Sparse Nonnegative Matrix Factorization for Multiple-Local-Community Detection

Local community detection algorithms are an important type of overlapping community detection methods. Local community detection methods identify local community structure through searching seeds and expansion process. In this paper, we propose a novel local community detection method on line graph through degree centrality and expansion (LCDDCE). We firstly employ line graph model to transfer edges into nodes of a new graph. Secondly, we evaluate edges relationship through a novel node similarity method on line graph. Thirdly, we introduce local community detection framework to identify local node community structure of line graph, combined with degree centrality and PageRank algorithm. Finally, we transfer them back into original graph. The experimental results on three classical benchmarks show that our LCDDCE method achieves a higher performance on normalized mutual information metric with other typical methods.

There have been more and more researches on community discovery in complex networks. Expanding of a source node into a community is one of the most successful methods for local community detection, especially when the global structure of the network is not accessible. In this paper, we propose CLOSE algorithm, Local Community Detection by LOcal Structure Expansion, based on the local expansion technique in the community detection. In CLOSE, we propose a novel connective function to identify a better source node. The node is in the center of a highly connected component of a graph. CLOSE selects a group of nodes instead of a single node to be the seed for the expansion of a local community. In addition, using the neighboring group can identify a suitable community for a hub node. Moreover, the expansion strategy is based on the label propagation technique instead of local community measurements. In experiments, we compare the performance of CLOSE with previous methods both on synthetic networks from the LFR Benchmark and real-world networks. We also examine the merit of the source node selection strategy. Both source node selection and community detection in CLOSE outperform previous algorithms.

HoSIM: Higher-order Structural Importance based method for multiple local community detection

Local community detection aims to find the communities that a given seed node belongs to. Most existing works on this problem are based on a very strict assumption that the seed node only belongs to a single community, but in real-world networks, nodes are likely to belong to multiple communities. In this paper, we first introduce a novel algorithm, HqsMLCD, that can detect multiple communities for a given seed node over static networks. HqsMLCD first finds the high-quality seeds which can detect better communities than the given seed node with the help of network representation, then expands the high-quality seeds one-by-one to get multiple communities, probably overlapping. Since dynamic networks also act an important role in practice, we extend the static HqsMLCD to handle dynamic networks and introduce HqsDMLCD. HqsDMLCD mainly integrates dynamic network embedding and dynamic local community detection into the static one. Experimental results on real-world networks demonstrate that our new method HqsMLCD outperforms the state-of-the-art multiple local community detection algorithms. And our dynamic method HqsDMLCD gets comparable results with the static method on real-world networks.

Recent years have witnessed the rapid growth of social network services. Real-world social networks are huge and changing over time. Consequently, the problems in this area have become more complex. Community detection is one of the most important problems in social networks. A good community can be defined as a group of nodes that are highly connected to each other and loosely connected to the nodes outside the community. Regarding the fact that social networks are huge in size, having complete information of the whole network is almost impossible. As a result, the problem of local community detection has become more popular in recent years. The problem of community detection in dynamic networks is well-investigated however, the local community detection is not widely addressed by researchers. In this paper, this problem is investigated by employing a number of existing local community detection algorithms in a dynamic structure. Results are reported in two different experiments. Experimental results show that one of the algorithms (algorithm P) outperforms other algorithms regarding the employed dynamic structure. The results indicate that algorithm P is much faster than other local community detection algorithms in dynamic networks.

Individuals are often involved in multiple online social networks. Considering that owners of these networks are unwilling to share their networks, some global algorithms combine information from multiple networks to detect all communities in multiple networks without sharing their edges. When data owners are only interested in the community containing a given node, it is unnecessary and computationally expensive for multiple networks to interact with each other to mine all communities. Moreover, data owners who are specifically looking for a community typically prefer to provide less data than the global algorithms require. Therefore, we propose the Local Collaborative Community Detection problem (LCCD). It exploits information from multiple networks to jointly detect the local community containing a given node without directly sharing edges between networks. To address the LCCD problem, we present a method developed from M method, called colM, to detect the local community in multiple networks. This method adopts secure multiparty computation protocols to protect each network’s private information. Our experiments were conducted on real-world and synthetic datasets. Experimental results show that colM method could effectively identify community structures and outperform comparison algorithms.

Local community detection aims to find the communities that a given seed node belongs to. Most existing works on this problem are based on a very strict assumption that the seed node only belongs to a single community, but in real-world networks, nodes are likely to belong to multiple communities. In this paper, we introduce a novel algorithm, HqsMLCD, that can detect multiple communities for a given seed node. HqsMLCD first finds the high-quality seeds which can detect better communities than the given seed node with the help of network representation, then expands the high-quality seeds one-by-one to get multiple communities, probably overlapping. Experimental results on real-world networks demonstrate that our new method HqsMLCD outperforms the state-of-the-art multiple local community detection algorithms.

Local community detection aims to obtain the local communities to which target nodes belong, by employing only partial information of the network. As a commonly used network model, bipartite applies naturally when modeling relations between two different classes of objects. There are three problems to be solved in local community detection, such as initial core node selection, expansion approach and community boundary criteria. In this work, a similarity based local community detection algorithm for bipartite networks (SLCDB) is proposed, and the algorithm can be used to detect local community structure by only using either type of nodes of a bipartite network. Experiments on real data prove that SLCDB algorithms output community structure can achieve a very high modularity which outperforms most existing local community detection methods for bipartite networks.

This paper is focused on the detection of communities in social networks. We propose and describe a novel method for detecting local communities. We have used this method in an experiment on student social networks in order to prove our hypothesis about the nature of student communities. The results of the experiment rationalized our hypothesis and confirmed the effectiveness of the described method of local community detection.

Local community detection aims to detect local communities that have expanded from the given node. Because of the convenience of obtaining the local information of the network and nearly linear time complexity, researchers have proposed many local community detection algorithms to discover the community structure of real-world networks and have obtained excellent results. Most existing local community detection algorithms expand from the given node to a community based on an expansion mechanism that can determine the membership of nodes. However, when determining the membership of neighboring nodes of a community, previous algorithms only considered the impact from the current community, but the impact from the potential communities around the node was neglected. As the name implies, a potential community is a community structure hidden in an unexplored network around a node. This paper gives the definition of potential communities of a node for the first time, that is, a series of connected components consisting of the node’s neighbors that are in the unexplored network. We propose a three-stage local expansion algorithm, named LCDPC, that performs Local Community Detection based on Potential Community exploration. First, we search for a suitable node to replace the given node as the seed by calculating the node importance and the node similarity. Second, we form the initial community by combining the seed and its suitable potential community. Finally, the eligible nodes are selected by comparing the similarities between potential communities and the expanding community and nodes and adding them to the initial community for community expansion. The proposed algorithm is compared with eight state-of-the-art algorithms on both real-world networks and artificial networks, and the experimental results show that the performance of the proposed algorithm is better than that of the comparison algorithms and that the application of potential community exploration can help identify the community structure of networks.

Sparse Non-negative Matrix Factorization (SNMF) based color unmixing for breast histopathological image analysis

Local community detection (or local clustering) is of fundamental importance in large network analysis. Random walk based methods have been routinely used in this task. Most existing random walk methods are based on the single-walker model. However, without any guidance, a single-walker may not be adequate to effectively capture the local cluster. In this paper, we study a multi-walker chain (MWC) model, which allows multiple walkers to explore the network. Each walker is influenced (or pulled back) by all other walkers when deciding the next steps. This helps the walkers to stay as a group and within the cluster. We introduce two measures based on the mean and standard deviation of the visiting probabilities of the walkers. These measures not only can accurately identify the local cluster, but also help detect the cluster center and boundary, which cannot be achieved by the existing single-walker methods. We provide rigorous theoretical foundation for MWC, and devise efficient algorithms to compute it. Extensive experimental results on a variety of real-world networks demonstrate that MWC outperforms the state-of-the-art local community detection methods by a large margin.

A method of image compression or dimensional reduction is proposed to avoid the dimension disaster in machine learning and the bottleneck in extracting and selecting sensitive characteristics, which can make it easier for automatic image recognition. Sparse non-negative matrix factorization (Sparse NMF) and the Probabilistic Neural Networks (PNN) are used to recognize the time-frequency images automatically for diesel valve trains. The comparison of the results of sparse NMF with the results of pane division have found that the former has higher correct recognition rate, which indicates that sparse NMF is an effective method for dimension reduction while preserving or even enhancing the information content. Thus the computation complexity is greatly reduced and the correct recognition rate is effectively improved.

Supervised single channel dual domains speech enhancement using sparse non-negative matrix factorization

Graph clustering, often addressed as community detection, is a prominent task in the domain of graph data mining with dozens of algorithms proposed in recent years. In this paper, we focus on several popular community detection algorithms with low computational complexity and with decent performance on the artificial benchmarks, and we study their behaviour on real-world networks. Motivated by the observation that there is a class of networks for which the community detection methods fail to deliver good community structure, we examine the assortativity coefficient of ground-truth communities and show that assortativity of a community structure can be very different from the assortativity of the original network. We then examine the possibility of exploiting the latter by weighting edges of a network with the aim to improve the community detection outputs for networks with assortative community structure. The evaluation shows that the proposed weighting can significantly improve the results of community detection methods on networks with assortative community structure.