Improving K-means clustering with enhanced Firefly Algorithms

State-of-the-art ensemble learning and unsupervised learning in fatigue crack recognition of glass fiber reinforced polyester composite (GFRP) using acoustic emission

Fatigue strength is one of the most important properties of composite materials because it directly relates to their lifespan. Acoustic emission (AE) is a passive structural health monitoring (SHM) technique that provides real-time damage detection based on stress waves generated by cracks in the structure. This study evaluates the damage progression on glass fiber reinforced polyester composite specimens using different approaches of machine learning. Different methodologies for damage detection and characterization of AE parameters are presented. Three different ensemble learning methods namely, XGboost, LightGBM, and CatBoost were chosen to predict damages and AE parameters. SHAP values were used to select AE key features and K-means algorithms were employed to classify damage severity. The accuracy of these approaches demonstrates the reliability of various machine learning techniques in predicting the fatigue life of composite materials using acoustic emission.

The research of a novel WOG-YOLO algorithm for autonomous driving object detection

Object detection has been one of the critical technologies in autonomous driving. To improve the detection precision, a novel optimization algorithm is presented to enhance the performance of the YOLOv5 model. First, by improving the hunting behavior of the grey wolf algorithm(GWO) and incorporating it into the whale optimization algorithm(WOA), a modified whale optimization algorithm(MWOA) is proposed. The MWOA leverages the population's concentration ratio to calculate [Formula: see text] for selecting the hunting branch of GWO or WOA. Tested by six benchmark functions, MWOA is proven to possess better global search ability and stability. Second, the C3 module in YOLOv5 is substituted by G-C3, and an extra detection head is added, thus a highly optimizable detection G-YOLO network is constructed. Based on the self-built dataset, 12 initial hyperparameters in the G-YOLO model are optimized by MWOA using a score fitness function of compound indicators, thus the final hyperparameters are optimized and the whale optimization G-YOLO (WOG-YOLO) model is obtained. In comparison with the YOLOv5s model, the overall mAP increases by 1.7[Formula: see text], the mAP of pedestrians increases by 2.6[Formula: see text] and the mAP of cyclists increases by 2.3[Formula: see text].

Detection and Classification of Skin Cancer Using Unmanned Transfer Learning Based Probabilistic Multi-Layer Dense Networks

Skin cancer is one of the most dangerous cancers that may occur for different age groups of people. As a result, early identification of skin cancer has the potential to save millions of lives. In Traditional machine learning approaches, there are various drawbacks in detection and classification of skin lesions. As a result, to achieve the robust performance, initially the joint trilateral and bilateral filter (JTBF) with convolutional auto encoder and decoder (CAED)-based preprocessing method is used to enhance the skin lesion and also removes hair from lesions. Then, transfer learning-based probabilistic multi-layer dense networks (PMDN) method-based unmanned Transfer learning segmentation method is adapted for accurately detecting the cancer region on skin lesions. Further, transfer learning convolution neural network (TL-CNN) is used to extract the features from the segmented region, which extracts the detailed inter-disease-dependent (IDD) and intra-disease specific (IDS) features. Finally, Alexa Net model is trained and tested with the IDD, IDS features and classifies the eight different skin cancer types. The complexity of the transfer learning networks is optimized by the using the Adam optimizer. Finally, the simulation results show that the proposed model resulted in superior segmentation, feature extraction, and classification performances as compared to conventional approaches. Further, the proposed method achieved 99.937% segmentation accuracy, 99.47% feature extraction accuracy, and 99.27% classification accuracy on ISIC-2019 public challenge dataset.

Improvement of K-Means Algorithm and Its Application in Air Passenger Grouping

The k-means is one of the most popular clustering analysis algorithm and widely used in various fields. Nevertheless, it continues to have some shortcomings, for example, extremely sensitive to the initial center points selection and the special points such as noise or outliers. Therefore, this paper proposed initial center points' selection optimization and phased assignment optimization to improve the k-means algorithm. The experimental results on 15 real-world and 10 synthetic datasets show that the improved k-means outperforms its main competitor k-means ++ and under the same setting conditions, namely, using the default parameters,its clustering performance is better than Affinity Propagation, Mean Shift, and DBSCAN. The proposed algorithm was applied to analyze the airline seat selection data to air passengers grouping. The clustering results, as well as absolute deviation rate analysis, realized customer grouping and found out suitable audience group for the recommendation of seat selection services.

NDPD: an improved initial centroid method of partitional clustering for big data mining

Purpose

The K-means (KM) clustering algorithm is extremely responsive to the selection of initial centroids since the initial centroid of clusters determines computational effectiveness, efficiency and local optima issues. Numerous initialization strategies are to overcome these problems through the random and deterministic selection of initial centroids. The random initialization strategy suffers from local optimization issues with the worst clustering performance, while the deterministic initialization strategy achieves high computational cost. Big data clustering aims to reduce computation costs and improve cluster efficiency. The objective of this study is to achieve a better initial centroid for big data clustering on business management data without using random and deterministic initialization that avoids local optima and improves clustering efficiency with effectiveness in terms of cluster quality, computation cost, data comparisons and iterations on a single machine.

Design/methodology/approach

This study presents the Normal Distribution Probability Density (NDPD) algorithm for big data clustering on a single machine to solve business management-related clustering issues. The NDPDKM algorithm resolves the KM clustering problem by probability density of each data point. The NDPDKM algorithm first identifies the most probable density data points by using the mean and standard deviation of the datasets through normal probability density. Thereafter, the NDPDKM determines K initial centroid by using sorting and linear systematic sampling heuristics.

Findings

The performance of the proposed algorithm is compared with KM, KM++, Var-Part, Murat-KM, Mean-KM and Sort-KM algorithms through Davies Bouldin score, Silhouette coefficient, SD Validity, S_Dbw Validity, Number of Iterations and CPU time validation indices on eight real business datasets. The experimental evaluation demonstrates that the NDPDKM algorithm reduces iterations, local optima, computing costs, and improves cluster performance, effectiveness, efficiency with stable convergence as compared to other algorithms. The NDPDKM algorithm minimizes the average computing time up to 34.83%, 90.28%, 71.83%, 92.67%, 69.53% and 76.03%, and reduces the average iterations up to 40.32%, 44.06%, 32.02%, 62.78%, 19.07% and 36.74% with reference to KM, KM++, Var-Part, Murat-KM, Mean-KM and Sort-KM algorithms.

Originality/value

The KM algorithm is the most widely used partitional clustering approach in data mining techniques that extract hidden knowledge, patterns and trends for decision-making strategies in business data. Business analytics is one of the applications of big data clustering where KM clustering is useful for the various subcategories of business analytics such as customer segmentation analysis, employee salary and performance analysis, document searching, delivery optimization, discount and offer analysis, chaplain management, manufacturing analysis, productivity analysis, specialized employee and investor searching and other decision-making strategies in business.

Use of mixed-type data clustering algorithm for characterizing temporal and spatial distribution of biosecurity border detections of terrestrial non-indigenous species

Appropriate inspection protocols and mitigation strategies are a critical component of effective biosecurity measures, enabling implementation of sound management decisions. Statistical models to analyze biosecurity surveillance data are integral to this decision-making process. Our research focuses on analyzing border interception biosecurity data collected from a Class A Nature Reserve, Barrow Island, in Western Australia and the associated covariates describing both spatial and temporal interception patterns. A clustering analysis approach was adopted using a generalization of the popular k-means algorithm appropriate for mixed-type data. The analysis approach compared the efficiency of clustering using only the numerical data, then subsequently including covariates to the clustering. Based on numerical data only, three clusters gave an acceptable fit and provided information about the underlying data characteristics. Incorporation of covariates into the model suggested four distinct clusters dominated by physical location and type of detection. Clustering increases interpretability of complex models and is useful in data mining to highlight patterns to describe underlying processes in biosecurity and other research areas. Availability of more relevant data would greatly improve the model. Based on outcomes from our research we recommend broader use of cluster models in biosecurity data, with testing of these models on more datasets to validate the model choice and identify important explanatory variables.

Estimation of Dairy Cow Survival in the First Three Lactations for Different Culling Reasons Using the Kaplan-Meier Method

The aims of the study were: (i) to compare survival curves for cows culled for different reasons over three successive lactations using the Kaplan-Meier estimator; (ii) to determine the effects of breeding documentation parameters on cow survival; (iii) to investigate the similarity between culling categories. The survival times for a subset of 347,939 Holstein-Friesian cows culled between 2017 and 2018 in Poland were expressed in months from calving to culling or the end of lactation. The survival tables were constructed for each culling category and lactation number. The survival curves were also compared. The main culling categories were reproductive disorders-40%, udder diseases-13 to 15%, and locomotor system diseases-above 10%. The survival curves for cows from individual culling categories had similar shapes. A low probability of survival curves for metabolic and digestive system diseases and respiratory diseases was observed in each of the three lactations. The contagious disease category was almost non-existent in the first lactation. The greatest influence on the relative culling risk was exerted by age at first calving, lactation length, calving interval, production subindex, breeding value for longevity, temperament, and average daily milk yield. A more accurate method of determining culling reasons would be required.

Binary Aquila Optimizer for Selecting Effective Features from Medical Data: A COVID-19 Case Study

Medical technological advancements have led to the creation of various large datasets with numerous attributes. The presence of redundant and irrelevant features in datasets negatively influences algorithms and leads to decreases in the performance of the algorithms. Using effective features in data mining and analyzing tasks such as classification can increase the accuracy of the results and relevant decisions made by decision-makers using them. This increase can become more acute when dealing with challenging, large-scale problems in medical applications. Nature-inspired metaheuristics show superior performance in finding optimal feature subsets in the literature. As a seminal attempt, a wrapper feature selection approach is presented on the basis of the newly proposed Aquila optimizer (AO) in this work. In this regard, the wrapper approach uses AO as a search algorithm in order to discover the most effective feature subset. S-shaped binary Aquila optimizer (SBAO) and V-shaped binary Aquila optimizer (VBAO) are two binary algorithms suggested for feature selection in medical datasets. Binary position vectors are generated utilizing S- and V-shaped transfer functions while the search space stays continuous. The suggested algorithms are compared to six recent binary optimization algorithms on seven benchmark medical datasets. In comparison to the comparative algorithms, the gained results demonstrate that using both proposed BAO variants can improve the classification accuracy on these medical datasets. The proposed algorithm is also tested on the real-dataset COVID-19. The findings testified that SBAO outperforms comparative algorithms regarding the least number of selected features with the highest accuracy.

Multidimensional Heterogeneous Network Link Adaptation Based on Mobile Environment

With the development of communication technology, train control operation system develops gradually, which significantly improves the reliability and efficiency of train operation. The current mobile Internet has gradually highlighted the many limitations of the mobile Internet in the high-speed mobile environment, which seriously deteriorate the service quality and user experience, and cause a waste of resources. In order to meet the real-time requirements of network communication resource scheduling in the mobile environment, aiming at the multidimensional dynamic adaptation framework constructed in a mobile environment, a service and network adaptation mechanism based on link failure state prediction is proposed in the paper. First, cross-layer theoretical analysis and actual data analysis are combined to construct a wireless link failure probability model. Then, reliable transmission requirements and transmission overhead are applied to optimize goals. Finally, simulation experiments are carried out according to the railway network data to evaluate the E-GCF adaptation algorithm. The experiment results show that compared with the current mainstream algorithms, the prediction accuracy of this adaptation algorithm is improved by 25%. The execution time of the algorithm is reduced by 9.6 seconds and the successful submission rate is as high as 99.99%. The advantages of the algorithm are significantly superior other algorithms. It proves that the research method of this paper can effectively improve the satisfaction rate and utility value of reliable transmission, as well as enhance the throughput performance. It solves the adaptation problems of frequent switching and low utilization of heterogeneous networks in a mobile environment, which contributes to the high-quality communication service of mobile network.

Color Image Contrast Enhancement Using Modified Firefly Algorithm

The image enhancement process is used for improving the standard of the image, it's inspired by the development of human perception pictorial information. Increasing the contrast of the image, removing the unwanted noise from the images is the picture enhancement process. A histogram of the low contrast images and depth image is employed to enhance image contrast. In the proposed work, a color image as input and extract out the red, green, and blue pixel matrixes from it, then obtain the optimized histogram using the modified firefly algorithm and compare the performance matrices like PSNR and Entropy, etc. with other optimization techniques.

Maxmin Data Range Heuristic-Based Initial Centroid Method of Partitional Clustering for Big Data Mining

The centroid-based clustering algorithm depends on the number of clusters, initial centroid, distance measures, and statistical approach of central tendencies. The initial centroid initialization algorithm defines convergence speed, computing efficiency, execution time, scalability, memory utilization, and performance issues for big data clustering. Nowadays various researchers have proposed the cluster initialization techniques, where some initialization techniques reduce the number of iterations with the lowest cluster quality, and some initialization techniques increase the cluster quality with high iterations. For these reasons, this study proposed the initial centroid initialization based Maxmin Data Range Heuristic (MDRH) method for K-Means (KM) clustering that reduces the execution times, iterations, and improves quality for big data clustering. The proposed MDRH method has compared against the classical KM and KM++ algorithms with four real datasets. The MDRH method has achieved better effectiveness and efficiency over RS, DB, CH, SC, IS, and CT quantitative measurements.

Hybrid Firefly-Ontology based Clustering Algorithm for Analyzing Tweets to Extract Causal Factors

Social media especially Twitter has become ubiquitous among people where they express their opinions on various domains. This paper presents a Hybrid Firefly – Ontology-based Clustering (FF-OC) algorithm which attempts to extract factors impacting a major public issue that is trending. In this research work, the issue of food price rise and disease which was trending during the time of the investigation is considered. The novelty of the algorithm lies in the fact that it clusters the association rules without any prior knowledge. The findings from the experimentation suggest different factors impacting the rise of price in food items and diseases such as diabetes, flu, zika virus. The empirical results show the significant improvement when compared with Artificial Bees Colony, Cuckoo Search Algorithm, Particle Swarm Optimization, and Ant Colony Optimization based clustering algorithms. The proposed method gives an improvement of 81% in terms of DB index, 79% in terms of silhouette index, 85% in terms of C index when compared to other algorithms.

K-Means-Based Nature-Inspired Metaheuristic Algorithms for Automatic Data Clustering Problems: Recent Advances and Future Directions

K-means clustering algorithm is a partitional clustering algorithm that has been used widely in many applications for traditional clustering due to its simplicity and low computational complexity. This clustering technique depends on the user specification of the number of clusters generated from the dataset, which affects the clustering results. Moreover, random initialization of cluster centers results in its local minimal convergence. Automatic clustering is a recent approach to clustering where the specification of cluster number is not required. In automatic clustering, natural clusters existing in datasets are identified without any background information of the data objects. Nature-inspired metaheuristic optimization algorithms have been deployed in recent times to overcome the challenges of the traditional clustering algorithm in handling automatic data clustering. Some nature-inspired metaheuristics algorithms have been hybridized with the traditional K-means algorithm to boost its performance and capability to handle automatic data clustering problems. This study aims to identify, retrieve, summarize, and analyze recently proposed studies related to the improvements of the K-means clustering algorithm with nature-inspired optimization techniques. A quest approach for article selection was adopted, which led to the identification and selection of 147 related studies from different reputable academic avenues and databases. More so, the analysis revealed that although the K-means algorithm has been well researched in the literature, its superiority over several well-established state-of-the-art clustering algorithms in terms of speed, accessibility, simplicity of use, and applicability to solve clustering problems with unlabeled and nonlinearly separable datasets has been clearly observed in the study. The current study also evaluated and discussed some of the well-known weaknesses of the K-means clustering algorithm, for which the existing improvement methods were conceptualized. It is noteworthy to mention that the current systematic review and analysis of existing literature on K-means enhancement approaches presents possible perspectives in the clustering analysis research domain and serves as a comprehensive source of information regarding the K-means algorithm and its variants for the research community.

Fault Feature Extraction for Reciprocating Compressors Based on Underdetermined Blind Source Separation

In practical engineering applications, the vibration signals collected by sensors often contain outliers, resulting in the separation accuracy of source signals from the observed signals being seriously affected. The mixing matrix estimation is crucial to the underdetermined blind source separation (UBSS), determining the accuracy level of the source signals recovery. Therefore, a two-stage clustering method is proposed by combining hierarchical clustering and K-means to improve the reliability of the estimated mixing matrix in this paper. The proposed method is used to solve the two major problems in the K-means algorithm: the random selection of initial cluster centers and the sensitivity of the algorithm to outliers. Firstly, the observed signals are clustered by hierarchical clustering to get the cluster centers. Secondly, the cosine distance is used to eliminate the outliers deviating from cluster centers. Then, the initial cluster centers are obtained by calculating the mean value of each remaining cluster. Finally, the mixing matrix is estimated with the improved K-means, and the sources are recovered using the least square method. Simulation and the reciprocating compressor fault experiments demonstrate the effectiveness of the proposed method.

Feature Selection Using Enhanced Particle Swarm Optimisation for Classification Models

In this research, we propose two Particle Swarm Optimisation (PSO) variants to undertake feature selection tasks. The aim is to overcome two major shortcomings of the original PSO model, i.e., premature convergence and weak exploitation around the near optimal solutions. The first proposed PSO variant incorporates four key operations, including a modified PSO operation with rectified personal and global best signals, spiral search based local exploitation, Gaussian distribution-based swarm leader enhancement, and mirroring and mutation operations for worst solution improvement. The second proposed PSO model enhances the first one through four new strategies, i.e., an adaptive exemplar breeding mechanism incorporating multiple optimal signals, nonlinear function oriented search coefficients, exponential and scattering schemes for swarm leader, and worst solution enhancement, respectively. In comparison with a set of 15 classical and advanced search methods, the proposed models illustrate statistical superiority for discriminative feature selection for a total of 13 data sets.

Study on frequency optimization and mechanism of ultrasonic waves assisting water flooding in low-permeability reservoirs

Water flooding is one of widely used technique to improve oil recovery from conventional reservoirs, but its performance in low-permeability reservoirs is barely satisfactory. Besides adding chemical agents, ultrasonic wave is an effective and environmental-friendly strategy to assist in water flooding for enhanced oil recovery (EOR) in unconventional reservoirs. The acoustic frequency plays a dominating role in the EOR performance of ultrasonic wave and is usually optimized through a series of time-consuming laboratory experiments. Hence, this study proposes an unsupervised learning method to group low-permeability cores in terms of permeability, porosity and wettability. This grouping algorithm succeeds to classify the 100 natural cores adopted in this study into five categories and the water flooding experiment certificates the accuracy and reliability of the clustering results. It is proved that ultrasonic waves can further improve the oil recovery yielded by water-flooding, especially in the oil-wet and weakly water-wet low-permeability cores. Furthermore, we investigated the EOR mechanism of ultrasonic waves in the low-permeability reservoir via scanning electron microscope observation, infrared characterization, interfacial tension and oil viscosity measurement. Although ultrasonic waves cannot ameliorate the components of light oil as dramatically as those of heavy oil, such compound changes still contribute to the oil viscosity and oil-water interfacial tension reductions. More importantly, ultrasonic waves may modify the micromorphology of low-permeability cores and improve the pore connectivity.