Efficient Soft-Constrained Clustering for Group-Based Labeling

Gaussian Mixture Density Modelling and Decomposition (GMDD) is a hierarchical clustering method based on robust statistical theory. Firstly, GMDD is assumed with a mixture group of Gaussian distribution in feature space, then by optimization algorithm the feature which mostly accord with the assumed distribution is hierarchically extracted from space until all of the features in the space are decomposed to a group of featuring pattern. Compared with conventional statistical clustering methods, GMDD's main outstanding superorities are:(1)Initial number of features does not need to be specified a priori; (2)The proportion of noisy data in the mixture can be large; (3)The parameters estimation of each feature is virtually initial independent; and (4)The variability in the shape and size of the feature densities in the mixture is taken into account. The article presented the model named the GMDD based remote sensing image feature estimation model (GIFEM) , and the model of GA space searching optimization is also presented out.

A new online clustering method based on not only reinforcement and competitive learning but also pursuit algorithm (Pursuit Reinforcement Competitive Learning: PRCL) as well as learning automata is proposed for reaching a relatively stable clustering solution in comparatively short time duration. UCI repository data which are widely used for evaluation of clustering performance in usual is used for a comparative study among the existing conventional online clustering methods of Reinforcement Guided Competitive Learning: RGCL, Sustained RGCL: SRGCL, Vector Quantization, and the proposed PRCL. The results show that the clustering accuracy of the proposed method is superior to the conventional methods. More importantly, it is found that the proposed PRCL is much faster than the conventional methods. The proposed method is then applied to the evacuation simulation study. It is found that the proposed method is much faster than the conventional method of vector quatization to find the most appropriate evacuation route. Due to the fact that the proposed PRCL method allows finding the most appropriate evacuation route, collisions among peoples who have to evacuate for the proposed method is much less than that of vector quatization.

Recently, kernel-based clustering in feature space has shown to perform better than conventional clustering methods in unsupervised classification. In this paper, a partitioning clustering method i...

This paper addresses the problem of identifying groups that satisfy the specific conditions for the means of feature variables. In this study, we refer to the identified groups as “target clusters” (TCs). To identify TCs, we propose a method based on the normal mixture model (NMM) restricted by a linear combination of means. We provide an expectation–maximization (EM) algorithm to fit the restricted NMM by using the maximum-likelihood method. The convergence property of the EM algorithm and a reasonable set of initial estimates are presented. We demonstrate the method's usefulness and validity through a simulation study and two well-known data sets. The proposed method provides several types of useful clusters, which would be difficult to achieve with conventional clustering or exploratory data analysis methods based on the ordinary NMM. A simple comparison with another target clustering approach shows that the proposed method is promising in the identification.

Wind power forecasting is significant for the security and economy of the power system. For the better robustness and accuracy, various clustering-based forecasting methods have been proposed. However, most of them utilize clustering on a daily sample basis usually called day clustering, which mainly exploits daily features of numerical weather prediction (NWP) and meets some limitations. In this paper, a single point clustering method is proposed to enhance the performance of day-ahead forecasting, which is conducted on a point sample basis through K-means algorithm. First, the extraction of clustering features is optimized. Being different from the commonly used clustering features, the high-order features extracted from NWP is used for the proposed method based on stacked denoising autoencoder (SDAE). Then, both the compactness within the single group and the separation among the multiple groups are enhanced, which is realized through choosing the number of clusters to maximize silhouette. In the case study, the proposed method is compared to the conventional day clustering method with two types of features and the single point clustering with the commonly used clustering features of NWP. Through the results, it is shown that the proposed method performs better than the other comparative methods.

Subarray partition is indispensable in large phased array radar system for reducing the manufacturing cost as well as realizing the system potentiality. The optimization of subarray partition for large planar phased array radar according to weighted K-means clustering method is mainly investigated in this paper. Based on the excitation matching technique, the optimization of subarray partition in monopulse application can be reformulated as a clustering of reference gain ratios to minimize the excitation matching error. However, when the element weights are non-uniform for specific intentions such as low sidelobes, the matching error could not be minimized completely by traditional K-means clustering. Therefore, in this paper, a weighted K-means clustering method is proposed to reduce the matching error by modifying the membership rule and cluster center of K-means clustering. The proposed method can provide smaller matching error compared with conventional clustering methods, especially when the elements are weighted non-uniformly. The effectiveness of proposed method is validated by numerical simulations and compared with several classical clustering methods.

Single-cell RNA sequencing (scRNAseq) makes it possible to analyze gene expression profiles at the individual cell scale and to discover intrinsic and extrinsic cellular processes in biological research. Cell clustering is one of the most important steps in analyzing scRNAseq data. With rapid developments of single cell sequencing technologies, scRNAseq data grow in size and heterogeneity. However, traditional clustering methods like Kmeans with or without dimension reduction methods, cannot handle high sparse and massive scRNAseq data. Although some deep learning based methods have been proposed to denoise the data and cluster cells simultaneously, learning informative representations of cells for accurate cell clustering is still a challenging problem to be solved. In this work, we propose a deep learning model that combines a deep graph convolutional network (GCN) and a self-supervised mechanism. The GCN considers not only the gene expressions but also the relationship between cells to represent cells. The self-supervised mechanism is employed to provide the clustering assignments of cells. Moreover, we utilize the negative log-likelihood of the negative binomial (NB) function as loss in the data reconstruction due to the assumption that genes expression values can be represented by the NB model. We compared the performance of our proposed method with those of the existing clustering methods for scRNAseq data and conventional clustering methods. Results show that our method achieves better performance in terms of accuracy, adjusted random index (ARI), and normalized mutual information (NMI).

Abstract. Accurate boundary layer structure and height are critical in the analysis of the features of air pollutants and local circulation. Although surface-based remote sensing instruments provide a high temporal resolution of the boundary layer structure, there are numerous uncertainties in terms of the accurate determination of the atmospheric boundary layer heights (ABLHs). In this study, an algorithm for an integrated system for ABLH estimation (ISABLE) was developed and applied to the vertical profile data obtained using a ceilometer and a microwave radiometer in Seoul city, Korea. A maximum of 19 ABLHs were estimated via the conventional time-variance, gradient, wavelet, and clustering methods using the backscatter coefficient from the ceilometer. Meanwhile, several stable boundary layer heights were extracted through near-surface inversion and environmental lapse rate methods using the potential temperature from the microwave radiometer. The ISABLE algorithm can find an optimal ABLH from post-processing, such as k-means clustering and density-based spatial clustering of applications with noise (DBSCAN) techniques. It was found that the ABLH determined using ISABLE exhibited more significant correlation coefficients and smaller mean bias and root mean square error between the radiosonde-derived ABLHs than those obtained using the most conventional methods. Clear skies exhibited higher daytime ABLH than cloudy skies, and the daily maximum ABLH was recorded in summer because of the more intense radiation. The ABLHs estimated by ISABLE are expected to contribute to the parameterization of vertical diffusion in the atmospheric boundary layer.

A Moving Shape-based Robust Fuzzy K-modes Clustering Algorithm for Electricity Profiles

Detecting biological associations between genes based on the theory of phase synchronization

BackgroundThe combination of systematic evolution of ligands by exponential enrichment (SELEX) and deep sequencing is termed high-throughput (HT)-SELEX, which enables searching aptamer candidates from a massive amount of oligonucleotide sequences. A clustering method is an important procedure to identify sequence groups including aptamer candidates for evaluation with experimental analysis. In general, aptamer includes a specific target binding region, which is necessary for binding to the target molecules. The length of the target binding region varies depending on the target molecules and/or binding styles. Currently available clustering methods for HT-SELEX only estimate clusters based on the similarity of full-length sequences or limited length of motifs as target binding regions. Hence, a clustering method considering the target binding region with different lengths is required. Moreover, to handle such huge data and to save sequencing cost, a clustering method with fast calculation from a single round of HT-SELEX data, not multiple rounds, is also preferred.ResultsWe developed fast string-based clustering (FSBC) for HT-SELEX data. FSBC was designed to estimate clusters by searching various lengths of over-represented strings as target binding regions. FSBC was also designed for fast calculation with search space reduction from a single round, typically the final round, of HT-SELEX data considering imbalanced nucleobases of the aptamer selection process. The calculation time and clustering accuracy of FSBC were compared with those of four conventional clustering methods, FASTAptamer, AptaCluster, APTANI, and AptaTRACE, using HT-SELEX data (>15 million oligonucleotide sequences). FSBC, AptaCluster, and AptaTRACE could complete the clustering for all sequence data, and FSBC and AptaTRACE performed higher clustering accuracy. FSBC showed the highest clustering accuracy and had the second fastest calculation speed among all methods compared.ConclusionFSBC is applicable to a large HT-SELEX dataset, which can facilitate the accurate identification of groups including aptamer candidates.Availability of data and materialsFSBC is available at http://www.aoki.ecei.tohoku.ac.jp/fsbc/.

The algorithm of medical image is an important part of special field image clustering. There are many problems of technical aspects and the problem of specific area, so that the study of this direction is very challenging. The existing algorithm of clustering has requirement about shape and density of data object, and it cannot get a good result to the application of medical image clustering. In view of the above problem and under the guidance of knowledge of medical image, at first, detects texture from image, and T-LBP method is put forward. Then divides the preprocessed image into many spaces, and calculates LBP value of spaces. At last build spatial sequence LBP histogram. Based on the LBP histogram, the clustering method of MCST is proposed. The result of experiment shows that there are good result at time complexity and clustering result in the algorithm of this paper.

Recently, the differential transcriptional responses of Mycobacterium tuberculosis to drug and growth-inhibitory conditions were monitored to generate a data set of 436 microarray profiles. These profiles were valuably used for grouping drugs, identifying drug targets and detecting related pathways, based on various conventional methods; such as Pearson correlation, hierarchical clustering, and statistical tests. These conventional clustering methods used the high dimensionality of gene space to reveal drug groups basing on the similarity of expression levels of all genes. In this study, we applied the factor analysis with these conventional methods for drug clustering, drug target detection and pathway detection. The latent variables or factors of gene expression levels in loading space from factor analysis allowed the hierarchical clustering to discover true drug groups. The t-test method was applied to identify drug targets which most significantly associated with each drug cluster. Then, gene ontology was used to detect pathway associations for each group of drug targets.

ABSTRACTAttempts to understand gene regulation by global transcription factors have largely been limited to expression studies under binary conditions of presence and absence of the transcription factor. Studies addressing genome-wide transcriptional responses to changing transcription factor concentration at high resolution are lacking. Here, we create a data set containing the entire Escherichia coli transcriptome in Luria-Bertani (LB) broth as it responds to 10 different cAMP concentrations spanning the biological range. We use the Hill’s model to accurately summarize individual gene responses into three intuitively understandable parameters, Emax, n, and k, reflecting the sensitivity, nonlinearity, and midpoint of the dynamic range. Our data show that most cAMP-regulated genes have an n of >2, with their k values centered around the wild-type concentration of cAMP. Additionally, cAMP receptor protein (CRP) affinity to a promoter is correlated with Emax but not k, hinting that a high-affinity CRP promoter need not ensure transcriptional activation at lower cAMP concentrations and instead affects the magnitude of the response. Finally, genes belonging to different functional classes are tuned to have different k, n, and Emax values. We demonstrate that phenomenological models are a better alternative for studying gene expression trends than classical clustering methods, with the phenomenological constants providing greater insights into how genes are tuned in a regulatory network.IMPORTANCE Different genes may follow different trends in response to various transcription factor concentrations. In this study, we ask two questions: (i) what are the trends that different genes follow in response to changing transcription factor concentrations and (ii) what methods can be used to extract information from the gene trends so obtained. We demonstrate a method to analyze transcription factor concentration-dependent genome-wide expression data using phenomenological models. Conventional clustering methods and principal-component analysis (PCA) can be used to summarize trends in data but have limited interpretability. The use of phenomenological models greatly enhances the interpretability and thus utility of conventional clustering. Transformation of dose-response data into phenomenological constants opens up avenues to ask and answer many different kinds of question. We show that the phenomenological constants obtained from the model fits can be used to generate insights about network topology and allows integration of other experimental data such as chromatin immunoprecipitation sequencing (ChIP-seq) to understand the system in greater detail.

Spatial clustering has been identified as an important technique in data mining owing to its various applications. In the conventional spatial clustering methods, data points are clustered mainly according to their geographic attributes. In real applications, however, the obtained data points consist of not only geographic attributes but also non-geographic ones. In general, geographic attributes indicate the data locations and non-geographic attributes show the characteristics of data points. It is thus infeasible, by using conventional spatial clustering methods, to partition the geographic space such that similar data points are grouped together. In this paper, we propose an effective and efficient algorithm, named incremental clustering toward the Bound INformation of Geography and Optimization spaces, abbreviated as BINGO, to solve the problem. The proposed BINGO algorithm combines the information in both geographic and non-geographic attributes by constructing a summary structure and possesses incremental clustering capability by appropriately adjusting this structure. Furthermore, most parameters in algorithm BINGO are determined automatically so that it is easy to be applied to applications without resorting to extra knowledge. Experiments on synthetic are performed to validate the effectiveness and the efficiency of algorithm BINGO.