@inproceedings{guerzizDomainAdversarialNeural2026,

title = {Domain Adversarial Neural Networks with Adversarial Robustness Evaluation for Intrusion Detection Systems},

author = {I. Guerziz and T. Falk and L. B. Le and Z. A. E. Houda},

editor = {Adi K. and Nguena Timo O. and Boulahia-Cuppens N. and Espes D. and Stakhanova N. and Omar M.},

url = {https://www.scopus.com/pages/publications/105046102448?origin=resultslist},

doi = {10.1007/978-3-032-20732-6_10},

isbn = {03029743 (ISSN); 978-303220731-9 (ISBN)},

year  = {2026},

date = {2026-01-01},

booktitle = {Lect. Notes Comput. Sci.},

volume = {16295 LNCS},

pages = {153–165},

publisher = {Springer Science and Business Media Deutschland GmbH},

abstract = {Modern Network Intrusion Detection Systems (NIDS) face the dual challenge of maintaining performance across diverse network environments while resisting adversarial manipulations. This paper investigates the intersection of domain adaptation and adversarial robustness in NIDS, a topic that has not been extensively studied. We implement a Domain-Adversarial Neural Network (DANN) with dynamic gradient reversal to adapt models from NSL-KDD to UNSW-NB15. To evaluate security, we assess the model under Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD) attacks. Additionally, we introduce FGSM-based adversarial training to enhance robustness. Our results show that while domain adaptation improves cross-domain detection, it also increases susceptibility to adversarial attacks. Incorporating adversarial training mitigates this vulnerability, improving resilience without compromising performance on clean data. These findings provide key insights for designing adaptive and secure intrusion detection systems. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.},

note = {Journal Abbreviation: Lect. Notes Comput. Sci.},

keywords = {Adversarial Attacks, Adversarial neural network, Adversarial Neural Networks, Attack Resilience, Attack resiliences, Computer crime, Domain adaptation, Fast gradient sign method, FGSM, Gradient-descent, Intrusion Detection, Intrusion Detection Systems, Network intrusion, Network intrusion detection systems, Network security, Neural networks, Neural-networks, PGD, Projected gradient, Projected gradient descent},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{zhangSelectiveShiftPersonalized2025,

title = {Selective Shift: Towards Personalized Domain Adaptation in Multi-Agent Collaborative Perception},

author = {H. Zhang and Y. Xu and Y. Tian and Y. Li and T. H. Falk and F. -Y. Wang},

url = {https://www.scopus.com/pages/publications/105024076483?origin=resultslist},

doi = {10.1145/3746027.3754723},

isbn = {979-840072035-2 (ISBN)},

year  = {2025},

date = {2025-01-01},

booktitle = {MM - Proc. ACM Int. Conf. Multimedia, Co-Located with MM},

pages = {886–895},

publisher = {Association for Computing Machinery, Inc},

abstract = {Given the scarcity of real data and the time-intensive nature of labeling, current multi-agent perception models often rely on simulated sensor data for training and validation. However, perception performance deteriorates significantly due to domain gap between simulated and real data. Existing adaptation methods focus on domain-generalized feature extraction while neglecting multi-agent shift uncertainty and relational semantic loss. To address this issue, we propose a Selective Shift Domain Adaptation method in multi-agent collaborative perception, called SSDA. SSDA incorporates two essential components: the frequency-decoupled feature shift adjustment (FSA) and the entropy-driven staged adaptive alignment (SAA). To mitigate the relational semantic loss, the FSA is proposed to simplify the representation of correlation features and remove redundant information from the source domain, thereby mitigating interference for domain adversarial scenarios. To tackle the shift uncertainty, the SAA is designed to achieve adaptive alignment from global to local guided by information entropy, which dynamically adjusts weights for samples according to their level of uncertainty. The results demonstrate that the SSDA is significantly superior to the SOTA, achieving up to 7.35% improvements on AP@0.7. © 2025 ACM.},

note = {Journal Abbreviation: MM - Proc. ACM Int. Conf. Multimedia, Co-Located with MM},

keywords = {3D object, 3D object detection, Adaptation methods, Adaptive alignment, Collaborative perception, Domain adaptation, Entropy, Intelligent agents, Multi agent, Multi agent systems, Object detection, Object recognition, Objects detection, Relational semantics, Semantics, uncertainty},

pubstate = {published},

tppubtype = {inproceedings}

}