Nora Cuppens

Lucas Schott, Josephine Delas, Hatem Hajri et al.

Deep Reinforcement Learning (DRL) is a subfield of machine learning for training autonomous agents that take sequential actions across complex environments. Despite its significant performance in well-known environments, it remains susceptible to minor condition variations, raising concerns about its reliability in real-world applications. To improve usability, DRL must demonstrate trustworthiness and robustness. A way to improve the robustness of DRL to unknown changes in the environmental conditions and possible perturbations is through Adversarial Training, by training the agent against well-suited adversarial attacks on the observations and the dynamics of the environment. Addressing this critical issue, our work presents an in-depth analysis of contemporary adversarial attack and training methodologies, systematically categorizing them and comparing their objectives and operational mechanisms.

Mohamed Amine Merzouk, Erwan Beurier, Reda Yaich et al.

The escalating sophistication of cyberattacks has encouraged the integration of machine learning techniques in intrusion detection systems, but the rise of adversarial examples presents a significant challenge. These crafted perturbations mislead ML models, enabling attackers to evade detection or trigger false alerts. As a reaction, adversarial purification has emerged as a compelling solution, particularly with diffusion models showing promising results. However, their purification potential remains unexplored in the context of intrusion detection. This paper demonstrates the effectiveness of diffusion models in purifying adversarial examples in network intrusion detection. Through a comprehensive analysis of the diffusion parameters, we identify optimal configurations maximizing adversarial robustness with minimal impact on normal performance. Importantly, this study reveals insights into the relationship between diffusion noise and diffusion steps, representing a novel contribution to the field. Our experiments are carried out on two datasets and against 5 adversarial attacks. The implementation code is publicly available.

Kéren Saint-Hilaire, Frédéric Cuppens, Nora Cuppens et al.

Attack graphs provide a representation of possible actions that adversaries can perpetrate to attack a system. They are used by cybersecurity experts to make decisions, e.g., to decide remediation and recovery plans. Different approaches can be used to build such graphs. We focus on logical attack graphs, based on predicate logic, to define the causality of adversarial actions. Since networks and vulnerabilities are constantly changing (e.g., new applications get installed on system devices, updated services get publicly exposed, etc.), we propose to enrich the attack graph generation approach with a semantic augmentation post-processing of the predicates. Graphs are now mapped to monitoring alerts confirming successful attack actions and updated according to network and vulnerability changes. As a result, predicates get periodically updated, based on attack evidences and ontology enrichment. This allows to verify whether changes lead the attacker to the initial goals or to cause further damage to the system not anticipated in the initial graphs. We illustrate the approach under the specific domain of cyber-physical security affecting smart cities. We validate the approach using existing tools and ontologies.