Explainable AI Planning (XAIP) is a pivotal research area focused on enhancing the transparency, interpretability, and trustworthiness of automated planning systems. This paper provides a comprehensive review of XAIP, emphasizing key techniques for plan explanation, such as contrastive explanations, hierarchical decomposition, and argumentative reasoning frameworks. We explore the critical role of argumentation in justifying planning decisions and address the challenges of replanning in dynamic and uncertain environments, particularly in high-stakes domains like healthcare, autonomous systems, and logistics. Additionally, we discuss the ethical and practical implications of deploying XAIP, highlighting the importance of human-AI collaboration, regulatory compliance, and uncertainty handling. By examining these aspects, this paper aims to provide a detailed understanding of how XAIP can improve the transparency, interpretability, and usability of AI planning systems across various domains.

Artificial Intelligence (AI) is revolutionizing information security, influencing both attack and defense strategies. Attackers leverage AI to automate cyberattacks and exploit vulnerabilities, while defenders utilize it for anomaly detection, predictive threat modeling, and automated responses. Federated Learning (FL), a privacy-preserving training method, remains vulnerable to inference attacks. To address this, we propose the Rényi Differential Privacy (RDP) based federated learning (RDP-FL) framework, which incorporates moment accounted noise scaling to dynamically regulate the privacy budget, achieving an optimal balance between privacy and utility. This method minimizes unnecessary noise addition while maintaining strong privacy guarantees, thereby preserving data integrity and enhancing model performance. Experimental validation on the Medical-MNIST and CIFAR-10 datasets demonstrates the effectiveness of RDP-FL, showing its ability to safeguard data privacy while ensuring high classification accuracy. This work advances the ongoing efforts to enhance cybersecurity in an AI-driven landscape.

Vehicular Ad hoc Networks (VANETs) play a crucial role in Intelligent Transportation Systems (ITSs) and Traffic Information Systems (TISs). However, the open nature of wireless communication exposes these networks to security threats from compromised nodes. In this work, we propose a trust-based traffic information system to enhance the security of vehicular networks. We introduce an infrastructure-based trust management model utilizing Smart Road Signs (SRSs) to address scalability challenges posed by the limited computational and storage capabilities of smaller devices such as connected vehicles and smartphones. By enabling multi-source data aggregation, we mitigate single-source failure risks and improve environmental visibility. To enhance the accuracy of trust evaluation, we propose a two-level trust evaluation process, combining Bayesian Inference and dynamic weighted summation. Additionally, a probabilistic graph is used to estimate prior information for Bayesian inference, learning causal relationships from previous traffic incident reports. Simulation results demonstrate the effectiveness of our model in detecting malicious attacks, including false injection and on-off attacks, even in environments with up to 40% malicious nodes.

Elderly fatigue, a critical issue affecting the health and well-being of the aging population worldwide, presents as a substantial decline in physical and mental activity levels. This widespread condition reduces the quality of life and introduces significant hazards, such as increased accidents and cognitive deterioration. Therefore, this study proposed a model to detect fatigue in the elderly with satisfactory accuracy. In our contribution, we use video and image processing through a video in order to detect the elderly’s face recognition in each frame. The model identifies facial landmarks on the detected face and calculates the Eye Aspect Ratio (EAR), Eye Fixation, Eye Gaze Direction, Mouth Aspect Ratio (MAR), and 3D head pose. Among the various methods evaluated in our study, the Extra Trees algorithm outperformed all others machine learning methods, achieving the highest results with a sensitivity of 98.24%, specificity of 98.35%, and an accuracy of 98.29%.

The rapid expansion of contextual information from smartphones and Internet of Things (IoT) devices paved the way for Context-Aware Recommendation Systems (CARS). This abundance of contextual data heralds a transformative era for traditional recommendation systems. In alignment with this trend, we propose a novel model which provides personalized recommendations based on context. Our approach uses K-means algorithm to cluster users based on contextual features. Then, the model performs collaborative filtering based on matrix factorization with enhanced contextual biases to provide relevant recommendations. We demonstrated the performance of our method through experiments conducted on the movie recommender dataset LDOS-CoMoDa. The experimental results showed the effective performance of our proposal compared to reference methods, achieving an RMSE of 0.7416 and an MAE of 0.6183.

Gestational Diabetes Mellitus (GDM) is a type of diabetes that develops during pregnancy. It is important for pregnant women to monitor their blood sugar levels regularly and follow a healthy diet. However, early intervention can greatly reduce risk of this type of diabetes. Machine Learning and Deep Learning techniques are utilized to predict this risk based on an individual's symptoms, lifestyle, and medical history. By identifying key features such as age, insulin, body mass index, and glucose levels, machine learning models such as Random Forest and XGBoost are used in this research work to classify patients at risk of a gestational diabetes. In addition, we propose an explainable feature selection approach to improve the accuracy of machine learning models for GDM prediction. This method involves iteratively eliminating features that exhibit a negative contribution as determined by the SHAP (Shapley Additive explanations) feature attribution explanations for the model’s predictions.

In recent years, the negative impacts of neglecting the environment, particularly global warming caused by greenhouse gases, have gained attention. Many countries and organizations are taking steps to reduce their greenhouse gas emissions and promote sustainable practices. In this paper, we aim to address the gap in the classical Vehicle Routing Problem (VRP) by taking into consideration the environmental effects of vehicles. To find a balance between cost-efficiency and environmental impact, we propose a Hybrid Genetic Algorithm (HGA) to address the Dynamic Vehicle Routing Problem with Time Windows (DVRPTW) and a heterogeneous fleet, taking into account new orders that arrive dynamically during the routing process. This approach takes into consideration the environmental effects of the solutions by optimizing the number and type/size of vehicles used to fulfill both static and dynamic orders. The goal is to provide a solution that is both cost-effective and environmentally friendly, addressing the issue of over-exploitation of energy and atmospheric pollution that threaten our ecological environment. Computational results prove that the hybridization of a genetic algorithm with a greedy algorithm can find high-quality solutions in a reasonable run time.

The paradigm of Social Internet of Things (SIoT) incorporates the concepts of social networking in the Internet of Things (IoT). The idea of SIoT is to allow objects to autonomously establish social relationships, which may facilitate network navigability and the discovery of information and services. The characteristics of the IoT such as the heterogeneity and the dynamicity of the network and the social relationships between devices lead to several challenges including how to build a reliable network. In this paper, we propose an adaptive trust management model that helps nodes seeking trusted service providers. The trustworthiness of a service provider is assessed on the basis of its past experiences with the requestor and of the recommendations of the requester’s neighbors. In our trust model, the trust parameters evolve dynamically in response to the change of the network context, the type of the demanded service and the nature of the relationships between the different nodes. The experiment results show that our proposed model achieves high accuracy and it is proved to be resilient against common attacks.

International audience

Services providers and requestors, in ubiquitous environments, are characterized by the heterogeneity of their devices and the constant changing of the network bandwidth. These characteristics pose a big challenge to service selection methods. This paper describes a new service selection method based on Multiple Linear Regression (MLR) techniques. In order to ensure the selection of the best service in terms of provider's quality of service (QoS), an MLR model is defined estimating especially Response time, returned from service provider. Response time is mainly influenced by three independent variables: bandwidth, processor speed and memory. We validate our model by testing five different types of devices (provider side) in five different types of network (i.e. different values for bandwidth). The results show that 97.24% of the variation in the response time can be explained by our model.

The workflow satisfiability problem has been studied by researchers in the security community using various approaches. The goal is to ensure that the user/role is authorised to execute the current task and that this permission doesn't prevent the remaining tasks in the workflow instance to be achieved. A valid authorisation plan consists in affecting authorised roles and users to workflow tasks in such a way that all the authorisation constraints are satisfied. Previous works are interested in workflow satisfiability problem by considering intra-instance constraints, i.e. constraints which are applied to a single instance. However, inter-instance constraints which are specified over multiple workflow instances are also paramount to mitigate the security frauds. In this paper, we present how ECA (Event-Condition-Action) paradigm and agent technology can be exploited to control authorisation plan in order to meet dynamic constraints, namely intra-instance and inter-instance constraints. We present a specification of a set of ECA rules that aim to achieve this goal. A prototype implementation of our proposed approach is also provided in this paper.

In this paper we present a specification of the most common static and dynamic workflow authorisation constraints. We propose an authorisation model that includes a planning phase, an execution phase and an adjustment phase. In addition, we focus on how the problems of role-task assignment and user-task assignment are respectively translated into CSP (Constraint Satisfaction Problem) and DyCSP (Dynamic constraint Satisfaction Problem) and solved using the explanation concept. In case of an inconsistent assignment problem, we propose to restore problem consistency based upon inconsistency explanation.

Role-Based Access Control (RBAC) model has been developed as an alternative to traditional approaches to handle access control in workflow systems. Accordingly, authorisation constraints must be defined to enforce the legal assignment of access privileges to roles and roles to users. The authorisation planning ensures that there is at least one way to complete the workflow instance without breaching any of the authorisation constraints. Authorisation planning with considering intra-instance constraints has been discussed in the research literature. However, the inter-instance constraints also need to be considered to mitigate the security fraud. In this paper, a novel authorisation system that incorporates intra-instance and inter-instance constraints is proposed. It includes the planning phase, the execution phase, and the adjustment phase. It is in charge of generating user/role assignment plans, verifying them and eventually updating them to take into account the dynamic (intra-instance and inter-instance) constraints. Besides, grounded upon agent technology and publish-subscribe communication model, a mechanism for the consideration of dynamic constraints (intra-instance and inter-intance) to generate valid assignment plans is demonstrated.

The need of interoperable e-government services is addressed through the use of web services where sensitive services need to be granted to only authorized subjects from different organizations. In this paper, we propose a Trust and Dynamic Role Based Access Control model (TDRBAC) which deals with the specific requirements of e-government services. It effectively enhances the access control level since it is based on the trust level notion. The trust level evaluation is based on contextual attributes to assign to user role the appropriate view during the active session. The TDRBAC model is sensitive to the internal or external arisen events and it incorporates them in the access decision which makes it suitable for e-government dynamic environment.

On the basis of a reference model for ADSL traffic on an IP backbone link, established in an earlier study, we show that it is possible to infer the characteristics of long flows by performing a deterministic 1/N packet sampling. By using the fact that the number of active long flows can be represented by means of the number of customers in an M/G//spl infin/ queue with Weibullian service times, we derive some probabilistic properties of the sampled data. These properties are then used to infer the characteristics of the original flows. The method is illustrated by considering an actual traffic trace captured in the France Telecom IP backbone network. Experimental data show that the method proves quite efficient.

To evaluate the impact of peer-to-peer applications on traffic in wide area IP networks, we analyze measurements from a high speed backbone link carrying TCP traffic towards several ADSL areas. The first observations are that the prevalent part of traffic is due to peer-to-peer applications (almost 80% of total traffic) and that the usage of network becomes symmetric in the sense that customers are not only clients but also servers. This latter point is observed by the significant proportion of long flows mainly composed of ACK segments. When analyzing the bit rate created by long flows, it turns out that those TCP connections due to peer-to-peer applications have a rather small bit rate and that there is no evidence for long range dependence. These facts are intimately related to the way peer-to-peer protocols are running.

Measurements from an Internet backbone link carrying TCP traffic towards different ADSL areas are analyzed. For traffic analysis, we adopt a flow based approach and the popular mice/elephants dichotomy, where mice refer to short traffic transfers and elephants to long transfers. The originality of the reported experimental data, when compared with previous measurements from very high speed backbone links, is that the commercial traffic includes a significant part generated by peer-to-peer applications. This kind of traffic exhibits some remarkable properties in terms of mice and elephants, as we describe. It turns out that by adopting a suitable level of aggregation, the bit rate of mice can be described by means of a Gaussian process. The bit rate of elephants is smoother than that of mice and can also be well approximated by a Gaussian process.