Adversarial Deep Learning: A Survey on Adversarial Attacks and Defense Mechanisms on Image Classification

Abstract

The popularity of adapting deep neural networks (DNNs) in solving hard problems has increased substantially. Specifically, in the field of computer vision, DNNs are becoming a core element in developing many image and video classification and recognition applications. However, DNNs are vulnerable to adversarial attacks, in which, given a well-trained image classification model, a malicious input can be crafted by adding mere perturbations to misclassify the image. This phenomena raise many security concerns in utilizing DNNs in critical life applications which attracts the attention of academic and industry researchers. As a result, multiple studies have proposed discussing novel attacks that can compromise the integrity of state-of-the-art image classification neural networks. The raise of these attacks urges the research community to explore countermeasure methods to mitigate these attacks and increase the reliability of adapting DDNs in different major applications. Hence, various defense strategies have been proposed to protect DNNs against adversarial attacks. In this paper, we thoroughly review the most recent and state-of-the-art adversarial attack methods by providing an in-depth analysis and explanation of the working process of these attacks. In our review, we focus on explaining the mathematical concepts and terminologies of the adversarial attacks, which provide a comprehensive and solid survey to the research community. Additionally, we provide a comprehensive review of the most recent defense mechanisms and discuss their effectiveness in defending DNNs against adversarial attacks. Finally, we highlight the current challenges and open issues in this field as well as future research directions.

Threats, Vulnerabilities, and Controls of Machine Learning Based Systems: A Survey and Taxonomy

“…We cannot completely remove a trained model's vulnerabilities by applying a state-of-the-art method of model development. For example, despite many studies on adversarial examples [50][51][52][53][54][55], no machine learning algorithm is known to produce models that behave correctly for all adversarial examples (Section 6.6).…”

“…Survey literature. For technical details of attacks and defenses, see previous papers, e.g., [50][51][52][53][54][55].…”

Leveraging linear mapping for model-agnostic adversarial defense

“…In recent years, the vulnerability of deep neural networks (DNNs) to adversarial attacks has sparked significant interest, leading to a growing body of research focused on interpreting adversarial attacks (Han et al, 2023) and devising defense and detection mechanisms (Khamaiseh et al, 2022). Various proposed methods include augmenting input images to enhance robustness against adversarial attacks (Frosio and Kautz, 2023), mapping adversarial images back to the clean distribution (Li et al, 2023), and using vector quantization (Dong and Mao, 2023).…”

DeepCNN: A Dual Approach to Fault Localization and Repair in Convolutional Neural Networks

“…These bugs, often termed as "model bugs" in the context of CNNs, can manifest in various forms, including structural bugs and training bugs [5]. Not surprisingly, the implications of such bugs can be severe, ranging from financial losses to amplified safety concerns [6], [7].…”