On the security of deep learning novelty detection

作者：

Highlights：

• Study the security aspects of novelty detection.

• Investigate the case of abstraction-based novelty detection.

• Show the feasibility of bypassing the novelty detection monitoring.

• Study the novelty detection configuration against the attack settings.

• Propose efficient defense mechanisms to protect novelty detection.

摘要

•Study the security aspects of novelty detection.•Investigate the case of abstraction-based novelty detection.•Show the feasibility of bypassing the novelty detection monitoring.•Study the novelty detection configuration against the attack settings.•Propose efficient defense mechanisms to protect novelty detection.

论文关键词：Novelty detection (ND),Anomaly detection (AD),Artificial intelligence (AI),Adversarial Machine Learning (AdvML),Auto-encoders

论文评审过程：Received 30 January 2022, Revised 11 May 2022, Accepted 21 June 2022, Available online 25 June 2022, Version of Record 8 July 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117964