The most successful MOAs are those which prioritize exploration by no more than 25% percent of the search process, while also manifesting a successful exchange from full exploration to full exploitation by the end of such process, with PSO and DE being among the most consistent examples.
Abstract
In this paper, we propose the implementation of a Fuzzy Inference System (FIS) for the systematic evaluation of exploration and exploitation capabilities in population-based Metaheuristic Optimization Algorithms (MOAs). Our approach involves an efficient feature extraction scheme applied through different stages of an algorithm’s search process, from which Dimension-Wise Diversity (DD), Average Distance to Global Best (DB), and Fitness Diversity (FD) are extracted and utilized as inputs for the proposed Mamdani-style FIS. These metrics feature zero additional objective function evaluations and are processed in real time by our proposed Mamdani-style inference system using symmetric triangular membership functions. Since the proposed FIS-based evaluation scheme relies on static membership boundaries and elementary min-max aggregation operators, constant time complexity is ensured, which contrasts with the expensive matrix operations that are frequently required in probabilistic models. To validate this framework, a comprehensive experimental design was performed. Such experiments involve testing with MOAs proposing unique and distinctive search strategies, including well-known examples such as Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Grey Wolf Optimizer (GWO), Differential Evolution (DE) and others. Our experiments also consider 14 benchmark functions (chosen by considering scalability and landscape variety) and three operational configurations (which consider varying numbers of decision variables, population size and iterations). Our results not only expose critical differences in exploration/exploitation capabilities among the compared strategies but also allow the visualization of underlying process characteristics, such as convergence speed and repeatability. Our results imply that the most successful MOAs are those which prioritize exploration by no more than 25% percent of the search process, while also manifesting a successful exchange from full exploration to full exploitation by the end of such process, with PSO and DE being among the most consistent examples.
The problem of insufficient classification accuracy in fuzzy clustering algorithms for multidimensional data is addressed in this paper. To tackle this issue, an improved genetic algorithm (GA)-based fuzzy controller is proposed, which combines the advantages of an improved genetic algorithm and a fuzzy C-means (FCM) clustering algorithm. The population initialization is performed using the Tent chaotic map, while the best individual retention strategy and last elimination selection operator are employed to adjust the population structure. Furthermore, an elitist crossover operator, an adaptive trial mutation operator, and a nonlinear convergence factor are introduced to mitigate the risk of falling into local optima. The controller algorithm integrates the intermediate parameters of FCM clustering into the fitness function of the genetic algorithm, and effectively improves the classification accuracy by screening the optimal feature subset and then fuzzy clustering. The experimental results on the Pistachio, WDBC, and Wine datasets show that the proposed method achieves competitive classification accuracy compared with other FCM-based feature-weighting optimization methods.
Zhi-Wen Li, Xing-Hua Wang, Yongtao Li et al.· Mathematics· 0 citations
Metaheuristic algorithms are a kind of optimization algorithms, which are widely used in reducing the number of variables, and choosing the most relevant, and identifiable ones in medium or high dimensional datasets. This topic, known as feature selection is widely researched and applied where binary metaheuristic algorithms are developed using as an evaluator of the quality of the feature subsets machine learning algorithms. The binary metaheuristics could be created by transforming the continuous space of the metaheuristic by using some operators named transfer function, into a binary one, since feature selection is a binary optimization problem. This paper aims to offer a comparative analysis of three distinct binary variants of metaheuristic optimization algorithms utilizing five different groups of transfer functions to assess their predictive efficacy. Four S-shaped, V-shaped, U-shaped, Q-shaped, and two X-shaped types are proposed for the binarization process. Salp Swarm Optimization, Ant Lion Optimization, and Teaching Learning-Based Optimization algorithms are chosen as the foundational algorithms for assessing each transfer function and comparing their impact on key metrics, including accuracy, F-score, sensitivity, average number of selected features, fitness, and time efficiency. The current work utilizes a medical dataset to predict the existence of gallstones. The experiment aims to compare and propose new solutions applied to the feature selection process. The optimal combinations of transfer functions and metaheuristic algorithms are produced using the combination of the Q3 transfer function together with a binary teaching-based learning method and the V2 transfer function combined with binary antlion optimization, achieving an accuracy of approximately 78%. The research emphasizes the necessity of appropriate binarization strategies and provides insights into selecting transfer functions to improve feature selection in medical data analysis.
Eris Zeqo, Edjola Naka, V. Guliashki· International journal of re...· 0 citations
The proliferation of online misinformation demands the development of highly accurate and computationally efficient automated systems for Fake News Detection. A primary impediment to system performance is the high dimensionality of textual features derived from techniques like TF-IDF, making optimal Feature Selection a critical step. This paper presents a detailed comparative experimental study of two prominent bio-inspired evolutionary metaheuristics, the Genetic Algorithm (GA) and Particle Swarm Optimisation (PSO) used as wrapper-based FS techniques for FND. The methodologies were rigorously tested across two distinct textual datasets: the complex, large-scale FakeNewsNet corpus and a moderate-scale general news dataset. The feature sets, once optimised, were evaluated using six standard Machine Learning (ML) classifiers. The GA-based FS approach, emphasising global exploration, achieved state-of-the-art accuracy of 99.91% with the Random Forest classifier on the FakeNewsNet dataset. In contrast, the PSO-based FS approach, valued for its rapid convergence, yielded a maximum accuracy of 93.29% with the Support Vector Machine (SVM) on the general news dataset. This analysis provides empirical evidence of the intrinsic trade-off between the algorithms: GA is superior for maximising accuracy in high-dimensional, complex textual spaces, while PSO offers a more efficient and practical solution for resource-constrained or moderate-scale FND tasks. The study confirms that evolutionary computation provides a robust, effective pathway for significantly enhancing ML classifier performance in this critical domain.
Nikita Garg, Pritam Singh Negi· International Journal of Edu...· 0 citations
AbstractThe action of combining the components from various algorithms is presently the utmost successful and effective trend in optimization. The primary goal of hybridizing disparate algorithmic concepts is to create better-performing systems that combine the advantages of many pure approaches–that is, hybrid systems that are meant to benefit from synergy. Actually, the secret to getting the best results while addressing a lot of challenging (complex) optimization issues is frequently to combine several algorithmic concepts in a proper way. However, developing a hybrid approach that is incredibly effective is not an easy undertaking. The hybridization of popular metaheuristics like Genetic algorithm, particle swarm optimization, evolutionary algorithms, simulated annealing, variable neighbourhood search, and ant colony optimization with techniques from other fields like artificial intelligence, operations research forms the basis of evolving more efficient and robust solutions. From the extensive review of the literature on hybrid Meta-heuristics algorithms in real life application, it is observed that their flexibility, effectiveness and robustness make them an increasing requirement for a variety of real-life optimization problems such as vehicle routing, traveling salesman problems, and supply chain network design, job scheduling.The goal of this review paper is, to cover the present state of hybrid metaheuristics, summarising their progress, applications, and a wide range of opportunities for further research in a fast-evolving world. different type of hybridization is available: (i) Combinations of using metaheuristics with other metaheuristic methods; (ii) Hybridization of metaheuristics and precise optimization methodologies; (iii) Hybridization of metaheuristics and constraint programming methods; (iv) Integration of machine learning and data mining techniques into metaheuristic techniques.
M. Raji, Arjan Singh, S. Singh et al.· Indian Journal of Industrial...· 0 citations
A solution methodology is developed which utilizes the concepts of weighting sum approach, centroid of fuzzy numbers and component wise optimization to equivalently derive a deterministic multi-objective linear optimization.
Since the cost function of the Quadratic Assignment Problem (QAP) is quadratic and its solution space expands linearly, this problem is computationally challenging. In other words, any problem that can be solved by an exact method on one medium- to large-scale instance should be practically intractable in general for the robust hybrid optimization algorithm, combining Genetic Algorithm (GA) with Cat Swarm Optimization (CSO), proposed in this study. It has both the characteristics of being aggressive with respect to exploration but conservative once a promising solution region is identified and introducing localized search operations for improving solution quality: A genetic algorithm is utilized to carry out recombination on the population. As for Cat Swarm Optimization, it adopts a two-way search model with the seeking mechanism and tracing mechanism applied. The new alternative structure, delta evaluation, uses permutational expression for the solutions. It can reduce computational complexity. The proposed algorithm was tested by running 30 independent trials on known QAPLIB benchmark instances, and then its results were analyzed using both Wilcoxon signed-rank tests and Friedman ranks. Experimental results confirm that in several cases, the hybrid GA (GA + CSO) framework provides good or even superior solution quality compared to classical metaheuristic methods. Median optimality gaps dropped and robustness increased. But the question of how to judge the overall performance, let alone prove that it’s not just a fluke (something relying on only a few tests), can only be decided after further study. This improvement—it is statistically confirmed at the 0.05 level—is significant. The proposed model is useful for solving large-scale NP-hard permutation-based combinatorial optimization. For similar problems and future research, it provides a flexible and scalable framework.
Firas Abdullah Attia, Iraq T. Abbas, R. A. Rahman· Ibn AL-Haitham Journal For P...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
