Search Results (224)

To address the limitations of the traditional Balancing Composite Motion Optimization (BCMO) algorithm—namely weak directional global exploration, insufficient local exploitation accuracy, and a tendency to fall into local optima with reduced population diversity in feature selection tasks—this paper proposes a Multi-Strategy Enhanced Balancing Composite Motion Optimization algorithm (MEBCMO). From a symmetry perspective, MEBCMO exploits the symmetric and asymmetric relationships among candidate solutions in the search space to achieve a better balance between exploration and exploitation. The performance of MEBCMO is enhanced through three complementary strategies. First, an adaptive heat-conduction search mechanism is introduced to simulate thermal transmission behavior, where a Sigmoid function adjusts the heat-conduction coefficient α_T from 0.9 to 0.2 during iterations. By utilizing the symmetric fitness–distance relationship between the current solution and the global best, this mechanism improves the directionality and efficiency of global exploration. Second, a quadratic interpolation search strategy is designed. By constructing a quadratic model based on the current individual, a randomly selected individual, and the global best, the algorithm exploits local symmetric characteristics of the fitness landscape to strengthen local exploitation and alleviate performance degradation in high-dimensional spaces. Third, an elite population genetic strategy is incorporated, in which the top three individuals generate new candidates through symmetric linear combinations with non-elite individuals and Gaussian perturbations, preserving population diversity and preventing premature convergence. To evaluate MEBCMO, extensive global optimization experiments are conducted on the CEC2017 benchmark suite with dimensions of 30, 50, and 100, and comparisons are made with eight mainstream algorithms, including PSO, DE, and GWO. Experimental results demonstrate that MEBCMO achieves superior performance across unimodal, multimodal, hybrid, and composite functions. Furthermore, MEBCMO is combined with LightGBM to form the MEBCMO-LightGBM model for feature selection on 14 public datasets, yielding lower fitness values, higher classification accuracy, and fewer selected features. Statistical tests and convergence analyses confirm the effectiveness, stability, and rapid convergence of MEBCMO in symmetric and complex optimization landscapes. Full article

With the explosive growth of data across various fields, effective data preprocessing has become increasingly critical. Evolutionary and swarm intelligence algorithms have shown considerable potential in feature selection. However, their performance often deteriorates in large-scale problems, due to premature convergence and limited exploration ability. To address these limitations, this paper proposes an algorithm named IHBOFS, a biomimetics-inspired optimization framework that integrates multiple adaptive strategies to enhance performance and stability. The introduction of the Good Point Set and Elite Opposition-Based Learning mechanisms provides the population with a well-distributed and diverse initialization. Furthermore, adaptive exploitation–exploration balancing strategies are designed for each subpopulation, effectively mitigating premature convergence. Extensive ablation studies on the CEC2022 benchmark functions verify the effectiveness of these strategies. Considering the discrete nature of feature selection, IHBOFS is further extended with continuous-to-discrete mapping functions and applied to six real-world datasets. Comparative experiments against nine metaheuristic-based methods, including Harris Hawk Optimization (HHO) and Ant Colony Optimization (ACO), demonstrate that IHBOFS achieves an average classification accuracy of 92.57%, confirming its superiority and robustness in high-dimensional feature selection tasks. Full article

Financial forecasting is a challenging task due to the complexity and nonlinear volatility that characterize modern financial markets. Machine learning algorithms are very effective at increasing prediction accuracy, thereby supporting data-driven decision making, optimizing pricing strategies, and improving financial risk management. In particular, combining machine learning techniques with metaheuristic algorithms often leads to significant performance improvements across various domains. This study proposes a hybrid framework for cryptocurrency price prediction, where Support Vector Regression (SVR) with radial basis function kernel is used to perform the prediction, while a Firefly algorithm is employed for correlation-based feature selection and hyperparameter tuning. To improve search performance, the parameters of the Firefly algorithm are optimized using the Fully Informed Search Algorithm (FISA) which is an improved version of the parameterless Rao algorithm. The model is applied to hourly data of Bitcoin, Ethereum, Binance, Solana and Ripple, separately. The model’s performance is evaluated by comparison with Gated Recurrent Unit (GRU), Multilayer Perceptron (MLP), and SVR methods using MSE, MAE, and MAPE metrics, along with statistical validation by Wilcoxon’s signed-rank test. The results show that the proposed model achieves a superior accuracy and demonstrate the critical importance of feature selection and hyperparameter tuning for achieving accurate predictions in volatile markets. Moreover, customizing both feature sets and model configurations for each cryptocurrency allows the model to capture distinct market characteristics and provides deeper insights into intra-day market dynamics. Full article

Polyphenol oxidase (PPO) is the primary biochemical driver of browning and the subsequent decline of market quality in harvested fruit. In this work, a fully non-invasive analytical framework was built using Visible/Near-Infrared (VIS/NIR) spectroscopy coupled with chemometric modeling in order to estimate PPO activity in two commercially relevant plum cultivars (Khormaei and Khoni). A comprehensive comparative study was conducted utilizing multiple machine learning and linear regression techniques, including Support Vector Regression (SVR), Decision Tree (DT), and Partial Least Squares Regression (PLSR). After acquiring the full VIS/NIR spectra, a suite of metaheuristic feature selection strategies was applied to compress the spectral space to roughly 10–15 highly informative wavelengths. SVR, DT, and PLSR models were then trained and benchmarked using (a) the complete spectral domain and (b) the reduced wavelength subsets. The results consistently demonstrated that non-linear models (DT and SVR) significantly outperformed the linear PLSR method, confirming the inherent complexity and non-linearity of the relationship between the spectra and PPO activity. Across all tests, DT consistently produced the strongest generalization. Under full spectra inputs, DT reached RPD values of 4.93 for Khormaei and 5.41 for Khoni. Even more importantly, the wavelength-reduced configuration further enhanced performance while substantially lowering computational cost, yielding RPDs of 3.32 (Khormaei) and 5.69 (Khoni). The results show that VIS/NIR combined with optimized key-wavelength DT modeling provides a robust, fast, and field-realistic route for quantifying PPO activity in plums without physical destruction of the product. Full article

Dam displacement monitoring is crucial for assessing structural safety; however, conventional models often prioritize single-task prediction, leading to an inherent difficulty in balancing monitoring data quality with model performance. To bridge this gap, this study proposes a novel two-stage analytical framework that synergistically integrates an improved isolation forest (iForest) with a metaheuristic-optimized random forest (RF). The first stage focuses on data cleaning, where Kalman filtering is applied for denoising, and a newly developed Dynamic Threshold Isolation Forest (DTIF) algorithm is introduced to effectively isolate noise and outliers amidst complex environmental loads. In the second stage, the model’s predictive capability is enhanced by first employing the LASSO algorithm for feature importance analysis and optimal subset selection, followed by an Improved Reptile Search Algorithm (IRSA) for fine-tuning RF hyperparameters, thereby significantly boosting the model’s robustness. The IRSA incorporates several key improvements: Tent chaotic mapping during initialization to ensure population diversity, an adaptive parameter adjustment mechanism combined with a Lévy flight strategy in the encircling phase to dynamically balance global exploration and convergence, and the integration of elite opposition-based learning with Gaussian perturbation in the hunting phase to refine local exploitation. Validated against field data from a concrete hyperbolic arch dam, the proposed DTIF algorithm demonstrates superior anomaly detection accuracy across nine distinct outlier distribution scenarios. Moreover, for long-term displacement prediction tasks, the IRSA-RF model substantially outperforms traditional benchmark models in both predictive accuracy and generalization capability, providing a reliable early risk warning and decision-support tool for engineering practice. Full article

The exponential growth of the Internet of Things (IoT) has made it increasingly vulnerable to cyberattacks, where malicious manipulation of network and sensor data can lead to incorrect data classification. IoT data are inherently heterogeneous, comprising sensor readings, network flow records, and device metadata that differ significantly in scale and structure. This diversity motivates transforming tabular IoT data into image-based representations to facilitate the recognition of intrusion patterns and the analysis of spatial correlations. Many deep learning models offer robust detection performance, including CNNs, LSTMs, CNN–LSTM hybrids, and Transformer-based networks, but many of these architectures are computationally intensive and require significant training resources. To address this challenge, this study introduces an evolutionary-driven framework that mathematically formalizes the transformation of tabular IoT data into image-encoded matrices and optimizes feature selection through metaheuristic algorithms. Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Variable Neighborhood Search (VNS) are employed to identify optimal feature subsets for Random Forest (RF) and Extreme Gradient Boosting (XGBoost) classifiers. The approach enhances discrimination by optimizing multi-objective criteria, including accuracy and sparsity, while maintaining low computational complexity suitable for edge deployment. Experimental results on benchmark IoT intrusion datasets demonstrate that VNS-XGBoost configurations performed better on the IDS2017 and IDS2018 benchmarks, achieving accuracies up to 0.99997 and a significant reduction in Type II errors (212 and 6 in tabular form, reduced to 4 and 1 using image-encoded representations). These results confirm that integrating evolutionary optimization with image-based feature modeling enables accurate, efficient, and robust intrusion detection across large-scale IoT systems. Full article

Accurate prediction of building energy loads is essential for smart buildings and sustainable energy management. While machine learning (ML) approaches outperform traditional statistical models at capturing nonlinear relationships, most studies primarily optimize prediction accuracy, overlooking the importance of computational efficiency and feature compactness, which are critical in real-time, resource-constrained environments. This study aims to evaluate whether hybrid nature-inspired feature-selection techniques can enhance the accuracy and computational efficiency of ML-based building energy load prediction. Using the UCI Energy Efficiency dataset, eight ML models (LightGBM, CatBoost, XGBoost, Decision Tree, Random Forest, Extra Trees, Linear Regression, Support Vector Regression) were trained under feature subsets obtained from the Butterfly Optimization Algorithm (BOA), Grey Wolf Optimization Algorithm (GWO), and a hybrid BOA–GWO approach. Model performance was evaluated using three metrics (MAE, RMSE, and R²), along with training time, prediction time, and the number of selected features. The results show that gradient-boosting models consistently yield the highest accuracy, with CatBoost achieving an R² of 0.99 or higher. The proposed hybrid BOA–GWO method achieved competitive accuracy with fewer features and reduced training time, demonstrating its suitability for efficient ML deployment in smart building environments. Rather than proposing a new metaheuristic algorithm, this study contributes by adapting a hybrid BOA–GWO feature-selection strategy to the building energy domain and evaluating its benefits under a multi-criteria performance framework. The findings support the practical adoption of hybrid feature-selection-supported ML pipelines for intelligent building systems, energy management platforms, and IoT-based real-time applications. Full article

This paper introduces an intelligent fault-diagnosis framework for power transformers that integrates hybrid machine-learning models with nature-inspired optimization. Current signals were acquired from a laboratory-scale three-phase transformer under both healthy and various fault conditions. A suite of 41 discriminative features was engineered from time–frequency and sparse representations generated via Discrete Wavelet Transform (DWT) and Matching Pursuit (MP). The resulting dataset of 2400 labeled segments was used to develop four hybrid models, PSO-SVM, PSO-RF, BA-SVM, and BA-RF, wherein Particle Swarm Optimization (PSO) and the Bees Algorithm (BA) served as wrapper optimizers for simultaneous feature selection and hyperparameter tuning. Rigorous evaluation with 5-fold and 10-fold cross-validation demonstrated the superior performance of Random Forest-based models, with the BA-RF hybrid achieving peak performance (98.33% accuracy, 99.09% precision). The results validate the proposed methodology, establishing that the fusion of wavelet- and MP-based feature extraction with metaheuristic optimization constitutes a robust and accurate paradigm for transformer fault diagnosis. Full article

Predicting fatigue behavior in steel components is highly challenging due to the nonlinear and uncertain nature of material degradation under cyclic loading. In this study, four hybrid machine learning models were developed—Histogram Gradient Boosting optimized with Prairie Dog Optimization (HGPD), Histogram Gradient Boosting optimized with Wild Geese Algorithm (HGGW), Categorical Gradient Boosting optimized with Prairie Dog Optimization (CAPD), and Categorical Gradient Boosting optimized with Wild Geese Algorithm (CAGW)—by coupling two advanced ensemble learning frameworks, Histogram Gradient Boosting (HGB) and Categorical Gradient Boosting (CAT), with two emerging metaheuristic optimization algorithms, Prairie Dog Optimization (PDO) and Wild Geese Algorithm (WGA). This integrated approach aims to enhance the accuracy, generalization, and robustness of predictive modeling for steel fatigue life assessment. Shapley Additive Explanations (SHAP) were employed to quantify feature importance and enhance interpretability. Results revealed that reduction ratio (RedRatio) and total heat treatment time (THT) exhibited the highest variability, with RedRatio emerging as the dominant factor due to its wide range and significant influence on model outcomes. The SHAP-driven analysis provided clear insights into complex interactions among processing parameters and fatigue behavior, enabling effective feature selection without loss of accuracy. Overall, integrating gradient boosting with novel optimization algorithms substantially improved predictive accuracy and robustness, advancing decision-making in materials science. Full article

Feature selection aims to identify a relevant subset of features from the original feature set to enhance the performance of machine learning models, which is crucial for improvig model accuracy. However, this task is highly challenging due to the enormous search space, often requiring the use of meta-heuristic algorithms to efficiently identify near-optimal feature subsets. This paper proposes an improved algorithm based on Northern Goshawk Optimization (NGO), called Elite-guided Hybrid Northern Goshawk Optimization (EH-NGO), for feature selection tasks. The algorithm incorporates an elite-guided strategy within the NGO framework, leveraging information from elite individuals to direct the population’s evolutionary trajectory. To further enhance population diversity and prevent premature convergence, a vertical crossover mutation strategy is adopted, which randomly selects two different dimensions of an individual for arithmetic crossover to generate new solutions, thereby improving the algorithm’s global exploration capability. Additionally, a boundary control strategy based on the global best solution is introduced to reduce ineffective searches and accelerate convergence. Experiments conducted on 30 benchmark functions from the CEC2017 and CEC2022 test set demonstrate the superiority of EH-NGO in global optimization, outperforming eight compared state-of-the-art algorithms. Furthermore, a novel feature selection method based on EH-NGO is proposed and validated on 22 datasets of varying scales. Experimental results show that the proposed method can effectively select feature subsets that contribute to improved classification performance. Full article

The widespread use of deepfake technologies has increased the demand for accurate and effective detection methods. This study presents a novel deepfake detection framework that utilizes meta-heuristic feature selection to enhance classification performance. The performance of the Artificial Hummingbird Algorithm (AHA), Polar Lights Optimization (PLO), and their hybrid model, AHA-PLO, is investigated. The hybrid model aims to conduct a more effective search in the feature space by combining AHA’s global exploration ability with PLO’s local exploitation precision. Experimental evaluations conducted on two benchmark datasets, FaceForensics++ (FF++) and Celeb-DF (CDF), demonstrate that the proposed AHA-PLO model consistently outperforms its individual components, achieving state-of-the-art AUC scores of 99.36% on FF++ and 98.78% on CDF. These findings support the hybrid model’s potential as a robust and generalizable solution for deepfake video detection. Full article

Feature selection in high-dimensional datasets presents significant computational challenges, particularly in domains with large feature spaces and limited sample sizes. This paper introduces FL-SBA, a novel metaheuristic algorithm integrating fractional calculus enhancements with Laguerre operators into the Secretary Bird Optimization Algorithm framework for binary feature selection. The methodology incorporates fractional opposition-based learning utilizing Laguerre operators for enhanced population initialization with non-local memory characteristics, and a Laguerre-based binary transformation function replacing conventional sigmoid mechanisms through orthogonal polynomial approximation. Fractional calculus integration introduces memory effects that enable historical search information retention, while Laguerre polynomials provide superior approximation properties and computational stability. Comprehensive experimental validation across ten high-dimensional gene expression datasets compared FL-SBA against standard SBA and five contemporary methods including BinCOA, BAOA, BJSO, BGWO, and BMVO. Results demonstrate FL-SBA’s superior performance, achieving 96.06% average classification accuracy compared to 94.41% for standard SBA and 82.91% for BinCOA. The algorithm simultaneously maintained exceptional dimensionality reduction efficiency, selecting 29 features compared to 40 for competing methods, representing 27% improvement while achieving higher accuracy. Statistical analysis reveals consistently lower fitness values (0.04924 averages) and stable performance with minimal standard deviation. The integration addresses fundamental limitations in integer-based computations while enhancing convergence behavior. These findings suggest FL-SBA represents significant advancement in metaheuristic-based feature selection, offering theoretical innovation and practical performance improvements for high-dimensional optimization challenges. Full article

In the field of industrial fault detection, accurate and timely fault identification is crucial for ensuring production safety and efficiency. Effective feature selection (FS) methods can significantly enhance detection performance in this process. However, the recently proposed Black-winged Kite Algorithm (BKA) tends to suffer from premature convergence and local optima when handling high-dimensional feature spaces. To address these limitations, this paper proposes an improved Black-winged Kite Algorithm (IBKA). This algorithm integrates two novel enhancement mechanisms: First, the Stagnation-Triggered Diversification Mechanism monitors the algorithm’s convergence state and applies mild perturbations to the worst-performing individuals upon detecting stagnation, effectively preventing traps in local optima. Second, the Adaptive Weak Guidance Mechanism employs a conditional elite guidance strategy during the late optimization phase to provide subtle directional guidance to underperforming individuals, thereby improving convergence efficiency. We comprehensively evaluated the proposed IBKA across 26 benchmark functions. Results demonstrate superior performance in solution quality, convergence speed, and robustness compared to the original BKA and other advanced meta-heuristics. Furthermore, fault detection applications on public datasets validate the practical applicability of the binary version of the IBKA (bIBKA), showcasing significant improvements in detection accuracy and reliability. Experimental results confirm that these enhancement mechanisms effectively balance exploration and exploitation capabilities while preserving algorithmic simplicity and computational efficiency. Full article

Nature has evolved sophisticated optimization strategies over billions of years, yet computational algorithms inspired by plants remain remarkably underexplored. We present a comprehensive systematic review following PRISMA 2020 guidelines, analyzing 175 studies (2000–2025) of plant-inspired metaheuristic optimization algorithms and their predominantly animal-inspired counterparts. Despite constituting only 9.7% of bio-inspired optimization literature, plant-inspired algorithms demonstrate competitive and often superior performance compared to animal-inspired approaches. Through a meta-analysis of empirical studies, we document that algorithms like Phototropic Growth and Binary Plant Rhizome Growth achieve 97% superiority on CEC2017 benchmarks and 81% accuracy on high-dimensional feature-selection tasks—significantly exceeding established animal-inspired methods like Particle Swarm Optimization and Genetic Algorithms (p < 0.05). However, our review reveals a critical gap: the majority of these algorithms lack the formal theoretical foundations of their counterparts. This paper systematically documents these theoretical deficiencies and positions them as a key area for future research. Our framework maps botanical processes to computational operators, providing structured guidance for future algorithm development. Plant-inspired approaches excel particularly in distributed optimization, resource allocation, and multi-objective problems by leveraging unique mechanisms evolved for survival in sessile, resource-limited environments. These findings establish plant-inspired approaches as a promising yet severely underexplored frontier in optimization theory, with immediate applications in sustainable computing, resilient network design, and resource-constrained artificial intelligence. Full article

Dengue is a life-threatening disease that is transmitted by mosquitoes. Dengue fever has no proper treatment. Early, proper diagnosis is essential to minimize complications and enhance outcomes in patients. This research uses a clinical and hematological dataset of dengue to assess the effectiveness of the Gradient Boosting (GB) classification model with and without feature selection. It initially employs a standalone GB model, achieving impeccable results for classification, at 100% accuracy, F1-score, precision, and recall. In addition, the Bird Swarm Algorithm (BSA)-based metaheuristic technique is implemented on the GB classifier to execute wrapper-based feature selection so that features are reduced and achieve better results. The BSA-GB model yielded an accuracy of 99.49%, F1-score of 99.62%, recall of 99.24%, and precision of 100%, but it only selected five features in total. An additional test with a five-fold cross-validation was employed for better performance and model evaluation. Folds 1 and 2 showed especially good results. Although fold 2 selected only four features, it still showed high results, compared to fold 1, which selected five features. In this context, fold 2 achieved an accuracy of 99.49%, F1-score of 99.65%, recall of 99.30%, and precision of 100%. Means of hyperparameters were also calculated across folds to make a generalized GB model, which maintained 99.49% of accuracy with just three features, namely, Hemoglobin, WBC Count, and Platelet Count. To enhance transparency, counterfactual explanations were performed to analyze the misclassified cases, which indicated that minimum changes in input features modify the predictions. Also, an evaluation of the SHAP value result designated WBC Count and Platelet Count as the most important features. Full article