hybrid cloud-IoT architectures are transforming healthcare by enhancing data management, improving patient outcomes, and enabling real-time decision-making. This paper presents a comprehensive review of the current state and future prospects of hybrid cloud-IoT architectures in healthcare. Our study identifies key trends, such as the increasing adoption of edge computing to reduce latency, the integration of AI for predictive analytics, and the emphasis on robust security measures to protect sensitive patient data. We address significant challenges, including data interoperability, scalability issues, and privacy concerns, and offer potential solutions and best practices. Furthermore, we provide a taxonomy-driven comparative framework evaluating five architectural paradigms edge, fog, mesh, hub-and-spoke, and serverless using healthcare-specific performance indicators such as cost, scalability, latency, security, and data confidentiality. Unlike prior studies, we synthesize existing research to produce a design recommendation. We propose a hybrid fog–mesh model that targets low latency, security, and scalability for healthcare delivery. This study links architectural theory to deployment practice. It focuses on dynamic VM selection and allocation, latency-aware computing, and data sovereignty. It also outlines key directions for future research to guide the implementation and optimization of hybrid cloud-IoT architectures in healthcare.

Code smells signal poor software design that can prevent maintainability and scalability. Identifying code smells is difficult because of the large volume of code, considerable detection expenses, and the substantial effort needed for manual tagging. Although current techniques perform well in within-project situations, they frequently struggle to adapt to cross-project environments that have varying data distributions. In this paper, we introduce CLADES (Cross-project Learning and Adaptation for Detection of Code Smells), a hybrid evolutionary approach consisting of three main modules: Initialization, Evolution, and Adaptation. The first module generates an initial population of decision tree detectors using labeled within-project data and evaluates their quality through fitness functions based on structural code metrics. The evolution module applies genetic operators (selection, crossover, and mutation) to create new offspring solutions. To handle cross-project scenarios, the adaptation module employs a clustering-based instance selection technique that identifies representative instances from new projects, which are added to the dataset and used to repair the decision trees through simulated annealing. These locally refined decision trees are then evolved using a genetic algorithm, thus enabling continuous adaptation to new project instances. The resulting optimized decision tree detectors are then employed to predict labels for the new unlabeled project instances. We assess CLADES across five open-source projects and we show that it has a better performance with respect to baseline techniques in terms of weighted F1-score and AUC-PR metrics. These results emphasize its capacity to effectively adjust to different project environments, facilitating precise and scalable detection of code smells while minimizing the need for manual review, contributing to more robust and maintainable software systems.

The use of serious games has shown immense importance in several fields such as education and in particular the learning of computer programming. Several experiences have shown the positive impact of integrating serious games into programming learning. Some programming concepts present many difficulties for learners, especially for beginners, such as variables and operator priority rules. In this paper, we present the design, development and evaluation of a serious game called "AppProg Game" dedicated to learning these two programming concepts: variables and the priority of arithmetic and logical operators. The evaluation took place with a final class scientific section in a secondary school in Tunisia at the level of learning. The result of this experience showed that the integration of serious games helped learners to assimilate the two concepts discussed. The experiment showed that learners did not progress in the same way. For this reason, in our future experiment, we plan to adapt the game to the learners' profile using artificial intelligence techniques.

Programming learning has become a 21st century skill, several countries have integrated it into their school curricula. Numerous studies have addressed the teaching of computational thinking and programming in primary schools and even nursery schools. These studies have shown that learning programming presents many difficulties for learners, especially beginners. This paper proposes a new classification of the difficulties encountered in learning programming for beginners, in particular for secondary school students. We also aim to study the impact of gender and level of learners in less developed regions on programming learning. We carried out two questionnaires with 79 students in a final year science class at a secondary school in Tunisia. The study showed that the majority of learners encountered difficulties of several types and with several concepts. The most difficult concepts were variables, operator priority rules and loops. The study also showed that gender had no effect on programming.

This paper introduces a comprehensive framework for designing Serious Games (SG) that integrate eXtended Reality (XR)-including virtual, mixed, and augmented reality systems, to create immersive and interactive learning experiences. Key features include a model to define game mechanics and learning content, a specification approach to ensure effective XR integration, alignment with educational goals, and iterative evaluation by designers, educators, and students to ensure continuous improvement. Our design approach empowers educators to define the purpose, mechanics, and XR features of the game while keeping them closely aligned with the pedagogical objectives. The proposed framework provides a step-by-step guide for participants in SG with XR (XR-SG) design, covering all stages from preparation to testing. This work contributes to the workshop’s mission by advancing the use of XR applications in education and fostering multidisciplinary collaboration for the creation of impactful, immersive learning solutions.

Urban traffic congestion remains an ongoing issue that requires advanced traffic management solutions. Accurate traffic forecasting plays a crucial role in Intelligent Transportation Systems, helping to mitigate congestion and improve mobility. Traditional machine learning approaches have been widely used for prediction tasks, often relying on large volumes of historical data for training. However, real-time adaptability is essential for dynamic traffic conditions. In this study, we leverage real-time traffic data and employ ADAptive GRAdient Descent, an online learning method that adaptively adjusts learning rates, allowing efficient updates as new data become available. To evaluate its performance, we implemented our approach on traffic data from a network of streets in Muscat, Oman, demonstrating its ability to provide accurate and timely congestion forecasts.

This paper tackles the Green Vehicle Routing Problem (GVRP), where vehicles with limited driving range must visit customers while recharging at Alternative Fuel Stations (AFSs). We propose a Multi-Start Tabu Search with Set Partitioning (MSTS-SP) approach structured in two phases. In the first phase, MSTS-SP uses a new constructive heuristic, Randomized Sectoring with Repair, to generate diverse initial solutions, which are then improved through multiple independent tabu search runs. The high-quality routes found during these runs are collected into a global pool. In the second phase, an exact set partitioning model is applied to this pool to select the best combination of routes. Computational experiments on 52 GVRP benchmark instances show that MSTS-SP matches 46 known best solutions (88%) and improves upon the best known solution for one large instance. These results demonstrate that MSTS-SP offers a competitive balance between solution quality and computational efficiency compared to state-of-the-art methods.

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 CO2emissions, 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.

Efficient patient admission scheduling is a key challenge in hospital management, as it directly impacts resource utilization and the quality of care. The Patient Admission Scheduling Problem (PASP) involves assigning patients to hospital beds over a planning horizon while considering medical constraints and hospital capacity. Due to its complexity, heuristic and metaheuristic approaches are often used to find high-quality solutions within a reasonable time. In this work, we propose a Variable Neighborhood Search (VNS) metaheuristic to solve the PASP. VNS systematically explores different neighborhoods to escape local optima and improve solution quality. To assess the impact of local search strategies, we implement four versions of VNS, each using a different method for modifying patient assignments. The proposed approach is evaluated on benchmark instances, where we conduct parameter tuning and analyze computational performance. Experimental results demonstrate the effectiveness of the method, showing that the appropriate choice of local search strategies significantly impact the quality of the results.

Home Health Care (HHC) aims to provide medical care and support services directly to patients in their own homes. The demand for HHC services is steadily increasing due to demographic trends, with a growing preference for receiving care in the home. This trend pushes organizations providing home health care services, to optimize their activities in order to meet this increasing demand efficiently. For this purpose, we propose in this work a new bi-level modeling of the problem, that we termed Bi-level Home Health Care Problem Considering Patients Preferences (Bi-HHCPP) aiming to find an efficient solution corresponding to this design. Existing research studies have focused on optimizing the problem considering only one decision-maker that optimizes both routing and scheduling entities imposed by the problem. This paper is the first to shed light on a new bi-level modeling of the problem involving two hierarchical decision entities: (1) a scheduling entity, and (2) a routing one. The proposed model primarily accounts for nurse qualification, travel costs, and patient preferences on visited nurses. Besides, the proposed mathematical formulation of the problem is tested using the CBC (Coin-or Branch and Cut) optimization solver.

Financial markets operate as dynamic systems susceptible to ongoing changes influenced by recent crises, such as geopolitical and health crises. Due to these factors, investor uncertainty has increased, making it challenging to identify trends in the stock markets. Predicting stock market prices enhances investors’ ability to make accurate investment decisions. This paper proposes an intelligent financial system named Pred-IFDSS, aiming to recommend the best model for accurate predictions of future stock market indexes. Pred-IFDSS includes seven machine learning models: (1) Linear Regression (LR), (2) Support Vector Regression (SVR), (3) eXtreme Gradient Boosting (XGBoost), (4) Simple Recurrent Neural Network (SRNN), (5) Gated Recurrent Unit (GRU), (6) Long Short-Term Memory (LSTM), and (7) Artificial Neural Network (ANN). Each model is tuned using the grid search strategy, trained, and evaluated. Experiments are conducted on three stock market indexes (NASDAQ, S&P 500, and NYSE). To measure the performance of these models, three standard strategic indicators are employed (MSE, RMSE, and MAE). The outcomes of the experiments demonstrate that the error rate in SRNN model is very low, and we recommend it to assist investors in foreseeing future trends in stock market prices and making the right investment decisions.

Abstract Serious games (SGs) are valuable tools for learning, training, and improving skills in various domains because they engage and motivate players to achieve planned processes to reach objectives. Several works provided methods, models, and frameworks to support SG development. However, designers, developers, teachers, and researchers face challenges in creating SG with entertainment and learning balance, and many designed games still do not fulfill the main intended objectives. This paper introduces an approach, called SGDA‐IE with phases and steps to follow during the entire SG design process. It was built on literature review and SG design challenges designers need to consider from the early stages when creating SG. The proposed approach is founded on three perspectives: software engineering best practices, video game industry practices, and SG success factors and provides means to overcome the investigated design challenges. These are characteristics taxonomy model, requirements specification approach, and artifacts iterative evaluation by designer, domain expert, and players. To assess our approach efficacy, we conceived a health, safety, and environment (HSE) training SG for workers on fuel storage sites and petroleum installations. The feedback received is positive and indicates a favorable specification method of the SG, effective participatory design, and control over requirements evolution. The SG playtesting reveals a significant involvement of participants and efficient tracking of the knowledge acquisition.

Malicious apps use various methods to spread viruses, take control of computers and/or IoT devices, and steal sensitive data such as credit card numbers or other personal information. Despite the numerous existing means of intrusion detection, malware code is not easily detectable. The primary issue with current malware detection approaches is their inability to identify novel attacks and obfuscated malware, as they rely on static bases of malware examples, making them susceptible to new unseen malware behaviors. To address this, we propose a new method for malware recognition, which consists of two processes: the first process creates new instances of malware using a memetic algorithm, and the second process detects these new instances of attacks through solid detectors produced by an artificial immune system-based algorithm. Our new malware recognition method has proven its merits through thorough experiments on widely used datasets and evaluation metrics, and has been compared to prominent state-of-the-art methods.

Feature construction represents a crucial data preprocessing technique in machine learning applications because it ensures the creation of new informative features from the original ones. This fact leads to the improvement of the classification performance and the reduction of the problem dimensionality. Since many feature construction methods require discrete data, it is important to perform discretization in order to transform the constructed features given in continuous values into their corresponding discrete versions. To deal with this situation, the aim of this paper is to jointly perform feature construction and feature discretization in a synchronous manner in order to benefit from the advantages of each process. Thus, we propose here to model the discretization-based feature construction task as a bi-level optimization problem in which the constructed features are evaluated based on their optimized sequence of cut-points. The resulting algorithm is termed Discretization-Based Feature Construction (Bi-DFC) where the proposed model is solved using an improved version of an existing co-evolutionary algorithm, named I-CEMBA that ensures the variation of concatenation trees. Bi-DFC performs the selection of original attributes at the upper level and ensures the creation and the evaluation of constructed features at the upper level based on their optimal corresponding sequence of cut-points. The obtained experimental results on ten high-dimensional datasets illustrate the ability of Bi-DFC in outperforming relevant state-of-the-art approaches in terms of classification results.

Multi-Objective Evolutionary Algorithms (MOEAs) is actually one of the most attractive and active research field in computer science. Significant research has been conducted in handling complex multi-objective optimization problems within this research area. The Non-Dominated Sorting Genetic Algorithm (NSGA-II) has garnered significant attention in various domains, emphasizing its specific popularity. However, the complexity of this algorithm is found to be$O(MN^{2})$with$M$objectives and$N$solutions, which is considered computationally demanding. In this paper, we are proposing a new variant of NSGA-II termed (Efficient-NSGA-II) based on our recently proposed quick non-dominated sorting algorithm with quasi-linear average time complexity; thereby making the NSGA-II algorithm efficient from a computational cost viewpoint. Experiments demonstrate that the improved version of the algorithm is indeed much faster than the previous one. Moreover, comparisons results against other multi-objective algorithms on a variety of benchmark problems show the effectiveness and the efficiency of this multi-objective version.

Drug target interaction is an important area of drug discovery, development, and repositioning. Knowing that in vitro experiments are time-consuming and computationally expensive, the development of an efficient predictive model is a promising challenge for Drug-Target Interactions (DTIs) prediction. Motivated by this problem, we propose in this paper a new prediction model called DeepCNN-DTI to efficiently solve such complex real-world activities. The main motivation behind this work is to explore the advantages of a deep learning strategy with feature extraction techniques, resulting in an advanced model that effectively captures the complex relationships between drug molecules and target proteins for accurate DTIs prediction. Experimental results generated based on a set of data in terms of accuracy, precision, sensitivity, specificity, and F1-score demonstrate the superiority of the model compared to other competing learning strategies.

Green manufacturing imposes higher expectations on manufacturing engineering, not only with respect to classic competitive factors such as cost, time and quality, but also with sustainable factors such as resources and energy. In this paper, we investigate green flexible job shop scheduling problem (GFJSP) with variable processing speeds. To solve the GFJSP problem, we propose an adaptive Variable Neighborhood Search to minimize the makespan and the total energy consumption. A number of experiments have been conducted to evaluate the performance of our proposed adaptive VNS algorithm. A comparative study was presented and have verified the out performance of the proposed algorithm against other VNS variants.

Financial markets are dynamic and open systems. They are subject to the influence of environmental changes. For this reason, predicting stock market prices is a difficult task for investors due to the volatility of the financial stock markets nature. Stock market forecasting leads investors to make decisions with more confidence based on the prediction of stock market price behavior. Indeed, a lot of analysts are greatly focused in the research domain of stock market prediction. Generally, the stock market prediction tools are categorized into two types of algorithms: (1) linear models like Auto Regressive (AR), Moving Average (MA), Auto-Regressive Integrated Moving Average (ARIMA), and (2) non-linear models like Autoregressive Conditionally Heteroscedastic (ARCH), Generalized Autoregressive Conditional Heteroskedasticity (GARCH) and recently Neural Network (NN)). This paper aspires to crucially predict the stock index movement for National Association of Securities Dealers Automated Quotations (NASDAQ) based on deep learning networks. We propose a hybrid stock price prediction model using Convolutional Neural Network (CNN) for feature selection and Neural Network models to perform the task of prediction. To evaluate the performance of the proposed models, we use five regression evaluation metrics: Root Mean Square Error (RMSE), Mean Square Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and R-Square (R2), and the Execution Time (ET) metric to calculate the necessary time for running each hybrid model. The results reveal that error rates in the CNN-BGRU model are found to be lower compared to CNN-GRU, CNN-LSTM, CNN-BLSTM and the the existing hybrid models. This research work produces a practical experience for decision makers on financial time series data.

The prediction of stock price movements is one of the most challenging tasks in financial market field. Stock price trends depended on various external factors like investor's sentiments, health and political crises which can make stock prices more volatile and chaotic. Lately, two crises affected the variation of stock prices, COVID-19 pandemic and Russia-Ukraine conflict. Investors need a robust system to predict future stock trends in order to make successful investments and to face huge losses in uncertainty situations. Recently, various machine learning (ML) models have been proposed to make accurate stock movement predictions. In this paper, a framework including five ML classifiers (Gaussian Naive Bayes (GNB), Random Forest (RF), Gradient Boosting (GB), Support Vector Machine (SVM), and K-Nearest Neighbors (kNN))) is proposed to predict the closing price trends. Technical indicators are calculated and used with historical stock data as input. These classifiers are hybridized with Principal Component Analysis method (PCA) for feature selection and Grid Search (GS) Optimization Algorithm for hyper-parameters tuning. Experimental results are conducted on National Association of Securities Dealers Automated Quotations (NASDAQ) stock data covering the period from 2018 to 2023. The best result was found with the Random Forest classifier model which achieving the highest accuracy (61%).

Discretization-based feature selection (DBFS) approaches have shown interesting results when using several metaheuristic algorithms, such as particle swarm optimization (PSO), genetic algorithm (GA), ant colony optimization (ACO), etc. However, these methods share the same shortcoming which consists in encoding the problem solution as a sequence of cut-points. From this cut-points vector, the decision of deleting or selecting any feature is induced. Indeed, the number of generated cut-points varies from one feature to another. Thus, the higher the number of cut-points, the higher the probability of selecting the considered feature; and vice versa. This fact leads to the deletion of possibly important features having a single or a low number of cut-points, such as the infection rate, the glycemia level, and the blood pressure. In order to solve the issue of the dependency relation between the feature selection (or removal) event and the number of its generated potential cut-points, we propose to model the DBFS task as a bilevel optimization problem and then solve it using an improved version of an existing co-evolutionary algorithm, named I-CEMBA. The latter ensures the variation of the number of features during the migration process in order to deal with the multimodality aspect. The resulting algorithm, termed bilevel discretization-based feature selection (Bi-DFS), performs selection at the upper level while discretization is done at the lower level. The experimental results on several high-dimensional datasets show that Bi-DFS outperforms relevant state-of-the-art methods in terms of classification accuracy, generalization ability, and feature selection bias.

Cost-sensitive learning is one of the most adopted approaches to deal with data imbalance in classification. Unfortunately, the manual definition of misclassification costs is still a very complicated task, especially with the lack of domain knowledge. To deal with the issue of costs' uncertainty, some researchers proposed the use of intervals instead of scalar values. This way, each cost would be delimited by two bounds. Nevertheless, the definition of these bounds remains as a very complicated and challenging task. Recently, some researches proposed the use of genetic programming to simultaneously build classification trees and search for optimal costs' bounds. As for any classification tree there is a whole search space of costs' bounds, we propose in this paper a bi-level evolutionary approach for interval-based cost-sensitive classification tree induction where the trees are constructed at the upper level while misclassification costs intervals bounds are optimized at the lower level. This ensures not only a precise evaluation of each tree but also an effective approximation of optimal costs intervals bounds. The performance and merits of our proposal are shown through a detailed comparative experimental study on commonly used imbalanced benchmark data sets with respect to several existing works.

To offer innovative malware evolution techniques, it is appealing to integrate approaches that handle imperfect data and knowledge.In fact, malware writers tend to target some precise features within the app's code to camouflage the malicious content.Those features may sometimes present conflictual information about the true nature of the content of the app (malicious/benign).In this paper, we show how the Variable Precision Rough Set (VPRS) model can be combined with optimization techniques, in particular Bilevel-Optimization-Problems (BLOPs), in order to establish a detection model capable of following the crazy race of malware evolution initiated among malware-developers.We propose a new malware detection technique, based on such hybridization, named Variable Precision Rough set Malware Detection (ProRSDet), that offers robust detection rules capable of revealing the new nature of a given app.ProRSDet attains encouraging results when tested against various state-of-the-art malware detection systems using common evaluation metrics.

Multi-objective Bi-Level Optimization Problem (MBLOP) is defined as a mathematical program where one multi-objective optimization task is constrained with another one. In this way, the evaluation of a single upper level solution necessitates the evaluation of the whole lower level problem. This fact brings new complexities to the bi-level framework, added to the conflicting objectives and their evaluation which need a large number of Function Evaluations (FEs). Despite the number of works dedicated to solve bi-level optimization problems, the number of methods applied to the multi-objective combinatorial case is much reduced. Motivated by these observations, we propose in this paper an approximation-based version of our recently proposed Bi-level Multi-objective Chemical Reaction Optimization (BMCRO), which we called BMCROII. The approximation technique is adopted here as a surrogate to the lower level leading then to generate efficiently the lower level optimality. Our choice is justified by two main arguments. First, BMCRO applies a Quick Non-Dominated Sorting Algorithm (Q-NDSA) with quasi-linear computational time complexity. Second, the number of FEs savings gained by the approximation technique can hugely improve the whole efficiency of the method. The proposed algorithm is applied to a new multi-objective formulation of the well-known Bi-level Multi Depot Vehicle Routing Problem (BMDVRP). The statistical analysis demonstrates the outperformance of our algorithm compared to prominent baseline algorithms available in literature. Indeed, a large number of savings are detected which confirms the merits of our proposal for solving such type of NP-hard problems.

Like the case of any software, Web Services (WSs) developers could introduce anti-patterns due to the lack of experience and badly-planned changes. During the last decade, search-based approaches have shown their outperformance over other approaches mainly thanks to their global search ability. Unfortunately, these approaches do not consider the uncertainty of class labels. In fact, two experts could be uncertain about the smelliness of a particular WS interface but also about the smell type. Currently, existing works reject uncertain data that correspond to WSs interfaces with doubtful labels. Motivated by this observation and the good performance of the possibilistic K-NN classifier in handling uncertain data, we propose a new evolutionary detection approach, named Web Services Anti-patterns Detection and Identification using Possibilistic Optimized K-NNs (WS-ADIPOK), which can cope with the uncertainty based on the Possibility Theory. The obtained experimental results reveal the merits of our proposal regarding four relevant state-of-the-art approaches.

A code smells detection rule is a combination of metrics with their corresponding crisp thresholds and labels. The goal of this paper is to deal with metrics' thresholds uncertainty; as usually such thresholds could not be exactly determined to judge the smelliness of a particular software class. To deal with this issue, we first propose to encode each metric value into a binary possibility distribution with respect to a threshold computed from a discretization technique; using the Possibilistic C-means classifier. Then, we propose ADIPOK-UMT as an evolutionary algorithm that evolves a population of PK-NN classifiers for the detection of smells under thresholds' uncertainty. The experimental results reveal that the possibility distribution-based encoding allows the implicit weighting of software metrics (features) with respect to their computed discretization thresholds. Moreover, ADIPOK-UMT is shown to outperform four relevant state-of-art approaches on a set of commonly adopted benchmark software systems.

Driving control in the car-following (CF) driving behavior has two aspects. First, in what measure an approximation distance is taken as a safe distance guaranteeing the safety of the follower drivers. Second, how to control the follower's vehicle velocities based on the stimulus of the leading vehicle. In this context, to resolve the driving control problem in the CF driving behavior, a bilevel optimization is presented in this paper, based on the behaviors of the follower and leader drivers. Bearing in mind that mathematics has contributed to the imitation of human behaviors, they are now reaching a level of complexity requiring the entry on the scene of a new player, which is artificial intelligence. Thus, in this paper; we used the fuzzy logic theory for modeling a follower driver with a nonnormative behavior. To validate our model, we used a data set from the program of the US Federal Highway Administration. Therefore, according to the experimental results, there is homogeneity between the actual and the simulated travel trajectories in terms of deviation. Besides, the driver's behavior adopted (normative or nonnormative) is reflected in his reactions to the various components of the road.

Feature selection and construction are important pre-processing techniques in data mining. They allow not only dimensionality reduction but also classification accuracy and efficiency improvement. While feature selection consists in selecting a subset of relevant features from the original feature set, feature construction corresponds to the generation of new high-level features, called constructed features, where each one of them is a combination of a subset of original features. However, different features can have different abilities to distinguish different classes. Therefore, it may be more difficult to construct a better discriminating feature when combining features that are relevant to different classes. Based on these definitions, feature construction could be seen as a BLOP (Bi-Level optimization Problem) where the feature subset should be defined in the upper level and the feature construction is applied in the lower level by performing mutliple followers, each of which generates a set class dependent constructed features. In this paper, we propose a new bi-level evolutionary approach for feature construction called BCDFC that constructs multiple features which focuses on distinguishing one class from other classes using Genetic Programming (GP). A detailed experimental study has been conducted on six high-dimensional datasets. The statistical analysis of the obtained results shows the competitiveness and the outperformance of our bi-level feature construction approach with respect to many state-of-art algorithms.

Bi-level optimization has gained a lot of interest during the last decade. This framework is suitable to model several real-life situations. Bi-level optimization problems refer to two related optimization tasks, each one is assigned to a decision level (i.e., upper and lower levels). In this way, the evaluation of an upper level solution requires the evaluation of the lower level. This hierarchical decision making necessitates the execution of a significant number of Function Evaluations (FEs). When dealing with a multi-objective optimization context, new complexities are added and imposed by the conflicting objectives and their evaluation techniques. In this paper, we aim to reduce the induced complexity using approximation techniques in order to obtain the lower level optimality. To this end, ideas from multi-objective optimization have been extracted, improved, and hybridized with evolutionary methods to build an efficient approach for Multi-objective Bi-Level Optimization Problems (MBLOPs). In this work, three techniques are suggested: (1) a complete lower level approximation Pareto front procedure, (2) a reference-based approximation selection procedure, and (3) a sub-set reference-based approximation selection one. The proposed variants are applied to a new multi-objective formulation of a well-known combinatorial problem integrating two systems in the supply chain management, namely, the Bi-level Multi Depot Vehicle Routing Problem (Bi-MDVRP). The statistical analysis demonstrates the efficiency of each algorithm according to a set of metrics. Indeed, a large number of savings are detected which confirms the efficiency of our proposals for solving combinatorial optimization problems.

The rise of terrorism over the past decade did not only hinder the development of some countries, but also it continues to destroy humanity. To face this concept of an emerging crisis, every country and every citizen is responsible for the fight against terrorism. As conventional plans became useless against terrorism, governments are required to establish innovative concepts and technologies to support units in this asymmetric war. In this paper, we propose a new multi-agent model for counter-terrorism characterized by a methodical process and a flexibility to handle different contingency scenarios. The division of labour in our multi-agent model improves decision making and the structuring of organisational plans.

Decomposition-based evolutionary algorithms using predefined reference points have shown good performance in many-objective optimization. Unfortunately, almost all experimental studies have focused on problems having regular Pareto fronts (PFs). Recently, it has been shown that the performance of such algorithms is deteriorated when facing irregular PFs, such as degenerate, discontinuous, inverted, strongly convex, and/or strongly concave fronts. The main issue is that the predefined reference points may not all intersect with the PF. Therefore, many researchers have proposed to update the reference points with the aim of adapting them to the discovered Pareto shape. Unfortunately, the adaptive update does not really solve the issue for two main reasons. On the one hand, there is a considerable difficulty to set the time and the frequency of updates. On the other hand, it is not easy to define how to update the search directions for an unknown PF shape. This article proposes to approximate irregular PFs using a set of predefined normal-boundary intersection (NBI) directions. The main motivation behind this article is that when using a set of well-distributed NBI directions, all these directions intersect with the PF regardless of its shape, except for the case of discontinuous and/or degenerate fronts. To handle the latter cases, a simple interaction mechanism between the decision maker (DM) and the algorithm is used. In fact, the DM is asked if the number of NBI directions needs to be increased in some stages of the evolutionary process. If so, the resolution of the NBI directions that intersect the PF is increased to properly cover discontinuous and/or degenerate PFs. Our experimental results on benchmark problems with regular and irregular PFs, having up to fifteen objectives, show the merits of our algorithm when compared to eight of the most representative state-of-the-art algorithms.

Bi-clustering is one of the main tasks in data mining with several application domains. It consists in partitioning a data set based on both rows and columns simultaneously. One of the main difficulties in bi-clustering is the issue of finding the number of bi-clusters, which is usually a user-specified parameter. Recently, in 2017, a new multi-objective evolutionary clustering algorithm, called MOCK-II, has shown its effectiveness in data clustering while automatically determining the number of clusters. Motivated by the promising results of MOCK-II, we propose in this paper a hybrid extension of this algorithm for the case of bi-clustering. Our new algorithm, called MOBICK, uses an efficient solution encoding, an effective crossover operator, and a heuristic mutation strategy. Similarly to MOCK-II, MOBICK is able to find automatically the number of bi-clusters. The outperformance of our algorithm is shown on a set of real gene expression data sets against several existing state-of-the-art works. Moreover, to be able to compare MOBICK to MOCK-I and MOCK-II, we have designed two basic extensions of MOCK-I and MOCK-II for the case of bi-clustering that we named B-MOCK-I and B-MOCK-II. Again, the experimental results confirm the merits of our proposal.

Feature selection and construction are important pre-processing techniques in machine learning and data mining. They may allow not only dimensionality reduction but also classifier accuracy and efficiency improvement. Feature selection aims at selecting relevant features from the original feature set, which could be less informative to achieve good performance. Feature construction may work well as it creates new highlevel features, but these features do not have the same degree of importance, which makes the use of weighted-features construction a very challenging topic. In this paper, we propose a bi-level evolutionary approach for efficient feature selection and simultaneous feature construction and feature weighting, called Bi-level Weighted-Features Construction (BWFC). The basic idea of our BWFC is to exploit the bi-level model for performing feature selection and weighted-features construction with the aim of finding an optimal subset of features combinations. Our approach has been assessed on six high-dimensional datasets and compared against three existing approaches, using three different classifiers for accuracy evaluation. Experimental results show that our proposed algorithm gives competitive and better results with respect to the state-of-the-art algorithms.

Social media are becoming widely used in the healthcare field as a patients-caregivers communication tool giving birth to new sources of information rich with the knowledge that may improve this field. Therefore, social media data analysis becomes a real business requirement for healthcare industrials and data scientists. However, regarding their complexity and unstructured character, existing natural language processing tools cannot succeed their exploitation. In the literature, a wide range of approaches appeared based on dictionaries, linguistic patterns and machine learning having their strengths and weaknesses. In this work, we propose a hybrid system combining the above approaches by taking the advantage of each of them to extract structured and salient drug abuse information from health-related tweets. We improve the system accuracy by real time update of the domain dictionary. We collected 1000000 tweets and we conducted different experiments showing the advantage of hybridization on efficient information extraction from social media data.

In the age of big data, entreprises’ information systems are ingested with data generated from social media which raises the need to integrate it in their business intelligence process for better decision making. However, these new data, streaming, voluminous, unstructured and variant, bring existing data warehousing systems and integration tools to their knees which motivated us to conduct this research work. In this paper, we propose a large scale system based on distributed storage and parallel processing to succeed social media data warehousing. In fact, we combine Storm and Hadoop for structured events extraction from social media data and their integration in the data warehouse. We take the advantage of real time analysis of streaming data offered by Storm and batch processing of large volumes of data of Hadoop which facilitated streaming social media data analysis task. For conceptual representation, we propose a customized multidimensional model in which we add an intermediate table to connect the social media data warehouse with the enterprise data warehouse. We implement it using Oracle 12c and we fed it with events extracted from 1000 000 tweets using Pentaho data integration tool.

Feature selection and construction are important pre-processing techniques in data mining. They may allow not only dimensionality reduction but also classifier accuracy and efficiency improvement. These two techniques are of great importance especially for the case of high-dimensional data. Feature construction for high-dimensional data is still a very challenging topic. This can be explained by the large search space of feature combinations, whose size is a function of the number of features. Recently, researchers have used Genetic Programming (GP) for feature construction and the obtained results were promising. Unfortunately, the wrapper evaluation of each feature subset, where a feature can be constructed by a combination of features, is computationally intensive since such evaluation requires running the classifier on the data sets. Motivated by this observation, we propose, in this paper, a hybrid multiobjective evolutionary approach for efficient feature construction and selection. Our approach uses two filter objectives and one wrapper objective corresponding to the accuracy. In fact, the whole population is evaluated using two filter objectives. However, only non-dominated (best) feature subsets are improved using an indicator-based local search that optimizes the three objectives simultaneously. Our approach has been assessed on six high-dimensional datasets and compared with two existing prominent GP approaches, using three different classifiers for accuracy evaluation. Based on the obtained results, our approach is shown to provide competitive and better results compared with two competitor GP algorithms tested in this study.

The evolving and nonstationary behavior of realworld data generally generated in streaming way creates serious challenges for learning models. Thus, changes may deteriorate previous decision models accuracy, which requires permanent adaptation strategies. Artificial neural networks have been among the popular choice of adaptation strategies to tackle concept drifting data streams, relying on their online learning capabilities. In this paper, the ability of most known neural networks of the literature to learn from data streams in presence of concept drift will be studied and compared using some meaningful criteria. Their limits will be highlighted using a case-study about the design of decision making aid model to improve the flexibility of electrical grids in presence of distributed Wind-PV renewable energy ressources. Finally, a self-adaptive scheme based on the use of neural networks is proposed in order to avoid these limits.

Bi-level optimization problems (BLOPs) are a class of challenging problems with two levels of optimization tasks. The usefulness of bi-level optimization in designing hierarchical decision processes prompted several researchers, in particular the evolutionary computation community, to pay more attention to such kind of problems. Several solution approaches have been proposed to solve these problems; however, most of them are restricted to the continuous case. Motivated by this observation, we have recently proposed a Co-evolutionary Decomposition-based Algorithm (CODBA-II) to solve combinatorial bi-level problems. CODBA-II scheme has been able to improve the bi-level performance and to bring down the computational expense significantly as compared to other competitive approaches within this research area. In this paper, we present an extension of the recently proposed CODBA-II algorithm. The improved version, called CODBA-IILS, further improves the algorithm by incorporating a local search process to both upper and lower levels in order to help in faster convergence of the algorithm. The improved results have been demonstrated on two different sets of test problems based on the bi-level production-distribution problems in supply chain management, and comparison results against the contemporary approaches are also provided.

Abstract In public bus transportation systems, several types of disturbances, such as accidents and congestion, may affect preestablished timetables and visiting hours at stations. Disturbances result in detrimental consequences, degraded performance, and poor quality of service in terms of extended delays and waiting times, punctuality, frequency, and efficiency of shuttles. Despite numerous research on the monitoring and control of public transportation systems by means of buses, distributed monitoring and control architectures that confer intelligence and decision‐making autonomy to buses to react to disturbances are still missing. This article addresses this gap by designing and developing a distributed architecture to monitor and control public transportation buses using multiagent systems. The design relies on biological immunity as a methodological framework that guides the development of knowledge models and decision‐making mechanisms. Knowledge models structure knowledge about disturbances and control decisions, whereas decision‐making mechanisms implement control and reaction strategies. Through experimental validation based on simulation, we show that the suggested immune multiagent distributed control architecture is not only able to maintain performance (average delay/earliness, average total time in simulated network) at acceptable levels but also to improve quality of service in terms of number of served passengers and stations by at least 15% in case of disturbances.

Multiobjective evolutionary algorithms (MOEAs) have proven their effectiveness and efficiency in solving problems with two or three objectives. However, recent studies show that MOEAs face many difficulties when tackling problems involving a larger number of objectives as their behavior becomes similar to a random walk in the search space since most individuals are nondominated with respect to each other. Motivated by the interesting results of decomposition-based approaches and preference-based ones, we propose in this paper a new decomposition-based dominance relation to deal with many-objective optimization problems and a new diversity factor based on the penalty-based boundary intersection method. Our reference point-based dominance (RP-dominance), has the ability to create a strict partial order on the set of nondominated solutions using a set of well-distributed reference points. The RP-dominance is subsequently used to substitute the Pareto dominance in nondominated sorting genetic algorithm-II (NSGA-II). The augmented MOEA, labeled as RP-dominance-based NSGA-II, has been statistically demonstrated to provide competitive and oftentimes better results when compared against four recently proposed decomposition-based MOEAs on commonly-used benchmark problems involving up to 20 objectives. In addition, the efficacy of the algorithm on a realistic water management problem is showcased.

Flexible job shop problems (FJSP) are among the most intensive combinatorial problems studied in literature. These latters cover two main difficulties, namely, machine assignment problem and operation sequencing problem. To reflect as close as possible the reality of this problem, two others constraints are taken into consideration which are: (1) The sequence dependent setup time and (2) the learning effects. For solving such complex problem, we propose an evolutionary algorithm (EA) based on genetic algorithm (GA) combined with two efficient local search methods, called, variable neighborhood search (VNS) and iterated local search (ILS). It is well known that the performance of EA is heavily dependent on the setting of control parameters. For that, our algorithm uses a self-adaptive strategy based on: (1) the current specificity of the search space, (2) the preceding results of already applied algorithms (GA, VNS and ILS) and (3) their associated parameter settings. We adopt this strategy in order to detect the next promising search direction and maintain the balance between exploration and exploitation. Computational results show that our algorithm is more effective and robust with respect to other well known effective algorithms.

During the few past years, decomposition has shown a high performance in solving Multi-objective Optimization Problems (MOPs) involving more than three objectives, called as Many-objective Optimization Problems (MaOPs). The performance of most of the existing decomposition-based algorithms has been assessed on the widely used DTLZ and WFG unconstrained test problems. However, the number of works that have been devoted to tackle the problematic of constrained many-objective optimization is relatively very small when compared to the number of works handling the unconstrained case. Recently there has been some interest to exploit infeasible isolated solutions when solving Constrained MaOPs (CMaOPs). Motivated by this observation, we firstly propose an IS-update procedure (Isolated Solution-based update procedure) that has the ability to: (1) handle CMaOPs characterized by various types of difficulties and (2) favor the selection of not only infeasible solutions associated to isolated sub-regions but also infeasible solutions with smaller Constraint Violation (CV) values. The IS-update procedure is subsequently embedded within the Multi-Objective Evolutionary Algorithm-based on Decomposition (MOEA/D). The new obtained algorithm, named ISC-MOEA/D (Isolated Solution-based Constrained MOEA/D), has been shown to provide competitive and better results when compared against three recent works on the CDTLZ benchmark problems.

The flexible job shop problem (FJSP) has an important significance in both fields of production management and combinatorial optimization. This problem covers two main difficulties, namely, machine assignment problem and operation sequencing problem. To reflect as close as possible the reality of this problem, the sequence dependent setup time is taken into consideration. For solving such a complex problem, we propose a hybrid algorithm based on a genetic algorithm (GA) combined with iterated local search (ILS). It is well known that the performance of an algorithm is heavily dependent on the setting of control parameters. For that, our algorithm uses a self-adaptive strategy based on : (1) the current specificity of the search space, (2) the preceding results of already applied algorithms (GA and ILS) and (3) their associated parameter settings. We adopt this strategy in order to detect the next promising search direction and maintain the balance between exploration and exploitation. Computational results show that our algorithm provides better solutions than other well known algorithms.

Bi-level optimization problems (BOPs) are a class of challenging problems with two levels of optimization tasks. The main goal is to optimize the upper level problem which has another optimization problem as a constraint. In these problems, the optimal solutions to the lower level problem become possible feasible candidates to the upper level one. Such a requirement makes the optimization problem difficult to solve, and has kept the researchers busy towards devising methodologies, which can efficiently handle the problem. Recently, a new research field, called EBO (Evolutionary Bi-Level Optimization) has appeared thanks to the promising results obtained by the use of EAs (Evolutionary Algorithms) to solve such kind of problems. However, most of these promising results are restricted to the continuous case. The number of existing EBO works for the discrete (combinatorial case) bi-level problems is relatively small when compared to the field of evolutionary continuous BOP. Motivated by this observation, we have recently proposed a Co-evolutionary Decomposition-Based Algorithm (CODBA) to solve combinatorial bi-level problems. The recently proposed approach applies a Genetic Algorithm to handle BOPs. Besides, a new recently proposed meta-heuristic called CRO has been successfully applied to several practical NP-hard problems. To this end, we propose in this work a CODBA-CRO (CODBA with Chemical Reaction Optimization) to solve BOP. The experimental comparisons against other works within this research area on a variety of benchmark problems involving various difficulties show the effectiveness and the efficiency of our proposal.

Bi-level optimization problems are a class of challenging optimization problems, that contain two levels of optimization tasks. In these problems, the optimal solutions to the lower level problem become possible feasible candidates to the upper level problem. Such a requirement makes the optimization problem difficult to solve, and has kept the researchers busy towards devising methodologies, which can efficiently handle the problem. In recent decades, it is observed that many efficient optimizations using modern advanced EAs have been achieved via the incorporation of domain specific knowledge. In such a way, the embedment of domain knowledge about an underlying problem into the search algorithms can enhance properly the evolutionary search performance. Motivated by this issue, we present in this paper a Memetic Evolutionary Algorithm for Bi-level Combinatorial Optimization (M-CODBA) based on a new recently proposed CODBA algorithm with transfer learning to enhance future bi-level evolutionary search. A realization of the proposed scheme is investigated on the Bi-CVRP and Bi-MDVRP problems. The experimental studies on well established benchmarks are presented to assess and validate the benefits of incorporating knowledge memes on bi-level evolutionary search. Most notably, the results emphasize the advantage of our proposal over the original scheme and demonstrate its capability to accelerate the convergence of the algorithm.

In parallel to optimizing energy consumption within houses, users' comfort is increasingly considered as an essential success criterion for automated smart home solutions. From the user perspective, balancing trade-offs between energy consumption and users' comfort when scheduling home appliances is a challenging task mainly within dynamic context (energy price, budget, user preferences, energy source, etc). To address this challenge, this paper has modeled appliances scheduling as a dynamic constrained multi-objective optimization problem and have leveraged a recently introduced dynamic evolutionary algorithm for the problem resolution. Moreover, there are typically multiple inhabitants in the same home who often share context-aware applications with various individual preferences which are likely to be conflicting. We propose a new comfort function to support multi-user conflictual preferences. Our experimental results have shown that our approach has a confirmed advantage on the user comfort while coping dynamically with the context changes.

Mining frequent itemsets is a very interesting issue in Data Streams handling, useful for several real world applications. This task reveals many challenges such the one-pass principle as well as performance problems due to the huge volumes of Data Streams. Performance is defined in terms of CPU and main memory consumption in terms of uncertainty management issues. In this paper, we introduce the concept of Evidential Data Streams and we present a new innovative algorithm for mining frequent itemsets from evidential data streams, based on the evidence theory concepts.

In recent years, social networks have become very popular. Twitter, a micro-blogging service, is estimated to have about 200 million registered users and these users create approximately 65 million tweets a day. Twitter constitutes a powerful medium today that people use to express their thoughts and intentions. The challenge is that each tweet is limited in 140 characters, and is hence very short. It may contain slang and misspelled words. Thus, it is difficult to apply traditional NLP techniques which are designed for working with formal languages, into Twitter domain. Another challenge is that the total volume of tweets is extremely high, and it takes a long time to process. In this paper, we describe a large-scale distributed system for intentions analysis process based on lexico semantic patterns using Hadoop Distributed File System (HDFS) and MapReduce functions. We conduct a case study of user intentions in the commercial field. The proposed method has stably performed data gathering and data loading. Besides, it has maintained stable load balancing of memory and CPU resources during data processing by the HDFS system. The proposed MapReduce functions have effectively performed intentions analysis in the experiments. Finally, obtained results show the importance and effectiveness of intentions detection using semantic patterns.

Since their emergence, social media have become a reliable source of social events which attracted the interest of research community to extract them for many business requirements. However, unlike formal sources like news articles, social data exploitation for events extraction is much harder regarding the complex character of social text. Many approaches, ranging from linguistic techniques to learning algorithms, were proposed to succeed this task. Nevertheless, achieved results are weak regarding the complexity and completeness of the task. In this paper, we focus on private events extraction from Twitter by tracking digital drug abusers. We propose a hybrid approach in which we combine strengths of linguistic rules and learning techniques looking for better performance. In fact, we use linguistic rules to build an automatically annotated training set and extract a set of features as well, to be used in a learning process in order to improve obtained results. The proposed approach outperforms the baseline by 24,8% thanks to combination of techniques.

Truck arrival management forms a very active stream of research and a crucial challenge for a cross-dock terminals. The study focuses on the truck congestion problem, which leads to a lower operation efficiency and a longer waiting time at the gate and at the yard. One of the operational measures to solve this problem is the truck appointment system. It is used to coordinate the major cross-dock planning activities and to regulate the arrival time of trucks at the cross-dock. When the trucker get an appointment time different to its preference time, then we are talking about a truck deviation time. Because the deviation will result in daily operations schedule, an optimization model for truck appointment was proposed in this paper. In the model, the truck deviation time was minimized subject to the constraints of resources availability including dock doors, yard zones, gate lanes, workforce and material handling systems. To solve the model, a method based multi-agent system to real-time truck scheduling, that take into account the uncertainty of arrival time as an operational characteristic, was designed. It ensures a negotiation among truck agents and resource agents. Lastly, a numerical experiments are provided to illustrate the validity of the model and to illustrate the working and benefit of our approach.

This paper presents a recommendation approach that proposes personalized queries to SOLAP users in order to enhance the exploitation of spatial data warehouses. The approach allows implicit extraction of the preferences and needs of SOLAP users using a spatial-semantic similarity measure between queries of different users. The proposal is defined theoretically and validated by experiments.

Several optimization problems encountered in practice have two levels of optimization instead of a single one. These BLOPs (Bi-Level Optimization Problems) are very computationally expensive to solve since the evaluation of each upper level solution requires finding an optimal solution for the lower level. Recently, a new research field, called EBO (Evolutionary Bi-Level Optimization) has appeared thanks to the promising results obtained by the use of EAs (Evolutionary Algorithms) to solve such kind of problems. Most of these promising results are restricted to the continuous case. Motivated by this observation, we propose a new bi-level algorithm, called CODBA (CO-Evolutionary Decomposition based Bi-level Algorithm), to tackle combinatorial BLOPs. The basic idea of our CODBA is to exploit decomposition, parallelism, and co-evolution within the lower level in order to cope with the high computational cost. CODBA is assessed on a set of instances of the bi-level MDVRP (MultiDepot Vehicle Routing Problem) and is confronted to two recently proposed bi-level algorithms. The statistical analysis of the obtained results shows the merits of CODBA from effectiveness and efficiency viewpoints.

Dynamic Multi-objective Optimization (DMO) is a challenging research topic since the objective functions, constraints, and problem parameters may change over time. Several evolutionary algorithms have been proposed to deal with DMO problems. Nevertheless, they were restricted to unconstrained or domain constrained problems. In this work, we focus on the dynamicty of problem constraints along with time-varying objective functions. As this is a very recent research area, we have observed a lack of benchmarks that simultaneously take into account these characteristics. To fill this gap, we propose a set of test problems that extend a suite of static constrained multi-objective problems. Moreover, we propose a new version of the Dynamic Non dominated Sorting Genetic Algorithm II to deal with dynamic constraints by replacing the used constraint-handling mechanism by a more elaborated and self-adaptive penalty function. Empirical results show that our proposal is able to: (1) handle dynamic environments and track the changing Pareto front and (2) handle infeasible solutions in an effective and efficient manner which allows avoiding premature convergence. Moreover, the statistical analysis of the obtained results emphasize the advantages of our proposal over the original algorithm on both aspects of convergence and diversity on most test problems.

We introduce and study the problem of obtaining a spatial or temporal configuration that maximizes the number of constraints satisfied in a qualitative constraint network (QCN). We call this problem the MAX-QCN problem and prove that it is NP-hard for most of the qualitative calculi. We also propose a complete generic branch and bound algorithm for solving the MAX-QCN problem. This algorithm builds on techniques used in the literature for solving the consistency checking problem and the minimal labeling problem of a given QCN. In particular, we make use of a tractable subclass of relations, a chordal graph provided by a triangulation of the input QCN, and the partial weak composition as a filtering method. The experimentation that we have conducted with QCNs from the Interval Algebra and the Region Connection Calculus shows the interest of our proposed algorithm.

Recently, a new metaheuristic called chemical reaction optimization was proposed. This search algorithm, inspired by chemical reactions launched during collisions, inherits several features from other metaheuristics such as simulated annealing and particle swarm optimization. This fact has made it, nowadays, one of the most powerful search algorithms in solving mono-objective optimization problems. In this paper, we propose a multiobjective variant of chemical reaction optimization, called nondominated sorting chemical reaction optimization, in an attempt to exploit chemical reaction optimization features in tackling problems involving multiple conflicting criteria. Since our approach is based on nondominated sorting, one of the main contributions of this paper is the proposal of a new quasi-linear average time complexity quick nondominated sorting algorithm; thereby making our multiobjective algorithm efficient from a computational cost viewpoint. The experimental comparisons against several other multiobjective algorithms on a variety of benchmark problems involving various difficulties show the effectiveness and the efficiency of this multiobjective version in providing a well-converged and well-diversified approximation of the Pareto front.

Dynamic multi-objective optimization problems involve the simultaneous optimization of several competing objectives where the objective functions and/or constraints may change over time. Evolutionary algorithms have been considered as popular approaches to solve such problems. Despite the considerable number of studies reported in evolutionary optimization in dynamic environments, most of them are restricted to the single objective case. Moreover, the majority of dynamic multi-objective optimization algorithms are based on the use of some techniques to detect or predict changes which is sometimes difficult or even impossible. In this work, we address the problem of dynamic multi-objective optimization with undetectable changes. To achieve this task, we propose a new algorithm called Multiple Reference Point-based Multi-Objective Evolutionary Algorithm (MRP-MOEA) which does not need to detect changes. Our algorithm uses a new reference point-based dominance relation ensuring the guidance of the search towards the Pareto optimal front. The performance of our proposed method is assessed using various benchmark problems. Furthermore, the comparative experiments show that MRP-MOEA outperforms serveral dynamic multi-objective optimization algorithms not only in tracking the Pareto front but also in maintainig diversity over time albeit the changes are undetectable.

Several engineering problems involve simultaneously several objective functions where at least one of them is expensive to evaluate. This fact has yielded to a new class of Multi-Objective Problems (MOPs) called expensive MOPs. Several attempts have been conducted in the literature with the goal to minimize the number of expensive evaluations by using surrogate models stemming from the machine learning field. Usually, researchers substitute the expensive objective function evaluation by an estimation drawn from the used surrogate. In this paper, we propose a new way to tackle expensive MOPs. The main idea is to use Neural Networks (NNs) within the Indicator-Based Evolutionary Algorithm (IBEA) in order to estimate the contribution of each generated offspring in terms of hypervolume. After that, only fit individuals with respect to the estimations are exactly evaluated. Our proposed algorithm called NN-SS-IBEA (Neural Networks assisted Steady State IBEA) have been demonstrated to provide good performance with a low number of function evaluations when compared against the original IBEA and MOEA/D-RBF on a set of benchmark problems in addition to the airfoil design problem.

In online services, making anonymous transactions is a crucial need in order to ensure the user's trust towards a particular service. In this context, anonymity quantification is required to provide at which level the e-service respects the user privacy regarding the link between his/her identity and actions. Most of the existing researches are limited to the anonymity quantification in a static way and based, mainly, on the user's set size. In this paper, the authors propose a new multi-agent based approach for anonymity quantification in e-services considering dynamic and mobile environment's characteristics. The authors' quantification is based on the fuzzy logic. It is based not only on the anonymity set size, which is always known in advance, but also on a set of other criteria such as the number of users and the priori and posteriori knowledge about internal and external attackers of an e-service. The carried out experimentations show competitive and better results when compared to other recently proposed anonymity quantification.

Autonomous software agents are increasingly used in a wide range of applications. Thus, testing these entities is extremely crucial. However, testing autonomous agents is still a hard task since they may react in different manners for the same input over time. To address this problem, Nguyen et al. [6] have introduced the first approach that uses evolutionary optimization to search for challenging test cases. In this paper, we extend this work by studying experimentally the effect of the number of objectives on the obtained test cases. This is achieved by proposing five additional objectives and solving the new obtained problem by means of a Preference-based Many-Objective Evolutionary Testing (P-MOET) method. The obtained results show that the hardness of test cases increases with the rise of the number of objectives.

During the two last decades, evolutionary algorithms have been successfully used to solve multiobjective optimization problems. Several works have been established to improve convergence and diversity. Recently, several multiobjective artificial immune systems have shown their ability to solve multiobjective optimization problems. However, in reality, decision makers are not interested with the whole optimal Pareto front rather than the portion of the Pareto front that matches at most their preferences, i.e., the region of interest. In this paper, we propose a new dominance relation inspired from several ideas of the danger theory, called Danger Zone-based dominance (DZ-dominance), which guides the search process towards the preferred part of the Pareto front. The DZ-dominance is incorporated within the Nondominated Neighbor Immune Algorithm (NNIA). The new preference-based algorithm, named DZ-NNIA, has demonstrated its ability to guide the search based on decision maker's preferences. Moreover, comparative experiments show that our algorithm outperforms the most recent preference-based immune algorithm HMIA and the preference-based multiobjective evolutionary algorithm g-NSGA-II.

Recent studies on Evolutionary Multi-objective Optimization (EMO) aim at focusing the search only on those portions of the front which satisfy the preferences of the Decision Maker (DM), i.e., the Regions Of Interest (ROIs), rather than approximating the whole Pareto front. Most studies assume the uniqueness of the DM which is not the case for several decision making situations. In this study, we address this problematic by providing the DMs with an agent-based negotiation support system to aggregate their conflicting preferences before the beginning of the evolutionary process. This negotiation system helps the DMs to confront and adjust their preferences through a number of negotiation rounds. The system output is a set of social preferences which will be injected subsequently in a preference-based EMO Algorithm (EMOA) in order to guide the search towards a satisfying social ROI. The usefulness of the proposed system is demonstrated through a case study.

Nadir point represents important information to multi-objective optimization practitioners. Along with the ideal point, the nadir point: (1) provides information about the ranges of the objectives at the Pareto optimality stage, (2) helps the decision maker to easily state his/her preferences, (3) facilitates the visualization of Pareto optimal solutions for highly dimension multi-objective problems, etc. Contrary to the ideal point which can be easily computed by optimizing each objective individually over the search space, the nadir point is constructed from worst objective function values of Pareto optimal solutions which makes the accurate estimation of the nadir objective values a difficult task especially when the number of objective functions increases. In this paper, we propose a new memetic preference-based multi-objective evolutionary algorithm, termed MR-NSGA-IIN, to estimate the nadir point. The basic idea is to use extreme solutions from the best non-dominated front as mobile reference points. The mobile reference points are updated in every generation by means of a gradient-based local search procedure in order to speed up the convergence towards the Pareto optimal extreme solutions. The performance assessment of MR-NSGA-IINis carried out on a set of three-to twenty-objective unconstrained/constrained linear/non-linear problems. The proposed approach has shown competitive and better results when compared to other recently proposed nadir point estimation approaches.

Evolutionary multiobjective optimization (EMO) methodologies have gained popularity in finding a representative set of Pareto optimal solutions in the past decade and beyond. Several techniques have been proposed in the specialized literature to ensure good convergence and diversity of the obtained solutions. However, in real world applications, the decision maker is not interested in the overall Pareto optimal front since the final decision is a unique solution. Recently, there has been an increased emphasis in addressing the decision-making task in searching for the most preferred alternatives. In this paper, we introduce a new variant of the Pareto dominance relation, called r-dominance, which has the ability to create a strict partial order among Pareto-equivalent solutions. This fact makes such a relation able to guide the search toward the interesting parts of the Pareto optimal region based on the decision maker's preferences expressed as a set of aspiration levels. After integrating the new dominance relation in the NSGA-II methodology, the efficacy and the usefulness of the modified procedure are assessed through two to ten-objective test problems a priori and interactively. Moreover, the proposed approach provides competitive and better results when compared to other recently proposed preference-based EMO approaches.

Lateral Transshipments afford a valuable mechanism for compensating unmet demands only with on-hand inventory. In this paper we investigate the case where locations have a limited storage capacity. The problem is to determine how much to replenish each period to minimize the expected global cost while satisfying storage capacity constraints. We propose a Real-Coded Genetic Algorithm (RCGA) with a new crossover operator to approximate the optimal solution. We analyze the impact of different structures of storage capacities on the system behaviour. We find that Transshipments are able to correct the discrepancies between the constrained and the unconstrained locations while ensuring low costs and system-wide inventories. Our genetic algorithm proves its ability to solve instances of the problem with high accuracy.

We introduce in this paper a new multi-objective memetic algorithm. This algorithm is a result of hybridization of the NSGA-II algorithm with a new designed local search procedure that we named Pareto Hill Climbing. Verification of our novel algorithm is carried out by testing it on two sets of multi-objective test problems and comparing it to other multi-objective evolutionary algorithms (MOEAs) and other multi-criterion memetic algorithms (MMAs). Simulation results show the algorithm ability in tackling continuous multi-objective problems in terms of convergence and diversity. Our hybrid algorithm (1) outperforms pure MOEAs, (2) is competent with other gradient based MMAs, and (3) can solve non differentiable problems.