Melanoma is one of the most aggressive forms of skin cancer, with a high mortality rate when not detected early. This public health challenge underscores the need for accurate and efficient diagnostic tools. Convolutional Neural Networks have shown strong performance in medical image analysis. However, their effectiveness relies heavily on optimal architectural and hyperparameter configurations, which are often designed without alignment to the target domain or transferred from unrelated domains, limiting adaptability to specific medical datasets. Existing hybrid CNN-metaheuristic approaches typically optimize only fixed network parameters. They often fail to explore how metaheuristics can adaptively shape the CNN architectures themselves.In this study, a comprehensive hybrid optimization framework is proposed that integrates CNNs with six nature-inspired metaheuristic algorithms that mimic biological or physical phenomena to solve complex problems. These include Cuckoo Search, Firefly Algorithm, Whale Optimization Algorithm, Particle Swarm Optimization, Grey Wolf Optimizer, and Crow Search Algorithm. Rather than tuning a predefined architecture, each optimizer searches the architectural and training space to identify high-performing CNN configurations, enabling emergent and data-driven network design. This unified framework allows a systematic cross-algorithm comparison under identical conditions, providing new insights into convergence stability, exploration-exploitation dynamics, and generalization behavior. A robust preprocessing and data augmentation pipeline, including brightness normalization, hair artifact removal, and geometric transformations, is incorporated to improve model generalization and enhance the optimizer's search landscape. Experiments on the HAM10000 dataset demonstrate that the metaheuristic-optimized CNNs outperform the baseline, achieving accuracies up to 91.25%. These findings confirm that population-based optimization is an efficient and reliable mechanism for guiding CNN architecture design. This approach achieves superior performance compared to traditional manual or other optimization-based strategies.

A discrete crow search algorithm for solving the uncapacitated facility location problem

For autonomous systems operating in complex environments, it is a critical challenge to develop effective MAS path planning control mechanism that enables agents to collaborate and avoid collisions with each other. In response to this crucial need, this paper introduces a novel hybrid algorithm for MAS path planning called Crow Swarm Optimization (CSO). This solution leverages the strengths of two distinct metaheuristic approaches, the Crow Search algorithm (CSA) and the Particle Swarm Optimization (PSO). Although CSA is effective during search space exploration, it has slower convergence and weaknesses in finding global optima due to random location updates. To enhance the local search capability of the CSA, this research leverages the strength of PSO in exploitation. To accomplish this, the primary objective of this paper is to leverage the functionally segregated mechanism to overcome the inherent weaknesses of single-approach algorithms, thereby achieving optimal MAS path planning. Beyond path optimization, the presented work addresses collision prevention employing a velocity adjustment technique. This approach minimizes the agent’s velocity to prevent collisions with other agents. To validate the findings, the study conducted a comparative analysis against various versions of both CSA and PSO. The results clearly demonstrate the efficacy of the proposed approach in solving the path-planning problem with collision avoidance.

Chronic kidney disease (CKD) is an advancing disease which significantly impacts global healthcare, requiring early detection and prompt treatment is required to prevent its advancement to end-stage renal disease. Conventional diagnostic methods tend to be invasive, lengthy, and costly, creating a demand for automated, precise, and efficient solutions. This study proposes a novel technique for identifying and classifying CKD from medical images by utilizing a Convolutional Neural Network based Crow Search (CNN based CS) algorithm. The method employs sophisticated pre-processing techniques, including Z-score standardization, min-max normalization and robust scaling to improve the input data's quality. Selection of features is carried out using the chi-square test, and the Crow Search Algorithm (CSA) further optimizes the feature set for the improvement of accuracy classification and effectivess. The CNN architecture is employed to capture complex patterns using deep learning methods to accurately classify CKD in medical pictures. The model optimized and examined using an open access Kidney CT Scan data set. It achieved 99.05% accuracy, 99.03% Area under the Receiver Operating Characteristic Curve (AUC-ROC), and 99.01% Area under the precision-recall curve (PR-AUC), along with high precision (99.04%), recall (99.02%), and F1-score (99.00%). The results show that the CNN-based CS method delivers high accuracy and improved diagnostic precision related to conventional machine learning techniques. By incorporating CSA for feature optimization, the approach minimizes redundancy and improves model interpretability. This makes it a promising tool for automated CKD diagnosis, contributing to the development of AI-driven medical diagnostics and providing a scalable solution for early detection and management of CKD.

Unmanned Aerial Vehicles (UAVs) equipped with Multiple-Input Multiple-Output (MIMO) communication systems are increasingly deployed to restore or extend connectivity in forested and remote regions where terrestrial infrastructure is unavailable. However, radio propagation through vegetation is strongly affected by polarization-dependent scattering, attenuation, and depolarization, which can severely degrade link reliability. This study investigates polarization-aware UAV deployment as a means to enhance air-to-ground communication performance under dense canopy conditions. A vegetation-aware propagation model is developed using the Debye relaxation framework combined with Kramers-Kronig relations to capture the dielectric response of moist foliage. Cross-Polarization Discrimination (XPD) is identified as a dominant factor influencing signal quality, exhibiting non-monotonic variations that complicate UAV positioning. To address this, the Crow Search Algorithm (CSA) is employed to determine optimal UAV locations that minimize XPD between orthogonal polarization channels. Simulation results demonstrate that polarization-aware optimization significantly improves link robustness compared to traditional path-loss-based strategies, particularly at higher frequencies. The findings highlight the importance of integrating polarization awareness into UAV communication planning for critical missions such as search-and-rescue and post-disaster recovery in vegetated environments.

To address the limitations of the Crow Search Algorithm regarding its susceptibility to local optima and slow convergence speed when handling high-dimensional complex problems, this study proposes six enhanced variants, encompassing ACSA-1 through ACSA-3 and ACSA-LEVY-1 through ACSA-LEVY-3, which integrate adaptive mechanisms, Lévy flight strategies, and a unified guidance strategy. These variants employ linear and piecewise functions to dynamically adjust flight length fl and awareness probability AP, incorporate Lévy flight perturbations, and utilize a probabilistic global-best guidance mechanism to synergistically balance global exploration and local exploitation. The performance of the proposed algorithms is rigorously evaluated across CEC benchmark functions, the CEC2019 test suite comprising Storn’s Chebyshev Polynomial Fitting and Inverse Hilbert Matrix problems, three classical engineering design problems, and a complex damper allocation problem based on nonlinear finite element time-history analysis. Experimental results demonstrate that the improved variants significantly outperform the original CSA and other well-established metaheuristic algorithms in terms of solution quality and robustness. Specifically, for the ill-conditioned Inverse Hilbert Matrix problem, the mean error is reduced from 511.46 to 161.30, representing a reduction of approximately 68.5%; in the pressure vessel design problem, the cost is lowered by 5.5%. Notably, in the large-scale seismic damper allocation problem, the proposed method achieves superior structural response indices in only 237s, standing in stark contrast to the 22,561s required by the Genetic Algorithm. The Wilcoxon signed-rank test with Bonferroni correction yielding a p-value less than 0.008333 confirms the statistical significance of these improvements. These findings highlight that the proposed enhancements effectively resolve stagnation issues without increasing asymptotic computational complexity, demonstrating strong potential for real-world engineering applications.

Power transformers, as oil-immersed equipment, are primarily diagnosed using dissolved gas analysis (DGA), a widely recognized and effective fault detection method. Over time, numerous approaches have been developed to identify transformer faults through DGA, with many incorporating artificial intelligence to enhance diagnostic accuracy. In this context, this study enhances the support vector machine and k-nearest neighbors algorithms by integrating them into the iterative processes of the artificial bee colony and crow search algorithms for fault identification. Chaos theory was embedded into the optimization framework to improve convergence and avoid local optima. Nine input vectors, derived from a dataset of 501 samples covering six fault types, were used to evaluate the enhanced algorithms. An improved binary crow search algorithm was employed to optimize the input vectors, effectively identifying the key features for fault diagnosis. The proposed enhancement achieved an accuracy exceeding 94%, demonstrating the potential of refined methodologies for practical transformer diagnostics.

This paper addresses concurrent scheduling of machines, automated guided vehicles (AGVs), tools, and Tool Transporter (TT) in a multi-machine Flexible Manufacturing System (FMS), considering tool and job transport times to minimize makespan (MKSN). However, due to financial constraints, only one copy of each tool type is offered. These tools are placed in a Central Tool Magazine (CTM), which is shared with numerous machines. The problem is the allocation of machines, tool assignment, job-operation sequencing, and trip operations, including TT and AGVs' empty and laden trip times for MKSN minimization. This study introduces a mixed nonlinear integer programing to model the problem, and a Crow Search Algorithm (CSA) based on the clever behavior of crows is used to solve it. Results are tabulated, examined, and compared to current algorithms.

The increasing adoption of solar energy as a clean and sustainable power source has intensified research efforts toward developing more efficient photovoltaic (PV) cells. These cells exhibit nonlinear characteristics that are significantly influenced by variations in irradiance and temperature. Accurate estimation of PV model parameters plays a crucial role in maximizing performance, particularly via precise maximum power point (MPP) tracking. This paper presents a new hybrid metaheuristic algorithm that combines the global exploration capability of the Bat Algorithm (BA) with local exploitation efficiency of the Crow Search Algorithm (CR) to optimize PV parameter estimation. The proposed approach is tested using Single-Diode (SDM), Double-Diode (DDM), and Triple-Diode (TDM) models based on the RTC France dataset. The hybrid model demonstrates better convergence behavior and robustness compared to conventional approaches. Qualitatively, it effectively manages parameter uncertainty; quantitatively, it achieves RMSE values of 0.00077299 (SDM), 0.0008215 (DDM), and 0.0008068 (TDM), outperforming traditional algorithms such as Particle Swarm Optimization (PSO) and Genetic Algorithm (GA).

Recently, researchers have attempted to develop a new algorithm by combining quantum systems and metaheuristics algorithms and are confirming its applicability in engineering optimization problems. This paper proposes a new QbCSA (quantum-based crow search algorithm) combining quantum systems and CSA (crow search algorithm). Unlike CSA, the initial matrix of QbCSA consists of qubits and performs operations through spin and measurement processes. Six benchmark functions were used to compare the convergence performance according to the parameter change used in the developed QbCSA, and the optimal parameter range is suggested. In addition, the CEC2019 benchmark functions and four engineering example problems were solved and compared with the results of previous studies. QbCSA demonstrated comparable performance to CSA, which uses decimal-based design variables, while achieving lower variance and more stable convergence than QbHSA. In particular, for multimodal optimization problems, QbCSA exhibited superior search efficiency and solution diversity. Furthermore, the four engineering examples confirmed the practical applicability of QbCSA, and these results indicate that qubit-based encoding can enhance the search efficiency of CSA and suggest broader applicability to engineering optimization problems.

This study presents the Comprehensive Learning Crow Search Algorithm (CL-CSA), a nature-inspired metaheuristic based on the intelligent behavior of crows in hiding and recovering a food source. The core of the CL-CSA is the improved learning strategy, which enables individuals to adaptively update their positions based on the recollection of multiple peers, rather than just one. This learning scheme will control the exploration versus exploitation balance, enabling the swarm to have memories from multiple sources to remember other food sources, which sustains greater diversity and avoids potentially trapping the swarm into local optima. The effectiveness of the proposed CL-CSA has been demonstrated based on complete tests involving a good number of benchmark functions. Several results are comparing the CL-CSA to many recent optimization algorithms and different CSA variants, which show clearly that the proposed framework is very effective for complex optimization problems. In addition, we applied the CL-CSA in real structural optimization problems to evaluate its overall performance and effectiveness in real-world engineering design applications.

Purpose This study aims to enhance the mechanical properties of 3D-printed acrylonitrile butadiene styrene (ABS) by reinforcing it with copper and optimizing key process parameters using Taguchi, artificial neural network (ANN) and metaheuristic optimization techniques. Design/methodology/approach Copper-reinforced ABS filaments were fabricated using a twin-screw extruder and printed via fused deposition modelling. A Taguchi L25 design was used to study the effects of printing temperature and layer height on tensile, compressive and flexural strengths. An ANN was trained on experimental data to model these properties, and a hybrid crow search algorithm–grey wolf optimizer (CSA–GWO) was used for multi-objective parameter optimization. Findings The Taguchi method identified printing temperature as the most influential factor. The ANN model demonstrated high predictive accuracy, achieving R² values exceeding 0.99 and maintaining prediction errors below 2%. The hybrid CSA–GWO algorithm effectively identified optimal parameters for maximizing each mechanical property, with a balanced setting of 245.68°C and 0.100 mm providing strong overall performance (947.97 N tensile, 4,173.61 N compressive and 176.06 N flexural). Originality/value The use of a hybrid CSA–GWO algorithm presents a novel approach within the additive manufacturing domain, offering enhanced exploration and convergence capabilities for optimizing mechanical properties of copper-reinforced ABS composites.

Reliable parameter identification is essential for the development and predictive use of non-linear bioprocess models. This study evaluates the recently proposed Dream Optimization Algorithm (DOA), a human-inspired metaheuristic based on memory retention, partial forgetting, and dream-sharing mechanisms, for the identification of kinetic parameters in an Escherichia coli fed-batch cultivation model. The algorithm’s performance is assessed using experimental cultivation data and compared with three widely employed metaheuristics: the genetic algorithm (GA), simulated annealing (SA), and the crow search algorithm (CSA). Results demonstrate that DOA achieves the lowest objective function value, the best mean performance across 30 independent runs, and substantially reduced computational time compared to SA and CSA. The model dynamics generated using DOA-identified parameters show excellent agreement with experimental biomass and substrate measurements, even in the presence of significant noise in the substrate data. These findings highlight the high accuracy, robustness, and computational efficiency of DOA, confirming its strong potential as an effective tool for bioprocess model parameter estimation and broader non-linear optimization tasks.

Research on the construction of English vocabulary learning recommendation system based on multi-objective crow search algorithm

The innumeracy of Internet of Things (IoT) devices has significantly raised Distributed Denial of Service (DDoS) attacks, posing a threat to the reliability and integrity of the network. To overcome this, this paper describes an intelligent detection framework, which is an optimized Long Short-Term Memory (LSTM) network that is trained with a Hybrid Crow Search Algorithm (CSA) with Centroid Opposition-Based Learning (COBL), and is called COCSA-LSTM. The hybrid optimization algorithm is effective in tuning the weights and biases of LSTM to reach higher convergence speed and accuracy in detection. COBL also enhances the exploration-exploitation ratio in CSA by creating opposition-based candidate solutions located at the centroid of the population to prevent premature convergence and enhance global search capability. This model was tested on four benchmark IoT intrusion detection datasets, such as BoT-IoT, CIC-IDS2017, CIC-DDoS2019, and IootID20, and the performance of the model was analyzed against standard and optimized versions of LSTM, such as GA-LSTM, PSO-LSTM, IPSO-LSTM, and CSA-LSTM. According to the experimental findings, COCSA-LSTM always performs better than other algorithms with the highest accuracy rates of 99.48, 99.05, 99.42, and 98.91 on the corresponding datasets. The ability of the model to attain a low Mean Squared Error (MSE) in a small number of iterations is also confirmed by convergence analysis, as the model is computationally efficient. The hybrid solution will offer a strong and scalable solution to real-time DDoS detection in the IoT network environment that can facilitate proactive network protection against emerging attack patterns.

An Accelerated Crow Search Algorithm for Multifactorial Optimization in Vehicle Routing Problem

This work proposes an intelligent strategy for the coordinated management of active and reactive power in Battery Energy Storage Systems (BESSs) within AC microgrids operating under both grid-connected (GCM) and islanded (IM) modes to minimize daily operational costs. The problem is formulated as a mixed-variable optimization model that explicitly leverages the control capabilities of BESS power converters. To solve it, a Parallel Particle Swarm Optimization (PPSO) algorithm is employed, coupled with a Successive Approximation (SA) power flow solver. The proposed approach was benchmarked against parallel implementations of the Crow Search Algorithm (PCSA) and the JAYA algorithm (PJAYA), both in parallel, using a realistic 33-node AC microgrid test system based on real demand and photovoltaic generation profiles from Medellín, Colombia. The strategy was evaluated under both deterministic conditions (average daily profiles) and stochastic scenarios (100 daily profiles with uncertainty). The proposed framework is evaluated on a 33-bus AC microgrid that operates in both grid-connected and islanded modes, with a battery energy storage system dispatched at both active and reactive power levels subject to network, state-of-charge, and power-rating constraints. Three population-based optimization algorithms are used to coordinate BESS schedules, and their performance is compared based on daily operating cost, BESS cycling, and voltage profile quality. Quantitatively, the PPSO strategy achieved cost reductions of 2.39% in GCM and 1.62% in IM under deterministic conditions, with a standard deviation of only 0.0200% in GCM and 0.2962% in IM. In stochastic scenarios with 100 uncertainty profiles, PPSO maintained its robustness, reaching average reductions of 2.77% in GCM and 1.53% in IM. PPSO exhibited consistent robustness and efficient performance, reaching the highest average cost reductions with low variability and short execution times in both operating modes. These findings indicate that the method is well-suited for real-time implementation and contributes to improving economic outcomes and operational reliability in grid-connected and islanded microgrid configurations. The case study results show that the different strategies yield distinct trade-offs between economic performance and computational effort, while all solutions satisfy the technical limits of the microgrid.

Skull-base brain metastases pose significant diagnostic and surgical challenges due to their proximity to vital neurological structures. We propose an enhanced Generative Adversarial Network (GAN) model optimised with the Crow Search Algorithm (CSA) to improve detection accuracy and intraoperative decision-making. The GAN framework facilitates high-fidelity image generation and segmentation, while CSA fine-tunes hyperparameters for improved model stability and accuracy. Trained on high-resolution brain MRI datasets with expert annotations, our model achieved a precision of 97.43%, surpassing existing approaches in accuracy and robustness. The system accurately delineates tumour margins and adjacent anatomical structures in real-time, enhancing surgical guidance and reducing operative risks. The inclusion of CSA significantly improved GAN convergence and reduced false positives. This integrated GAN-CSA approach shows promise for revolutionizing neuro-oncology practices by enabling safer and more precise skull base surgeries. As an initial proof-of-concept, the evaluation was limited to 156 MR volumes from a single scanner, and future cross-centre studies will be pursued to establish robustness across varying field strengths, coils, and imaging protocols.

With the gradual deepening of coal mining, the surrounding rock pressure significantly increases, and the risk of gas release and accumulation also increases, increasing the likelihood of coal and gas outburst hazards. This study used boxplot and data interpolation method to preprocess data and used correlation to screen out highly correlated influencing factors as disaster prediction indicators. Build an initial prediction model framework using Convolutional Neural Network (CNN), optimize model hyperparameters using Chaos Mapping and Levy Flight Improved Crow Search Algorithm (ICSA), and establish a coal and gas outburst prediction model based on ICSA-CNN. Finally, a comparative model was established to compare the evaluation indicators and confusion matrix. According to the results, the ICSA-CNN model stood out as the most accurate in its predictive capabilities, better robustness and generalization ability, and higher security.

Handwritten and printed character recognition is crucial for various real-world applications, including pattern recognition and communication technologies. Precise character recognition is challenging due to the inability of conventional models to effectively capture long-range dependencies and sequential variations in character structures. Moreover, high dimensionality of extracted features complicates the identification of relevant information, leading to suboptimal classification performance. In order to address the aforementioned challenges, a Firefly Algorithm-based Improved Crow Search Algorithm (FAICSA) is proposed for feature optimization, integrated with a Dynamic Long Short-Term Memory (DLSTM) network for character recognition. The region of interest (ROI) containing the required character is then segmented by applying morphological operations of erosion and dilation. Feature extraction is carried out on the segmented image by integrating five techniques: ResNet-18, contour extraction, skeletonization, zoning, and statistical features. Subsequently, the dimensionality of the original extracted features is reduced using FAICSA. The features selected by FAICSA are input into the DLSTM network for character recognition. Furthermore, the Exponential Linear Unit (ELU) activation function is incorporated into the DLSTM to stabilize training by avoiding exploding gradients, thereby enhancing generalization. The FAICSA-DLSTM model achieves impressive classification accuracies of 94.60% and 99.72% on real-time and Chars74K datasets, respectively. The results obtained by the FAICSA-DLSTM model significantly outperforms existing models, including Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), sparse autoencoder, and Gated Recurrent Unit (GRU).