Search Algorithm For Selection Research Articles

SAM-Induced Pseudo Fully Supervised Learning for Weakly Supervised Object Detection in Remote Sensing Images

Weakly supervised object detection (WSOD) in remote sensing images (RSIs) aims to detect high-value targets by solely utilizing image-level category labels; however, two problems have not been well addressed by existing methods. Firstly, the seed instances (SIs) are mined solely relying on the category score (CS) of each proposal, which is inclined to concentrate on the most salient parts of the object; furthermore, they are unreliable because the robustness of the CS is not sufficient due to the fact that the inter-category similarity and intra-category diversity are more serious in RSIs. Secondly, the localization accuracy is limited by the proposals generated by the selective search or edge box algorithm. To address the first problem, a segment anything model (SAM)-induced seed instance-mining (SSIM) module is proposed, which mines the SIs according to the object quality score, which indicates the comprehensive characteristic of the category and the completeness of the object. To handle the second problem, a SAM-based pseudo-ground truth-mining (SPGTM) module is proposed to mine the pseudo-ground truth (PGT) instances, for which the localization is more accurate than traditional proposals by fully making use of the advantages of SAM, and the object-detection heads are trained by the PGT instances in a fully supervised manner. The ablation studies show the effectiveness of the SSIM and SPGTM modules. Comprehensive comparisons with 15 WSOD methods demonstrate the superiority of our method on two RSI datasets.

A binary sparrow search algorithm for feature selection on classification of X-ray security images

A binary sparrow search algorithm for feature selection on classification of X-ray security images

A Machine-Learning Strategy to Detect Mura Defects in a Low-Contrast Image by Piecewise Gamma Correction.

A detection and classification machine-learning model to inspect Thin Film Transistor Liquid Crystal Display (TFT-LCD) Mura is proposed in this study. To improve the capability of the machine-learning model to inspect panels' low-contrast grayscale images, piecewise gamma correction and a Selective Search algorithm are applied to detect and optimize the feature regions based on the Semiconductor Equipment and Materials International Mura (SEMU) specifications. In this process, matching the segment proportions to gamma values of piecewise gamma is a task that involves derivative-free optimization which is trained by adaptive particle swarm optimization. The detection accuracy rate (DAR) is approximately 93.75%. An enhanced convolutional neural network model is then applied to classify the Mura type through using the Taguchi experimental design method that identifies the optimal combination of the convolution kernel and the maximum pooling kernel sizes. A remarkable defect classification accuracy rate (CAR) of approximately 96.67% is ultimately achieved. The entire defect detection and classification process can be completed in about 3 milliseconds.

Hybrid whale optimized crow search algorithm and multi-SVM classifier for effective system level test case selection

A novel approach to enhance software testing through intelligent test case selection is proposed in this work. The proposed method combines feature extraction, clustering, and a hybrid optimization algorithm to improve testing effectiveness while reducing resource overhead. It employs a context encoder to extract relevant features from software code, enhancing the accuracy of subsequent testing. Through the use of Fuzzy C-means (FCM) clustering, the test cases are classified into groups, streamlining the testing process by identifying similar cases. To optimize feature selection, a Hybrid Whale Optimized Crow Search Algorithm (HWOCSA), which intelligently combines the strengths of both Whale Optimization Algorithm (WOA) and Crow Search Algorithm (CSA) is introduced. This hybrid approach mitigates limitations while maximizing the selection of pertinent features for testing. The ultimate contribution of this work lies in the proposal of a multi-SVM classifier, which refines the test case selection process. Each classifier learns specific problem domains, generating predictions that guide the selection of test cases with unprecedented precision. Experimental results demonstrate that the proposed method achieves remarkable improvements in testing outcomes, including enhanced performance metrics, reduced computation time, and minimized training data requirements. By significantly streamlining the testing process and accurately selecting relevant test cases, this work paves the way for higher quality software updates at a reduced cost.

Improved reptile search algorithm with sequential assignment routing based false packet forwarding scheme for source location privacy protection on wireless sensor networks

Source Location Privacy (SLP) in Wireless Sensor Networks (WSNs) refers to a set of techniques and strategies used to safeguard the anonymity and confidentiality of the locations of sensor nodes (SNs) that are the source of transmitted data within the network. This protection is important in different WSN application areas like environmental monitoring, surveillance, and healthcare systems, where the revelation of the accurate location of SNs can pose security and privacy risks. Therefore, this study presents metaheuristics with sequential assignment routing based false packet forwarding scheme (MSAR-FPFS) for source location privacy protection (SLPP) on WSN. The contributions of the MSAR-FPFS method revolve around enhancing SLP protection in WSNs through the introduction of dual-routing, SAR technique with phantom nodes (PNs), and an optimization algorithm. In the presented MSAR-FPFS method, PNs are used for the rotation of dummy packets using the SAR technique, which helps to prevent the adversary from original data transmission. Next, the MSAR-FPFS technique uses an improved reptile search algorithm (IRSA) for the optimal selection of routes for real packet transmission. Moreover, the IRSA technique computes a fitness function (FF) comprising three parameters namely residual energy (RE), distance to BS (DBS), and node degree (ND). The experimental evaluation of the MSAR-FPFS system was investigated under different factors and the outputs show the promising achievement of the MSAR-FPFS system compared to other existing models.

Pantograph–Catenary Interaction Prediction Model Based on SCSA-RBF Network

As a traditional numerical simulation method for pantograph–catenary interaction research, the pantograph–catenary finite element model cannot be applied to the real-time monitoring of pantograph–catenary contact force, and the computational cost required for the multi-parameter joint optimization of the pantograph–catenary system with the finite element model is very high. In this paper, based on the selective crow search algorithm–radial basis function (SCSA-RBF) network, the time-domain signal of the panhead acceleration, which can be obtained in real-time through non-contact test technology, is taken as the boundary condition to directly solve the pantograph dynamic equation and a data-physics coupling model that can quickly predict the pantograph–catenary interaction is proposed. The prediction model is trained and verified using the dataset generated through the finite element model. Furthermore, the prediction model is applied to the multi-parameter joint optimization of six pantograph dynamic parameters and nine pantograph dynamic parameters, considering nonlinear panhead stiffness, and optimization suggestions under various speeds and filtering frequencies are given.

Contact Force Surrogate Model and Its Application in Pantograph–Catenary Parameter Optimization

The significant increase in the speed of high-speed trains has made the optimization of pantograph–catenary parameters aimed at improving current collection quality become one of the key issues that urgently need to be addressed. In this paper, a method and solutions are proposed for optimizing multiple pantograph–catenary parameters, taking into account the speed levels and engineering feasibility, for pantograph–catenary systems that contain dozens of parameters and exhibit strong nonlinear coupling characteristics. Firstly, a surrogate model capable of accurately predicting the standard deviation of contact force based on speed and 14 pantograph–catenary parameters was constructed by using the pantograph–catenary finite element model and feedforward neural network. Secondly, sensitivity analysis and rating of the pantograph–catenary parameters under different speeds were conducted using the variance-based method and the surrogate model. Finally, by combining the sensitivity analysis results and the Selective Crow Search Algorithm, joint optimization of 10 combinations of the pantograph–catenary parameters across the entire speed range was performed, providing efficient pantograph–catenary parameter optimization solutions for various engineering conditions.

A hybrid variable selection method combining Fisher's linear discriminant combined population analysis and an improved binary cuckoo search algorithm.

In this paper, a novel hybrid variable selection method for model building by near-infrared (NIR) spectroscopy is proposed for composition measurement in industrial processes. A double-layer structure is designed for variable selection by combining Fisher's linear discriminant combined population analysis (FCPA) and an improved binary cuckoo search algorithm (IBCS). The Fisher classifier combined with model population analysis is used to select the variable interval wherein the useful variables are roughly located even when strong multicollinearity exists among spectral variables. Opposition-based learning (OBL) and jumping genes (JG) are introduced to improve the binary cuckoo search algorithm for the fine selection of key variables, thus avoiding the loss of excellent solutions due to randomness and the local optimum. Different variable selection methods were used to select variables for beer, corn, and diesel fuel datasets, and the partial least squares (PLS) algorithms were used to build calibration models to predict the original extract concentration of beer, the protein and starch content of corn, and the boiling point of diesel fuel, respectively. The results showed that the proposed PLS modeling method based on FCPA-IBCS has higher fitting accuracy and smaller prediction errors.

MOBCSA: Multi-Objective Binary Cuckoo Search Algorithm for Features Selection in Bioinformatics

MOBCSA: Multi-Objective Binary Cuckoo Search Algorithm for Features Selection in Bioinformatics

Defects Detection on 110 MW AC Wind Farm’s Turbine Generator Blades Using Drone-Based Laser and RGB Images with Res-CNN3 Detector

An effective way to perform maintenance on the wind turbine generator (WTG) blades installed in grid-connected wind farms is to inspect them using Unmanned Aerial Vehicles (UAV). The ability to detect wind turbine blade defects from these laser and RGB images captured by drones has been the subject of numerous studies. The issue that most applied techniques battle with is being able to locate different wind turbine blade defects with high confidence scores and precision. The accuracy of these models’ defect detection decreases due to varying testing image scales. This article proposes the Res-CNN3 technique for detecting wind turbine blade defects. In Res-CNN3, defect region detection is achieved through a bipartite process that processes the laser delta and RGB delta structure of a wind turbine blade image with an integration of residual networks and concatenated CNNs to determine the presence of typical defect regions in the image. The loss function is logistic regression, and a Selective Search (SS) algorithm is used to predict the regions of interest (RoI) of the input images for defects detection. Several experiments are conducted, and the outcomes prove that the proposed model has a high prospect for accuracy in solving the problem of defect detection in a manner similar to the advanced benchmark methods.

Assessing performance of spatial statistical stream network models in predicting alkalinity: A comparative study of two river basins in Türkiye

Developing and implementing water quality models that utilize sophisticated processes and plenty data in large river basins can be a challenging task. Factors such as sampling design, prediction at unsampled sites, and derivation of new spatial information in monitoring studies can restrict the methodologies applicable to water quality modeling. In recent years, spatial stream network (SSN) models, incorporating hydrologic distance and topological data structures, have emerged as a novel approach for predicting water quality in extensive study areas. These models have integrated conventional geostatistical methodologies and have demonstrated their effectiveness in analyzing water quality within large river basins. In this study, we aimed to investigate the feasibility of using SSN to predict alkalinity through a comparative analysis of two distinct river basins in Türkiye. We evaluated the prediction performance of SSN and the spatial characterization of the basins using spatial data sets as model components. The models are based on moving-average (MA) functions and constructed by evaluating alkalinity observations at 90 locations in Kucuk Menderes River Basin and 224 locations in the Coruh River Basin, using torgegram, correlogram, principal component analysis, and exhaustive search algorithm for covariate selection. In this research, we evaluated the optimal combination of covariates obtained from 20 different variables within the context of five spatial data set classes: topography, geology, climate, land use, and anthropology. The outcomes of the model indicate that impact of anthropogenic activities on autocorrelation based on hydrologic distance, impact of spatial covariates on variance, and root mean square prediction error (RMSPE) values of the predictions in Kucuk Menderes are more effective than those in the Coruh. The findings from the model predictions show that modeling water quality on stream network scale by using topological relationships and autocorrelation based on hydrologic distance can be recommended for rivers affected by complex and detailed processes.

Dynamic Parameter Optimization of High-Speed Pantograph Based on Swarm Intelligence and Machine Learning

Good pantograph–catenary interaction quality is a fundamental premise for ensuring stable and reliable current collection of high-speed trains, and the optimization of dynamic parameters of high-speed pantographs provides an effective approach to improve the current collection quality of the pantograph–catenary system. In this paper, with the objective of minimizing the standard deviation of the pantograph–catenary contact force, the multi-parameter joint optimization for pantograph at different filtering frequencies and running speeds was carried out by using swarm intelligence optimization algorithm and artificial neural network method. First, the selection operator in genetic algorithm (GA) was introduced into crow search algorithm (CSA), and the selective CSA was proposed, which can effectively improve the solution accuracy and convergence rate of multi-parameter optimization. Second, a radial basis function (RBF) neural network was used to construct a surrogate model of the standard deviation of contact force with respect to the running speed and pantograph dynamic parameters, and a method for optimizing the upper limit of mapping interval of the decision variables by the selective crow search algorithm (SCSA) was proposed, which effectively improves the generalization ability of the surrogate model. Finally, by combining the surrogate model and SCSA, optimization iterations for a total of 630 combined conditions such as cut-off frequency, running speed and pantograph dynamic parameters were conducted, and an optimization method for high-speed pantograph dynamic parameters with universal applicability was proposed.

A novel artificial intelligence approach to detect the breast cancer using KNNet technique with EPM gene profiling.

Women's most frequent type of cancer is breast cancer, second only to lung cancer. This paper summarizes changes in genomics and epigenetics and incremental biological activities. A tumour develops through a series of phases involving a separate abnormal gene. Even though many diseases cause DNA mutations, most treatments are designed to relieve symptoms rather than change the DNA. Clustering short palindromic repeats (CRISPR) or Cas9 is the primary approach for discovering and confirming tumorigenic genomic targets. A Kohonen neural network with an expression programming model was developed for gene selection. The main problem in genetic selection is reducing the number of features chosen while maintaining accuracy. This purpose is accomplished systematically. In the end, the approach method performed better than the existing quantum squirrel-inspired algorithm and the recurrent neural network oppositional call search algorithm for genetic selection. The KNNet-EPM model used an expression programming approach to identify gene biomarkers for breast cancer. This method was achieved with RAE of 42%, sensitivity of 93%, f1 score of 88%, accuracy of 98%, kappa score of 83%, specificity of 92% and MAE of 30%.

Multi-objective squirrel search algorithm for EEG feature selection

Feature selection plays a critical role in the application of Brain Computer Interface (BCI) systems. Many methods have been used to solve the feature selection problem, but they model it as a single-objective problem, considering only classification accuracy or number of features. To close this critical gap, we improve the squirrel search algorithm by combining it with the grid method, and propose a Multi-Objective Squirrel Search Algorithm (MOSSA) to solve the feature selection problem in BCI. We conduct experiments on three publicly available motion imagery datasets, and the experimental results reveal the best classification results of the method on dataset 1. The average classification accuracy of dataset 2 is 96.71%, with the number of selected features reduced to 18 on average. The highest classification accuracy of dataset 3 is 83.57% on the training set and 82.86% on the test set. In addition, we compare MOSSA with other algorithms and the results show the superiority of our proposed method in solving the feature selection problem. Finally, we combine MOSSA with an online application of BCI, where subjects visualize controlling the robot to perform the corresponding actions by the left and right hand movements. The average recognition rate of the three subjects is approximately 70%. In summary, the MOSSA is an effective method for solving the feature selection problem and is useful for the development of online applications of BCI.

ReliefF based feature selection and Gradient Squirrel search Algorithm enabled Deep Maxout Network for detection of heart disease

Detecting heart disease is challenging in clinical settings, leading to an increase in mortality rates. Current detection processes often rely on Electrocardiography (ECG) signal analysis, which requires accurate data processing and feature extraction. Traditional methods have limitations like processing time and accuracy. To address these issues, a novel approach called Gradient Squirrel Search Algorithm-Deep Maxout Network (GSSA-DMN) is proposed for more effective heart disease detection. The proposed GSSA-DMN approach involves several steps. Initially, input data is obtained from a specific database and subjected to data pre-processing, including log scaling for pattern transformation. Feature selection is then performed using ReliefF on the pre-processed data. The core of the approach lies in the Deep Maxout Network (DMN) trained by the Gradient Squirrel Search Algorithm (GSSA), which combines Gradient Descent Optimization (GDO) with the Squirrel Search Algorithm (SSA). The GSSA-DMN approach demonstrates remarkable performance. It achieves high accuracy, sensitivity, and specificity values of approximately 93.2%, 93%, and 91.5%, respectively. These results indicate its effectiveness in heart disease detection.Comparatively, the proposed GSSA-DMN method outperforms existing techniques. Its accuracy surpasses those of other methods by margins of 6.97%, 5.79%, 4.50%, 3.43%, and 1.93% when compared to BF-PSO, Bi-LSTM-CRF, XGBoost, RLNNC, and DMOA-SqueezeNet for K-value. This suggests that GSSA-DMN provides superior accuracy in detecting heart disease. In summary, the GSSA-DMN approach presents a promising solution for improving the accuracy and efficiency of heart disease detection compared to traditional methods and existing state-of-the-art techniques.

Intelligent Monitoring System to Assess Plant Development State Based on Computer Vision in Viticulture

Plant health plays an important role in influencing agricultural yields and poor plant health can lead to significant economic losses. Grapes are an important and widely cultivated plant, especially in the southern regions of Russia. Grapes are subject to a number of diseases that require timely diagnosis and treatment. Incorrect identification of diseases can lead to large crop losses. A neural network deep learning dataset of 4845 grape disease images was created. Eight categories of common grape diseases typical of the Black Sea region were studied: Mildew, Oidium, Anthracnose, Esca, Gray rot, Black rot, White rot, and bacterial cancer of grapes. In addition, a set of healthy plants was included. In this paper, a new selective search algorithm for monitoring the state of plant development based on computer vision in viticulture, based on YOLOv5, was considered. The most difficult part of object detection is object localization. As a result, the fast and accurate detection of grape health status was realized. The test results showed that the accuracy was 97.5%, with a model size of 14.85 MB. An analysis of existing publications and patents found using the search “Computer vision in viticulture” showed that this technology is original and promising. The developed software package implements the best approaches to the control system in viticulture using computer vision technologies. A mobile application was developed for practical use by the farmer. The developed software and hardware complex can be installed in any vehicle. Such a mobile system will allow for real-time monitoring of the state of the vineyards and will display it on a map. The novelty of this study lies in the integration of software and hardware. Decision support system software can be adapted to solve other similar problems. The software product commercialization plan is focused on the automation and robotization of agriculture, and will form the basis for adding the next set of similar software.

Interoperable IoMT Approach for Remote Diagnosis with Privacy-Preservation Perspective in Edge Systems.

The emergence of the Internet of Medical Things (IoMT) has brought together developers from the Industrial Internet of Things (IIoT) and healthcare providers to enable remote patient diagnosis and treatment using mobile-device-collected data. However, the utilization of traditional AI systems raises concerns about patient privacy. To address this issue, we present a privacy-enhanced approach for illness diagnosis within the IoMT framework. Our proposed interoperable IoMT implementation focuses on optimizing IoT network performance, including throughput, energy consumption, latency, packet delivery ratio, and network longevity. We achieve these improvements using techniques such as device authentication, energy-efficient clustering, environmental monitoring using Circular-based Hidden Markov Model (C-HMM), data verification using Awad's Entropy-based Ten-Fold Cross Entropy Verification (TCEV), and data confidentiality using Twine-LiteNet-based encryption. We employ the Search and Rescue Optimization algorithm (SRO) for optimal route selection, and the encrypted data are securely stored in a cloud server. With extensive network simulations using ns-3, our approach demonstrates substantial enhancements in the specified performance metrics compared with previous works. Specifically, we observe a 20% increase in throughput, a 15% reduction in packet drop rate (PDR), a 35% improvement in network lifetime, and a 10% decrease in energy consumption and delay. These findings underscore the efficacy of our approach in enhancing IoT network interoperability and protection, fostering improved patient care and diagnostic capabilities.

A weighted-sum chaotic sparrow search algorithm for interdisciplinary feature selection and data classification

In today’s data-driven digital culture, there is a critical demand for optimized solutions that essentially reduce operating expenses while attempting to increase productivity. The amount of memory and processing time that can be used to process enormous volumes of data are subject to a number of limitations. This would undoubtedly be more of a problem if a dataset contained redundant and uninteresting information. For instance, many datasets contain a number of non-informative features that primarily deceive a given classification algorithm. In order to tackle this, researchers have been developing a variety of feature selection (FS) techniques that aim to eliminate unnecessary information from the raw datasets before putting them in front of a machine learning (ML) algorithm. Meta-heuristic optimization algorithms are often a solid choice to solve NP-hard problems like FS. In this study, we present a wrapper FS technique based on the sparrow search algorithm (SSA), a type of meta-heuristic. SSA is a swarm intelligence (SI) method that stands out because of its quick convergence and improved stability. SSA does have some drawbacks, like lower swarm diversity and weak exploration ability in late iterations, like the majority of SI algorithms. So, using ten chaotic maps, we try to ameliorate SSA in three ways: (i) the initial swarm generation; (ii) the substitution of two random variables in SSA; and (iii) clamping the sparrows crossing the search range. As a result, we get CSSA, a chaotic form of SSA. Extensive comparisons show CSSA to be superior in terms of swarm diversity and convergence speed in solving various representative functions from the Institute of Electrical and Electronics Engineers (IEEE) Congress on Evolutionary Computation (CEC) benchmark set. Furthermore, experimental analysis of CSSA on eighteen interdisciplinary, multi-scale ML datasets from the University of California Irvine (UCI) data repository, as well as three high-dimensional microarray datasets, demonstrates that CSSA outperforms twelve state-of-the-art algorithms in a classification task based on FS discipline. Finally, a 5%-significance-level statistical post-hoc analysis based on Wilcoxon’s signed-rank test, Friedman’s rank test, and Nemenyi’s test confirms CSSA’s significance in terms of overall fitness, classification accuracy, selected feature size, computational time, convergence trace, and stability.

Expected polynomial-time randomized algorithm for graph coloring problem

Given a graph G=(V,E) with n=|V| vertices, the graph coloring problem is defined as Find a color vectorC=(c(v)), where c(v)∈{1,2,…,n} denotes the color of vertex v∈V, such that no monochromatic edge exists in G and the span, the total number of distinct colors in C, is minimized. Since the problem is NP-complete, a greedy coloring is commonly used for solving it. Greedy coloring visits the vertices of G following an order S, and while visiting a vertex v, it puts the minimum color absent in all neighbors of v. We show that the orders producing span ≤k, where k≤n is a positive integer, can be partitioned into disjoint subsets of equivalent orders. Next, we propose a selective search (SS) algorithm, which takes ρ as an input parameter, selects ≥ρ orders each from a different set of equivalent orders with high probability, applies greedy coloring on them, and returns the color vector with minimum span. We analytically show that SS performs better than greedy coloring with high probability by evaluating the same number of orders. We propose an incremental search heuristic (ISH), which ρ1 times execute SS with parameter ρ2 and returns the color vector with minimum span. A parallel version of ISH called PISH is also proposed, executing ρ1 SS calls in parallel. We show that ISH hits an optimum coloring for a graph G=(V,E) with χ(G)(Δ(G)+1)≤ne in expected O(|V|+|E|) time and space complexities. We have evaluated ISH and PISH on 136 challenging benchmarks and shown that they significantly outperform 10 existing state-of-the-art algorithms. Finally, we validated our theoretical findings by evaluating ISH and greedy coloring on random graphs.

Cognitively Enhanced Versions of Capuchin Search Algorithm for Feature Selection in Medical Diagnosis: a COVID-19 Case Study.

Feature selection (FS) is a crucial area of cognitive computation that demands further studies. It has recently received a lot of attention from researchers working in machine learning and data mining. It is broadly employed in many different applications. Many enhanced strategies have been created for FS methods in cognitive computation to boost the performance of the methods. The goal of this paper is to present three adaptive versions of the capuchin search algorithm (CSA) that each features a better search ability than the parent CSA. These versions are used to select optimal feature subset based on a binary version of each adapted one and the k-Nearest Neighbor (k-NN) classifier. These versions were matured by applying several strategies, including automated control of inertia weight, acceleration coefficients, and other computational factors, to ameliorate search potency and convergence speed of CSA. In the velocity model of CSA, some growth computational functions, known as exponential, power, and S-shaped functions, were adopted to evolve three versions of CSA, referred to as exponential CSA (ECSA), power CSA (PCSA), and S-shaped CSA (SCSA), respectively. The results of the proposed FS methods on 24 benchmark datasets with different dimensions from various repositories were compared with other k-NN based FS methods from the literature. The results revealed that the proposed methods significantly outperformed the performance of CSA and other well-established FS methods in several relevant criteria. In particular, among the 24 datasets considered, the proposed binary ECSA, which yielded the best overall results among all other proposed versions, is able to excel the others in 18 datasets in terms of classification accuracy, 13 datasets in terms of specificity, 10 datasets in terms of sensitivity, and 14 datasets in terms of fitness values. Simply put, the results on 15, 9, and 5 datasets out of the 24 datasets studied showed that the performance levels of the binary ECSA, PCSA, and SCSA are over 90% in respect of specificity, sensitivity, and accuracy measures, respectively. The thorough results via different comparisons divulge the efficiency of the proposed methods in widening the classification accuracy compared to other methods, ensuring the ability of the proposed methods in exploring the feature space and selecting the most useful features for classification studies.