Slide
Centre Interdisciplinaire
de Recherche et d’Innovation
en Cybersécurité et Société
de Recherche et d’Innovation
en Cybersécurité et Société
1.
Ameyoud, S. Mohamed; Allili, M. Saïd
Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training Article de journal
Dans: Journal of Information Security and Applications, vol. 99, 2026, ISSN: 22142134 (ISSN).
Résumé | Liens | BibTeX | Étiquettes: Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers
@article{mohamed_ameyoud_multi-modal_2026,
title = {Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training},
author = {S. Mohamed Ameyoud and M. Saïd Allili},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-105031186108&doi=10.1016%2Fj.jisa.2026.104429&partnerID=40&md5=2425da4ab40f9043ba4e67d223a1bdd9},
doi = {10.1016/j.jisa.2026.104429},
issn = {22142134 (ISSN)},
year = {2026},
date = {2026-01-01},
journal = {Journal of Information Security and Applications},
volume = {99},
abstract = {The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.},
keywords = {Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers},
pubstate = {published},
tppubtype = {article}
}
The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.
2.
Laamari, A.; Moudoud, H.; Houda, Z. A. El
Lightweight LLM Adaptation for Intrusion Detection via Token-Efficient Flow Representation Article d'actes
Dans: IEEE Conf. Artif. Intell., CAI, p. 2122–2127, Institute of Electrical and Electronics Engineers Inc., 2026, ISBN: 979-833156039-3 (ISBN), (Journal Abbreviation: IEEE Conf. Artif. Intell., CAI).
Résumé | Liens | BibTeX | Étiquettes: Classification (of information), Data flow analysis, Decoder-only large language model, Decoder-only LLMs, decoding, Flow classification, Intrusion Detection, Intrusion-Detection, Language model, Large language model, LLMs, LoRA, Low-rank adaptation, Network Flow Classification, Network intrusion, Network security, Networks flows, Qwen2.5, Signal encoding, T5-Small, Token-oriented object notation, Tokenization, TOON
@inproceedings{laamariLightweightLLMAdaptation2026,
title = {Lightweight LLM Adaptation for Intrusion Detection via Token-Efficient Flow Representation},
author = {A. Laamari and H. Moudoud and Z. A. El Houda},
url = {https://www.scopus.com/pages/publications/105042133342?origin=resultslist},
doi = {10.1109/CAI68641.2026.11536533},
isbn = {979-833156039-3 (ISBN)},
year = {2026},
date = {2026-01-01},
booktitle = {IEEE Conf. Artif. Intell., CAI},
pages = {2122–2127},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {Large language models (LLMs) are emerging as a promising approach for intrusion detection using structured network flow data. However, their practical deployment is constrained by context window limitations and the excessive token overhead introduced by conventional tabular serialization formats such as JSON. Verbose data representations inflate sequence lengths, often exceeding model input limits and causing feature truncation. Additionally, it remains unclear which LLM architecture is more suitable for structured intrusion detection tasks under limited training resources. To tackle this issue, we propose a novel representation-aware intrusion detection framework based on Token-Oriented Object Notation (TOON), a compact serialization format that maximizes token efficiency while preserving schema structure. Also, we integrate a Low-Rank Adaptation (LoRA) scheme to enable parameter-efficient fine-tuning. Finally, we evaluate the proposed framework as an encoder-decoder model (T5-Small) with a decoder-only model (Qwen2.5) on three benchmark datasets, including NSL-KDD, UNSW-NB15, and CIC-IDS2018 for binary and multi-class classification scenarios. The results show that our tokenizer-aligned, representation-aware preprocessing combined with lightweight encoder-decoder adaptation provides a practical and resource-efficient foundation for LLM-based intrusion detection. © 2026 IEEE.},
note = {Journal Abbreviation: IEEE Conf. Artif. Intell., CAI},
keywords = {Classification (of information), Data flow analysis, Decoder-only large language model, Decoder-only LLMs, decoding, Flow classification, Intrusion Detection, Intrusion-Detection, Language model, Large language model, LLMs, LoRA, Low-rank adaptation, Network Flow Classification, Network intrusion, Network security, Networks flows, Qwen2.5, Signal encoding, T5-Small, Token-oriented object notation, Tokenization, TOON},
pubstate = {published},
tppubtype = {inproceedings}
}
Large language models (LLMs) are emerging as a promising approach for intrusion detection using structured network flow data. However, their practical deployment is constrained by context window limitations and the excessive token overhead introduced by conventional tabular serialization formats such as JSON. Verbose data representations inflate sequence lengths, often exceeding model input limits and causing feature truncation. Additionally, it remains unclear which LLM architecture is more suitable for structured intrusion detection tasks under limited training resources. To tackle this issue, we propose a novel representation-aware intrusion detection framework based on Token-Oriented Object Notation (TOON), a compact serialization format that maximizes token efficiency while preserving schema structure. Also, we integrate a Low-Rank Adaptation (LoRA) scheme to enable parameter-efficient fine-tuning. Finally, we evaluate the proposed framework as an encoder-decoder model (T5-Small) with a decoder-only model (Qwen2.5) on three benchmark datasets, including NSL-KDD, UNSW-NB15, and CIC-IDS2018 for binary and multi-class classification scenarios. The results show that our tokenizer-aligned, representation-aware preprocessing combined with lightweight encoder-decoder adaptation provides a practical and resource-efficient foundation for LLM-based intrusion detection. © 2026 IEEE.
3.
Ameyoud, S. Mohamed; Allili, M. Saïd
Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training Article de journal
Dans: Journal of Information Security and Applications, vol. 99, 2026, ISSN: 22142134 (ISSN).
Résumé | Liens | BibTeX | Étiquettes: Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers
@article{mohamedameyoudMultimodalMalwareClassification2026,
title = {Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training},
author = {S. Mohamed Ameyoud and M. Saïd Allili},
url = {https://www.scopus.com/pages/publications/105031186108?origin=resultslist},
doi = {10.1016/j.jisa.2026.104429},
issn = {22142134 (ISSN)},
year = {2026},
date = {2026-01-01},
journal = {Journal of Information Security and Applications},
volume = {99},
publisher = {Elsevier Ltd},
abstract = {The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.},
keywords = {Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers},
pubstate = {published},
tppubtype = {article}
}
The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.
4.
Nouboukpo, A.; Allili, M. S.
Semi-supervised flow-augmented Gaussian mixture VAEs with weighted contrastive learning for out-of-distribution detection Article de journal
Dans: Knowledge-Based Systems, vol. 347, 2026, ISSN: 09507051 (ISSN).
Résumé | Liens | BibTeX | Étiquettes: Auto encoders, Benchmarking, Classification (of information), Clusterings, Contrastive Learning, Flow based, Flow-based GMM clustering, Gaussian distribution, Gaussian-mixtures, Learning algorithms, Learning systems, Out-of-distribution detection, Semantics, Semi-supervised, Semi-supervised learning, uncertainty, Variational autoencoder, Variational autoencoders (VAEs)
@article{nouboukpoSemisupervisedFlowaugmentedGaussian2026,
title = {Semi-supervised flow-augmented Gaussian mixture VAEs with weighted contrastive learning for out-of-distribution detection},
author = {A. Nouboukpo and M. S. Allili},
url = {https://www.scopus.com/pages/publications/105040505017?origin=resultslist},
doi = {10.1016/j.knosys.2026.116312},
issn = {09507051 (ISSN)},
year = {2026},
date = {2026-01-01},
journal = {Knowledge-Based Systems},
volume = {347},
publisher = {Elsevier B.V.},
abstract = {Out-of-distribution (OOD) detection is essential for the safe deployment of machine learning systems, particularly in high-stakes domains where identifying inputs that deviate from the in-distribution (ID) is critical. However, in real-world scenarios, true OOD data are typically unavailable at training time, and only auxiliary datasets acting as imperfect proxies can be used. This makes fully supervised approaches impractical and potentially biased toward specific anomaly types. To address this limitation, we propose FLoW-ssGMVAE, a semi-supervised generative–discriminative framework that leverages limited proxy anomaly supervision together with abundant unlabeled data for robust OOD detection. Our model integrates a flow-augmented Gaussian Mixture VAE, enabling flexible latent modeling that captures nonlinear and multimodal intra-class distributions. Unlike standard VAEs with unimodal Gaussian priors, it uses class-conditional normalizing flows to better represent complex ID and OOD variability. To further refine the latent space and strengthen ID/OOD separation, our model integrates a weighted contrastive learning objective guided by sample-wise attention scores derived from likelihood uncertainty. This mechanism emphasizes ambiguous or hard-to-classify instances, reinforcing semantic boundaries and reducing false detections. By combining expressive generative modeling with uncertainty-aware discriminative training, our method constructs a structured and interpretable latent space that reliably identifies OOD samples under limited supervision. Experiments on standard benchmarks validate the effectiveness and scalability of our approach, demonstrating state-of-the-art performance with reasonable computational cost. Our implementation will be available at: FLoW-ssGMVAE. © 2026},
keywords = {Auto encoders, Benchmarking, Classification (of information), Clusterings, Contrastive Learning, Flow based, Flow-based GMM clustering, Gaussian distribution, Gaussian-mixtures, Learning algorithms, Learning systems, Out-of-distribution detection, Semantics, Semi-supervised, Semi-supervised learning, uncertainty, Variational autoencoder, Variational autoencoders (VAEs)},
pubstate = {published},
tppubtype = {article}
}
Out-of-distribution (OOD) detection is essential for the safe deployment of machine learning systems, particularly in high-stakes domains where identifying inputs that deviate from the in-distribution (ID) is critical. However, in real-world scenarios, true OOD data are typically unavailable at training time, and only auxiliary datasets acting as imperfect proxies can be used. This makes fully supervised approaches impractical and potentially biased toward specific anomaly types. To address this limitation, we propose FLoW-ssGMVAE, a semi-supervised generative–discriminative framework that leverages limited proxy anomaly supervision together with abundant unlabeled data for robust OOD detection. Our model integrates a flow-augmented Gaussian Mixture VAE, enabling flexible latent modeling that captures nonlinear and multimodal intra-class distributions. Unlike standard VAEs with unimodal Gaussian priors, it uses class-conditional normalizing flows to better represent complex ID and OOD variability. To further refine the latent space and strengthen ID/OOD separation, our model integrates a weighted contrastive learning objective guided by sample-wise attention scores derived from likelihood uncertainty. This mechanism emphasizes ambiguous or hard-to-classify instances, reinforcing semantic boundaries and reducing false detections. By combining expressive generative modeling with uncertainty-aware discriminative training, our method constructs a structured and interpretable latent space that reliably identifies OOD samples under limited supervision. Experiments on standard benchmarks validate the effectiveness and scalability of our approach, demonstrating state-of-the-art performance with reasonable computational cost. Our implementation will be available at: FLoW-ssGMVAE. © 2026
5.
Sadfi, Y.; Amamou, H.; Davoust, A.; Avila, A. R.
Comparative Analysis of Machine Learning, LLMs, and RAG for Fake News Detection Article d'actes
Dans: IEEE Conf. Artif. Intell., CAI, p. 2128–2133, Institute of Electrical and Electronics Engineers Inc., 2026, ISBN: 979-833156039-3 (ISBN), (Journal Abbreviation: IEEE Conf. Artif. Intell., CAI).
Résumé | Liens | BibTeX | Étiquettes: Classification (of information), Comparative analyzes, Comprehensive assessment, decision making, Decision-making process, Fake detection, Generalisation, High-accuracy, Language model, Learning algorithms, Learning systems, Machine learning, Machine learning algorithms, Machine-learning, Network platforms, Performance
@inproceedings{sadfiComparativeAnalysisMachine2026,
title = {Comparative Analysis of Machine Learning, LLMs, and RAG for Fake News Detection},
author = {Y. Sadfi and H. Amamou and A. Davoust and A. R. Avila},
url = {https://www.scopus.com/pages/publications/105042025800?origin=resultslist},
doi = {10.1109/CAI68641.2026.11536343},
isbn = {979-833156039-3 (ISBN)},
year = {2026},
date = {2026-01-01},
booktitle = {IEEE Conf. Artif. Intell., CAI},
pages = {2128–2133},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {The rapid spread of misinformation on social network platforms poses a serious threat to the decision-making processes in democratic societies. To mitigate the problem and cope with new misinformation scenarios several datasets have been proposed, followed by the emergence of new methods for detecting fake news. While several comparative analyses have been proposed to benchmark these solutions, the ever-growing development of new methods requires up-to-date analysis. For instance, recent approaches based on retrieval-augmented generation (RAG) have yet to be assessed and compared to previous methods used for fake news detection. In this study, we aim to fill this gap by presenting a comprehensive assessment of classical machine learning (ML) algorithms, Transformer-based models, such as BERT, large language models (LLMs), and RAG-based classifiers. We adopt four traditional fake news datasets, including a Portuguese one. Results indicate that classical ML and BERT deliver strong performance on in-domain tasks, demonstrating efficiency and high accuracy. RAG-augmented LLMs, on the other hand, offer enhanced generalization and robustness in cross-domain and cross-dataset settings, but are outperformed by traditional approaches for in-domain evaluation. © 2026 IEEE.},
note = {Journal Abbreviation: IEEE Conf. Artif. Intell., CAI},
keywords = {Classification (of information), Comparative analyzes, Comprehensive assessment, decision making, Decision-making process, Fake detection, Generalisation, High-accuracy, Language model, Learning algorithms, Learning systems, Machine learning, Machine learning algorithms, Machine-learning, Network platforms, Performance},
pubstate = {published},
tppubtype = {inproceedings}
}
The rapid spread of misinformation on social network platforms poses a serious threat to the decision-making processes in democratic societies. To mitigate the problem and cope with new misinformation scenarios several datasets have been proposed, followed by the emergence of new methods for detecting fake news. While several comparative analyses have been proposed to benchmark these solutions, the ever-growing development of new methods requires up-to-date analysis. For instance, recent approaches based on retrieval-augmented generation (RAG) have yet to be assessed and compared to previous methods used for fake news detection. In this study, we aim to fill this gap by presenting a comprehensive assessment of classical machine learning (ML) algorithms, Transformer-based models, such as BERT, large language models (LLMs), and RAG-based classifiers. We adopt four traditional fake news datasets, including a Portuguese one. Results indicate that classical ML and BERT deliver strong performance on in-domain tasks, demonstrating efficiency and high accuracy. RAG-augmented LLMs, on the other hand, offer enhanced generalization and robustness in cross-domain and cross-dataset settings, but are outperformed by traditional approaches for in-domain evaluation. © 2026 IEEE.
6.
Ameyoud, S. Mohamed; Allili, M. Saïd
Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training Article de journal
Dans: Journal of Information Security and Applications, vol. 99, 2026, ISSN: 22142134 (ISSN).
Résumé | Liens | BibTeX | Étiquettes: Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers
@article{mohamedameyoudMultimodalMalwareClassification2026a,
title = {Multi-modal malware classification with hierarchical consistency and saliency-constrained adversarial training},
author = {S. Mohamed Ameyoud and M. Saïd Allili},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-105031186108&doi=10.1016%2Fj.jisa.2026.104429&partnerID=40&md5=2425da4ab40f9043ba4e67d223a1bdd9},
doi = {10.1016/j.jisa.2026.104429},
issn = {22142134 (ISSN)},
year = {2026},
date = {2026-01-01},
journal = {Journal of Information Security and Applications},
volume = {99},
abstract = {The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.},
keywords = {Adversarial training, Capability of detection, Classification (of information), Convolution, convolutional neural network, Convolutional neural networks, Detection system, Hierarchical consistency, Hierarchical systems, Malware, Malware classification, Malware classifications, Malware families, Malwares, Multi-modal, Multi-modal learning, Semantics, Vision transformer, Vision transformers},
pubstate = {published},
tppubtype = {article}
}
The increasing complexity of malware, including polymorphic, obfuscated, and adversarial variants, continues to outpace the capabilities of detection systems. Here, we introduce a robust multi-modal hierarchical framework that jointly leverages visual and code-level semantics to enhance malware family and type classification. Our architecture fuses convolutional and transformer-based encoders to extract complementary representations from raw malware binaries and decompiled control-flow functions, enabling a rich, cross-modal understanding of malicious behavior. The classification pipeline follows a two-stage hierarchical protocol, where the predicted malware type informs the family-level classification. This enforces ontological consistency between type and family prediction levels. To further bolster robustness against adversarial and obfuscated malware, we integrate a novel adversarial training strategy that generates plausible perturbations guided by attention distributions. Evaluation on multiple large-scale benchmarks including BODMAS, Malimg, Microsoft BIG 2015, and a curated set of from MalwareBazaar, demonstrate that our framework consistently outperforms state-of-the-art baselines, including ResNet, Swin Transformer, and MalBERTv2, across both malware type and family prediction tasks. Notably, our model exhibits outstanding generalization to unpacked, obfuscated, and previously unseen samples, with minimal performance degradation. It achieves accuracy gains of +3-6% over leading methods and exhibits superior resilience under adversarial threat models. These results highlight the effectiveness of hierarchical conditioning, adversarial robustness, and multi-modal fusion in tackling the evolving landscape of malware. The proposed framework thus offers a scalable and generalizable approach for next-generation malware classification in real-world cybersecurity environments. © 2026 Elsevier Ltd.
7.
Soltani, N.; Nejadshamsi, S.; Houda, Z. A. El; Khoury, R.; Costa, K. A. P.; Falk, T. H.; Avila, A. R.
Enhancing Network Intrusion Detection Systems: A Multi-Layer Ensemble Approach to Mitigate Adversarial Attacks Article d'actes
Dans: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern., p. 39–44, Institute of Electrical and Electronics Engineers Inc., 2025, ISBN: 1062922X (ISSN); 979-833153358-8 (ISBN), (Journal Abbreviation: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.).
Résumé | Liens | BibTeX | Étiquettes: Adversarial machine learning, Adversarial networks, Classification (of information), Computer crime, Fast gradient sign method, Fast Gradient Sign Method (FGSM), Generative adversarial network, Generative Adversarial Network (GAN), Generative adversarial networks, Generative model, Generative Models, Intrusion Detection, Intrusion-Detection, Learning algorithms, Learning systems, Machine-learning, Multi-layers, Network intrusion, Network intrusion detection systems, Network layers, Network security, Second layer, Stackings
@inproceedings{soltaniEnhancingNetworkIntrusion2025,
title = {Enhancing Network Intrusion Detection Systems: A Multi-Layer Ensemble Approach to Mitigate Adversarial Attacks},
author = {N. Soltani and S. Nejadshamsi and Z. A. El Houda and R. Khoury and K. A. P. Costa and T. H. Falk and A. R. Avila},
url = {https://www.scopus.com/pages/publications/105033159769?origin=resultslist},
doi = {10.1109/SMC58881.2025.11342479},
isbn = {1062922X (ISSN); 979-833153358-8 (ISBN)},
year = {2025},
date = {2025-01-01},
booktitle = {Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.},
pages = {39–44},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {Adversarial examples can represent a serious threat to machine learning (ML) algorithms. If used to manipulate the behaviour of ML-based Network Intrusion Detection Systems (NIDS), they can jeopardize network security. In this work, we aim to mitigate such risks by increasing the robustness of NIDS towards adversarial attacks. To that end, we explore two adversarial methods for generating malicious network traffic. The first method is based on Generative Adversarial Networks (GAN) and the second one is the Fast Gradient Sign Method (FGSM). The adversarial examples generated by these methods are then used to evaluate a novel multilayer defense mechanism, specifically designed to mitigate the vulnerability of ML-based NIDS. Our solution consists of one layer of stacking classifiers and a second layer based on an autoencoder. If the incoming network data are classified as benign by the first layer, the second layer is activated to ensure that the decision made by the stacking classifier is correct. We also incorporated adversarial training to further improve the robustness of our solution. Experiments on two datasets, namely UNSW-NB15 and NSL-KDD, demonstrate that the proposed approach increases resilience to adversarial attacks. © 2025 IEEE.},
note = {Journal Abbreviation: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.},
keywords = {Adversarial machine learning, Adversarial networks, Classification (of information), Computer crime, Fast gradient sign method, Fast Gradient Sign Method (FGSM), Generative adversarial network, Generative Adversarial Network (GAN), Generative adversarial networks, Generative model, Generative Models, Intrusion Detection, Intrusion-Detection, Learning algorithms, Learning systems, Machine-learning, Multi-layers, Network intrusion, Network intrusion detection systems, Network layers, Network security, Second layer, Stackings},
pubstate = {published},
tppubtype = {inproceedings}
}
Adversarial examples can represent a serious threat to machine learning (ML) algorithms. If used to manipulate the behaviour of ML-based Network Intrusion Detection Systems (NIDS), they can jeopardize network security. In this work, we aim to mitigate such risks by increasing the robustness of NIDS towards adversarial attacks. To that end, we explore two adversarial methods for generating malicious network traffic. The first method is based on Generative Adversarial Networks (GAN) and the second one is the Fast Gradient Sign Method (FGSM). The adversarial examples generated by these methods are then used to evaluate a novel multilayer defense mechanism, specifically designed to mitigate the vulnerability of ML-based NIDS. Our solution consists of one layer of stacking classifiers and a second layer based on an autoencoder. If the incoming network data are classified as benign by the first layer, the second layer is activated to ensure that the decision made by the stacking classifier is correct. We also incorporated adversarial training to further improve the robustness of our solution. Experiments on two datasets, namely UNSW-NB15 and NSL-KDD, demonstrate that the proposed approach increases resilience to adversarial attacks. © 2025 IEEE.
8.
Soultana, O. A.; Moudoud, H.
Adaptive Heterogeneous Ensemble Learning for Attack Detection in IoT Networks Article d'actes
Dans: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern., p. 27–32, Institute of Electrical and Electronics Engineers Inc., 2025, ISBN: 1062922X (ISSN); 979-833153358-8 (ISBN), (Journal Abbreviation: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.).
Résumé | Liens | BibTeX | Étiquettes: Attack detection, Classification (of information), Computational efficiency, Detection accuracy, Ensemble learning, Ensemble techniques, Heterogeneous ensembles, Internet of thing security, Internet of things, Intrusion Detection, Intrusion-Detection, IoT Security, Learning systems, Nearest neighbor search, Security vulnerabilities, Stackings, Support vector machines, Zero-day attack, Zero-day detection
@inproceedings{soultanaAdaptiveHeterogeneousEnsemble2025,
title = {Adaptive Heterogeneous Ensemble Learning for Attack Detection in IoT Networks},
author = {O. A. Soultana and H. Moudoud},
url = {https://www.scopus.com/pages/publications/105033149093?origin=resultslist},
doi = {10.1109/SMC58881.2025.11343130},
isbn = {1062922X (ISSN); 979-833153358-8 (ISBN)},
year = {2025},
date = {2025-01-01},
booktitle = {Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.},
pages = {27–32},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {The proliferation of Internet of Things (IoT) devices has introduced significant security vulnerabilities, particularly in detecting zero-day attacks within highly dynamic and heterogeneous environments. Traditional machine learning models often fall short due to their static nature and computational demands. In this paper, we propose an adaptive ensemble learning framework that dynamically selects optimal detection models on a per-attack-class basis to improve detection accuracy while maintaining computational efficiency. Our approach combines multiple base classifiers (Random Forest, K-Nearest Neighbors, and Support Vector Machine) using ensemble techniques including bagging, boosting, and stacking. Ensemble techniques such as Bagging, Boosting, Voting, and Stacking. The key innovation lies in a class-aware model selection mechanism that identifies the most effective classifier-ensemble combination for each specific attack category, rather than applying a single model across all threat types. This targeted approach recognizes that different attack patterns exhibit distinct characteristics that may be better captured by different algorithmic approaches. Finally, we propose a decision-rule mechanism that selects the best-performing model for each attack class to improve detection accuracy. The proposed framework is evaluated through extensive experiments. The results show that our approach significantly enhances classification performance, especially for complex and rare attack types. © 2025 IEEE.},
note = {Journal Abbreviation: Conf. Proc. IEEE Int. Conf. Syst. Man Cybern.},
keywords = {Attack detection, Classification (of information), Computational efficiency, Detection accuracy, Ensemble learning, Ensemble techniques, Heterogeneous ensembles, Internet of thing security, Internet of things, Intrusion Detection, Intrusion-Detection, IoT Security, Learning systems, Nearest neighbor search, Security vulnerabilities, Stackings, Support vector machines, Zero-day attack, Zero-day detection},
pubstate = {published},
tppubtype = {inproceedings}
}
The proliferation of Internet of Things (IoT) devices has introduced significant security vulnerabilities, particularly in detecting zero-day attacks within highly dynamic and heterogeneous environments. Traditional machine learning models often fall short due to their static nature and computational demands. In this paper, we propose an adaptive ensemble learning framework that dynamically selects optimal detection models on a per-attack-class basis to improve detection accuracy while maintaining computational efficiency. Our approach combines multiple base classifiers (Random Forest, K-Nearest Neighbors, and Support Vector Machine) using ensemble techniques including bagging, boosting, and stacking. Ensemble techniques such as Bagging, Boosting, Voting, and Stacking. The key innovation lies in a class-aware model selection mechanism that identifies the most effective classifier-ensemble combination for each specific attack category, rather than applying a single model across all threat types. This targeted approach recognizes that different attack patterns exhibit distinct characteristics that may be better captured by different algorithmic approaches. Finally, we propose a decision-rule mechanism that selects the best-performing model for each attack class to improve detection accuracy. The proposed framework is evaluated through extensive experiments. The results show that our approach significantly enhances classification performance, especially for complex and rare attack types. © 2025 IEEE.
9.
Cote, S. S. -P.; Paquette, G. R.; Neveu, S. -M.; Chartier, S.; Labbe, D. R.; Renaud, P.
Combining electroencephalography with plethysmography for classification of deviant sexual preferences. Article d'actes
Dans: Proceedings - 9th International Workshop on Biometrics and Forensics, IWBF 2021, Institute of Electrical and Electronics Engineers Inc., 2021, ISBN: 978-172819556-8 (ISBN), (Journal Abbreviation: Proc. - Int. Workshop Biom. Forensics, IWBF).
Résumé | Liens | BibTeX | Étiquettes: Biometrics, Classification, Classification (of information), Decision trees, Deviant sexual preferences, Dimensionality reduction, Electroencephalography, Electrophysiology, extraction, Extraction method, Machine learning, Plethysmography, Proof of concept, Psychophysiological measures, Standard protocols, Variable selection and extraction
@inproceedings{cote_combining_2021,
title = {Combining electroencephalography with plethysmography for classification of deviant sexual preferences.},
author = {S. S. -P. Cote and G. R. Paquette and S. -M. Neveu and S. Chartier and D. R. Labbe and P. Renaud},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85113855965&doi=10.1109%2fIWBF50991.2021.9465078&partnerID=40&md5=b545b2a6d22e32115ac179399188960e},
doi = {10.1109/IWBF50991.2021.9465078},
isbn = {978-172819556-8 (ISBN)},
year = {2021},
date = {2021-01-01},
booktitle = {Proceedings - 9th International Workshop on Biometrics and Forensics, IWBF 2021},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {Evaluating sexual preferences is a difficult task. Past researchrelied mostly on penile plethysmography (PPG). Even though this technique is the standard protocol used in most currentforensic settings, its usage showed mixed results. One way to improve PPG is the addition of other psychophysiological measures such as electroencephalography (EEG). However, EEG generates significant amount of data that hinders classification. Machine learning (ML) is nowadays an excellent tool to identify most discriminating variables and for classification. Therefore, it is proposed to use ML selection and extraction methods for dimensionality reduction and then to classify sexual preferences. Evidence from this proof of concept shows that using EEG and PPG together leads to better classification (85.6%) than using EEG (82.2%) or PPG individually (74.4%). The Random Forest (RF) classifier combined with the Principal Component Analysis (PCA) extraction method achieves a slightly higher general performance rate. This increase in performances opens the door for using more reliable biometric measures in the assessment of deviant sexual preferences. © 2021 IEEE.},
note = {Journal Abbreviation: Proc. - Int. Workshop Biom. Forensics, IWBF},
keywords = {Biometrics, Classification, Classification (of information), Decision trees, Deviant sexual preferences, Dimensionality reduction, Electroencephalography, Electrophysiology, extraction, Extraction method, Machine learning, Plethysmography, Proof of concept, Psychophysiological measures, Standard protocols, Variable selection and extraction},
pubstate = {published},
tppubtype = {inproceedings}
}
Evaluating sexual preferences is a difficult task. Past researchrelied mostly on penile plethysmography (PPG). Even though this technique is the standard protocol used in most currentforensic settings, its usage showed mixed results. One way to improve PPG is the addition of other psychophysiological measures such as electroencephalography (EEG). However, EEG generates significant amount of data that hinders classification. Machine learning (ML) is nowadays an excellent tool to identify most discriminating variables and for classification. Therefore, it is proposed to use ML selection and extraction methods for dimensionality reduction and then to classify sexual preferences. Evidence from this proof of concept shows that using EEG and PPG together leads to better classification (85.6%) than using EEG (82.2%) or PPG individually (74.4%). The Random Forest (RF) classifier combined with the Principal Component Analysis (PCA) extraction method achieves a slightly higher general performance rate. This increase in performances opens the door for using more reliable biometric measures in the assessment of deviant sexual preferences. © 2021 IEEE.
10.
Cote, S. S. -P.; Paquette, G. R.; Neveu, S. -M.; Chartier, S.; Labbe, D. R.; Renaud, P.
Combining electroencephalography with plethysmography for classification of deviant sexual preferences. Article d'actes
Dans: Proceedings - 9th International Workshop on Biometrics and Forensics, IWBF 2021, Institute of Electrical and Electronics Engineers Inc., 2021, ISBN: 978-172819556-8 (ISBN), (Journal Abbreviation: Proc. - Int. Workshop Biom. Forensics, IWBF).
Résumé | Liens | BibTeX | Étiquettes: Biometrics, Classification, Classification (of information), Decision trees, Deviant sexual preferences, Dimensionality reduction, Electroencephalography, Electrophysiology, extraction, Extraction method, Machine learning, Plethysmography, Proof of concept, Psychophysiological measures, Standard protocols, Variable selection and extraction
@inproceedings{coteCombiningElectroencephalographyPlethysmography2021,
title = {Combining electroencephalography with plethysmography for classification of deviant sexual preferences.},
author = {S. S. -P. Cote and G. R. Paquette and S. -M. Neveu and S. Chartier and D. R. Labbe and P. Renaud},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85113855965&doi=10.1109%2fIWBF50991.2021.9465078&partnerID=40&md5=b545b2a6d22e32115ac179399188960e},
doi = {10.1109/IWBF50991.2021.9465078},
isbn = {978-172819556-8 (ISBN)},
year = {2021},
date = {2021-01-01},
booktitle = {Proceedings - 9th International Workshop on Biometrics and Forensics, IWBF 2021},
publisher = {Institute of Electrical and Electronics Engineers Inc.},
abstract = {Evaluating sexual preferences is a difficult task. Past researchrelied mostly on penile plethysmography (PPG). Even though this technique is the standard protocol used in most currentforensic settings, its usage showed mixed results. One way to improve PPG is the addition of other psychophysiological measures such as electroencephalography (EEG). However, EEG generates significant amount of data that hinders classification. Machine learning (ML) is nowadays an excellent tool to identify most discriminating variables and for classification. Therefore, it is proposed to use ML selection and extraction methods for dimensionality reduction and then to classify sexual preferences. Evidence from this proof of concept shows that using EEG and PPG together leads to better classification (85.6%) than using EEG (82.2%) or PPG individually (74.4%). The Random Forest (RF) classifier combined with the Principal Component Analysis (PCA) extraction method achieves a slightly higher general performance rate. This increase in performances opens the door for using more reliable biometric measures in the assessment of deviant sexual preferences. © 2021 IEEE.},
note = {Journal Abbreviation: Proc. - Int. Workshop Biom. Forensics, IWBF},
keywords = {Biometrics, Classification, Classification (of information), Decision trees, Deviant sexual preferences, Dimensionality reduction, Electroencephalography, Electrophysiology, extraction, Extraction method, Machine learning, Plethysmography, Proof of concept, Psychophysiological measures, Standard protocols, Variable selection and extraction},
pubstate = {published},
tppubtype = {inproceedings}
}
Evaluating sexual preferences is a difficult task. Past researchrelied mostly on penile plethysmography (PPG). Even though this technique is the standard protocol used in most currentforensic settings, its usage showed mixed results. One way to improve PPG is the addition of other psychophysiological measures such as electroencephalography (EEG). However, EEG generates significant amount of data that hinders classification. Machine learning (ML) is nowadays an excellent tool to identify most discriminating variables and for classification. Therefore, it is proposed to use ML selection and extraction methods for dimensionality reduction and then to classify sexual preferences. Evidence from this proof of concept shows that using EEG and PPG together leads to better classification (85.6%) than using EEG (82.2%) or PPG individually (74.4%). The Random Forest (RF) classifier combined with the Principal Component Analysis (PCA) extraction method achieves a slightly higher general performance rate. This increase in performances opens the door for using more reliable biometric measures in the assessment of deviant sexual preferences. © 2021 IEEE.