Revisiting ensemble adversarial attack

作者：

Highlights：

• A finding of the drawback of existing ensemble attacks.

• A new insight into ensemble adversarial attacks from viewpoint of gradient magnitudes.

• Two simple yet effective methods to boosting ensemble adversarial attacks.

摘要

•A finding of the drawback of existing ensemble attacks.•A new insight into ensemble adversarial attacks from viewpoint of gradient magnitudes.•Two simple yet effective methods to boosting ensemble adversarial attacks.

论文关键词：Adversarial attack,Ensemble strategies,Gradient-based methods,Deep neural networks,Image classification

论文评审过程：Received 22 April 2021, Revised 28 February 2022, Accepted 27 May 2022, Available online 4 June 2022, Version of Record 16 June 2022.

论文官网地址：https://doi.org/10.1016/j.image.2022.116747