Vahid Heydari
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
Research Subjects
Vahid Heydari has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
Kofi Nyarko
4 shared publications- Kofi Nyarko
Morgan State University, USA
4 shared publications Seong-Moo Yoo
Department of Electrical and Computer Engineering, The University of Alabama in Huntsville, Huntsville, AL, USA
3 shared publications- Seong-Moo Yoo
University of Alabama in Huntsville
3 shared publications John Stranahan
Rowan University
2 shared publications- Sun-Il Kim
Department of Computer Science, North Central College, Naperville, IL, USA
2 shared publications
Latest Publications
- Open Access
Shortcut learning and identifier/composition stress testing in frame-level CAN intrusion detection
- Open Access
- 21 March 2026
- Published by The Colloquium for Information Systems Security Education in Journal of The Colloquium for Information Systems Security Education
- Vol. 13 (1) , 11
Integrating Artificial Intelligence into Undergraduate Cybersecurity Education
This paper introduces an undergraduate course, Artificial Intelligence Applications in Cybersecurity, designed to equip students with Artificial Intelligence (AI) and Machine Learning (ML) skills to address modern cyber threats. The curriculum integrates supervised and unsupervised learning, deep learning, explainable AI (XAI), adversarial ML, and ethical considerations. Using accessible tools (Python, Google Colab) and real-world datasets (e.g., NSL-KDD, CICIDS2017, malware corpora), students complete phased projects progressing from classical ML baselines to deep learning with interpretability (SHAP/LIME) and robustness against adversarial attacks (FGSM/PGD with mitigation). The course aligns with data science and cybersecurity workforce frameworks, emphasizing reproducibility, communication, and responsible AI practices.
- 21 March 2026
- Published by The Colloquium for Information Systems Security Education in Journal of The Colloquium for Information Systems Security Education
- Vol. 13 (1) , 11
Open AccessIntegrating Artificial Intelligence into Undergraduate Cybersecurity Education
This paper introduces an undergraduate course, Artificial Intelligence Applications in Cybersecurity, designed to equip students with Artificial Intelligence (AI) and Machine Learning (ML) skills to address modern cyber threats. The curriculum integrates supervised and unsupervised learning, deep learning, explainable AI (XAI), adversarial ML, and ethical considerations. Using accessible tools (Python, Google Colab) and real-world datasets (e.g., NSL-KDD, CICIDS2017, malware corpora), students complete phased projects progressing from classical ML baselines to deep learning with interpretability (SHAP/LIME) and robustness against adversarial attacks (FGSM/PGD with mitigation). The course aligns with data science and cybersecurity workforce frameworks, emphasizing reproducibility, communication, and responsible AI practices.
- Open Access
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 17585-17609
Alarm-Budgeted Event-Level Evaluation Of Ics Anomaly Detection: Lessons From Swat And Wadi
Industrial control systems (ICS) operate critical infrastructure; however, anomaly detectors for these environments are typically reported using unconstrained, threshold-free metrics. Operators have a limited capacity during deployment. A detector that appears strong under such metrics can be unusable if it generates too many alarm events. This study examines budgeted, event-level evaluations of industrial anomaly detection. Each detector was calibrated on the nominal stream to satisfy the target alarm event rate (events/h) with a duty-cycle guardrail. An attacked stream is then evaluated, reporting the event-level missed-incident rate and detection delay using attack-interval uncertainty. Six representative unsupervised and weakly supervised detectors are compared on two water-sector benchmarks (a water-treatment testbed and a water-distribution testbed). Under matched alert budgets, no detector dominated consistently: on SWaT, Miss rates ranged from 0.62 to 0.97 at B = 0.5, while on WADI some methods achieved near-zero Miss but incurred over 10× the intended alert workload under attack. Accordingly, rankings that appear clear under unconstrained metrics can be reversed once workload constraints are enforced. The study further provides Pareto trade–off analysis, a stakeholder–weighted decision layer for selecting actionable operating points, and sensitivity checks for common alarm post–processing choices. These analyses caution against over–interpreting marginal performance differences, particularly in small–attack or intermediate–budget regimes.
- 28 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 17585-17609
Open AccessAlarm-Budgeted Event-Level Evaluation Of Ics Anomaly Detection: Lessons From Swat And Wadi
Industrial control systems (ICS) operate critical infrastructure; however, anomaly detectors for these environments are typically reported using unconstrained, threshold-free metrics. Operators have a limited capacity during deployment. A detector that appears strong under such metrics can be unusable if it generates too many alarm events. This study examines budgeted, event-level evaluations of industrial anomaly detection. Each detector was calibrated on the nominal stream to satisfy the target alarm event rate (events/h) with a duty-cycle guardrail. An attacked stream is then evaluated, reporting the event-level missed-incident rate and detection delay using attack-interval uncertainty. Six representative unsupervised and weakly supervised detectors are compared on two water-sector benchmarks (a water-treatment testbed and a water-distribution testbed). Under matched alert budgets, no detector dominated consistently: on SWaT, Miss rates ranged from 0.62 to 0.97 at B = 0.5, while on WADI some methods achieved near-zero Miss but incurred over 10× the intended alert workload under attack. Accordingly, rankings that appear clear under unconstrained metrics can be reversed once workload constraints are enforced. The study further provides Pareto trade–off analysis, a stakeholder–weighted decision layer for selecting actionable operating points, and sensitivity checks for common alarm post–processing choices. These analyses caution against over–interpreting marginal performance differences, particularly in small–attack or intermediate–budget regimes.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 134454-134478
RIPPLE: Differential Testing of Exact-Value Restoration Boundaries in Privacy-Preserving LLM Pipelines
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 134454-134478
Open AccessRIPPLE: Differential Testing of Exact-Value Restoration Boundaries in Privacy-Preserving LLM Pipelines
- Open Access
- 15 August 2025
- Published by MDPI AG in Electronics
- Vol. 14 (16) , 3249
Enhancing Adversarial Robustness in Network Intrusion Detection: A Novel Adversarially Trained Neural Network Approach
Machine learning (ML) has greatly improved intrusion detection in enterprise networks. However, ML models remain vulnerable to adversarial attacks, where small input changes cause misclassification. This study evaluates the robustness of a Random Forest (RF), a standard neural network (NN), and a Transformer-based Network Intrusion Detection System (NIDS). It also introduces ADV_NN, an adversarially trained neural network designed to improve resilience. Model performance is tested using the UNSW-NB15 dataset under both clean and adversarial conditions. The attack types include Projected Gradient Descent (PGD), Fast Gradient Sign Method (FGSM), and Black-Box transfer attacks. The proposed ADV_NN achieves 86.04% accuracy on clean data. It maintains over 80% accuracy under PGD and FGSM attacks, and exceeds 85% under Black-Box attacks at . In contrast, the RF, NN, and Transformer-based models suffer significant degradation under adversarial perturbations. These results highlight the need to incorporate adversarial defenses into ML-based NIDS for secure deployment in real-world environments.- 15 August 2025
- Published by MDPI AG in Electronics
- Vol. 14 (16) , 3249
Open AccessEnhancing Adversarial Robustness in Network Intrusion Detection: A Novel Adversarially Trained Neural Network Approach
Machine learning (ML) has greatly improved intrusion detection in enterprise networks. However, ML models remain vulnerable to adversarial attacks, where small input changes cause misclassification. This study evaluates the robustness of a Random Forest (RF), a standard neural network (NN), and a Transformer-based Network Intrusion Detection System (NIDS). It also introduces ADV_NN, an adversarially trained neural network designed to improve resilience. Model performance is tested using the UNSW-NB15 dataset under both clean and adversarial conditions. The attack types include Projected Gradient Descent (PGD), Fast Gradient Sign Method (FGSM), and Black-Box transfer attacks. The proposed ADV_NN achieves 86.04% accuracy on clean data. It maintains over 80% accuracy under PGD and FGSM attacks, and exceeds 85% under Black-Box attacks at . In contrast, the RF, NN, and Transformer-based models suffer significant degradation under adversarial perturbations. These results highlight the need to incorporate adversarial defenses into ML-based NIDS for secure deployment in real-world environments. - Open Access
- 20 April 2025
- Published by The Colloquium for Information Systems Security Education in Journal of The Colloquium for Information Systems Security Education
- Vol. 12 (1) , 8
Empowering the Next Generation: A Strategic Roadmap for AI in Cybersecurity Education
The integration of artificial intelligence (AI) into cybersecurity is revolutionizing how institutions address increasingly complex cyber threats. As the demand for expertise in both AI and cybersecurity grows, Historically Black Colleges and Universities (HBCUs) have a unique opportunity to develop programs that equip students with the necessary skills to meet these evolving challenges. This paper presents a strategic roadmap for developing AI in Cybersecurity programs at HBCUs, emphasizing interdisciplinary collaboration, hands-on learning, adversarial defense, explainability, ethical leadership, and diversity. To strengthen the roadmap, this paper incorporates real-world case studies from existing AI-cybersecurity programs and proposes strategies to overcome key challenges such as faculty expertise gaps, funding limitations, and resource scalability. Additionally, a framework for evaluating program effectiveness is introduced, offering measurable metrics such as student outcomes, industry collaboration, and curriculum adaptability. By implementing this roadmap, HBCUs can establish sustainable AI in Cybersecurity programs that align with industry needs while fostering leadership and innovation in the cybersecurity workforce.
- 20 April 2025
- Published by The Colloquium for Information Systems Security Education in Journal of The Colloquium for Information Systems Security Education
- Vol. 12 (1) , 8
Open AccessEmpowering the Next Generation: A Strategic Roadmap for AI in Cybersecurity Education
The integration of artificial intelligence (AI) into cybersecurity is revolutionizing how institutions address increasingly complex cyber threats. As the demand for expertise in both AI and cybersecurity grows, Historically Black Colleges and Universities (HBCUs) have a unique opportunity to develop programs that equip students with the necessary skills to meet these evolving challenges. This paper presents a strategic roadmap for developing AI in Cybersecurity programs at HBCUs, emphasizing interdisciplinary collaboration, hands-on learning, adversarial defense, explainability, ethical leadership, and diversity. To strengthen the roadmap, this paper incorporates real-world case studies from existing AI-cybersecurity programs and proposes strategies to overcome key challenges such as faculty expertise gaps, funding limitations, and resource scalability. Additionally, a framework for evaluating program effectiveness is introduced, offering measurable metrics such as student outcomes, industry collaboration, and curriculum adaptability. By implementing this roadmap, HBCUs can establish sustainable AI in Cybersecurity programs that align with industry needs while fostering leadership and innovation in the cybersecurity workforce.
- 13 April 2021
- Published by Springer Nature
- p. 736-755
SCADA Testbed Implementation, Attacks, and Security Solutions
- 13 April 2021
- Published by Springer Nature
- p. 736-755
SCADA Testbed Implementation, Attacks, and Security Solutions
- 9 September 2020
- Published by Springer Nature
- p. 139-153
Using Least-Significant Bit and Random Pixel Encoding with Encryption for Image Steganography
- 9 September 2020
- Published by Springer Nature
- p. 139-153
Using Least-Significant Bit and Random Pixel Encoding with Encryption for Image Steganography