Journal of Artificial Intelligence and Metaheuristics

The growing complexity of engineering optimization problems has revealed significant limitations in traditional mathematical programming approaches, necessitating the development of innovative metaheuristic algorithms capable of handling high-dimensional, multi-modal, and discontinuous objective functions. This paper presents the Somersaulting Spider Optimizer (SSO), a novel bio-inspired metaheuristic algorithm that draws inspiration from the extraordinary locomotion mechanisms of the somersaulting spider, a desert-dwelling arachnid species renowned for its acrobatic somersaulting capabilities. The proposed algorithm incorporates dual movement mechanisms that effectively balance global exploration through somersaulting behavior and local exploitation via controlled rolling movements. A distinctive feature of SSO is its adaptive energy-management system, which dynamically regulates exploration–exploitation transitions based on solution-improvement patterns and stagnation detection. Complementary adaptive parameters balance global search and local refinement throughout optimization. The algorithm was evaluated on four engineering design benchmarks: pressure vessel, welded beam, three-bar truss, and cantilever beam. Comparisons with the Genetic Algorithm (GA), Whale Optimization Algorithm (WOA), Harris Hawks Optimization (HHO), and Bat Algorithm (BAT) report that SSO outperforms these methods on the test problems. ANOVA and Wilcoxon signed-rank tests assess the significance of the reported performance differences. The results identify SSO as an effective optimization method, with improvements in solution quality, convergence stability, and computational efficiency.

The importance of network security has greatly increased in the continuously changing modern digital environment. Network security is a multi-layered defense mechanism that protects networks, data, and systems from unauthorized access and other malicious activities. As cyber threats become more advanced, traditional protective measures may no longer be adequate, and intrusion-detection systems must rapidly evolve. This research proposes BBERSC (Balance Between Al-Biruni Earth Radius Optimization and Sine Cosine Algorithm), which combines the local-search capability of Al-Biruni Earth Radius Optimization with the probabilistic improvement of the Sine Cosine Algorithm. The approach tunes Random Forest hyperparameters, including the number of trees, maximum depth, and minimum samples, to improve pattern recognition in network traffic. ANOVA and Wilcoxon signed-rank tests validate the effectiveness of the models and algorithms and assess differences between them. The Random Forest model within the BBERSC framework achieves an accuracy of 0.9901719, exceeding the other evaluated machine-learning algorithms.

A brain stroke represents a deadly health condition that emerges from poor blood flow to the brain. Brain tissue affected by a stroke will completely cease regular operations. Immediate detection of a brain stroke leads to better treatment success. Images from computed tomography (CT) provide a quick diagnosis of stroke. But time is passing quickly as physicians examine each brain CT scan. This situation could cause therapy to be delayed and mistakes to be made. Thus, we focused on using a practical artificial intelligence algorithm for stroke detection. This paper proposes several deep neural network models, such as DenseNet121, ResNet50, Xception, and EfficientNetV2S, for transfer learning to study the features of stroke lesions and achieve complete intelligent automatic detection by classifying CT images into two categories (stroke and normal). The dataset comprises 437 testing, 235 validation, and 1843 training photos. Using the same dataset, the experimental findings outperform all state-of-the-art. The optimal model utilizing the EfficientNetV2S model for transfer learning has an overall accuracy of 99.57% and the same value for precision and recall.

This study presents a comprehensive evaluation of metaheuristic-optimized machine learning models for automated apple quality classification, addressing the critical need for accurate and consistent fruit grading systems in agricul-tural applications. The research integrates four bio-inspired optimization algorithms—Whale Optimization Algorithm (WOA), Salp Swarm Algorithm (SSA), Cuckoo Search (CS), and Bat Algorithm (BAT)—with Multi-Layer Percep-tron (MLP) classifiers to enhance fruit quality assessment performance. Experimental validation was conducted using a comprehensive apple quality dataset containing seven key attributes: size, weight, sweetness, crunchiness, juiciness, ripeness, and acidity. The results demonstrate that WOA-MLPClassifier achieves superior performance with 95.37%accuracy, 95.99% sensitivity, and balanced effectiveness across all evaluation metrics including specificity, positive predictive value, negative predictive value, and F1 Score. Statistical validation through one-way ANOVA and Wilcoxon signed-rank tests confirms significant performance improvements over baseline models and alternative optimization approaches, with p-values less than 0.001. The proposed framework exhibits remarkable consistency across multiple evaluation runs, with perfect positive rank sums indicating reliable optimization behavior. These findings establish a new benchmark for automated fruit quality classification systems and provide valuable insights for deploying bio-inspired optimization techniques in agricultural machine learning applications where both accuracy and reliability are essential for commercial viability.

Retinal diseases are a major cause of vision loss, making timely and accurate diagnosis important for effective treatment. This study compares convolutional neural networks (CNN), k-nearest neighbors (K-NN), and support vector machines (SVM) for classifying digital fundus images into four categories: normal retina, cataract, glaucoma, and other retinal diseases. The models are evaluated using accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F-score. CNN achieves the highest reported accuracy (94.05%) and sensitivity (0.9977), with an F-score of 0.9693. The results indicate that CNN can be useful for automated retinal screening, while its low specificity (0.0178) highlights the need to reduce false-positive predictions. The comparison illustrates both the potential and the limitations of machine-learning-based retinal disease classification.