Center-based Clustering Algorithms Research Articles

On efficient model selection for sparse hard and fuzzy center-based clustering algorithms

The class of center-based clustering algorithms offers methods to efficiently identify clusters in data sets, making them applicable to larger data sets. While a data set may contain several features, not all of them may be equally informative or helpful towards cluster detection. Therefore, sparse center-based clustering methods offer a way to select only those features that may be useful in identifying the clusters present in a data set. However, to automatically determine the degree to which features should be selected, these methods use the Permutation Method which involves generating and clustering multiple randomly permuted data sets, leading to much higher computation costs. In this paper, we propose an improved approach towards model selection for sparse clustering by using expressions of Bayesian Information Criterion (BIC) derived for the center-based clustering methods of k-Means and Fuzzy c-Means. The derived expressions of BIC require significantly lower computation costs, yet allow us to compare and select a suitable sparse clustering among several possible sparse partitions that may have selected different subsets of features. Experiments on synthetic and real-world data sets show that using BIC for model selection leads to remarkable improvements in the identification of sparse clusterings for both Sparse k-Means and Sparse Fuzzy c-Means.

DISCERN: diversity-based selection of centroids for k-estimation and rapid non-stochastic clustering

One of the applications of center-based clustering algorithms such as K-Means is partitioning data points into K clusters. In some examples, the feature space relates to the underlying problem we are trying to solve, and sometimes we can obtain a suitable feature space. Nevertheless, while K-Means is one of the most efficient offline clustering algorithms, it is not equipped to estimate the number of clusters, which is useful in some practical cases. Other practical methods which do are simply too complex, as they require at least one run of K-Means for each possible K. In order to address this issue, we propose a K-Means initialization similar to K-Means++, which would be able to estimate K based on the feature space while finding suitable initial centroids for K-Means in a deterministic manner. Then we compare the proposed method, DISCERN, with a few of the most practical K estimation methods, while also comparing clustering results of K-Means when initialized randomly, using K-Means++ and using DISCERN. The results show improvement in both the estimation and final clustering performance.

A Multi-Objective Optimization Algorithm for Center-Based Clustering

Center-based clustering is a set of clustering problems that require finding a single element, a center, to represent an entire cluster. The algorithms that solve this type of problems are very efficient for clustering large and high-dimensional datasets. In this paper, we propose a similar heuristic used in Lloyd's algorithm to approximately solve (EMAX algorithm) a more robust variation of the k-means problem, namely the EMAX problem. Also, a new center-based clustering algorithm (SSO-C) is proposed, which is based on a swarm intelligence technique called Social Spider Optimization. This algorithm minimizes a multi-objective optimization function defined as a weighted combination of the objective functions of the k-means and EMAX problems. Also, an approximation algorithm for the discrete k-center problem is used as a local search strategy for initializing the population. Results of the experiments showed that SSO-C algorithm is suitable for finding maximum best values, however EMAX algorithm is better in finding median and mean values.

A Center-Based Secure and Stable Clustering Algorithm for VANETs on Highways

Currently, communications in the vehicular ad hoc network (VANET) can be established via both Dedicated Short Range Communication (DSRC) and mobile cellular networks. To make use of existing Long Term Evolution (LTE) network in data transmissions, many methods are proposed to manage VANETs. Grouping the vehicles into clusters and organizing the network by clusters are one of the most universal and most efficacious ways. Since the high mobility of vehicles makes VANETs different from other mobile ad hoc networks (MANETs), the previous cluster-based methods for MANETs may have trouble for VANETs. In this paper, we introduce a center-based clustering algorithm to help self-organized VANETs forming stable clusters and decrease the status change frequency of vehicles on highways and two metrics. A novel Cluster Head (CH) selection algorithm is also proposed to reduce the impact of vehicle motion differences. We also introduce two metrics to improve the security of VANETs. A simulation is conducted to compare our mechanism to some other mechanisms. The results show that our mechanism obtains high stability and lower packet loss rate.

A Central Force Optimization Approach for Data Clustering

This research presents a Central Force Optimization (CFO) method for finding the optimal cluster centers in a given dataset. The data clustering is one of the most important research topics of data mining. The objective of the clustering algorithm is to group the similar data objects together and has become an important issue in many fields and applications of sciences and engineering problems like pattern recognition, bioinformatics, machine learning, and etc. Due to the nature of the deterministic and population-based search, CFO can overcome the drawbacks of traditional clustering methods like K-means that easily converge to local optima. A novel and effective clustering based on CFO is proposed in this paper. CFO is an efficient and powerful population-based intelligence algorithm for optimization problems. It is a deterministic nature-inspired metaheuristic, which is based on an analogy to classical particle kinematics in a gravitational field. CFO uses probes as its basic population. The probes are scattered throughout the search space and as time progresses they move slowly towards the probe that has achieved the highest mass or fitness. The performance of the presented algorithm is evaluated on several well-known benchmark datasets and compared with other well-known center-based clustering algorithms. CFO performs well against the other clustering techniques. The comparisons revealed that the proposed algorithm delivers more efficient and effective results than the other clustering techniques.

Natural neighbor-based clustering algorithm with local representatives

Clustering by identifying cluster centers is important for detecting patterns in a data set. However, many center-based clustering algorithms cannot process data sets containing non-spherical clusters. In this paper, we propose a novel clustering algorithm called NaNLORE based on natural neighbor and local representatives. Natural neighbor is a new neighbor concept and introduced to compute local density and find local representatives which are points with local maximum density. We first find local representatives and then select cluster centers from the local representatives. The density-adaptive distance is introduced to measure the distance between local representatives, which helps to solve the problem of clustering data sets with complex manifold structure. Cluster centers are characterized by higher density than their neighbors and a relatively large density-adaptive distance from any local representatives with higher density. In experiments, we compare the proposed algorithm NaNLORE with existing algorithms on synthetic and real data sets. Results show that NaNLORE performs better than existing algorithm, especially on clustering non-spherical data and manifold data.

Anr An Algorithm To Recommend Initial Cluster Centers For K-means Algorithm

Clustering is one of the widely used knowledge discovery techniques to detect structure of datasets and can be extremely useful to the analyst. In center based clustering algorithms such as k-means, choosing initial cluster centers is really important as it has an important impact on the clustering result. It is desirable to select initial centers which are well separated. In this paper, we have proposed an algorithm to find initial cluster centers based on choosing two attributes that can describe the data space better and using the number of neighbors in a specific radius in data space. The proposed Attribute and Neighborhood Radius based (ANR) initial cluster center computing algorithm is applied to several well-known datasets .experimental results shows that it prevents form choosing noise data points as cluster center and tries to choose data points from dense areas in data space.

An Efficient Intelligent Clustering Tool based on Hybrid Fuzzified Algorithm for Electrical Data

Fuzzified optimization based data clustering is one of the important data mining tool which is active research of real world problems. This paper proposed Fuzzified Particle Swarm Optimization and K-Harmonic Means algorithm (FPSO+KHM) for clustering the electrical data systems. The partitioned clustering algorithms are more suitable for clustering large datasets. The K-Harmonic means algorithm is center based clustering algorithm and very insensitive to the selection of initial partition using built in boost function, but easily trapped in global optima. The proposed algorithm uses Fuzzified PSO and KHarmonic means to generate more accurate, robust , better clustering results. This algorithm can generate the solution in few number of iterations, and get faster convergence when compare to K-Harmonic Means and hybrid PSO+ K-Harmonic Means algorithms. This algorithm is applied for two different set of IEEE standard electrical bus data systems.

Anomaly Detection Using Metaheuristic Firefly Harmonic Clustering

The performance of communication networks can be affected by a number of factors including misconfiguration, equipments outages, attacks originated from legitimate behavior or not, software errors, among many other causes. These factors may cause an unexpected change in the traffic behavior and create what we call anomalies that may represent a loss of performance or breach of network security. Knowing the behavior pattern of the network is essential to detect and characterize an anomaly. Therefore, this paper presents an algorithm based on the use of Digital Signature of Network Segment (DSNS), used to model the traffic behavior pattern. We propose a clustering algorithm, K-Harmonic means (KHM), combined with a new heuristic approach, named Firefly Algorithm (FA), for network volume anomaly detection. The KHM calculate the weighting function of each point to calculate new centroids and circumventing the initialization problem present in most center based clustering algorithm and exploits the search capability of FA from escaping local optima. Processing the DSNS data and real traffic data is possible to detect and classify intervals considered anomalous with a trade-off between the 80% true-positive rate and 20% false-positive rate.