Sharif Digital Repository

Clustering is the problem of partitioning a (large) set of data using unsupervised techniques. Today, there exist many clustering techniques. The most important characteristic of a clustering technique is the shape of the cluster it can find. In this paper, we propose a method that is capable to find arbitrary shaped clusters and uses simple geometric constructs, Circlusters. Circlusters are different radius sectored circles. Circlusters can be used to create many hybrid approaches in mixture with density based or partitioning based methods. We also proposed two new clustering methods that are capable to find complex clusters in O(n), where n is the size of the data set. Both of the methods... 

One of the important problems in knowledge discovery from data is clustering. Clustering is the problem of partitioning a set of data using unsupervised techniques. An important characteristic of a clustering technique is the shape of the cluster it can find. Clustering methods which are capable to find simple cluster shapes are usually fast but inaccurate for complex data sets. Ones capable to find complex cluster shapes are usually not fast but accurate. In this paper, we propose a simple clustering technique named circlusters. Circlusters are circles partitioned into different radius sectors. Circlusters can be used to create hybrid approaches with density based or partitioning based... 

It has long been thought that intermediate-mass black holes (IMBHs) might be located in the cores of globular clusters. However, studies attempting to confirm this possibility have been inconclusive. To refine the search for these objects, Baumgardt et al. completed a series of N-body simulations to determine the observational properties that a host globular cluster should possess. Keys to revealing the presence of an IMBH were found to be the shape of the cluster's core proper motion dispersion profile and its surface density profile. Among the possible host clusters identified by Baumgardt et al., NGC6266 was found to be the most suitable object to search. Hubble Space Telescope Wide Field... 

Purpose: The main purpose of this study is to review the knowledge management literature from a content-related perspective using cluster analysis. Design/methodology/approach: A critical analysis of previous review articles in KM provided a conceptual framework with nine dimensions. A survey was then administered to 120 KM authors asking them to review which dimensions they considered in their own research. Findings: Three clusters of KM research were identified as follows: the socialization school, the collaboration school, and the codification school. Research limitations/implications: The study does not consider the dimension of strategic versus operational KM issues nor does it consider... 

In many popular applications like peer-to-peer systems, large amounts of data are distributed among multiple sources. Analysis of this data and identifying clusters is challenging due to processing, storage, and transmission costs. In this paper, we propose GDCluster, a general fully decentralized clustering method, which is capable of clustering dynamic and distributed data sets. Nodes continuously cooperate through decentralized gossip-based communication to maintain summarized views of the data set. We customize GDCluster for execution of the partition-based and density-based clustering methods on the summarized views, and also offer enhancements to the basic algorithm. Coping with... 

Tumor metastasis is responsible for the vast majority of cancer-associated morbidities and mortalities. Recent studies have disclosed the higher metastatic potential of circulating tumor cell (CTC) clusters than single CTCs. Despite long-term study on metastasis, the characterizations of its most potent cellular drivers, i.e., CTC clusters have only recently been investigated. The analysis of CTC clusters offers new intuitions into the mechanism of tumor metastasis and can lead to the development of cancer diagnosis and prognosis, drug screening, detection of gene mutations, and anti-metastatic therapeutics. In recent years, considerable attention has been dedicated to the development of... 

Identifying clusters is an important aspect of analyzing large datasets. Clustering algorithms classically require access to the complete dataset. However, as huge amounts of data are increasingly originating from multiple, dispersed sources in distributed systems, alternative solutions are required. Furthermore, data and network dynamicity in a distributed setting demand adaptable clustering solutions that offer accurate clustering models at a reasonable pace. In this paper, we propose GoScan, a fully decentralized density-based clustering algorithm which is capable of clustering dynamic and distributed datasets without requiring central control or message flooding. We identify two major... 

We study the statistical and computational properties of a network Lasso method for local graph clustering. The clusters delivered by nLasso can be characterized elegantly via network flows between cluster boundaries and seed nodes. While spectral clustering methods are guided by a minimization of the graph Laplacian quadratic form, nLasso minimizes the total variation of cluster indicator signals. As demonstrated theoretically and numerically, nLasso methods can handle very sparse clusters (chain-like) which are difficult for spectral clustering. We also verify that a primal-dual method for non-smooth optimization allows to approximate nLasso solutions with optimal worst-case convergence... 

Recently, data stream has become popular in many contexts of data mining. Due to the high amount of incoming data, traditional clustering algorithms are not suitable for this family of problems. Many data stream clustering algorithms proposed in recent years considered the scalability of data, but most of them did not attend the following issues: (1) The quality of clustering can be dramatically low over the time. (2) Some of the algorithms cannot handle arbitrary shapes of data stream and consequently the results are limited to specific regions. (3) Most of the algorithms have not been evaluated in multi-density environments. Identifying appropriate clusters for data stream by handling the... 

This paper addresses the density based clustering problem in data mining where clusters are established based on density of regions. The most well-known algorithm proposed in this area is DBSCAN [1] which employs two parameters influencing the shape of resulted clusters. Therefore, one of the major weaknesses of this algorithm is lack of ability to handle clusters in multi-density environments. In this paper, a new density based grid clustering algorithm, GDCLU, is proposed which uses a new definition for dense regions. It determines dense grids based on densities of their neighbors. This new definition enables GDCLU to handle different shaped clusters in multi-density environments. Also... 

Clustering Web data is one important technique for extracting knowledge from the Web. In this paper, a novel method is presented to facilitate the clustering. The method determines the appropriate number of clusters and provides suitable representatives for each cluster by inference from a Bayesian network. Furthermore, by means of the Bayesian network, the contents of the Web pages are converted into vectors of lower dimensions. The method is also extended for hierarchical clustering, and a useful heuristic is developed to select a good hierarchy. The experimental results show that the clusters produced benefit from high quality  

Clustering approaches are going to efficiently define the activated regions of the brain in fMRI studies. However, choosing appropriate clustering algorithms and defining optimal number of clusters are still key problems of these methods. In this paper, we apply an improved version of Growing Neural Gas algorithm, which automatically operates on the optimal number of clusters. The decision criterion for creating new clusters at the heart of this online clustering is taken from MB cluster validity index. Comparison with other so-called clustering methods for fMRI data analysis shows that the proposed algorithm outperforms them in both artificial and real datasets. ©2009 IEEE  

In this paper, we propose a fuzzification method for clusters produced from a clustering process, based on Active Learning Method (ALM). ALM is a soft computing methodology which is based on a hypothesis claiming that human brain interprets information in pattern-like images. The proposed fuzzification method is applicable to all non-fuzzy clustering algorithms as a post process. The most outstanding advantage of this method is the ability to determine the membership degrees of each data to all clusters based on the density and shape of the clusters. It is worth mentioning that for existing fuzzy clustering algorithms such as FCM the membership degree is usually determined as a function of... 

Until now, many algorithms have been introduced for finding arbitrary shaped clusters, but none of these algorithms is able to identify all sorts of cluster shapes and structures that are encountered in practice. Furthermore, the time complexity of the existing algorithms is usually high and applying them on large dataseis is time-consuming. In this paper, a novel fast clustering algorithm is proposed. This algorithm distinguishes clusters of different shapes using a twostage clustering approach. In the first stage, the data points are grouped into a relatively large number of fuzzy ellipsoidal sub-clusters. Then, connections between sub-clusters are established according to the Bhatiacharya... 

Recently, several algorithms have been introduced for enhancing clustering quality by using supervision in the form of constraints. These algorithms typically utilize the pair wise constraints to either modify the clustering objective function or to learn the clustering distance measure. Very few of these algorithms show the ability of discovering clusters of different shapes along with satisfying the provided constraints. In this paper, a novel semi-supervised clustering algorithm is introduced that uses the side information and finds clusters of arbitrary shapes. This algorithm uses a two-stage clustering approach satisfying the pair wise constraints. In the first stage, the data points... 

Today's world needs new methods to deal with and analyze the ever-increasingly generated data streams. Two of the most challenging aspects of data streams are (i) concept drift, i.e. evolution of data stream over time, which requires the ability to make timely decisions against the high speed of receiving new data; (ii) limited memory storage and the impracticality of using memory due to the large amount of data. Clustering is one of the common methods to process data streams. In this paper, we propose a novel, fully-online, density-based method for clustering evolving data streams. In recent years, a number of methods have been proposed, which also have the ability to cluster data streams.... 

Clustering is a major task in data mining which is used in many applications. However, general clustering is inappropriate for many applications where some constraints should be applied. One category of these constraints is the cluster size constraint. In this paper, we propose a new algorithm for solving the clustering with balancing constraints by using the minimum matching. We compare our algorithm with the method proposed by Banerjee and Ghosh that uses stable matching and show that our algorithm converge to the final solution in fewer iterations. ©2009 IEEE  

We determine thermal entanglement in mean-field clusters of N spin one-half particles interacting via the anisotropic Heisenberg interaction, with and without external magnetic field. For the xxx cluster in the absence of magnetic field we prove that only the N=2 ferromagnetic cluster shows entanglement. An external magnetic field B can only entangle xxx antiferromagnetic clusters in certain regions of the B-T plane. On the other hand, the xxz clusters of size N>2 are entangled only when the interaction is ferromagnetic. Detailed dependence of the entanglement on various parameters is investigated in each case. © 2006 The American Physical Society  

In recent years, the active learning algorithms have focused on combining correlation criterion and uncertainty criterion for evaluating instances. Although these criteria might be useful, applying these measures on whole input space globally may lead to inefficient selected instances for active learning. The proposed method takes advantage of clustering to partition input space to subspaces. Then it exploits both labeled and unlabeled data locally for selection of instances by using a graph-based active learning. We define a novel utility score for selecting clusters by combining uncertainty criterion, local entropy of clusters and the factor of contribution of each cluster in queries....