Feature selection is a process used during machine learning and data analysis, aimed at selecting the best features to increase model efficiency, decrease complexity, and increase readability. Metaheuristic algorithms are suited to provide solutions to feature selection problems because these problems are combinatorial and require an effective and efficient search through large solution spaces. Over the last decade (2014–2024), numerous approaches have been explored, each with its own optimization strengths and constraints. Swarm intelligence and evolutionary algorithms—including genetic algorithms, particle swarm optimization, and the zebra optimization algorithm—have operated effectively in this area. This study seeks to discuss the metaheuristic algorithms for feature selection and responds to questions in the process. In this study, a case study is provided using datasets from the University of California, Irvine repository, where various metaheuristic algorithms are applied to identify optimal feature subsets.
The vehicle routing problem with time windows is an NP-hard optimization problem vital to logistics and supply chain management. It involves optimizing vehicle routes to serve customers within time windows and capacity limits. This study proposes a hybrid genetic algorithm combining a nearest neighbor-based initialization with advanced mutation operators. The nearest neighbor method ensures high-quality initial solutions by prioritizing proximity and constraints, while multiple mutation operators enhance exploration and exploitation. Tested on the Solomon 100-customer dataset, NN-IHGA outperformed benchmarks, especially on random and mixed datasets, reducing travel costs and vehicle counts. Results highlight NN-IHGA's robustness and adaptability, offering a practical solution for real-world logistics optimization.
The vehicle routing problem with time windows is an NP-hard optimization problem vital to logistics and supply chain management. It involves optimizing vehicle routes to serve customers within time windows and capacity limits. This study proposes a hybrid genetic algorithm combining a nearest neighbor-based initialization with advanced mutation operators. The nearest neighbor method ensures high-quality initial solutions by prioritizing proximity and constraints, while multiple mutation operators enhance exploration and exploitation. Tested on the Solomon 100-customer dataset, NN-IHGA outperformed benchmarks, especially on random and mixed datasets, reducing travel costs and vehicle counts. Results highlight NN-IHGA’s robustness and adaptability, offering a practical solution for real-world logistics optimization.
Algorithms are fundamental to solving complex problems in science and engineering. However, conventional methods often struggle with nonlinear, high-dimensional landscapes. The biologically inspired dung beetle optimization algorithm has offered promising solutions but is still prone to premature convergence and falling into local optima. This article presents a multi-strategy enhanced dung beetle optimization algorithm that integrated tent chaotic mapping for population initialization, a golden sine strategy for position updating, Lévy flights to escape local minima, and dynamic weighting coefficients for adaptive search balancing. These strategies collectively enhanced population diversity and improved the balance between exploration and exploitation. Benchmarking on the CEC2017 test suite and real-world engineering problems demonstrated that the multi-strategy enhanced dung beetle optimization algorithm achieved superior convergence speed, solution accuracy, and robustness when compared with the standard dung beetle optimization algorithm and other state-of-the-art metaheuristics.
This study examines the performance of genetic algorithms, the particle swarm optimization (PSO) algorithm, and the artificial bee colony (ABC) algorithm in procedural content generation for game map layouts. A series of experiments evaluated each algorithm's efficiency based on convergence speed, content quality, and overall map structure. The results showed that the genetic algorithms with tournament selection outperformed the PSO and the ABC algorithms in generating high-quality maps, though the PSO and the ABC algorithms demonstrated competitive performance in specific scenarios. This research highlights the importance of task-specific optimization, suggesting that hybrid approaches could improve game content generation by combining the strengths of different algorithms.
The Multidimensional Knapsack Problem (MDKP) stands as a prominent challenge in combinatorial optimization, with diverse applications across various domains. The Artificial Bee Colony (ABC) algorithm is a swarm intelligence optimization algorithm inspired by the foraging behavior of bees. The aim of this paper is to develop an ABC with the goal of improving the solution quality in comparison to previous studies for the MDKP. In the proposed ABC algorithm, a heuristic method is presented to make employed bees. The roulette wheel and k-tournament methods are investigated for selecting employed bees by onlooker bees. For crossing over, two methods including one-point and uniform are studied. To tune the parameters, the Design of Experiment (DOE) method has been applied. The well-known benchmark test problems have been used to evaluate the proposed algorithm. The results show the absolute superiority of the solutions generated by the proposed algorithm in compared with the previous studies.
When it comes to making financial decisions, choosing stocks is crucial to building a successful portfolio. Stocks are evaluated according to lower risk, and the best stocks are chosen to produce assets that are then utilized to construct the portfolio. In this study, we have compared the integrated particle swarm method to four other algorithms for stock selection and optimization: the genetic algorithm, the Pareto search algorithm, the pattern search algorithm, and quadratic programming in the Matlab toolbox. Six particular stock firms are taken into consideration for this reason during a given time period. First, we will use the aforementioned Matlab toolbox techniques to conduct Markowitz's mean variance model. Additionally, the usual embedded particle swarm methodology and the penalty function approach will be used to create this model. The difference between the averages at ten distinct levels of predicted values will be examined in the next research based on the returns in the chosen portfolios. Statistical tests will be employed to differentiate noteworthy distinctions between the suggested approach and the alternative algorithms. MSC: 65K10,91B05.
Procedural Content Generation (PCG) aims to automatically generate the content of games using algorithmic approaches, as this can reduce the cost of game design and development. PCG algorithms can be applied to all elements of a game, including terrain, maps, stories, dialogues, quests, and characters. A wide variety of search algorithms can be applied to PCG problems; however, those most often used are variations of evolutionary algorithms. This study focuses on comparing three metaheuristic approaches applied to racetrack games, with the specific goal of evaluating the effectiveness of different algorithms in producing game content. To that end, a Genetic Algorithm (GA), Artificial Bee Colony (ABC), and Particle Swarm Optimization (PSO) are applied to a game-level design task to attempt to identify any discernible differences in their performance and identify whether alternative algorithms offer desirable performance characteristics. The results of the study indicate that both the ABC and PSO approaches offer potential advantages to Genetic Algorithm implementation.
The Multidimensional Knapsack Problem (MDKP) stands as a prominent challenge in combinatorial optimization, with diverse applications across various domains. The Artificial Bee Colony (ABC) algorithm is a swarm intelligence optimization algorithm inspired by the foraging behavior of bees. The aim of this paper is to develop an ABC with the goal of improving the solution quality in comparison to previous studies for the MDKP. In the proposed ABC algorithm, a heuristic method is presented to make employed bees. The roulette wheel and k-tournament methods are investigated for selecting employed bees by onlooker bees. For crossing over, two methods including one-point and uniform are studied. To tune the parameters, the Design of Experiment (DOE) method has been applied. The well-known benchmark test problems have been used to evaluate the proposed algorithm. The results show the absolute superiority of the solutions generated by the proposed algorithm in compared with the previous studies.
The fault detection and isolation (FDI) procedure increases the assurance of quality, reliability, and safety of industrial systems. Several faults may appear simultaneously, and the purpose of multi-fault diagnosis is to identify and locate these multiple faults. This work is particularly interested in the diagnosis based on the structural analysis of the system; residuals can be generated and used as fault indicators by model-based fault detection techniques. The isolation is dependent on the structure of the fault signature matrix. A new fault signature that represents the superposition of the fault is produced by simultaneous fault effects, resulting in an additional column in an extended signature matrix. This remedy is rather combinatorial. This research focuses on two methods to isolate multiple faults: (1) A modified enumerative method; (2) A hybrid ant colony optimization algorithm-genetic algorithm (Hybrid ACO-GA) is adapted to the MFD problem which has the advantage of a better research as well as the hybridization with GA.
Arithmetic optimization algorithm (AOA) is a recent population-based metaheuristic widely used for solving optimization problems. However, the emerging large-scale optimization problems pose a great challenge for AOA due to its prohibitive computational cost to traverse the huge solution space effectively. This article proposes a parallel Spark-AOA using Scala on Apache Spark computing platform. Spark-AOA leverages the intrinsic parallel nature of the population-based AOA and the native iterative in-memory computation support of Spark through resilient distributed datasets (RDD) to accelerate the optimization process. Spark-AOA divides the solutions population into several subpopulations that are distributed into multiple RDD partitions and manipulated concurrently. Simulation experiments on different benchmark functions with up to 1,000-dimension and three engineering design problems demonstrate that Spark-AOA outperforms considerably standard AOA and Spark-based implementations of two recent metaheuristics both in terms of run-time and solution quality.
In this article, the authors propose a maximum flow algorithm based on flow matrix. The algorithm only requires the effort to reduce the capacity of the underutilized arcs to that of the respective flow. The optimality of the algorithm is proved by the max-flow min-cut theorem. The algorithm is table-based, thus avoiding augmenting path and residual network concepts. The authors used numerical examples and computational comparisons to demonstrate the efficiency of the algorithm. These examples and comparisons revealed that the proposed algorithm is capable of computing exact solutions while using few iterations as compared to some existing algorithms.
In this paper, we consider an extension of the Dynamic Vehicle Routing Problem with Backhauls integrated with two-dimensional loading problem called DVRPB with 2D loading constraints (2L-DVRPB). In the VRPB, a vehicle can deliver (Linehaul) then collect goods from customers (backhaul) and bring back to the depot. Once customer demand is formed by a set of two-dimensional items the problem will be treat as a 2L-VRPB. The 2L-VRPB has been studied on the static case. However, in most real-life application, new customer requests can be happen over time of backhaul and thus perturb the optimal routing schedule that was originally invented. This problem has not been analysed sofar in the literature. The 2L-DVRPB is an NP-Hard problem, so, we propose to use a Genetic algorithm for routing and a packing problems. We applied our approach in a real case study of the Regional Post Office of the city of Jendouba in the North of Tunisia. Results indicate that the AGA approach is considered as the best approach in terms of solutions quality for a real world routing system.
In this article whale optimization algorithm (WOA) has been applied to solve the combined heat and power economic dispatch (CHPED) problem. The CHPED is energy system which provides both heat and power. Due to presence of valve point loading and the prohibited working region, the CHPED problems become more complex one. The main objective of CHPED problem is to minimize the total cost of fuel as well as heat with fulfill the load demand. This optimization technique shows several advantages like having few input variables, best quality of solution with rapid computational time. The recommended approach is carried out on three test systems. The simulation results of the present work certify the activeness of the proposed technique.
In any construction projects,assessment of liquefaction potential induced due to seismic excitation during earthquake is a critical concern.The objective of present model development is to classify and assess liquefaction potential of soil.This paper addresses Emotional Neural Network(ENN), Cultural Algorithm(CA) and biogeography optimized(BBO) based adaptive neuro-fuzzy inference system (ANFIS) for liquefaction study.The performance of neural emotional network and cultural algorithm has been also discussed. BBO-ANFIS combines the biogeography features to optimize the ANFIS parameters to achieve higher prediction accuracy.The model is trained with case history of liquefaction databases.Two parameters are used as input such as the cyclic stress ratio and standard penetration test (SPT) value.The performance of these models was assessed using different indexes i.e. sensitivity, specificity, FNR, FPR and accuracy rate.The performance of all models is compared.Among the models, the BBO-ANFIS model has been outperformed and can be adopted as new reliable technique for liquefaction study.
The standard Whale Optimization Algorithm (WOA) involves exploitation and exploration operations which require to be balanced for improved performance. This paper suggests a new enhanced WOA to improve the convergence speed and enhance the global optimum by balancing exploitation and exploration processes. New stages have been suggested at the hunting stages of the WOA to increase the exploitation capability. The performance of the modified algorithm has been analyzed on few commonly used test functions and Reactive Power Optimization (RPO) problem. The numerical studies have shown that the proposed WOA variant has outperformed other compared optimization algorithms in terms of global optimum and convergence speed.
Usually, the One-Class Support Vector Machine (OC-SVM) requires a large dataset for modeling effectively the target class independently to other classes. For finding the OC-SVM model, the available dataset is subdivided into two subsets namely training and validation, which are used for training and validating the optimal parameters. This approach is effective when a large dataset is available. However, when training samples are reduced, parameters of the OC-SVM are difficult to find in absence of the validation subset. Hence, this paper proposes various techniques for selecting the optimal parameters using only a training subset. The experimental evaluation conducted on several real-world benchmarks proves the effective use of the new selection parameter techniques for validating the model of OC-SVM classifiers versus the standard validation techniques
This paper discusses the U-shaped assembly line balancing problem in case of stochastic processing time. The problem is formulated using chance-constrained programming and the greedy randomized adaptive search procedure is used to solve the problem. In order to prove the efficiency of the proposed algorithm, 71 problems taken from well-known benchmarks are solved and compared with the theoretical lower bound and 13 of them were compared with another approach used to solve the same problem in another paper, which is beam search. The results show that 59 problems are the same as the theoretical aspiration lower bound. In addition, the results of 11 of 13 problems compared with beam search are the same and the results of 2 problems are better than beam search. The t-test statistics is applied and showed that there is no significance difference between the proposed algorithm and the theoretical lower bound thus, the proposed algorithm shows efficiency when compared with the aspired values of the theoretical lower bound.
This paper plans to develop a novel image compression model with four major phases. (i) Segmentation (ii) Feature Extraction (iii) ROI classification (iv) Compression. The image is segmented into two regions by Adaptive ACM. The result of ACM is the production of two regions, this model enables separate ROI classification phase. For performing this, the features corresponding to GLCM are extracted from the segmented parts. Further, they are subjected to classification via NN, in which new training algorithm is adopted. As a main novelty JA and WOA are merged together to form J-WOA with the aim of tuning the ACM (weighting factor and maximum iteration), and training algorithm of NN, where the weights are optimized. This model is referred as J-WOA-NN. This classification model exactly classifies the ROI regions. During the compression process, the ROI regions are handled by JPEG-LS algorithm and the non-ROI region are handled by wavelet-based lossy compression algorithm. Finally, the decompression model is carried out by adopting the same reverse process.
Social network analysis (SNA) is an active research domain that mainly deals with large social graphs and their properties. Community detection (CD) is one of the active research topics belonging to this domain. Social graphs in real-time are huge, complex, and require more computational resources to process. In this paper, the authors present a CPU-based hybrid parallelization architecture that combines both master-slave and island models. They use particle swarm optimization (PSO)-based clustering approach, which models community detection as an optimization problem and finds communities based on concepts of PSO. The proposed model is scalable, suitable for large datasets, and is tested on real-time social networking datasets with node attributes belonging to all three sizes (small, medium, and large). The model is tested on standard benchmark functions and evaluated on well-known evaluation strategies related to both community clusters and parallel systems to show its efficiency.
