Incremental Clustering in Geography and Optimization Spaces

Incremental clustering or one-pass clustering is very useful when we work with data stream or dynamic data. In each incremental clustering algorithm, two process including insertion and deletion for new data points are used for updating the current clusters. In fact, for traditional clustering such as K-Means, Fuzzy C-Means, DBSCAN, etc., many versions of incremental clustering have been developed. However, to the best of our knowledge, there are no incremental semi-supervised clustering in literature. This paper introduces a new incremental semi-supervised clustering which was based on a graph of k-nearest neighbor using seeds, namely IncrementalSSGC. Experiments conducted on some data sets from UCI and the 802.11 network data set (AWID) show the effectiveness of our new IncrementalSSGC.

A kernel-based subtractive clustering method

Dimension reduction and (spatial) clustering is usually performed sequentially; however, the low-dimensional embeddings estimated in the dimension-reduction step may not be relevant to the class labels inferred in the clustering step. We therefore developed a computation method, Dimension-Reduction Spatial-Clustering (DR-SC), that can simultaneously perform dimension reduction and (spatial) clustering within a unified framework. Joint analysis by DR-SC produces accurate (spatial) clustering results and ensures the effective extraction of biologically informative low-dimensional features. DR-SC is applicable to spatial clustering in spatial transcriptomics that characterizes the spatial organization of the tissue by segregating it into multiple tissue structures. Here, DR-SC relies on a latent hidden Markov random field model to encourage the spatial smoothness of the detected spatial cluster boundaries. Underlying DR-SC is an efficient expectation-maximization algorithm based on an iterative conditional mode. As such, DR-SC is scalable to large sample sizes and can optimize the spatial smoothness parameter in a data-driven manner. With comprehensive simulations and real data applications, we show that DR-SC outperforms existing clustering and spatial clustering methods: it extracts more biologically relevant features than conventional dimension reduction methods, improves clustering performance, and offers improved trajectory inference and visualization for downstream trajectory inference analyses.

Previous chapter Next chapter Full AccessProceedings Proceedings of the 2007 SIAM International Conference on Data Mining (SDM)Incremental Spectral Clustering With Application to Monitoring of Evolving Blog CommunitiesHuazhong Ning, Wei Xu, Yun Chi, Yihong Gong, and Thomas HuangHuazhong Ning, Wei Xu, Yun Chi, Yihong Gong, and Thomas Huangpp.261 - 272Chapter DOI:https://doi.org/10.1137/1.9781611972771.24PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract In recent years, spectral clustering method has gained attentions because of its superior performance compared to other traditional clustering algorithms such as K-means algorithm. The existing spectral clustering algorithms are all off-line algorithms, i.e., they can not incrementally update the clustering result given a small change of the data set. However, the capability of incrementally updating is essential to some applications such as real time monitoring of the evolving communities of websphere or blogsphere. Unlike traditional stream data, these applications require incremental algorithms to handle not only insertion/deletion of data points but also similarity changes between existing items. This paper extends the standard spectral clustering to such evolving data by introducing the incidence vector/matrix to represent two kinds of dynamics in the same framework and by incrementally updating the eigenvalue system. Our incremental algorithm, initialized by a standard spectral clustering, continuously and efficiently updates the eigenvalue system and generates instant cluster labels, as the data set is evolving. The algorithm is applied to a blog data set. Compared with recomputation of the solution by standard spectral clustering, it achieves similar accuracy but with much lower computational cost. Close inspection into the blog content shows that the incremental approach can discover not only the stable blog communities but also the evolution of the individual multi-topic blogs. Previous chapter Next chapter RelatedDetails Published:2007ISBN:978-0-89871-630-6eISBN:978-1-61197-277-1 https://doi.org/10.1137/1.9781611972771Book Series Name:ProceedingsBook Code:PR127Book Pages:xiv + 648Key words:Incremental clustering, Spectral Clustering, Incidence Vector/Matrix, Web-blogs

This paper presents two fuzzy clustering algorithms for categorical multivariate data based on q-divergence. First, this study shows that a conventional method for vectorial data can be explained as regularizing another conventional method using q-divergence. Second, based on the known results that Kullback-Leibler (KL)-divergence is generalized into the q-divergence, and two conventional fuzzy clustering methods for categorical multivariate data adopt KL-divergence, two fuzzy clustering algorithms for categorical multivariate data that are based on q-divergence are derived from two optimization problems built by extending the KL-divergence in these conventional methods to the q-divergence. Through numerical experiments using real datasets, the proposed methods outperform the conventional methods in term of clustering accuracy.

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.

This paper provides a new similarity detection algorithm. Given an input set of multi-dimensional data points, where each data point is assumed to be multi-dimensional, and an additional reference data point for similarity finding, the algorithm uses kernel method that embeds the data points into a low dimensional manifold. Unlike other kernel methods, which consider the entire data for the embedding, our method selects a specific set of kernel eigenvectors. The eigenvectors are chosen to separate between the data points and the reference data point so that similar data points can be easily identified as being distinct from most of the members in the dataset.

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.

Clustering techniques typically group similar instances underlying individual attributes by supposing that similar instances have similar attributes characteristic. On contrary, clustering similar instances given a specific behavior is framed through supervised learning. For instance, which fashion products have similar behavior in term of sales. Unfortunately, conventional clustering methods cannot tackle this case, since they handle attributes by a same manner. In fact, conventional clustering approaches do not consider any response, and moreover they assume attributes act by the same importance. However, clustering instances with respect to responses leads to a better data analytics. In this research, the authors introduce an approach for the goal supervised clustering and show its advantage in terms of data analytics as well as prediction. To verify the feasibility and the performance of this approach the authors conducted several experiments on a real dataset derived from an apparel industry.

A major outstanding challenge in developing unconventional wells is determining the optimal cluster spacing. The spacing between perforation clusters influences hydraulic fracture geometry, drainage volume, production rates, and the estimated ultimate recovery (EUR) of a well. This paper systematically examines the impact of cluster spacing in the Eagle Ford shale wells by calibrating fracture geometry and fracture/reservoir properties using field injection and production data and evaluating the optimal cluster spacing under different reservoir conditions. We explore a sequential technique to evaluate and optimize cluster spacing using a controlled field test at the Eagle Ford field. This study first identifies the fracture geometry by history matching the field injection treatment pressure. Using the rapid Fast Marching Method based flow simulation and Pareto-based multi-objective history matching, we match the well drainage volume and the cumulative production to calibrate the fracture and SRV properties. The impact of cluster spacing on the EUR are examined using the calibrated models. We run injection and production forecasts for various cluster spacing to investigate optimal completion under different reservoir conditions. The unique set of injection and production data used for this study includes two horizontal wells completed side by side. The well with tighter cluster spacing has larger drainage volume and better production performance. This is because of the increased fracture complexity in spite of the impact of stress shadow effects leading to shorter fractures. The calibrated models suggest that most of the fractures are planar in the Eagle Ford shale. The well with wider cluster spacing tends to develop longer fractures but the well with tighter cluster spacing has better stimulated reservoir volume with enhanced permeability, thus resulting in better drainage volume and production performance. From the optimization runs under different reservoir conditions, our results seem to indicate that when natural fractures are present or when stress anisotropy is high with no natural fractures, the wells with tighter cluster spacing tend to outperform the wells with wider cluster spacing. However, severe stress shadow effect is observed when stress anisotropy is low with no natural fractures, likely making tighter cluster spacing wells less favorable. The calibrated fracture geometries and properties with a unique set of Eagle Ford field data explain the performance variation for completions using different cluster spacing within the reservoir and provides insight into optimal cluster spacing under different reservoir conditions (low vs high stress anisotropy and with/without natural fractures).

Classifying genes into clusters depending on their expression profiles is one of the most important analysis techniques for microarray data. Because temporal gene expression profiles are indicative of the dynamic functional properties of genes, the application of clustering analysis to time-course data allows the more precise division of genes into functional classes. Conventional clustering methods treat the sampling data at each time point as data obtained under different experimental conditions without considering the continuity of time-course data between time periods t and t+1. Here, we propose a method designated mathematical model-based clustering (MMBC). The proposed method, designated MMBC, was applied to artificial data and time-course data obtained using Saccharomyces cerevisiae. Our method is able to divide data into clusters more accurately and coherently than conventional clustering methods. Furthermore, MMBC is more tolerant to noise than conventional clustering methods. Software is available upon request. taizo@brs.kyushu-u.ac.jp.

Improving neighborhood construction with Apollonius region algorithm based on density for clustering

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...

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.

Clustering by fast search and find of density peak (CFSFDP) is a simple and crisp density-clustering algorithm. The original algorithm is not suitable for direct application to anomaly detection. Its clustering results have a high level of redundant density information. If used directly as behavior profiles, the computation and storage costs of anomaly detection are high. Therefore, an improved algorithm based on CFSFDP is proposed for anomaly detection. The improved algorithm uses a few data points and their radius to support behavior profiles, and deletes the redundant data points without supporting profiles. This method not only reduces the large amount of data storage and distance calculation in the process of generating profiles, but also reduces the search space of profiles in the detection process. Numerous experiments show that the improved algorithm generates profiles faster than density-based spatial clustering of application with noise (DBSCAN), and has better profile precision than adaptive real-time anomaly detection with incremental clustering (ADWICE). The improved algorithm inherits the arbitrary shape clusters of CFSFDP, and improves the storage and computation performance. Compared with DBSCAN and ADWICE, the improved anomaly-detection algorithm based on CFSFDP has more balanced detection precision and real-time performance.