Cluster Information Research Articles

Joint Slot Scheduling and Power Allocation in Clustered Underwater Acoustic Sensor Networks

Dear Editor This letter deals with the joint slot scheduling and power allocation in clustered underwater acoustic sensor networks (UASNs), based on the known clustering and routing information, to maximize the network's energy efficiency (EE). Based on the block coordinated decent (BCD) method, the formulated mixed-integer non-convex problem is alternatively optimized by leveraging the Kuhn-Munkres algorithm, the Dinkelbach's method and the successive convex approximation (SCA) technique. Numerical results show that the proposed scheme has a better performance in maximizing EE compared to the separate optimization methods.

Wafer map defect pattern detection method based on improved attention mechanism

Wafer map defect pattern recognition is an indispensable part in the semiconductor manufacturing process. Wafer map defect pattern recognition and classification can locate the cause of failures in the manufacturing process. In this study, we propose a deep convolutional neural network architecture incorporating improved convolutional block attention module (I-CBAM) for pattern recognition and classification of single-type wafer map defects. The proposed method consists of three steps: First, a multi-branch attention module is designed, which focuses more on the main information of defect clusters, extracts richer feature information, and improves feature extraction capabilities. Secondly, I-CBAM is integrated into the ResneXt50 backbone network to extract wafer map defect features. Finally, the extracted features are fed to a support vector machine combined with an error-correcting output code to classify the wafer map defect pattern. The experimental results on the real-world datasets WM-811K show that the accuracy rate of the I-CBAM-ResneXt50 can reach 96.96%. Compared with the existing algorithms, the proposed model can achieve better recognition and classification results.

Scalable incomplete multi-view clustering via tensor Schatten p-norm and tensorized bipartite graph

Graph-based incomplete multi-view clustering (IMVC) methods have drawn considerable attention due to their good performance in exploring the nonlinear structure of data. However, they still have the following shortcomings. First, graph construction and eigen decomposition of the Laplacian matrix included in the IMVC methods generally have high computational complexity. Second, most methods do not consider the impact of missing views and neglect the potential relationships between different views. Third, few algorithms consider both intra-view and inter-view information for clustering. Therefore, we innovatively propose a scalable incomplete multi-view clustering via the tensor Schatten p-norm and tensorized bipartite graph (SIMVC/TSTBG) method, which combines tensorized bipartite graph, graph completion, and tensor low-rank constraint into a joint framework. Concretely, we first construct bipartite graphs based on the selected m anchor points and the n data points, reducing the size of the graph from n×n to n×m(m<<n), which considerably reduces the computational complexity. Second, we adaptively complete the missing bipartite graph, which reduces the effect of missing view information on the clustering results. Third, to explore connections between missing views and mine high-order information between views, we splice the bipartite graphs into a tensor and impose a tensor low-rank constraint, i.e., the tensor Schatten p-norm, on it. At the same time, we also design an efficient algorithm to solve SIMVC/TSTBG. To our knowledge, we are the first successful practice to integrate the tensor technique with the scalable IMVC method. Compared with other IMVC methods, the results on seven datasets fully show the high efficiency and effectiveness of SIMVC/TSTBG.

Toward Noise-Resistant Graph Embedding With Subspace Clustering Information.

Most existing approaches of attributed network embedding often combine topology and attribute information based on the homophily assumption. In many real-world networks, such an assumption does not hold since the nodes are usually associated with many noisy or irrelevant attributes. To tackle this issue, we propose a noise-resistant graph embedding method, called NGE, by leveraging the subspace clustering information (i.e., the formation of communities is driven by different latent features in distinct subspaces). Specifically, we first construct a tensor to represent a given attributed network and then map it into different feature subspaces to capture community structure via tensor decomposition. For structure embedding, the link-level and community-level constraints are imposed. For attribute embedding, the feature-selection constraint is used to reinforce the relationship between topology and noise-removal attributes. By learning structure and attribute embedding with subspace clustering information, NGE can benefit both community detection, link prediction, and node classification. Extensive experimental results have demonstrated the superiority of NGE over many state-of-the-art approaches.

Trust-based distributed state estimation for microgrids with encoding-decoding mechanisms

In this article, a trust-based estimation algorithm in a distributed paradigm is developed for microgrids whose information transmission obeys the predetermined encoding and decoding rule. Because of weak communication protection, and harsh electromagnetic or geographical environments, the output of sensors in microgrids could be subject to outliers or cyberattacks, which will lead to anomalous estimates. According to such a challenge, a simple clustering method is employed to distinguish between trusted and untrusted data. In light of distinguished data, a novel distributed estimator under an encoding and decoding scheme is then proposed where the correction of untrusted data is executed by the cluster information. Furthermore, a performable algorithm is developed by optimizing the upper bound of error covariance and a novel decomposition of system matrices, and a condition is derived to check the boundedness of the developed iterative algorithm. In the end, a simulation example is given to verify the effectiveness.

Consensus and fuzzy partition of dendrograms from a three-way dissimilarity array

In cluster analysis, a problem often addressed is finding a consensus on a set of hierarchical classifications of the same set of objects, named primary hierarchies (dendrograms). A unique consensus of the primary hierarchies, called a secondary hierarchy, is sufficient to synthesize the relevant clustering information only when the primary dendrograms are similar to one another. In contrast, when several differences are observed, more than one consensus secondary hierarchy is required to clearly synthesize the different primary hierarchies.Our methodology, PARoDENo3WD (PARtition of DENdrograms of a 3-Way Data array), aims to obtain a secondary fuzzy partition of the primary hierarchies, where hierarchies belonging to the same class are perceived as similar. Each class is associated with a consensus hierarchy.The fuzzy approach allows each primary hierarchy to contribute to the definition of all classes of the secondary partition, according to different membership degrees. In this way, the ‘clustering uncertainty’ is taken into account. Indeed, in several realistic applications, primary dendrograms belonging to different classes of the secondary partition may share some similar features.The performance of PARoDENo3WD is evaluated by an extended simulation study, generating 1800 three-way datasets. An application of PARoDENo3WD on a real data set is given.

CCR-Net: Consistent contrastive representation network for multi-view clustering

Recently, researchers have focused on utilizing given heterogeneous features to explore obvious discrimination information for clustering. Most of the current work exploits consistency using some fusion metrics, but the complementarity of multi-view features is not well leveraged. In this paper, we propose an efficient consistent contrastive representation network (CCR-Net) for multi-view clustering, which provides a generalized framework for multi-view learning tasks. First, the proposed model explores the complementarity by a designed contrastive fusion module to learn a shared fusion weight. Second, the proposed method utilizes a consistent representation module to ensure consistency and obtains a consistent graph. Furthermore, we also extend the proposed method to incomplete multi-view scenarios. The designed contrastive fusion module utilizes the complementarity of multiple views to fill in the missing view graphs. Moreover, the consistent feature representation module adds a maxpooling layer on CCR-Net to explore a shared local structure and extract a latent low-dimensional embedding. Finally, the proposed method presents end-to-end training and flexible task interfaces for multi-view learning. Comprehensive evaluations on challenging multi-view tasks demonstrate that the proposed method achieves outstanding performance.

KGLRR: A low-rank representation K-means with graph regularization constraint method for Single-cell type identification

Single-cell RNA sequencing technology provides a tremendous opportunity for studying disease mechanisms at the single-cell level. Cell type identification is a key step in the research of disease mechanisms. Many clustering algorithms have been proposed to identify cell types. Most clustering algorithms perform similarity calculation before cell clustering. Because clustering and similarity calculation are independent, a low-rank matrix obtained only by similarity calculation may be unable to fully reveal the patterns in single-cell data. In this study, to capture accurate single-cell clustering information, we propose a novel method based on a low-rank representation model, called KGLRR, that combines the low-rank representation approach with K-means clustering. The cluster centroid is updated as the cell dimension decreases to better from new clusters and improve the quality of clustering information. In addition, the low-rank representation model ignores local geometric information, so the graph regularization constraint is introduced. KGLRR is tested on both simulated and real single-cell datasets to validate the effectiveness of the new method. The experimental results show that KGLRR is more robust and accurate in cell type identification than other advanced algorithms.

Using clustering algorithms and GPM data to identify spatial precipitation patterns over southeastern Brazil

Southeastern Brazil comprises an important geoeconomic and populous region in South America. Consequently, it is essential to analyze and understand the precipitation profiles in this region. Among different data sources and techniques available to perform such study, the use of clustering algorithms and information from the Global Precipitation Measurement (GPM) project rises as a convenient yet few exploited alternative. Precisely, this study employs the K-Means, the Hierarchical Ward, and the Self-Organizing Maps methods to cluster the annual and seasonal precipitation data from GPM project recorded from 2001 to 2019. The adopted methods are compared in terms of quantitative measures and the number of clusters defined through a well-established rule. The results demonstrate that the annual and seasonal periods are organized according to different number of clusters. Moreover, the results allow: identify the presence of a spatially heterogeneous distribution in the study area; to conclude that the K-Means algorithm is a suitable clustering method in the context of this investigation when compared to Ward’s Hierarchical and Self-Organizing Maps methods in terms of the Calinski-Harabasz and Davies-Bouldin measures; and that the spatial precipitation distribution over Southeastern Brazil is represented by 10 clusters in annual and summer periods, 11 clusters in autumn and spring and 9 clusters in winter period.

Performance comparison of three scaling algorithms in NMR-based metabolomics analysis.

Unit variance (UV) scaling, mean centering (CTR) scaling, and Pareto (Par) scaling are three commonly used algorithms in the preprocessing of metabolomics data. Based on our NMR-based metabolomics studies, we found that the clustering identification performances of these three scaling methods were dramatically different as tested by the spectra data of 48 young athletes' urine samples, spleen tissue (from mice), serum (from mice), and cell (from Staphylococcus aureus) samples. Our data suggested that for the extraction of clustering information, UV scaling could serve as a robust approach for NMR metabolomics data for the identification of clustering analysis even with the existence of technical errors. However, for the purpose of discriminative metabolite identification, UV scaling, CTR scaling, and Par scaling could equally extract discriminative metabolites efficiently based on the coefficient values. Based on the data presented in this study, we propose an optimal working pipeline for the selection of scaling algorithms in NMR-based metabolomics analysis, which has the potential to serve as guidance for junior researchers working in the NMR-based metabolomics research field.

Change-detection-assisted multiple testing for spatiotemporal data

This paper considers a large-scale multiple testing problem for spatiotemporal data with multiple change points. A data-driven procedure that aims to fully utilize the clustering information is proposed. Specifically, we first develop a new change-point detection algorithm that integrates the kernel-based aggregation of spatial observations with a global loss function at the temporal level to group data into several sets, and then derive an FDR control scheme for set-wise multiple testing. Under some mild conditions on the spatiotemporal dependence structure, FDR is shown to be strongly controlled. Theoretical analysis and numerical studies demonstrate the advantages of the algorithm over competing methods.

Time Cluster Personalized Ranking Recommender System in Multi-Cloud

Recommender systems have become a vital tool to identify items for users based on personalized preferences. The personalized ranking or item recommendation generates a ranked list of items for the users. Clustering methods offer better scalability than collaborative filtering (CF) methods since they make predictions within small clusters. The major challenges of recommender systems are accuracy and scalability. Traditionally, recommendation systems are based on a centralized framework that restrains quick scalability for enormous data volumes. The emergence of cloud technology resolves this issue as it handles vast data and supports massive processing. This paper proposes a time cluster personalized ranking recommender system (TCPRRS) in a multi-cloud environment. TCPRRS is a five-stage system that generates recommendations based on temporal information of user consumption and clustering with personalized ranking. Particle swarm optimization (PSO) is utilized for optimizing the solution. The efficiency of TCPRRS is estimated using similarity metrics.

RAC-GAN-Based Scenario Generation for Newly Built Wind Farm

Due to the lack of historical output data of new wind farms, there are difficulties in the scheduling and planning of power grid and wind power output scenario generation. The randomness and uncertainty of meteorological factors lead to the results of traditional scenario generation methods not having the ability to accurately reflect their uncertainty. This article proposes a RAC-GAN-based scenario generation method for a new wind farm output. First, the Pearson coefficient is adopted in this method to screen the meteorological factors and obtain the ones that have larger impact on wind power output; Second, based on the obtained meteorological factors, the Grey Relation Analysis (GRA) is used to analyze the meteorological correlation between multiple wind farms with sufficient output data and new wind farms (target power stations), so that the wind farm with high meteorological correlation is selected as the source power station. Then, the K-means method is adopted to cluster the meteorological data of the source power station, thus generating the target power station scenario in which the cluster information serves as the label of the robust auxiliary classifier generative adversarial network (RAC-GAN) model and the output data of the source power station is considered as the basis. Finally, the actual wind farm output and meteorological data of a region in northeast China are employed for arithmetic analysis to verify the effectiveness of the proposed method. It is proved that the proposed method can effectively reflect the characteristics of wind power output and solve the problem of insufficient historical data of new wind farm output.

CTNRL

Network representation learning is one of the important works of analyzing network information. Its purpose is to learn a vector for each node in the network and map it into the vector space, and the resulting number of node dimensions is much smaller than the number of nodes in the network. Most of the current work only considers local features and ignores other features in the network, such as attribute features. Aiming at such problems, this paper proposes novel mechanisms of combining network topology, which models node text information and node clustering information on the basis of network structure and then constrains the learning process of network representation to obtain the optimal network node vector. The method is experimentally verified on three datasets: Citeseer (M10), DBLP (V4), and SDBLP. Experimental results show that the proposed method is better than the algorithm based on network topology and text feature. Good experimental results are obtained, which verifies the feasibility of the algorithm and achieves the expected experimental results.

Graph representation learning via redundancy reduction

In recent years, self-supervised learning methods based on mutual information maximization have achieved remarkable success on graph data tasks. However, most of them heavily rely on a large number of negative samples, which is computationally expensive. These methods also fail to extract the semantic cluster information of the data. To overcome these problems, we propose a novel self-supervised approach called Graph Representation Learing via Redundancy Reduction (GRRR) to learn node representations based on the redundancy-reduction principle. The proposed GRRR preserves as much topological information of the graph as possible, and minimizes the redundancy of representation in terms of node instance and semantic cluster information. Specifically, we first design three graph data augmentation strategies to construct two augmented views. Then, to filter the redundant information in each augmented view, we propose the self-redundancy reduction module which implements the structural reconstruction. Finally, we propose the joint redundancy reduction module to further filter undesirable information via a cross-view approach. It preserves the most essential instance features and semantic cluster information by maximizing the agreement across different views not only on node instance features but also on cluster assignments. Results on several benchmark datasets show that GRRR outperforms state-of-the-art methods in downstream node classification, link prediction, and node clustering tasks.

A comparison of methods for training population optimization in genomic selection

Key messageMaximizing CDmean and Avg_GRM_self were the best criteria for training set optimization. A training set size of 50–55% (targeted) or 65–85% (untargeted) is needed to obtain 95% of the accuracy. With the advent of genomic selection (GS) as a widespread breeding tool, mechanisms to efficiently design an optimal training set for GS models became more relevant, since they allow maximizing the accuracy while minimizing the phenotyping costs. The literature described many training set optimization methods, but there is a lack of a comprehensive comparison among them. This work aimed to provide an extensive benchmark among optimization methods and optimal training set size by testing a wide range of them in seven datasets, six different species, different genetic architectures, population structure, heritabilities, and with several GS models to provide some guidelines about their application in breeding programs. Our results showed that targeted optimization (uses information from the test set) performed better than untargeted (does not use test set data), especially when heritability was low. The mean coefficient of determination was the best targeted method, although it was computationally intensive. Minimizing the average relationship within the training set was the best strategy for untargeted optimization. Regarding the optimal training set size, maximum accuracy was obtained when the training set was the entire candidate set. Nevertheless, a 50–55% of the candidate set was enough to reach 95–100% of the maximum accuracy in the targeted scenario, while we needed a 65–85% for untargeted optimization. Our results also suggested that a diverse training set makes GS robust against population structure, while including clustering information was less effective. The choice of the GS model did not have a significant influence on the prediction accuracies.

Implementation of a Noise Filter for Grouping in Bibliographic Databases using Latent Semantic Indexing

Clustering algorithms can assist in scientific research by presenting themes related to some topics from which we can extract information more easily. However, it is common for many of these clusters to have documents that have no relevance to the topic of interest, thereby reducing the quality of the information. We can manage the reduced quality of information of clusters for a bibliographic database by dealing with noise in the semantic space that represents the relations between the grouped documents. In this work, we sustain the hypothesis of using the Latent Semantic Indexing (LSI) technique as an efficient instrument to reduce noise and promote better group quality. Using a database of 90 scientific publications from different areas, we pre-processed the documents by LSI and grouped them using six clustering algorithms. The results were significantly improved compared to our initial results that did not use LSI-based pre-processing. From the perspective of individual performance of the algorithms demonstrating the best results, CMeans was the one that got the highest average gain, with approximately 25%, followed by K-Means and SKmeans, with 17% each; PAM, with 16.5%; and EM, with 15%. The conclusion is that Latent Semantic Indexing has proven to be a helpful tool for noise reduction. We recommend its use to improve the cluster quality of bibliographic databases significantly.

LRT-CLUSTER: A New Clustering Algorithm Based on Likelihood Ratio Test to Identify Driving Genes.

Somatic mutations often occur at high relapse sites in protein sequences, which indicates that the location clustering of somatic missense mutations can be used to identify driving genes. However, the traditional clustering algorithm has such problems as the background signal over-fitting, the clustering algorithm is not suitable for mutation data, and the performance of identifying low-frequency mutation genes needs to be improved. In this paper, we propose a linear clustering algorithm based on likelihood ratio test knowledge to identify driver genes. In this experiment, firstly, the polynucleotide mutation rate is calculated based on the prior knowledge of likelihood ratio test. Then, the simulation data set is obtained through the background mutation rate model. Finally, the unsupervised peak clustering algorithm is used to, respectively, evaluate the somatic mutation data and the simulation data to identify the driver genes. The experimental results show that our method achieves a better balance of precision and sensitivity. It can also identify the driver genes missed by other methods, making it an effective supplement to other methods. We also discover some potential linkages between genes and between genes and mutation sites, which is of great value to target drug therapy research. Method framework: Our proposed model framework is as follows. a. Counting mutation sites and the number of mutations in tumor gene elements. b. The nucleotide context mutation frequency is counted based on the likelihood ratio test knowledge, and the background mutation rate model is obtained. c. Based on Monte Carlo simulation method, data sets with the same number of mutations as gene elements are randomly sampled to obtain simulated mutation data, and the sampling frequency of each mutation site is related to the mutation rate of polynucleotide. d. The original mutation data and the simulated mutation data after random reconstruction are clustered by peak density, respectively, and the corresponding clustering scores are obtained. e. We can obtain the clustering information statistics in each gene segment and score of each gene segment from the original single nucleotide mutation data through step d. f. According to the observed score and the simulated clustering score, the p-value of the corresponding gene fragment is calculated. g. We can obtain the clustering information statistics in each gene segment and score of each gene segment from the simulated single nucleotide mutation data through step d.

GRACE: A General Graph Convolution Framework for Attributed Graph Clustering

Attributed graph clustering (AGC) is an important problem in graph mining as more and more complex data in real-world have been represented in graphs with attributed nodes. While it is a common practice to leverage both attribute and structure information for improved clustering performance, most existing AGC algorithms consider only a specific type of relations, which hinders their applicability to integrate various complex relations into node attributes for AGC. In this article, we propose GRACE, an extended graph convolution framework for AGC tasks. Our framework provides a general and interpretative solution for clustering many different types of attributed graphs, including undirected, directed, heterogeneous and hyper attributed graphs. By building suitable graph Laplacians for each of the aforementioned graph types, GRACE can seamlessly perform graph convolution on node attributes to fuse all available information for clustering. We conduct extensive experiments on 14 real-world datasets of four different graph types. The experimental results show that GRACE outperforms the state-of-the-art AGC methods on the different graph types in terms of clustering quality, time, and memory usage.

Millimeter Channel Clustering by Self-Organizing Maps With Time-Varying Topological Structure

Cluster-based channel model plays an important role in the fifth-generation (5G) and beyond 5G (B5G) wireless communication systems. A well-designed clustering algorithm can achieve a good compromise between the accuracy and complexity of the wireless channel model. The dependence on prior information of cluster, such as the number of clusters, is a common dilemma of current algorithms. In this article, a self-organizing map with time-varying topological structure (SOM-TVS) clustering algorithm is proposed. First, we combine the locations and weights of the neurons in competition layer and initialize the SOM according to multipath components (MPCs), in which the topology of the SOM is changed with the iterations of the algorithm. Then, a new MPCs’ power weighted Gaussian–Sinc function is proposed to optimize the competitiveness performance of the algorithm. Finally, both the measured and simulated channels in millimeter wave (mmWave) are used to validate the performance of the proposed SOM-TVS algorithm. The SOM-TVS algorithm can perform great effectiveness and robustness based on numerical simulations and experimental validation results. More significantly, the proposed clustering algorithm does not require any prior knowledge of cluster and has a fairly low complexity.