Mohammadreza Mohammadi

Machine learning and deep learning are transformative forces reshaping our networks, industries, services, and ways of life. However, the susceptibility of these intelligent systems to adversarial attacks remains a significant issue. On the one hand, recent studies have demonstrated the potential transferability of adversarial attacks across diverse models. On the other hand, existing defense mechanisms are vulnerable to advanced attacks or are often limited to certain attack types. This study proposes a random deep feature selection approach to mitigate such transferability and improve the robustness of models against adversarial manipulations. Our approach is designed to strengthen deep models against poisoning (e.g., label flipping) and exploratory (e.g., DeepFool, BIM, FGSM, I-FGSM, L-BFGS, C&W, JSMA, and PGD) attacks that are applied in both the training and testing stages, and Transfer Learning-Based Adversarial Attacks. We consider scenarios involving perfect and semi-knowledgeable attackers. The performance of our approach is evaluated through extensive experiments on the renowned UNSW-NB15 dataset, including both real-world and synthetic data, covering a wide range of modern attack behaviors and benign activities. The results indicate that our approach boosts the effectiveness of the target network to over 80% against labelflipping poisoning attacks and over 60% against all major types of exploratory attacks.

Anomaly detection is essential for ensuring the safe and efficient operation of industrial systems like smart grids. Smart grid stations handle sensitive data and are often hesitant to share it with third-party servers for centralized anomaly detection. Federated Learning (FL) offers a viable solution to this issue by enhancing anomaly detection in smart grids without compromising data privacy. We present a method for developing an unsupervised anomaly detection system using FL applied to a synthetic dataset that mimics a real-world smart grid system's behavior. We focus on utilizing FL's long short-term memory autoencoder in short, LSTM-AE for anomaly detection. However, there are concerns about potential privacy breaches in the FL system. Hence, to address this issue, we propose to integrate differential privacy (DP) with FL for anomaly detection by adding artificial noise to parameters at the client side before aggregation. This method ensures data privacy while maintaining the convergence of federated learning algorithms. Moreover, this research determines the optimal privacy level to balance noise scale and model accuracy. Our findings suggest a criterion for selecting the right privacy budget of DP based on the requirement of the system to provide good level of privacy in the system while maintaining the f1-score of FL-based anomaly detection system greater than 90%.

In the past decades, the rise of artificial intelligence has given us the capabilities to solve the most challenging problems in our day-to-day lives, such as cancer prediction and autonomous navigation. However, these applications might not be reliable if not secured against adversarial attacks. In addition, recent works demonstrated that some adversarial examples are transferable across different models. Therefore, it is crucial to avoid such transferability via robust models that resist adversarial manipulations. In this paper, we propose a feature randomization-based approach that resists eight adversarial attacks targeting deep learning models in the testing phase. Our novel approach consists of changing the training strategy in the target network classifier and selecting random feature samples. We consider the attacker with a Limited-Knowledge and Semi-Knowledge conditions to undertake the most prevalent types of adversarial attacks. We evaluate the robustness of our approach using the well-known UNSW-NB15 datasets that include realistic and synthetic attacks. Afterward, we demonstrate that our strategy outperforms the existing state-of-the-art approach, such as the Most Powerful Attack, which consists of fine-tuning the network model against specific adversarial attacks. Further, we demonstrate the practicality of our approach using the VIPPrint dataset through a comprehensive set of experiments. Finally, our experimental results show that our methodology can secure the target network and resists adversarial attack transferability by over 60%.

Anomaly detection is critical in industrial systems such as smart grid systems to guarantee their safe and effective operation. The smart grid stations contain sensitive data, and they are concerned about sharing it with a third-party server to establish a centralized anomaly detection system. Federated Learning (FL) is a feasible solution to these problems for enhancing anomaly detection in smart grid systems. This study describes a method for developing an unsupervised anomaly detection based on FL system using a synthetic dataset based on real-world grid system behavior. The paper investigates the usage of FL's long short-term memory autoencoder (LSTM-AE) for anomaly detection. For more accurate identification, this research explores the performance of integrating LSTM-AE with one-class support vector machine (OC-SVM) and isolation forest (IF) and compares their results with a threshold-based anomaly detection approach. Moreover, an approach is described for generating synthetic anomalies with different levels of difficulty to evaluate the robustness of the anomaly detection FL model. The FL models results are compared with the centralized version of the models as a baseline and the results show that FL models outperformed the centralized approach by detecting higher outlier data by achieving 99% F1-Score.

In the past few years, Convolutional Neural Networks (CNN) have demonstrated promising performance in various real-world cybersecurity applications, such as network and multimedia security. However, the underlying fragility of CNN structures poses major security problems, making them inappropriate for use in security-oriented applications including such computer networks. Protecting these architectures from adversarial attacks necessitates using security-wise architectures that are challenging to attack. In this study, we present a novel architecture based on an ensemble classifier that combines the enhanced security of 1-Class classification (known as 1C) with the high performance of conventional 2-Class classification (known as 2C) in the absence of attacks.Our architecture is referred to as the 1.5-Class (SPRITZ-1.5C) classifier and constructed using a final dense classifier, one 2C classifier (i.e., CNNs), and two parallel 1C classifiers (i.e., auto-encoders). In our experiments, we evaluated the robustness of our proposed architecture by considering eight possible adversarial attacks in various scenarios. We performed these attacks on the 2C and SPRITZ-1.5C architectures separately. The experimental results of our study showed that the Attack Success Rate (ASR) of the I-FGSM attack against a 2C classifier trained with the N-BaIoT dataset is 0.9900. In contrast, the ASR is 0.0000 for the SPRITZ-1.5C classifier.