the Art Robustness Targeted DeepFool

Targeted DeepFool

Papers

Tailoring Adversarial Attacks on Deep Neural Networks for Targeted Class Manipulation Using DeepFool Algorithm

Revisiting DeepFool: generalization and improvement