clustering

Metaheuristics have proved highly effective in addressing optimization challenges. Various algorithms address the clustering problem to find optimal centers for the clusters. One of the disadvantages of some of these algorithms is stagnation in local optima, especially for big data. If this problem is not properly solved, the clustering process will suffer. This research introduces a new hybrid method by merging the capabilities of two metaheuristic algorithms: Harris hawks optimization algorithm (HHO) and slime mould algorithm (SMA). These metaheuristic methods are employed to determine the best location for the cluster centers. Optimization aims to reduce intra-cluster distance. In other words, the data points of each cluster should be close to its cluster center and also to avoid local optima. The effectiveness of these techniques is assessed and contrasted with the SMA and HHO algorithms on Iris, Vowel and Wine data sets. Compared to mentioned algorithms, our proposed method exhibits significantly improved convergence speed. The results also proved this method can properly find the optimal centers for clustering which finally improves the performance of the proposed method.

This paper is the second in the series celebrating the mathematical works of Professor Themba Dube. In this sequel, we give prominence to Dube's pivotal contributions on pointfree convergence at the unstructured frame level, in the category of locales, and on his noteworthy conceptions on extensions and frame quotients. We distill and draw attention to particular studies of Dube on filters and his novel characterizations of certain conservative pointfree properties by filter and ultrafilter convergence, notably normality, almost realcompactness, and pseudocompactness. We also feature Dube's joint work on convergence and clustering of filters in Loc and coconvergence and coclustering of ideals in the category Frm.

The efficiency of Independent Component Analysis ($\rm ICA$) algorithms relies heavily on the choice of objective function and optimization algorithms. The design of objective functions for $\rm ICA$ algorithms necessitate a foundation built upon specific dependence criteria. This paper will investigate a general class of dependency criteria based on the copula density function. One of the aims of this study is to characterize the independence between two random variables and investigate their properties. Additionally, this paper introduces a novel algorithm for $\rm ICA$ based on estimators derived from the proposed criteria. To compare the performance of the proposed algorithm against existing methods, a Monte Carlo simulation-based approach was employed. The results of this simulation revealed significant improvements in the algorithm's outputs. Finally, the algorithm was tested on a batch of time series data related to the international tourism receipts index. It served as a pre-processing procedure within a hybrid clustering algorithm alongside ${\tt PAM}$. The obtained results demonstrated that the utilization of this algorithm led to improved performance in clustering countries based on their international tourism receipts index.

Wireless Sensor Networks (WSNs) encounter considerable challenges in terms of energy efficiency and network longevity due to their limited energy resources. This paper proposes a novel hybrid clustering-based routing protocol that addresses these challenges by integrating fuzzy logic for dynamic and adaptive cluster head (CH) selection based on residual energy, node degree, and proximity, and genetic algorithms (GA) for optimising cluster formation by balancing energy consumption and minimising communication distances. The protocol's objectives are threefold: to minimise energy consumption, extend network lifespan, and enhance Quality of Service (QoS).The proposed method was simulated in MATLAB and benchmarked against the LEACH and TEEN protocols. The results demonstrated the protocol's superior performance, achieving a 30% reduction in energy consumption, a 25% increase in network longevity, and higher data reliability. The primary factors contributing to this enhanced performance are the integrated use of fuzzy logic for optimised cluster head selection and genetic algorithms for optimal cluster formation. The findings substantiate the protocol's capacity to substantially enhance the energy efficiency and scalability of WSNs, providing a resilient and pragmatic solution for practical applications in real-world settings.

A well-known clustering problem called Density-Based Spatial Clustering of Applications with Noise~(DBSCAN) involves computing the solutions of at least one disk range query per input point, computing the connected components of a graph, and bichromatic fixed-radius nearest neighbor. MapReduce class is a model where a sublinear number of machines, each with sublinear memory, run for a polylogarithmic number of parallel rounds. Most of these problems either require quadratic time in the sequential model or are hard to compute in a constant number of rounds in MapReduce. In the Euclidean plane, DBSCAN algorithms with near-linear time and a randomized parallel algorithm with a polylogarithmic number of rounds exist. We solve DBSCAN in the Euclidean plane in a constant number of rounds in MapReduce, assuming the minimum number of points in range queries is constant and each connected component fits inside the memory of a single machine and has a constant diameter.

Cluster analysis in data envelopment analysis (DEA) is determining clusters for the units under evaluation regarding to their similarity. which measure of distances define their similarities. Over the years, researches have been carried out in the field of clustering of DMUs. In this paper, an algorithm for clustering units using projecting them on the frontier is presented. In fact, we gained for every decision making unit (DMU), nearest most productive scale size (MPSS) as target, to find number of clusters 2 method applied. Silhouette index was used to measure similarity value for our clustering. Numerical examples are provided to illustrate the proposed method and its results.

This paper proposes an observation-driven finite mixture model for clustering high-dimension data. A simple algorithm using static hidden variables statically clusters the data into separate model components. The model accommodates normal and skew-normal distributed mixtures with time-varying component means, covariance matrices and skewness coefficient. These parameters are estimated using the EM algorithm and updated with the Generalized Autoregressive Scale (GAS) approach. Our proposed model is preferably clustered using a skew-normal distribution rather than a normal distribution when dealing with real data that may be skewed and asymmetrical. Finally, our proposed model will be evaluated using a simulation study and the results will be discussed using a real data set.

The smart city model on multi-agent systems and the Internet of Things using a wireless sensor network is designed to improve the quality of life for citizens, increase resource efficiency, and reduce costs. This model enables the collection, analysis, and sharing of information by connecting and coordinating devices and systems within the smart city. In this model, intelligent agents act as sensors, and the smart gateway plays the role of a base station. The main goal of this model is to reduce energy consumption. To achieve this goal, intelligent agents are divided into clusters, with each cluster having a cluster head. The cluster head’s task is to collect and aggregate information from the intelligent agents within its cluster and send it to the smart gateway. In the proposed method, each intelligent agent selects a cluster in a distributed manner. An intelligent agent may choose another intelligent agent as its cluster head or select itself as a cluster head and directly send the data to the smart gateway. Each intelligent agent chooses the cluster head after calculating the importance level of neighboring intelligent agents. By using this model, cities can experience increased resource efficiency and cost reduction by leveraging innovative technologies. The proposed method has been implemented in different scenarios of smart cities, such as sparse and crowded smart cities with varying message sizes. In all simulations, the proposed method demonstrated good capabilities in optimizing energy consumption management.

With the advancements in science and technology‎, ‎the industrial and aviation sectors have witnessed a significant increase in data‎. ‎A vast amount of data is generated and utilized continuously‎. ‎It is imperative to employ data mining techniques to extract and uncover knowledge from this data‎. ‎Data mining is a method that enables the extraction of valuable information and hidden relationships from datasets‎. ‎However‎, ‎the current aviation data presents challenges in effectively extracting knowledge due to its large volume and diverse structures‎. ‎Air Traffic Management (ATM) involves handling Big data‎, ‎which exceeds the capacity of conventional acquisition‎, ‎matching‎, ‎management‎, ‎and processing within a reasonable timeframe‎. ‎Aviation Big data exists in batch forms and streaming formats‎, ‎necessitating the utilization of parallel hardware and software‎, ‎as well as stream processing‎, ‎to extract meaningful insights‎. ‎Currently‎, ‎the map-reduce method is the prevailing model for processing Big data in the aviation industry‎. ‎This paper aims to analyze the evolving trends in aviation Big data processing methods‎, ‎followed by a comprehensive investigation and discussion of data analysis techniques‎. ‎We implement the map-reduce optimization of the K-Means algorithm in the Hadoop and Spark environments‎. ‎The K-Means map-reduce is a crucial and widely applied clustering method‎. ‎Finally‎, ‎we conduct a case study to analyze and compare aviation Big data related to air traffic management in the USA using the K-Means map-reduce approach in the Hadoop and Spark environments‎. ‎The analyzed dataset includes flight records‎. ‎The results demonstrate the suitability of this platform for aviation Big data‎, ‎considering the characteristics of the aviation dataset‎. ‎Furthermore‎, ‎this study presents the first application of the designed program for air traffic management‎.

Unmanned Aerial Vehicles (UAVs) bring both potential and difficulties for emergency applications, including packet loss and changes in network topology. UAVs are also quickly taking up a sizable portion of the airspace, allowing Flying Ad-hoc NETworks (FANETs) to conduct effective ad hoc missions. Therefore, building routing protocols for FANETs is difficult due to flight restrictions and changing topology. To solve these problems, a bio-inspired route selection technique is proposed for FANET. A combined trustworthy and bioinspired-based transmission strategy is developed as a result of the growing need for dynamic and adaptable communications in FANETs. The fitness theory is used to assess direct trust and evaluate credibility and activity to estimate indirect trust. In particular, assessing UAV behavior is still a crucial problem in this field. It recommends fuzzy logic, one of the most widely utilized techniques for trusted route computing, for this purpose. Fuzzy logic can manage complicated settings by classifying nodes based on various criteria. This method combines geocaching and unicasting, anticipating the location of intermediate UAVs using 3-D estimates. This method guarantees resilience, dependability, and an extended path lifetime, improving FANET performance noticeably. Two primary features of FANETs that shorten the route lifetime must be accommodated in routing. First, the collaborative nature necessitates communication and coordination between the flying nodes, which uses a lot of energy. Second, the flying nodes' highly dynamic mobility pattern in 3D space may cause link disconnection because of their potential dispersion. Using ant colony optimization, it employs trusted leader drone selection within the cluster and safe routing among leaders. a fuzzy‐based UAV behavior analytics is presented for trust management in FANETs. Compared to existing protocols, the simulated results demonstrate improvements in delay routing overhead in FANET.

The fuzzy-C-means (FCM) algorithm is one of the most famous fuzzy clus-tering algorithms, but it gets stuck in local optima. In addition, this algo-rithm requires the number of clusters. Also, the density-based spatial of the application with noise (DBSCAN) algorithm, which is a density-based clus-tering algorithm, unlike the FCM algorithm, should not be pre-numbered. If the clusters are specific and depend on the number of clusters, then it can determine the number of clusters. Another advantage of the DBSCAN clus-tering algorithm over FCM is its ability to cluster data of different shapes. In this paper, in order to overcome these limitations, a hybrid approach for clustering is proposed, which uses FCM and DBSCAN algorithms. In this method, the optimal number of clusters and the optimal location for the centers of the clusters are determined based on the changes that take place according to the data set in three phases by predicting the possibility of the problems stated in the FCM algorithm. With this improvement, the values of none of the initial parameters of the FCM algorithm are random, and in the first phase, it has been tried to replace these random values to the optimal in the FCM algorithm, which has a significant effect on the convergence of the algorithm because it helps to reduce iterations. The proposed method has been examined on the Iris flower and compared the results with basic FCM   algorithm and another algorithm. Results shows the better performance of the proposed method.

Due to the spread of the Internet and its pervasiveness, ``big data" is created daily. Processing this amount of data requires a system with high processing power. In fact, the production and collection of data from a wide range of different equipment and tools lead to the creation of large-scale databases. In dealing with large and unstructured databases and their management, there are always challenges. This study aims to present a model to increase the clustering accuracy of big data using a fuzzy clustering system based on data mining in a MatLab programming environment. For this purpose, first, the importance of each variable in the decision tree models in SPSSModeler software is determined, then with the help of these results, fuzzy rules are explained and a fuzzy inference system is formed in MATLAB software. This study uses data mining techniques such as C\&R Tree, Chaid and C5.0 to study the development of the FCM method to increase clustering accuracy in high volume data and related factors such as data preparation indicators, data type Data quality, data dimensions, data volume and number of clusters were evaluated as inputs and clustering accuracy index was evaluated as output. Then, with the help of these results, the rules of forming a fuzzy inference system were determined and by explaining the membership functions of the decision model, it showed what effect each input index has on the output index.

The user cold start challenge, occurs when a user joins the system which used recommender systems, for the first time. Since the recommender system has no knowledge of the user preferences at first, it will be difficult to make appropriate recommendations. In this paper, users’ demographics information are used for clustering to find the users with similar preferences in order to improve the cold start challenge by employing the kmeans, k-medoids, and k-prototypes algorithms. The target user’s neighbors are determined by using a hybrid similarity measure including a combination of users’ demographics information similarity and users rating similarity. The asymmetric Pearson correlation coefficient utilized to calculate the user rating similarity, whereas GMR (i.e., global most rated) and GUC(i.e., global user local clustering) strategies are adopted to make recommendations. The proposed method was implemented on MovieLens dataset. The results of this research shows that the MAE of the proposed method has improved the accuracy of the proposals up to about 26% compared to the GMR method and up to about 34% compared to the GUC method. Also, the results show about 60% improvement in terms of rating coverage compared to the GMR and GUC methods.

One of well-known techniques in data mining is clustering. Clustering method which is very popular is K-means cluster because its algorithm is very easy and simple. However, K-means cluster has some weaknesses, one of which is that the cluster result is sensitive towards centroid initialization so that the cluster result tends to local optimal. This paper explains the modification of K-means cluster, that is, K-means hybridization with ant colony optimization (K-ACO). Ant Colony Optimization (ACO) is optimization algorithm based on ant colony behavior. Through K-ACO, the weaknesses of cluster result which tends to local optimal can be overcome well. The application of hybrid method of K-ACO with the use of R program gives better accuracy compared to K-means cluster. K-means cluster accuracy yielded by Minitab, Mathlab, and SAS at iris data is 89%. Meanwhile, K-ACO hybrid clustering with R program simulated on 38 treatments with 3-time repetitions gives accuracy result of 93,10%.

This study combines Data Envelopment Analysis (DEA) with machine learning clustering method in datamining for finding the most efficient Decision Making Unit (DMU) and the best clustering algorithm, respectively. The problem of assessment of units by using DEA may not be straightforward due to the data uncertainty. Several scholars have been attracted to develop methods which incorporate uncertainty into input/output values in the DEA literature. On the other hand, in many real world applications, the data is reported in the form of intervals. This means that each input/output value is selected from a symmetric box. In the DEA literature, this type of uncertainty has been addressed as Interval DEA approaches. The main goal of this study is to evaluate the efficiency of banks in the case of data uncertainty with cross-efficiency method in the DEA literature. For this purpose, we consider the BCC-CCR and CCR-BCC models in the presence of uncertain data to find the superior model. After applying the optimization models, in machine learning step, clustering method is applied. Clustering is a procedure for grouping similar items together which this group is called the cluster. Also, the different clustering algorithms can be used according to the behavior of data. In this study, we apply the farthest first and expectation maximization algorithms and show that, in the case of data uncertainty, the BCC-CCR and farthest first algorithms are as a superior optimization model and machine learning algorithm, respectively.

The Internet of Things (IoT) is an emerging phenomenon in the field of communication, in which smart objects communicate with each other and respond to user requests. The IoT provides an integrated framework providing interoperability across various platforms. One of the most essential and necessary components of IoT is wireless sensor networks. Sensor networks play a vital role in the lowest level of IoT. Sensors in sensor networks use batteries which are not replaceable, and hence, energy consumption becomes of great importance. For this reason, many algorithms have been recently proposed to reduce energy consumption. In this paper, a meta-heuristic method called whale optimization algorithm(WOA) is used to clustering and select the optimal cluster head in the network. Factors such as residual energy, shorter distance, and collision reduction have been considered to determine the optimal cluster head. To prove the optimal performance of the proposed method, it is simulated and compared with three other methods in the same conditions. It outperforms the other methods in terms of energy consumption and the number of dead nodes.

Using smart methods for the automatic generation of keywords in legislative documents has  attracted the attention of many researchers over the past few decades. With the increasing  evelopment of legislative documents and the large volume of unstructured texts, the need for rapid  access to these documents has become more significant. Extracting the keywords in legislative documents will accel-erate the legislative process and reduce costs. Nowadays, many methods are presented dynamically for generating keywords. The present study attempted to extract more meaningful keywords from texts by using the thesaurus, which has a structured system to improve the  classification of legislative documents. In this method, the semantic relationships in the thesaurus and document clustering were used and the statistical features of different words were calculated to extract some words as keywords. After pre-processing the texts, first the keywords in the text were selected using statisti-cal methods. Then, the phrases derived from the keywords were extracted using semantic terms in the thesaurus. After that, a numerical weight was assigned to each word to determine the relative importance of the words and indicate the effect of the word in relation to the text and compared to other words. Finally, the final keywords were selected using the relationships in the thesaurus and clustering methods. In order to evaluate this method, the tested text was compared to educational texts and the similarity between them was used. The results of testing various texts and subjects indicated the high accuracy of the proposed method. The data from the Parliament of Iran and the Deputy for Presidential Laws were used to evaluate the proposed model. This model could provide a very high accuracy and performance in these two bases in comparison to other methods.