Abstract

In modern smart cities, user request handling has evolved beyond conventional software solutions and electronic services. Companies now leverage a diverse range of cutting-edge technologies, such as Artificial Intelligence, the Internet of Things (IoT), Edge/Fog computing, and Cloud computing, to deliver various smart services. These services cover domains such as smart healthcare, smart buildings, smart transportation, smart living, and smart administration. However, a significant challenge hindering the widespread adoption and effective deployment of these services is the limited awareness among citizens of their availability and features, including aspects like pricing models, security policies, and service-level agreements. This issue largely stems from the lack of centralized public repositories where companies can showcase their smart offerings. Consequently, citizens often rely on traditional Web search tools (e.g., search engines, social networks), which limits their ability to fully benefit from the advantages of smart services. In addition, users are often connected to multiple service providers, meaning that their related knowledge (e.g., profiles, service usage, ratings) is distributed across multiple information sources.
To bridge the gap between smart service providers and citizens, and to effectively utilize user and service information across multiple networks, we propose CrossRecSmart, a cross-network recommender system built upon a multiplex network of smart services. This system can be conceptualized as a large-scale, distributed, domain-specific marketplace for available smart services. Our approach integrates Graph Neural Networks (GNNs) with cross-network representation learning to construct this multiplex network. The complex and distributed nature of this task is addressed through a cross-network learning model that facilitates collaboration among multiple smart service providers and enables the aggregation of their knowledge. This is achieved by interconnecting multiple networks through bridge nodes, also referred to as anchor entities. Furthermore, we introduce an algorithm for the cross-network recommendation of top-rated smart services. Comparative analyses with existing recommender systems designed for smart cities demonstrate the superiority of our proposed approach, owing to the concept of anchor links.

Abstract

Recently, drones have emerged as an effective and rapid mean to deliver various packages as services. Exhibiting the same behavior and principles of the service paradigm, drones’ delivery tasks can be abstracted as services, also called DaaS (Drone-as-a-Service). However, drones operate in a dynamic and uncertain environment (flight regulations, weather conditions, charging resources). In addition, the drones’ limited capacity (e.g., flying range, battery capacity, payload) and the delivery constraints (e.g., cost, delivery time) make the delivery missions beyond one drone’s capacity. To cope with the above challenges, and to meet the customers’ and providers’ delivery constraints, composing drone services is a natural solution. We propose a DaaS composition approach that takes advantage of a recent and powerful technology, called knowledge graph. First, we model the drones’ environment (charging stations, skyway segments, drone services) as a heterogeneous information network. Then, we combine knowledge graph embedding with subgraph-aware proximity measure, to arrange drones and charging stations in a low-dimentional vector space. This pre-processing step helped select high-flight capacity drone services, and generate low-congestion delivery paths. Experimental results have proven the approach’s performance and the quality of DaaS compositions, compared to recent state-of-the-art approaches.

Abstract

Vehicle-as-a-Service (VaaS) refers to on-demand rides and the sharing/offering of various kinds of transportation facilities (e.g., smart buses, electric vehicles, business jets, autonomous cars), to move from a source to a destination across one or several regions. Coupled with smart transportation services, which are critical for addressing road network issues (e.g., traffic congestion, parking, safety), VaaS is increasingly adopted in smart cities. For example, a user may express his needs in terms of source/destination regions, time and cost constraints, in addition to preferred transportation modes and services (e.g., only connected buses and electric cars). This user profile is evaluated against each available VaaS, as well as the smart urban network (SUN) conditions. However, current solutions do not consider the customization of VaaS combinations and treat user requests as a traditional vehicle routing problem; while in a SUN context, processing VaaS requests not only consists of finding the optimal transportation path that meets user constraints (e.g., time, cost), but also selecting the top-rated combination of available VaaSs (e.g., sequence of smart buses) that meet the user profile (e.g., connectivity, specific facilities, transit restrictions) and trip constraints. To cope with these challenges, the goal of this paper is to develop an orchestration engine that allows the personalized selection and composition of VaaS services. We applied Fuzzy Relational Concept Analysis as an effective data analysis and clustering technique to solve the following issues: (i) representing VaaS-related knowledge and excluding low-quality VaaSs, (ii) clustering of VaaS services based on their common features and covered regions, (iii) clustering of smart urban regions based on their neighborhood and their frontier stations. At a second stage, a search method based on A* algorithm is proposed to find the optimal local paths in each traversed region of the smart urban network. Finally, a bottom-up lattice parsing method is defined to determine the VaaSs with high quality and high coverage capacity. Comparative experiments with state-of-the-art solutions proved the efficiency of our approach.

Abstract

In recent years, smart mobility networks have experienced significant growth due to the integration of key technologies such as cloud computing, edge intelligence, and the Internet of Things (IoT) into transportation infrastructure. When combined with the principles of service-oriented computing (SOC), various transportation modes now feature intelligent capabilities, including eco-driving assistance, emergency service integration, V2X communication, environmental sensors, in-vehicle infotainment, Over-the-Air (OTA) updates, driver behavior monitoring, and AI-powered assistance. This has led to the emergence of Connected Vehicle as a Service (CVaaS) as a new paradigm for smart vehicles and transportation services.
However, with the increasing complexity of AI-driven features and integration with smart city infrastructure, traditional recommender systems can no longer meet user requirements such as personalized connectivity preferences and eco-friendly route optimization. CVaaS recommendations also inherit challenges from traditional transportation systems, including multi-modal integration (e.g., coordinating smart buses and autonomous vehicles), environmental considerations (e.g., smart parking and dedicated lanes for autonomous cars), uncertain demand, user trust, regulatory compliance, and data privacy concerns.
In this paper, we address the challenges of multi-modal transportation and environmental uncertainty, such as traffic congestion and VaaS demand fluctuations. By modeling Smart Urban Network (SUN) traffic and VaaS demand, we predict congestion patterns and VaaS availability using a Long Short-Term Memory (LSTM) model. Additionally, we apply Spectral hyper-graph Theory to cluster the SUN into closely connected regions, identifying traversed areas for trip requests. These preprocessing steps help eliminate high-congestion zones and low-demand VaaS services, improving trip efficiency.
Finally, inspired by the combinatorial nature of VaaS selection, we propose a Quantum-Inspired variant of the Gravitational Search Algorithm (Q-GSA) to explore and evaluate possible VaaS combinations, ultimately selecting an optimal set of smart transportation services. Experimental comparisons with four benchmark methods confirm the superiority of our approach in terms of efficiency and solution quality.

Abstract

Abstract—The increasing complexity of global supply chains,
combined with the need for fast, cost-effective, and environmentally
friendly deliveries, has reinforced the importance of
multimodal freight transportation(MFT) as a key solution to
meet modern demands. One of the main challenges in MFT
is to develop an innovative optimization model to plan and
manage the supply chain. In this work, we consider four
modes of transportation (air, road, rail, and sea) and propose
an innovative multi-objective optimization model, designed to
simultaneously minimize transportation costs, transit times,
and CO2 emissions, while integrating the complex operational
constraints inherent in current logistic systems. To address
this problem, we adopt two well-known algorithms : Non-
Dominated Sorting Genetic Algorithm III (NSGA-III) and
Teaching-Learning Optimization (TLBO), through an experimental
study demonstrating the effectiveness of these evolutionary
solution methods in solving these complex optimization
problem.
The results show that TLBO optimization effectively reduces
costs and environmental impact, while the NSGAIII algorithm
improves delivery times.

Abstract

As drones continue to see widespread adoption across commercial, private, healthcare, and education sectors, their commercial use is experiencing rapid growth. To enhance drone-based package delivery efficiency and improve customer experience, the vast amount of flight and recharging data collected from drones and stations offers valuable opportunities for predicting both resource availability within the sky network and drones’ capacity to complete delivery missions. However, the variations in regional regulations and privacy restrictions enforced by drone service providers lead to data heterogeneity, making centralized processing of flight and charging history impractical. Running machine learning models locally at the service provider level (drones and stations) addresses privacy concerns, yet processing the large volume and diversity of raw data remains a significant challenge. To deal with these issues, a collaborative learning approach based on historical delivery data presents an elegant solution. Aiming to offer predictive scheduling of drone delivery missions, while taking into consideration the complexity, heterogeneity, and dynamic nature of their flight environment, we propose a predictive and federated approach for the resilient selection and composition of drone delivery services, leveraging the strengths of federated learning (FL) to handle data privacy and heterogeneity. Our method utilizes a federated Recurrent Neural Network (FL-RNN) model that combines predictive capabilities with federated behavior, enabling collaborative forecasting and efficient mission scheduling in low-congestion regions based on the most reliable drone services. Additionally, an enhanced A* search algorithm is defined to identify the optimal delivery path by factoring in station overload probabilities. Besides computational efficiency, experimental results demonstrate the effectiveness of our approach, achieving a 16.66% improvement in prediction accuracy and an 8.89% reduction in delivery costs compared to non-federated and non-predictive solutions.

Abstract

Contextual recommendation has become attainable through the massive amounts of contextual information generated by smartphones and Internet of Things (IoT) devices. The availability of a huge amount of contextual data paves the way for a revolution in recommendation systems. It overcomes the static nature of personalization, which does not allow the discovery of new items and interests, toward a contextualization of the user’s tastes, which are in constant evolution. In this paper, we proposed CoD-MaF, a Context-Driven Collaborative Filtering using Contextual biased Matrix Factorization. Our approach employs feature selection methods to extract the most influential contextual features that will be used to cluster the users using K-means algorithm. The model then performs a collaborative filtering based on matrix factorization with improved contextual biases to suggest relevant personalized recommendations. We highlighted the performance of our method through experiments on four datasets (LDOS-CoMoDa, STS-Travel, IncarMusic and Frappe). Our model enhanced the accuracy of predictions and achieved competitive performance compared to baseline methods in metrics of RMSE and MAE.

Abstract

Classification of imbalanced multi-class data is still so far one of the most challenging issues in machine learning and data mining. This task becomes more serious when classes containing fewer instances are located in overlapping regions. Several approaches have been proposed through the literature to deal with these two issues such as the use of decomposition, the design of ensembles, the employment of misclassification costs, and the development of ad-hoc strategies. Despite these efforts, the number of existing works dealing with the imbalance in multi-class data is much reduced compared to the case of binary classification. Moreover, existing approaches still suffer from many limits. These limitations include difficulties in handling imbalances across multiple classes, challenges in adapting sampling techniques, limitations of certain classifiers, the need for specialized evaluation metrics, the complexity of data representation, and increased computational costs. Motivated by these observations, we propose a multi-objective evolutionary induction approach that evolves a population of NLM-DTs (Non-Linear Multivariate Decision Trees) using the -NSGA-III (-Non-dominated Sorting Genetic Algorithm-III) as a search engine. The resulting algorithm is termed EMO-NLM-DT (Evolutionary Multi-objective Optimization of NLM-DTs) and is designed to optimize the construction of NLM-DTs for imbalanced multi-class data classification by simultaneously maximizing both the Macro-Average-Precision and the Macro-Average-Recall as two possibly conflicting objectives. The choice of these two measures as objective functions is motivated by a recent study on the appropriateness of performance metrics for imbalanced data classification, which suggests that the mAURPC (mean Area Under Recall Precision Curve) satisfies all necessary conditions for imbalanced multi-class classification. Moreover, the NLM-DT adoption as a baseline classifier to be optimized allows the generation non-linear hyperplanes that are well-adapted to the classes ‘boundaries’ geometrical shapes. The statistical analysis of the comparative experimental results on more than twenty imbalanced multi-class data sets reveals the outperformance of EMO-NLM-DT in building NLM-DTs that are highly effective in classifying imbalanced multi-class data compared to seven relevant and recent state-of-the-art methods.

Abstract

Abstract

Multi-output Regression (MOR) is a machine learning technique that aims to predict several values simultaneously. Some existing approaches addressed this problem by decomposing the MOR problem into separate single-target ones. However, in real-world applications, it is more advantageous to exploit the inter-target correlations in the prediction task. Some other approaches proposed simultaneous prediction but they are based on greedy algorithms and are prone to fall easily into local optima. In order to solve these issues, we propose a novel approach called Bi-level Evolutionary Model TreeChain Induction (BEMTCI) which is able to deal with multi-output datasets using a bi-level evolutionary algorithm. BEMTCI evolves a population of Model Tree Chains (MTCs) where each Model Tree (MT) focuses on the prediction of one single target. The upper-level explores different orderings of the MTs of each MTC to find the best chaining order which is able to express the relationships among the output variables. A further optimization is performed in the lower-level of BEMTCI which concerns the linear models at the leaves of the MTs. The experimental study showed the effectiveness of our approach compared to the existing ones when applied on sixteen MOR datasets. The genetic operators employed in our BEMTCI ensure the variation of the population and guarantee a fair and a precise prediction due to the evaluation process. The obtained results prove the performance of our BEMTCI in solving MOR problems.