Optimal Cluster Centers Research Articles

K-Means text clustering method based on Decision Grey Wolf Optimization

Aiming at the problem that the traditional algorithm is easy to fall into local optimum in the process of text clustering, which leads to inaccurate text clustering results, a text clustering method based on decision grey wolf optimization K-Means is proposed to cluster the text data set and the standard UCI data set respectively. Afterword segmentation, stop words removal, feature extraction, and text vectorization of text data, the powerful optimization ability of the Decision Gray Wolf Optimization (DGWO) algorithm is used for global optimization, and the clustering center in K-Means algorithm is replaced by the location of wolves. The position of the wolf group is updated by iterative optimization to obtain the optimal clustering center, to perform text clustering. The experimental results show that compared with the traditional method, the precision, recall, and F-Measure of the text data clustering are improved by 49.22%, 51.15%, and 48.98% respectively. The precision, recall, and F-Measure of UCI data clustering are increased by 23.92%, 25.40%, and 24.70% respectively, and the text clustering results are more reliable.

A novel BiGRU multi-step wind power forecasting approach based on multi-label integration random forest feature selection and neural network clustering

Accurate wind power forecasting helps to carry out effective scheduling and scientific management of wind power, and improve the security and reliability of the power grid. However, the intermittency, volatility and instability of wind energy make wind power forecasting challenging. Therefore, in order to improve the accuracy and stability of wind power forecasting, this paper proposes a bidirectional gated recurrent unit (BiGRU) multi-step wind power forecasting approach based on multi-label integration random forest (MLRF) feature selection and neural network clustering (NNClustering). The proposed MLRF method extends the applicability of random forest through multiple criteria and enables feature selection for multi-step forecasting tasks of multi-factor time series to obtain optimal input features and time steps, which reduces the computational cost and improves the generalization ability of the model. The proposed NNClustering method establishes a novel convolution-based clustering structure and adjusts the parameters by gradient descent method to obtain the optimal clustering centers, and the robust data applicability of the method is validated in multiple seasonal experiments. The WOA-BiGRU forecasting model is constructed separately for each cluster, which reduces the modeling difficulty and better extracts the characteristics. The BiGRU model extracts more efficient characteristics by processing sequences in both directions and the important parameters of BiGRU are optimized by the whale optimization algorithm (WOA) to obtain the optimal forecasting model. Experimental results over multiple seasons show that the proposed hybrid approach has good forecasting performance and robustness, which provides a novel and efficient solution for wind power forecasting.

Multi-kernel subspace stable clustering with exact rank constraints

The identification of cancer subtypes plays a critical role in the early diagnosis of cancer and the delivery of appropriate treatment. Clustering based on comprehensive multi-omic data often yields superior results than clustering using a single data type since each omic type may contain complementary information. However, existing clustering methods cannot fully explore the internal structural information within multi-omic data and can lead to unstable clustering results. This paper proposes a novel multi-omic clustering framework, Multi-Kernel Subspace Stable Clustering with Exact Rank Constraints (MKSSC-ERC), which can effectively explore the correlation and complementary information between different omics types. Specifically, we design a kernel selection criterion that considers both accuracy and diversity and each omic data gets a base kernel. Then, we extract a consistent affinity matrix for these base kernels using subspace segmentation with exact rank constraints, which can refine the extracted shared structures and make them robust to sample noises. In addition, we devise a strategy for selecting the optimal initial cluster centers, which significantly enhances the stability of clustering results. Simulation studies on benchmark multi-omic datasets illustrate substantial gains over existing state-of-the-art methods in survival analysis. MKSSC-ERC is available at https://github.com/xuzihan66/MKSSC-ERC.

Online English Resource Integration Algorithm based on high-dimensional Mixed Attribute Data Mining

To improve the scalability of resources and ensure the effective sharing and utilization of online English resources, an online English resource integration algorithm based on high-dimensional mixed-attribute data mining is proposed. First, an integration structure based on high-dimensional mixed-attribute data mining is constructed. According to this structure, the characteristics of online English resources are extracted, and historical data mining is carried out in combination with the spatial distribution characteristics of resources. In this way, the spatial mapping function of features is established, and the optimal clustering center is designed according to the clustering and fusion structure of online English resources. At this node, the clustering and fusion of online English resources are carried out. According to the fusion results, the distribution structure model of online English resources is constructed, and the optimization research of the integration algorithm of online English resources is carried out. The experimental results show that the integration optimization efficiency of the proposed algorithm is 89%, and the packet loss rate is 0.19%. It has good integration performance, and can realize the integration of multi-channel and various forms of online English resources.

The Application of SPOC Teaching Model in Japanese Language Teaching

Abstract Currently, Japanese language courses in some colleges and universities are still based on didactic teaching, with a lack of interaction between teachers and students, which makes it challenging to meet the needs of students’ communicative ability development. This paper constructs a blended teaching model of the Japanese language in colleges and universities based on the SPOC teaching platform, selects two Japanese language parallel classes in the second year of a university in H city as the research object, and designs the relevant teaching process. Based on the students’ learning data in the SPOC teaching platform, the learners’ behavioral characteristics are analyzed. The K-Means algorithm is improved to obtain the optimal clustering center by using the density weights, and the learning characteristics are clustered. Then, the effectiveness of the SPOC teaching model was responded to by the teaching effect of students’ Japanese speaking and reading skills. In addition, the correlation between learner behavioral characteristics was analyzed using Pearson’s correlation coefficient. It was found that SPOC-based Japanese language teaching increased students’ willingness to learn independently by 5% to 31%, and students’ Japanese reading posttest scores in the experimental group increased by 5.123 points compared with the control group, and online learning behaviors showed a high degree of correlation with learning achievements. The application of the SPOC teaching mode in Japanese language teaching can effectively enhance students’ learning engagement, promote the enhancement of students’ intercultural communication skills, and provide a new path to improve the quality of Japanese language teaching.

Dual Construction of “Composition” and “Analysis” in Composition and Technical Theory in the Context of Big Data

Abstract Based on big data technology, this paper firstly collects and pre-processes educational resources and builds a dual teaching system based on two basic concepts of “creation” and “analysis”, combining technology and theory. Secondly, the FCM clustering algorithm is used to cluster the student learning data and get the optimal clustering center, and the Lagrange multiplier method is used to optimize the objective function, identify the student learning characteristics, and conduct a dynamic pre-testing assessment. Finally, the quantum optimization algorithm dynamically assigns the assessment data to eliminate redundant data and construct a feature set library for individual student learning. The practical analysis and application results show that the effect feedback in the two courses of composition and orchestration reached 88% and 93%. In the music analysis and music composition assessments, the percentage of the number of students with performance range values of 100-90 averaged 43% and 42.75%, respectively. In addition, 72.97% of the students found the system helpful in independent learning ability. It indicates that the teaching system constructed in this paper can better serve the teaching of composition technique and further improve the teaching model and process.

Dominant Partitioning of Discontinuities of Rock Masses Based on DBSCAN Algorithm

In the analysis of rock slope stability and rock mass hydraulics, the dominant partitioning of discontinuities of rock masses is a very important concept, and it is still a key for establishing the three-dimensional (3-D) network model of random discontinuities. The traditional graphical analysis method is inadequate and greatly influenced by subjective experience. A new method using density-based spatial clustering of applications with noise (DBSCAN) algorithm is proposed for the dominant partitioning of discontinuities of rock mass. In the proposed method, we do not need to determine the centers of every cluster before clustering, and the acnodes or outliers can be eliminated effectively after clustering. Firstly, the spatial coordinate transformation of the discontinuity occurrence is carried out and the objective function is established by using the sine value of the angle of the unit normal vector as the similarity measure standard. The DBSCAN algorithm is used to establish the optimal clustering centers by searching the global optimal solution of the objective function, and the fuzzy C-means clustering algorithm is optimized and the mathematical model of the advantage grouping of rock discontinuities is established. The new method and the fuzzy C-means method are compared and verified by using the artificially randomly generated discontinuity occurrence data. The proposed method is a better method than the fuzzy C-means method in general cases, and it can provide more accurate results by eliminating the acnodes or outliers. Finally, the proposed method is applied to discontinuity orientation partition data at Maji dam site, Nujiang River, and there is good agreement with the in situ measurement.

Optimasi Fuzzy C-Means Menggunakan Particle Swarm Optimization Untuk Pengelompokan Kabupaten/Kota Di Pulau Kalimantan (Studi Kasus : Data Indikator Kesejahteraan Rakyat Tahun 2020)

Fuzzy C-Means (FCM) is a method of grouping data based on the degree of membership whose observation object is based on the information found in the data describing the object. The FCM method has weaknesses in the initial cluster center determination, so it can be overcome by the Particle Swarm Optimization (PSO) method that can be applied to find the optimal solution of the optimal cluster center determination. The purpose of this research is to determine the optimal number of clusters based on the validity indexes of Partition Coefficient (PC) and Modified Partition Coefficient (MPC), and obtain the results of grouping regencies/cities using the FCMPSO method. Based on the FCMPSO method with a validity index of PC and MPC, it produces an optimal cluster of two clusters, the first cluster consisting of 33 regencies/cities on Kalimantan Island and the second cluster consisting of 23 regencies/cities on Kalimantan Island.

Improved interval type-2 fuzzy K-means clustering based on adaptive iterative center with new defuzzification method

Interval Type-2 Fuzzy K-means (IT2FKM) is an efficient clustering algorithm, which mainly focuses on further describes the uncertainty of the data sets by introducing interval type-2 fuzzy membership function. Since the estimation center obtained by two membership functions is an interval type-1 fuzzy set, it is necessary to use the defuzzification method to obtain a crisp cluster center. The existing IT2FKM algorithms adopt the generalized centroid method (GC) in the process of type-reduction and defuzzification for iterative clustering center. But unfortunately, this method simply uses a single numerical value 0.5 as the type-reduction coefficient to balance and synthesize for the left value and the right value of an interval cluster center, while ignoring the distribution of the clusters. This defect makes IT2FKM fail to update the optimal cluster center when facing with imbalanced clusters, which affect the accuracy in some cases. To address this failing, in this paper, some adverse effects caused by generalized centroid method when facing with imbalanced clusters are analyzed in detail. Thereafter, an improved type-reduction method of iterative cluster centers for IT2FKM clustering is proposed considering the imbalanced clustering distributions. On the basis of which, an enhanced IT2FKM algorithm is further developed. The enhanced IT2FKM algorithm has an average classification accuracy of 97.5% when faced with four datasets with different degrees imbalance, which is about 3.25% higher than other clustering algorithms. The effectiveness and superiority of the proposed algorithm are further demonstrated by some experimental analysis of UCI standard datasets.

Multi-index dominant grouping of rock mass discontinuities based on the combined weighting method: A case study for the Huayang tunnel

The properties of discontinuities are determined by their occurrence, trace length, aperture, roughness, and the degree of particle fill; therefore, a multi-index dominant grouping method for rock mass discontinuities based on the combined weighting method was proposed. Firstly, the combined weighting method was used to calculate the weighted value of each index participating in the similarity measurement, and a reasonable elimination of outlier discontinuities is achieved based on the principle of possibility clustering. The adaptive chaotic annealing particle swarm optimization algorithm was then used to search the global optimal initial clustering centers of the fuzzy C-means clustering algorithm. The objective function was established by minimizing the weighted sum of the distance between the discontinuities to be grouped, and the clustering centers to complete the multi-index dominant grouping of rock mass discontinuities. 230 artificial discontinuities with outliers were grouped and compared with other methods, which indicated that this method has higher grouping accuracy. The proposed method was applied to group the measured discontinuities of the Huayang tunnel in Chongqing City, China. Based on the grouping results, a discrete element model for tunnel excavation was established, and the construction was guided according to the calculated results from the model. The engineering applicability of the method was further verified.

Image Segmentation Method Based on Improved PSO Optimized FCM Algorithm and Its Application

<p>In image segmentation, FCM clustering algorithm can not find the optimal initial clustering center and fall into local extremum, which leads to the decrease of image segmentation accuracy. The PSO algorithm has strong optimization ability, so a new method based on improved PSO algorithm is proposed to optimize the FCM clustering center selection. Firstly, the optimization performance of the PSO algorithm is improved. The distance difference between each particle and the optimal particle is calculated, and the maximum distance difference is selected. The ratio of the distance difference to the maximum distance difference and the aggregation degree of particles are used to construct the natural exponential function. This natural exponential function is used to improve the calculation method of inertia weight value of PSO algorithm, so that the farther the particle is away from the optimal position, the larger the inertia weight value it will get, the stronger the global search ability of particle; on the contrary, the smaller the inertia weight value, the stronger the local search ability of particle, so as to improve the optimization ability of PSO algorithm. The improved PSO algorithm is called DDPSO (Distance Difference PSO). Then the optimized FCM algorithm is applied to the segmentation of standard image and eggshell damaged image to improve the accuracy of image segmentation. Finally, the experimental results show that the FCM algorithm optimized by DDPSO has higher segmentation accuracy than the traditional method.</p> <p> </p>

Sparse representation and modified density peak clustering-based state identification for multimode processes

Industrial processes with high-dimensional data are generally operated with mixed normal/faulty states in different modes, making it difficult to automatically and accurately identify the faults. In this paper, a state identification framework is proposed for multimode processes. First, a key variable selection approach is presented based on sparse representation to eliminate redundant variables. Then, modified density peak clustering is proposed to identify different states, in which a distance measurement with a time factor is constructed to select all the possible cluster centers. Then, the sum of squared errors-based approach is developed to determine the optimal cluster centers automatically. Further, considering that the mode attributes may be mixed with the fault attributes, a two-step ‘coarse-to-fine identification’ strategy is designed to precisely identify the modes and the faults in each mode. Finally, three cases including a numerical simulation, Tennessee Eastman benchmark process and an actual semiconductor manufacturing process are presented to show the feasibility of the proposed method.

Effective User Clustering and Power Control for Multiantenna Uplink NOMA Transmission

This paper investigates the user clustering and power control in the uplink multiple-input single-output non-orthogonal multiple access (MISO-NOMA) networks. A joint optimization problem is formulated to minimize the system transmit power. The formulated optimization problem is prohibitively complicated, especially when the number of users is large. Alternatively, a two-step user clustering and power control algorithm is proposed. First, a K-means-based algorithm is proposed for user clustering, where both channel gain and channel correlation among users are taken into account for the distance measurement to reduce the intra- and inter-cluster interference. Then, a semi-orthogonal user selection (SUS) algorithm is designed, with which the optimal cluster number and cluster centers can be dynamically obtained. Further, the closed-form expression of the optimal intra-cluster power control is derived, and the resulting inter-cluster power control problem is solved by designing an efficient iterative algorithm. Simulation results show that the proposed K-means-based iterative power control scheme outperforms other reference methods, and can approach the optimal performance in terms of power consumption and energy efficiency at a much lower computational complexity.

Computer-aided diagnosis for early cancer detection using Adaptive Kernel Based Fuzzy Cuckoo Search Optimization Clustering from mammogram images

Early detection of breast cancer is major issue that must be addressed globally. Segmentation process play key role in detection. We propose a new hybrid approach called Adaptive Kernel Based Fuzzy Cuckoo Search Optimization Clustering to segment mammogram images. It is diversified into two stages. A kernel-based fuzzy c-means algorithm with cuckoo search optimization is used to obtain optimal cluster centers and obtained optimum centers act as centers to the KFCM clustering process to get different segmented regions in pre processing stage. In post processing stage, a level set method is employed to minimize boundary leakages. The simulation was performed on RIDER data set. The proposed approach is validated by calculating similarity indexing parameters. The simulation was performed on 12 distinct images from the RIDER mammography database. The suggested method's experimental results were compared to existing level set approaches such as IVC2010, IVC2013, and ESA2021.

An improved Fuzzy based GWO algorithm for predicting the potential host receptor of COVID-19 infection

Coronavirus disease (COVID-19) is caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and has infected millions worldwide. SARS-CoV-2 spike protein uses Angiotensin-converting enzyme 2 (ACE2) and Transmembrane serine protease 2 (TMPRSS2) for entering and fusing the host cell membrane. However, interaction with spike protein receptors and protease processing are not the only factors determining coronaviruses’ entry. Several proteases mediate the entry of SARS-CoV-2 virus into the host cell. Identifying receptor factors helps understand tropism, transmission, and pathogenesis of COVID-19 infection in humans. The paper aims to identify novel viral receptor or membrane proteins that are transcriptionally and biologically similar to ACE2 and TMPRSS2 through a fuzzy clustering technique that employs the Grey wolf optimizer (GWO) algorithm for finding the optimal cluster center. The exploratory and exploitation capability of GWO algorithm is improved by hybridizing mutation and crossover operators of the evolutionary algorithm. Also, the genetic diversity of the grey wolf population is enhanced by eliminating weak individuals from the population. The proposed clustering algorithm’s effectiveness is shown by detecting novel viral receptors and membrane proteins associated with the pathogenesis of SARS-CoV-2 infection. The expression profiles of ACE2 protein and its co-receptor factor are analyzed and compared with single-cell transcriptomics profiling using the Seurat R toolkit, mass spectrometry (MS), and immunohistochemistry (IHC). Our advanced clustering method infers that cell that expresses high ACE2 level are more affected by SARS-CoV-infection. So, SARS-CoV-2 virus affects lung, intestine, testis, heart, kidney, and liver more severely than brain, bone marrow, skin, spleen, etc.We have identified 58 novel viral receptors and 816 membrane proteins, and their role in the pathogenicity mechanism of SARS-CoV-2 infection has been studied. Besides, our study confirmed that Neuropilins (NRP1), G protein-coupled receptor 78 (GPR78), C-type lectin domain family 4 member M (CLEC4M), Kringle containing transmembrane protein 1 (KREMEN1), Asialoglycoprotein receptor 1 (ASGR1), A Disintegrin and metalloprotease 17 (ADAM17), Furin, Neuregulin-1,(NRG1), Basigin or CD147 and Poliovirus receptor (PVR) are the potential co-receptors of SARS-CoV-2 virus. A significant finding is that heparin derivative glycosaminoglycans could block the replication of SARS-CoV-2 virus inside the host cytoplasm. The membrane protein N-Deacetylase/N-Sulfotransferase-2 (NDST2), Extostosin protein (EXT1, EXT2, and EXT3), Glucuronic acid epimerase (GLCE), and Xylosyltransferase I, II (XYLT1, XYLT2) could act as the therapeutic target for inhibiting the spread of SARS-CoV-2 infection. Drugs such as carboplatin and gemcitabine are effective in such situations.

The Clustering Rice Plant Diseases Using Fuzzy C-Means and Genetic Algorithm

Rice is an agricultural sector that is very important for Indonesia's economy. The main problem with rice plants is pest and disease control which has a very dangerous impact as well as economic losses for farmers. The characteristics that are very visible on rice leaves have a greater area than other plant structures, rice leaves can be applied for early diagnosis of rice plant diseases. Fuzzy C-Means (FCM) and Genetic Algorithm-Fuzzy C-Means are the approaches employed (GA-FCM). The center of the cluster is obtained while adopting genetic algorithms for optimization. The primary dataset used in this research is Teaching Sawah Farm IPB, and the secondary dataset is UCI Rice Leaf Diseases. According to the results of the comparison the GA-FCM optimization results in a higher level of clustering precision with a 65% optimal cluster center point on the silhoutte coefficient value compared to just 60% for FCM. This research shows the results that the proposed method can add 5% accuracy to the clustering results in terms of identifying the types of rice plant diseases properly.

Subspace Decomposition Based Adaptive Density Peak Clustering for Radar Signals Sorting

Radar signal sorting (RSS) plays an important role in the electronic support measurement system. However, the existing clustering-based RSS methods depend heavily on prior knowledge to achieve excellent performance, which may bring severe challenges to RSS in actual scenarios. This letter proposes a novel subspace decomposition based adaptive density peak clustering (SD-ADPC) method to address the problems of low accuracy and high computational cost in RSS. First, the original complex radar signal data is directly decomposed into two-dimensional (2D) subspace by t-distributed random neighborhood embedding (t-SNE). Then, based on the outlier detection of the products between the peak density and the distance of the data points, ADPC is used to adaptively determine the optimal clustering centers of the original data in 2D subspace. Finally, the reminding data is assigned to its nearest cluster with Euclidean distance in one step. The experimental results of the simulated RSS dataset and the open baselines show that our proposed method does not require any knowledge and can achieve better or competitive performance in terms of accuracy and computational cost, compared to existing state-of-the-art methods.

A novel quantum inspired genetic algorithm to initialize cluster centers in fuzzy C-means

We propose a new quantum-inspired genetic algorithm (QGA) aimed to find the optimal initial cluster centers of the Fuzzy C-Means (FCM). The quantum chromosomes of the quantum population are collapsed to conventional chromosomes formed by binary strings encoding the cluster centers. We set as fitness function to minimize a fuzzy entropy function used to measure the fuzziness of clustering. A quantum rotation gate operator is applied in any generation to evolve the quantum population. The best conventional chromosome is used to set the initial cluster centers running FCM. We execute comparative tests of our QGA with other quantum-inspired method proposed in literature in order to find the best initial cluster centers: the results show that our method provides the best trade-off between the accuracy and precision of results and execution times.

A Modified Cuckoo Search Algorithm for Data Clustering

Clustering of data is one of the necessary data mining techniques, where similar objects are grouped in the same cluster. In recent years, many nature-inspired based clustering techniques have been proposed, which have led to some encouraging results. This paper proposes a Modified Cuckoo Search (MoCS) algorithm. In this proposed work, an attempt has been made to balance the exploration of the Cuckoo Search (CS) algorithm and to increase the potential of the exploration to avoid premature convergence. This algorithm is tested using fifteen benchmark test functions and is proved as an efficient algorithm in comparison to the CS algorithm. Further, this method is compared with well-known nature-inspired algorithms such as Ant Colony Optimization (ACO), Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO), Particle Swarm Optimization with Age Group topology (PSOAG) and CS algorithm for clustering of data using six real datasets. The experimental results indicate that the MoCS algorithm achieves better results as compared to other algorithms in finding optimal cluster centers.

Random Fuzzy Granular Decision Tree

In this study, the classification problem is solved from the view of granular computing. That is, the classification problem is equivalently transformed into the fuzzy granular space to solve. Most classification algorithms are only adopted to handle numerical data; random fuzzy granular decision tree (RFGDT) can handle not only numerical data but also nonnumerical data like information granules. Measures can be taken in four ways as follows. First, an adaptive global random clustering (AGRC) algorithm is proposed, which can adaptively find the optimal cluster centers and maximize the ratio of interclass standard deviation to intraclass standard deviation, and avoid falling into local optimal solution; second, on the basis of AGRC, a parallel model is designed for fuzzy granulation of data to construct granular space, which can greatly enhance the efficiency compared with serial granulation of data; third, in the fuzzy granular space, we design RFGDT to classify the fuzzy granules, which can select important features as tree nodes based on information gain ratio and avoid the problem of overfitting based on the pruning algorithm proposed. Finally, we employ the dataset from UC Irvine Machine Learning Repository for verification. Theory and experimental results prove that RFGDT has high efficiency and accuracy and is robust in solving classification problems.