@inproceedings{71fa9fee14b444e8ab0e0bf91d2b571c,
title = "AdvEdge: Optimizing Adversarial Perturbations Against Interpretable Deep Learning",
abstract = "Deep Neural Networks (DNNs) have achieved state-of-the-art performance in various applications. It is crucial to verify that the high accuracy prediction for a given task is derived from the correct problem representation and not from the misuse of artifacts in the data. Hence, interpretation models have become a key ingredient in developing deep learning models. Utilizing interpretation models enables a better understanding of how DNN models work, and offers a sense of security. However, interpretations are also vulnerable to malicious manipulation. We present AdvEdge and AdvEdge +, two attacks to mislead the target DNNs and deceive their combined interpretation models. We evaluate the proposed attacks against two DNN model architectures coupled with four representatives of different categories of interpretation models. The experimental results demonstrate our attacks{\textquoteright} effectiveness in deceiving the DNN models and their interpreters.",
keywords = "Adversarial image, Deep learning, Interpretability",
author = "Eldor Abdukhamidov and Mohammed Abuhamad and Firuz Juraev and Eric Chan-Tin and Tamer AbuHmed",
note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 10th International Conference on Computational Data and Social Networks, CSoNet 2021 ; Conference date: 15-11-2021 Through 17-11-2021",
year = "2021",
doi = "10.1007/978-3-030-91434-9\_9",
language = "English",
isbn = "9783030914332",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "93--105",
editor = "David Mohaisen and Ruoming Jin",
booktitle = "Computational Data and Social Networks - 10th International Conference, CSoNet 2021, Proceedings",
}