Outlier detection is of much importance in preprocessing of data collected from complex industry system, for the data has strong nonlinearity and poor stability, involving much noise. Outlier detection based on cluste...
详细信息
ISBN:
(纸本)9781467313988
Outlier detection is of much importance in preprocessing of data collected from complex industry system, for the data has strong nonlinearity and poor stability, involving much noise. Outlier detection based on clustering, rejects abnormal data points which have significant difference from others according to the definition of similarity. Self-organizing Feature Map (SOM) Neural Network algorithm has the self-study and adaptive functions of neural networks, so as to be a hot research in clustering analysis recently. This paper first introduces Self-organizing Feature Map algorithm based on artificial neural network, and then improves the algorithm by using weighted Euclidean distance, finally uses the software of MATLAB to analyze some actual data of electrical power. The result shows that SOM algorithm achieves a very good effect in clustering, and the MATLAB toolbox shows favorable visual effects.
Recently, extensive research efforts have been devoted to the design of efficient clustering algorithms to divide all the nodes in a mobile ad hoc network into multiple clusters to form a clustered architecture. A clu...
详细信息
Recently, extensive research efforts have been devoted to the design of efficient clustering algorithms to divide all the nodes in a mobile ad hoc network into multiple clusters to form a clustered architecture. A clustered architecture is more stable if it can hold for a longer period of time. In a clustered architecture, due to node mobility, a node may depart from its original cluster and enter another cluster dynamically. Such a change may cause the clustered architecture to be reconfigured, leading to the instability of the network. Frequent information exchanges among the participating nodes and re-computation of clusters involve high communication and computation overheads. Therefore, it is obvious that a more stable clustered architecture will directly lead to the performance improvement of the whole network. In this paper, we propose an efficient clustering algorithm that can establish a more stable clustered architecture by keeping a node with many weak links from being selected as a clusterhead. Computer simulations show that the clustered architectures generated by our clustering algorithm are more stable than those generated by other clustering algorithms.
Data clustering partitions a dataset into clusters where each cluster contains similar data. clustering algorithms usually require users to set the number of clusters, e.g., k-means or fuzzy c-means. However, it is di...
详细信息
ISBN:
(纸本)9781467311830
Data clustering partitions a dataset into clusters where each cluster contains similar data. clustering algorithms usually require users to set the number of clusters, e.g., k-means or fuzzy c-means. However, it is difficult to determine a meaningful number of clusters if users lack prior knowledge of the data. Data clustering may use a validity index to grade the clustering quality. Most validity indices are based on clustering compactness and separation, but other criteria are also used for clustering. Therefore, no individual validity index is applicable to data with different properties. This paper presents a novel dynamic clustering based on particle swarm optimization. The proposed algorithm is compared with other dynamic clustering algorithms based on particle swarm optimization using artificial and real data sets. The experimental results showed that our proposed algorithm not only determines the appropriate number of clusters with correct cluster centers but can also be applied to data with different properties using various validity indices.
An important class of data mining problems is clustering. The traditional clustering algorithm is a kind of hard partition and it parts strictly each object into some cluster. But the real object is not always having ...
详细信息
ISBN:
(纸本)9781424417339
An important class of data mining problems is clustering. The traditional clustering algorithm is a kind of hard partition and it parts strictly each object into some cluster. But the real object is not always having distinct attributes, so fuzzy theory is introduced to deal with it. In this paper an improved algorithm is introduced based on the research of fuzzy theory and other clustering algorithms. It is introduced on the base of improving the FCM algorithm and used to analyzing the data that are from the environment detecting system. The result is encouraging.
In MANETs,the scalability problem has been solved by the clustering ***,current clustering algorithms consider on the network stability only in terms of some metrics affecting innercluster structure's stability,an...
详细信息
In MANETs,the scalability problem has been solved by the clustering ***,current clustering algorithms consider on the network stability only in terms of some metrics affecting innercluster structure's stability,and neglect some metrics affecting intercluster structure's stability which are more favorable to global *** solve this problem,a clustering algorithm is proposed in this *** gives a comprehensive measurement on stability metrics of the innercluster structure and the intercluster *** a better comprehension of our algorithm,an explanatory example is *** compare the performance of our algorithm to that of clustering algorithms with clusterheads,we simulate the structural adjusting times and network overheads during the process of the cluster formation and *** conclusion shows that our algorithm is more favorable to the stability of the global hierarchical structure and reduces network overheads a lot,which improves the global network performance.
clustering analysis is an active and challenge research direction in the field of data mining. In this paper we propose a new clustering algorithm based on dimensional reduction approach and K-harmonic means algorithm...
详细信息
clustering analysis is an active and challenge research direction in the field of data mining. In this paper we propose a new clustering algorithm based on dimensional reduction approach and K-harmonic means algorithm. Numerical results illustrate that the new hybrid clustering algorithm has advantages in the computation time, iteration numbers and clustering results in most cases, and it is also an algorithm which is suitable for large scale data sets.
Environmental pollution affects the earth and the health of human *** this paper,we will study the various environmental poisonous substances that are pathogenic together with the chemicals,mechanical facilities, work...
详细信息
Environmental pollution affects the earth and the health of human *** this paper,we will study the various environmental poisonous substances that are pathogenic together with the chemicals,mechanical facilities, work environment,work pressure and other related cause. We will study the relationship and treat to the human health so that prevention strategies can be made from early *** experimental testing is performed on the health checkup data for the year 2006 with permission from a medical organization in central Taiwan.A sample data of required feature values is extracted by integration, simplification and transformation *** clustering algorithm is then applied to the occupation and working environment of the *** aim is to study the relationship between the occupation and abnormal factors affecting the *** medical diagnoses can be used as warning alerts for high risk working environment.
As a fundamental problem in data mining, pattern recognition and machine learning, clustering algorithm has been studied for decades, and has been improved in many aspects. However, parameter-free clustering algorithm...
详细信息
As a fundamental problem in data mining, pattern recognition and machine learning, clustering algorithm has been studied for decades, and has been improved in many aspects. However, parameter-free clustering algorithms are still quite weak, which makes their potential generalization to a lot of promising applications rather difficult. A parameter-free clustering algorithm based on density model is proposed in this paper. This algorithm explores in a dynamically constructed nearest neighbor graph to detect which points are of the same density model, and then agglomerates them into the same cluster. It requires neither previously nor interactively setting of pivotal parameters via range scaling and proportional criterion technique. Its overall computational complexity is ???n nlog ??. And the experimental results demonstrate that the proposed algorithm can correctly recognize the arbitrary shaped clusters.
A novel clustering algorithm based on location aware in wireless sensor network is proposed in order to solve the problem of collecting spatial correlated data with energy validity. According to the user-provided erro...
详细信息
A novel clustering algorithm based on location aware in wireless sensor network is proposed in order to solve the problem of collecting spatial correlated data with energy validity. According to the user-provided error-tolerance threshold and sensing data similarity matrix based on location information, the algorithm performs unsupervised data mining task and then divides the monitoring range into several equivalent information ranges. Only one node per equivalent range is selected as the cluster-head by present maximal remainder energy. A mobile agent is used to collect sensing information of cluster-heads. This mechanism reduces the amount of transmission data and saves energy efficiently.
This paper proposes a multiple criteria decision making (MCDM)-based framework to address two fundamental issues in cluster validation: 1) evaluation of clustering algorithms and 2) estimation of the optimal cluster n...
详细信息
This paper proposes a multiple criteria decision making (MCDM)-based framework to address two fundamental issues in cluster validation: 1) evaluation of clustering algorithms and 2) estimation of the optimal cluster number for a given data set. Since both issues involve more than one criterion, they can be modeled as multiple criteria decision making (MCDM) problems. The proposed framework is examined by an experimental study. The results suggest that MCDM methods are practical tools for the evaluation of clustering algorithms. In addition, the selected MCDM method, PROMETHEE II can estimate the optimal numbers of clusters for ten out of fifteen datasets by adjusting the weights of criteria.
暂无评论