Magecart skimming attacks have emerged as a significant threat to client-side security and user trust in online payment systems. This paper addresses the challenge of achieving robust and explainable detection of Magecart attacks through a comparative study of Machine Learning (ML) models with a real-world dataset. Tree-based, linear, and kernel-based models were applied with hyperparameter tuning and feature selection, to distinguish between benign and malicious scripts. The models are supported by a Behavior Deterministic Finite Automaton (DFA), which captures structural behavior patterns in scripts, helping to analyze and classify client-side script execution logs. To ensure robustness against adversarial evasion attacks, adversarial training and evaluations were performed using attacks from Adversarial Robustness Toolbox and Adaptative Perturbation Pattern Method. In addition, concise explanations of ML decisions are provided, supporting transparency and user trust. Experimental validation demonstrated high detection performance and interpretable reasoning, demonstrating that traditional ML models can be effective in real-world web security contexts.

Magecart skimming attacks pose a significant threat to client-side security and user trust in online payment systems. This poster details an approach for the robust and explainable detection of these attacks, with a deep focus on the critical role of feature engineering in optimizing Machine Learning model performance. A suite of classifiers was evaluated under three distinct conditions to determine the most effective feature set: a comprehensive set of 103 features, a 60-feature subset selected via Random Forest importance scores, and an 80-feature subset identified through Pearson’s correlation. The findings demonstrate that the importance-based selection strategy yields a superior trade-off between model performance and computational efficiency. This is best exemplified by the Support Vector Machine model, which achieved a peak F1-score and recall when trained on the 60 most important features. The resilience of this optimized model was subsequently confirmed through extensive adversarial testing against a range of evasion attacks, where it consistently outperformed other models.

Network intrusion detection systems are fundamental to the early detection of anomalous behaviour in networks. Modern versions of these tools take advantage of Machine Learning to process large amounts of data, identify patterns, and make predictions. Their development relies on the ability to access good historical network data. Therefore, the research community has been actively working on creating new datasets, and network traffic analysis tools are frequently used in this context. This study provides a comprehensive review of existing tools for network traffic analysis, highlighting their main advantages and drawbacks. A categorisation for these tools is introduced, as well as an overview of the dataset creation process by combining one or more of these categories. An updated analysis of existing datasets is also provided, along with details regarding their creation, highlighting the progression in dataset production. Finally, the impact of dataset features is discussed, underscoring their role in enhancing the effectiveness of network intrusion detection systems.

The increasing concerns about data privacy and the stringent enforcement of data protection laws are placing growing pressure on organizations to secure large datasets. The challenge of ensuring data privacy becomes even more complex in the domains of Artificial Intelligence and Machine Learning due to their requirement for large amounts of data. While approaches like differential privacy and secure multi-party computation allow data to be used with some privacy guarantees, they often compromise data integrity or accessibility as a tradeoff. In contrast, when using encryption-based strategies, this is not the case. While basic encryption only protects data during transmission and storage, Homomorphic Encryption (HE) is able to preserve data privacy during its processing on a centralized server. Despite its advantages, the computational overhead HE introduces is notably challenging when integrated into Neural Networks (NNs), which are already computationally expensive. In this work, we present a framework called ENNigma, which is a Private Neural Network (PNN) that uses HE for data privacy preservation. Unlike some state-of-the-art approaches, ENNigma guarantees data security throughout every operation, maintaining this guarantee even if the server is compromised. The impact of this privacy preservation layer on the NN performance is minimal, with the only major drawback being its computational cost. Several optimizations were implemented to maximize the efficiency of ENNigma, leading to occasional computational time reduction above 50%. In the context of the Network Intrusion Detection System application domain, particularly within the sub-domain of Distributed Denial of Service attack detection, several models were developed and employed to assess ENNigma’s performance in a real-world scenario. These models demonstrated comparable performance to non-private NNs while also achiev ing the two-and-a-half-minute inference latency mark. This suggests that our framework is approaching a state where it can be effectively utilized in real-time applications. The key takeaway is that ENNigma represents a significant advancement in the field of PNN as it ensures data privacy with minimal impact on NN performance. While it is not yet ready for real-world deployment due to its computational complexity, this framework serves as a milestone toward realizing fully private and efficient NNs.

Healthcare data contains some of the most sensitive information about an individual, yet sharing this data with healthcare practitioners can significantly enhance patient care and support research efforts. However, current systems for sharing health data between patients and caregivers do not fully address the critical security requirements of privacy, confidentiality, and consent management. Furthermore, compliance with regulatory laws such as GDPR and HIPAA is often deficient, largely because patients typically are asked to provide general consent for healthcare entities to access their data. Recognizing the limitations of existing systems, we present a novel approach to sharing health data that provides patients with control over who accesses their data, what data is accessed, and when. Our system ensures end-to-end privacy by integrating a Proxy ReEncryption Scheme with a Searchable Encryption Scheme, utilizing Homomorphic Encryption to enable healthcare practitioners to easily and securely search and access patients’ documents. A time performance analysis is also presented, which allowed us to observe that the number of keywords has a much greater impact on the running time of the different processes than the number of files.

Cybersecurity threats highlight the need for robust network intrusion detection systems to identify malicious behaviour. These systems rely heavily on large datasets to train machine learning models capable of detecting patterns and predicting threats. In the past two decades, researchers have produced a multitude of datasets, however, some widely utilised recent datasets generated with CICFlowMeter contain inaccuracies. These result in flow generation and feature extraction inconsistencies, leading to skewed results and reduced system effectiveness. Other tools in this context lack ease of use, customizable feature sets, and flow labelling options. In this work, we introduce HERA, a new open-source tool that generates flow files and labelled or unlabelled datasets with user-defined features. Validated and tested with the UNSW-NB15 dataset, HERA demonstrated accurate flow and label generation.

Machine Learning (ML) can be incredibly valuable to automate anomaly detection and cyber-attack classification, improving the way that Network Intrusion Detection (NID) is performed. However, despite the benefits of ML models, they are highly susceptible to adversarial cyber-attack examples specifically crafted to exploit them. A wide range of adversarial attacks have been created and researchers have worked on various defense strategies to safeguard ML models, but most were not intended for the specific constraints of a communication network and its communication protocols, so they may lead to unrealistic examples in the NID domain. This Systematization of Knowledge (SoK) consolidates and summarizes the state-of-the-art adversarial learning approaches that can generate realistic examples and could be used in real ML development and deployment scenarios with real network traffic flows. This SoK also describes the open challenges regarding the use of adversarial ML in the NID domain, defines the fundamental properties that are required for an adversarial example to be realistic, and provides guidelines for researchers to ensure that their future experiments are adequate for a real communication network.

Complementary and alternative medicine are commonly used concomitantly with conventional medications leading to adverse drug reactions and even fatality in some cases. Furthermore, the vast possibility of herb-drug interactions prevents health professionals from remembering or manually searching them in a database. Decision support systems are a powerful tool that can be used to assist clinicians in making diagnostic and therapeutic decisions in patient care. Therefore, an original and hybrid decision support system was designed to identify herb-drug interactions, applying artificial intelligence techniques to identify new possible interactions. Different machine learning models will be used to strengthen the typical rules engine used in these cases. Thus, using the proposed system, the pharmacy community, people’s first line of contact within the Healthcare System, will be able to make better and more accurate therapeutic decisions and mitigate possible adverse events.