Alireza Izaddoost
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Alireza Izaddoost has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Shahram Shah Heydari
Faculty of Business and IT, University of Ontario Institute of Technology, Oshawa, ON L1G 0C5, Canada
2 shared publications - Amlan Chatterjee
California State University Domínguez Hills
1 shared publication - Bhrigu Celly
Department of Computer Technology Information System, Merced College, Merced, CA 95348, USA
1 shared publication - Jennifer Kawakami
Department of Computer Science, California State University-Dominguez Hills (CSUDH), Carson, CA 90747, USA
1 shared publication - Nazila Safari Yazd
RISE Research Institutes of Sweden
1 shared publication - Patrice Mégret
University of Mons, Mons, Belgium
1 shared publication
Latest Publications
- Open Access
- 15 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (18) , 9157
Cloud-Based Distributed Deep Learning for Credit Card Fraud Detection: A Scalability and Data Partitioning Analysis
Recent advances in artificial intelligence have significantly improved credit card fraud detection, with deep learning emerging as an effective approach for learning complex transaction patterns. However, as financial transaction datasets continue to grow, training deep learning models on a single computing node becomes increasingly computationally expensive, motivating the adoption of cloud-based distributed learning. Because distributed deep learning partitions training data across multiple worker nodes, this study evaluates how different data partitioning strategies influence predictive performance, computational efficiency, and scalability. The proposed framework was evaluated under both independently and identically distributed (IID) and non-independent and identically distributed (non-IID) data partitioning strategies, including random, stratified, label skew, quantity skew, temporal skew, and amount skew, using single-node, three-worker, and seven-worker configurations. Experimental results demonstrate a maximum training speedup of 4.181× and a 76.084% reduction in average epoch training time while maintaining consistently high recall across all partitioning strategies; however, the -score decreased from 0.696 to 0.587 (approximately 16%), due to increased false-positive predictions. The evaluated partitioning strategies exhibited different trade-offs between predictive performance and computational efficiency. These findings demonstrate that the proposed framework provides a scalable solution for cloud-based credit card fraud detection while offering practical insights into the influence of data partitioning strategies on distributed deep learning performance.- 15 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (18) , 9157
Open AccessCloud-Based Distributed Deep Learning for Credit Card Fraud Detection: A Scalability and Data Partitioning Analysis
Recent advances in artificial intelligence have significantly improved credit card fraud detection, with deep learning emerging as an effective approach for learning complex transaction patterns. However, as financial transaction datasets continue to grow, training deep learning models on a single computing node becomes increasingly computationally expensive, motivating the adoption of cloud-based distributed learning. Because distributed deep learning partitions training data across multiple worker nodes, this study evaluates how different data partitioning strategies influence predictive performance, computational efficiency, and scalability. The proposed framework was evaluated under both independently and identically distributed (IID) and non-independent and identically distributed (non-IID) data partitioning strategies, including random, stratified, label skew, quantity skew, temporal skew, and amount skew, using single-node, three-worker, and seven-worker configurations. Experimental results demonstrate a maximum training speedup of 4.181× and a 76.084% reduction in average epoch training time while maintaining consistently high recall across all partitioning strategies; however, the -score decreased from 0.696 to 0.587 (approximately 16%), due to increased false-positive predictions. The evaluated partitioning strategies exhibited different trade-offs between predictive performance and computational efficiency. These findings demonstrate that the proposed framework provides a scalable solution for cloud-based credit card fraud detection while offering practical insights into the influence of data partitioning strategies on distributed deep learning performance. - Open Access
- 13 September 2022
- Published by MDPI AG in Future Internet
- Vol. 14 (9) , 263
Trade-offs between Risk and Operational Cost in SDN Failure Recovery Plan
We consider the problem of SDN flow optimization in the presence of a dynamic probabilistic link failures model. We introduce a metric for path risk, which can change dynamically as network conditions and failure probabilities change. As these probabilities change, the end-to-end path survivability probability may drop, i.e., its risk may rise. The main objective is to reroute at-risk end-to-end flows with the minimum number of flow operation so that a fast flow recovery is guaranteed. We provide various formulations for optimizing network risk versus operational costs and examine the trade-offs in flow recovery and the connections between operational cost, path risk, and path survival probability. We present our suboptimal dynamic flow restoration methods and evaluate their effectiveness against the Lagrangian relaxation approach. Our results show a significant improvement in operational cost against a shortest-path approach.
- 13 September 2022
- Published by MDPI AG in Future Internet
- Vol. 14 (9) , 263
Open AccessTrade-offs between Risk and Operational Cost in SDN Failure Recovery Plan
We consider the problem of SDN flow optimization in the presence of a dynamic probabilistic link failures model. We introduce a metric for path risk, which can change dynamically as network conditions and failure probabilities change. As these probabilities change, the end-to-end path survivability probability may drop, i.e., its risk may rise. The main objective is to reroute at-risk end-to-end flows with the minimum number of flow operation so that a fast flow recovery is guaranteed. We provide various formulations for optimizing network risk versus operational costs and examine the trade-offs in flow recovery and the connections between operational cost, path risk, and path survival probability. We present our suboptimal dynamic flow restoration methods and evaluate their effectiveness against the Lagrangian relaxation approach. Our results show a significant improvement in operational cost against a shortest-path approach.
- Open Access
- 12 December 2021
- Published by MDPI AG in Journal of Cybersecurity and Privacy
- Vol. 1 (4) , 767-792
Polymorphic Adversarial Cyberattacks Using WGAN
Intrusion Detection Systems (IDS) are essential components in preventing malicious traffic from penetrating networks and systems. Recently, these systems have been enhancing their detection ability using machine learning algorithms. This development also forces attackers to look for new methods for evading these advanced Intrusion Detection Systemss. Polymorphic attacks are among potential candidates that can bypass the pattern matching detection systems. To alleviate the danger of polymorphic attacks, the IDS must be trained with datasets that include these attacks. Generative Adversarial Network (GAN) is a method proven in generating adversarial data in the domain of multimedia processing, text, and voice, and can produce a high volume of test data that is indistinguishable from the original training data. In this paper, we propose a model to generate adversarial attacks using Wasserstein GAN (WGAN). The attack data synthesized using the proposed model can be used to train an IDS. To evaluate the trained IDS, we study several techniques for updating the attack feature profile for the generation of polymorphic data. Our results show that by continuously changing the attack profiles, defensive systems that use incremental learning will still be vulnerable to new attacks; meanwhile, their detection rates improve incrementally until the polymorphic attack exhausts its profile variables.
- 12 December 2021
- Published by MDPI AG in Journal of Cybersecurity and Privacy
- Vol. 1 (4) , 767-792
Open AccessPolymorphic Adversarial Cyberattacks Using WGAN
Intrusion Detection Systems (IDS) are essential components in preventing malicious traffic from penetrating networks and systems. Recently, these systems have been enhancing their detection ability using machine learning algorithms. This development also forces attackers to look for new methods for evading these advanced Intrusion Detection Systemss. Polymorphic attacks are among potential candidates that can bypass the pattern matching detection systems. To alleviate the danger of polymorphic attacks, the IDS must be trained with datasets that include these attacks. Generative Adversarial Network (GAN) is a method proven in generating adversarial data in the domain of multimedia processing, text, and voice, and can produce a high volume of test data that is indistinguishable from the original training data. In this paper, we propose a model to generate adversarial attacks using Wasserstein GAN (WGAN). The attack data synthesized using the proposed model can be used to train an IDS. To evaluate the trained IDS, we study several techniques for updating the attack feature profile for the generation of polymorphic data. Our results show that by continuously changing the attack profiles, defensive systems that use incremental learning will still be vulnerable to new attacks; meanwhile, their detection rates improve incrementally until the polymorphic attack exhausts its profile variables.
- Open Access
Effect of Peak Tracking Methods on FBG Calibration Derived by Factorial Design of Experiment
We present a calibration procedure for a humidity sensor made of a fiber Bragg grating covered by a polyimide layer. FBGs being intrinsically sensitive to temperature and strain, the calibration should tackle three variables, and, therefore, consists of a three-variable, two-level factorial design tailored to assess the three main sensitivities, as well as the five cross-sensitivities. FBG sensing information is encoded in the reflection spectrum from which the Bragg wavelength should be extracted. We tested six classical peak tracking methods on the results of the factorial design of the experiment applied to a homemade FBG humidity sensor. We used Python programming to compute, from the raw spectral data with six typical peak search algorithms, the temperature, strain and humidity sensitivities, as well as the cross-sensitivities, and showed that results are consistent for all algorithms, provided that the points selected to make the computation are correctly chosen. The best results for this particular sensor are obtained with a 3 threshold, whatever the peak search method used, and allow to compute the effective humidity sensitivity taking into account the combined effect of temperature and strain. The calibration procedure presented here is nevertheless generic and can thus be adapted to other sensors.Open AccessEffect of Peak Tracking Methods on FBG Calibration Derived by Factorial Design of Experiment
We present a calibration procedure for a humidity sensor made of a fiber Bragg grating covered by a polyimide layer. FBGs being intrinsically sensitive to temperature and strain, the calibration should tackle three variables, and, therefore, consists of a three-variable, two-level factorial design tailored to assess the three main sensitivities, as well as the five cross-sensitivities. FBG sensing information is encoded in the reflection spectrum from which the Bragg wavelength should be extracted. We tested six classical peak tracking methods on the results of the factorial design of the experiment applied to a homemade FBG humidity sensor. We used Python programming to compute, from the raw spectral data with six typical peak search algorithms, the temperature, strain and humidity sensitivities, as well as the cross-sensitivities, and showed that results are consistent for all algorithms, provided that the points selected to make the computation are correctly chosen. The best results for this particular sensor are obtained with a 3 threshold, whatever the peak search method used, and allow to compute the effective humidity sensitivity taking into account the combined effect of temperature and strain. The calibration procedure presented here is nevertheless generic and can thus be adapted to other sensors.