Graph Clustering Algorithm Research Articles

Network Clustering Analysis Using Mixture Exponential-Family Random Graph Models and Its Application in Genetic Interaction Data.

Epistatic miniarrary profile (EMAP) studies have enabled the mapping of large-scale genetic interaction networks and generated large amounts of data in model organisms. It provides an incredible set of molecular tools and advanced technologies that should be efficiently understanding the relationship between the genotypes and phenotypes of individuals. However, the network information gained from EMAP cannot be fully exploited using the traditional statistical network models. Because the genetic network is always heterogeneous, for example, the network structure features for one subset of nodes are different from those of the left nodes. Exponential-family random graph models (ERGMs) are a family of statistical models, which provide a principled and flexible way to describe the structural features (e.g., the density, centrality, and assortativity) of an observed network. However, the single ERGM is not enough to capture this heterogeneity of networks. In this paper, we consider a mixture ERGM (MixtureEGRM) networks, which model a network with several communities, where each community is described by a single EGRM. EM algorithm is a classical method to solve the mixture problem, however, it will be very slow when the data size is huge in the numerous applications. We adopt an efficient novel online graph clustering algorithm to classify the graph nodes and estimate the ERGM parameters for the MixtureERGM. In comparison studies, the MixtureERGM outperforms the role analysis for the network cluster in which the mixture of exponential-family random graph model is developed for many ego-network according to their roles. One genetic interaction network of yeast and two real social networks (provided as supplemental materials, which can be found on the Computer Society Digital Library at http://doi.ieeecomputersociety.org/10.1109/TCBB.2017.2743711) show the wide potential application of the MixtureERGM.

Hyperspectral Image Classification Using Graph Clustering Methods

Hyperspectral imagery is a challenging modality due to the dimension of the pixels which can range from hundreds to over a thousand frequencies depending on the sensor. Most methods in the literature reduce the dimension of the data using a method such as principal component analysis, however this procedure can lose information. More recently methods have been developed to address classification of large datasets in high dimensions. This paper presents two classes of graph-based classification methods for hyperspectral imagery. Using the full dimensionality of the data, we consider a similarity graph based on pairwise comparisons of pixels. The graph is segmented using a pseudospectral algorithm for graph clustering that requires information about the eigenfunctions of the graph Laplacian but does not require computation of the full graph. We develop a parallel version of the Nyström extension method to randomly sample the graph to construct a low rank approximation of the graph Laplacian. With at most a few hundred eigenfunctions, we can implement the clustering method designed to solve a variational problem for a graph-cut-based semi-supervised or unsupervised classification problem. We implement OpenMP directive-based parallelism in our algorithms and show performance improvement and strong, almost ideal, scaling behavior. The method can handle very large datasets including a video sequence with over a million pixels, and the problem of segmenting a data set into a pre-determined number of classes.

A fast heuristic detection algorithm for visualizing structure of large community

With the increase in number of users, social networks data is growing more big and complex to examine mutual information between different objects. Different graph visualization algorithms are used to explore such a big and complex network data. Network graphs are naturally complex and can have overlapping contents. In this paper, a novel clustering based visualization algorithm is proposed to draw network graph with reduced visual complexity. The proposed algorithm neither comprises of any random element nor it requires any pre-determined number of communities. Because of its less computational time i.e. O(nlogn), it can be applied effectively on large scale networks. We tested our algorithm on thirteen different types and scales of real-world complex networks ranging from N=101 to N=106 vertices. The performance of the proposed algorithms is compared with six existing widely used graph clustering algorithms. The experimental results show superiority of our algorithm over existing algorithms in terms of execution speed, accuracy, and visualization.

Detection of Complexes in Biological Networks Through Diversified Dense Subgraph Mining.

Protein-protein interaction (PPI) networks, providing a comprehensive landscape of protein interaction patterns, enable us to explore biological processes and cellular components at multiple resolutions. For a biological process, a number of proteins need to work together to perform a job. Proteins densely interact with each other, forming large molecular machines or cellular building blocks. Identification of such densely interconnected clusters or protein complexes from PPI networks enables us to obtain a better understanding of the hierarchy and organization of biological processes and cellular components. However, most existing graph clustering algorithms on PPI networks often cannot effectively detect densely connected subgraphs and overlapped subgraphs. In this article, we formulate the problem of complex detection as diversified dense subgraph mining and introduce a novel approximation algorithm to efficiently enumerate putative protein complexes from biological networks. The key insight of our algorithm is that instead of enumerating all dense subgraphs, we only need to find a small diverse subset of subgraphs that cover as many proteins as possible. The problem is modeled as finding a diverse set of maximal dense subgraphs where we develop highly effective pruning techniques to guarantee efficiency. To scale up to large networks, we devise a divide-and-conquer approach to speed up the algorithm in a distributed manner. By comparing with existing clustering and dense subgraph-based algorithms on several yeast and human PPI networks, we demonstrate that our method can detect more putative protein complexes and achieves better prediction accuracy.

Outlier edge detection using random graph generation models and applications

Outliers are samples that are generated by different mechanisms from other normal data samples. Graphs, in particular social network graphs, may contain nodes and edges that are made by scammers, malicious programs or mistakenly by normal users. Detecting outlier nodes and edges is important for data mining and graph analytics. However, previous research in the field has merely focused on detecting outlier nodes. In this article, we study the properties of edges and propose effective outlier edge detection algorithm. The proposed algorithms are inspired by community structures that are very common in social networks. We found that the graph structure around an edge holds critical information for determining the authenticity of the edge. We evaluated the proposed algorithms by injecting outlier edges into some real-world graph data. Experiment results show that the proposed algorithms can effectively detect outlier edges. In particular, the algorithm based on the Preferential Attachment Random Graph Generation model consistently gives good performance regardless of the test graph data. More important, by analyzing the authenticity of the edges in a graph, we are able to reveal underlying structure and properties of a graph. Thus, the proposed algorithms are not limited in the area of outlier edge detection. We demonstrate three different applications that benefit from the proposed algorithms: (1) a preprocessing tool that improves the performance of graph clustering algorithms; (2) an outlier node detection algorithm; and (3) a novel noisy data clustering algorithm. These applications show the great potential of the proposed outlier edge detection techniques. They also address the importance of analyzing the edges in graph mining—a topic that has been mostly neglected by researchers.

Detection of protein complex from protein-protein interaction network using Markov clustering

Detection of complexes, or groups of functionally related proteins, is an important challenge while analysing biological networks. However, existing algorithms to identify protein complexes are insufficient when applied to dense networks of experimentally derived interaction data. Therefore, we introduced a graph clustering method based on Markov clustering algorithm to identify protein complex within highly interconnected protein-protein interaction networks. Protein-protein interaction network was first constructed to develop geometrical network, the network was then partitioned using Markov clustering to detect protein complexes. The interest of the proposed method was illustrated by its application to Human Proteins associated to type II diabetes mellitus. Flow simulation of MCL algorithm was initially performed and topological properties of the resultant network were analysed for detection of the protein complex. The results indicated the proposed method successfully detect an overall of 34 complexes with 11 complexes consisting of overlapping modules and 20 non-overlapping modules. The major complex consisted of 102 proteins and 521 interactions with cluster modularity and density of 0.745 and 0.101 respectively. The comparison analysis revealed MCL out perform AP, MCODE and SCPS algorithms with high clustering coefficient (0.751) network density and modularity index (0.630). This demonstrated MCL was the most reliable and efficient graph clustering algorithm for detection of protein complexes from PPI networks.

Combining Structured Node Content and Topology Information for Networked Graph Clustering

Graphs are popularly used to represent objects with shared dependency relationships. To date, all existing graph clustering algorithms consider each node as a single attribute or a set of independent attributes, without realizing that content inside each node may also have complex structures. In this article, we formulate a new networked graph clustering task where a network contains a set of inter-connected (or networked) super-nodes, each of which is a single-attribute graph. The new super-node representation is applicable to many real-world applications, such as a citation network where each node denotes a paper whose content can be described as a graph, and citation relationships between papers form a networked graph (i.e., a super-graph). Networked graph clustering aims to find similar node groups, each of which contains nodes with similar content and structure information. The main challenge is to properly calculate the similarity between super-nodes for clustering. To solve the problem, we propose to characterize node similarity by integrating structure and content information of each super-node. To measure node content similarity, we use cosine distance by considering overlapped attributes between two super-nodes. To measure structure similarity, we propose an Attributed Random Walk Kernel (ARWK) to calculate the similarity between super-nodes. Detailed node content analysis is also included to build relationships between super-nodes with shared internal structure information, so the structure similarity can be calculated in a precise way. By integrating the structure similarity and content similarity as one matrix, the spectral clustering is used to achieve networked graph clustering. Our method enjoys sound theoretical properties, including bounded similarities and better structure similarity assessment than traditional graph clustering methods. Experiments on real-world applications demonstrate that our method significantly outperforms baseline approaches.

A parallel fuzzy clustering algorithm for large graphs using Pregel

Large graphs are scale free and ubiquitous having irregular relationships. Clustering is used to find existent similar patterns in graphs and thus help in getting useful insights. In real-world, nodes may belong to more than one cluster thus, it is essential to analyze fuzzy cluster membership of nodes. Traditional centralized fuzzy clustering algorithms incur high communication cost and produce poor quality of clusters when used for large graphs. Thus, scalable solutions are obligatory to handle huge amount of data in less computational time with minimum disk access. In this paper, we proposed a parallel fuzzy clustering algorithm named ‘PGFC’ for handling scalable graph data. It will be advantageous from the viewpoint of expert systems to develop a clustering algorithm that can assure scalability along with better quality of clusters for handling large graphs.The algorithm is parallelized using bulk synchronous parallel (BSP) based Pregel model. The cluster centers are initialized using degree centrality measure, resulting in lesser number of iterations. The performance of PGFC is compared with other state of art clustering algorithms using synthetic graphs and real world networks. The experimental results reveal that the proposed PGFC scales up linearly to handle large graphs and produces better quality of clusters when compared to other graph clustering counterparts.

Analysis and Implementation of Graph Clustering for Digital News Using Star Clustering Algorithm

Since Web 2.0 notion emerged and is used extensively by many services in the Internet, we see an unprecedented proliferation of digital news. Those digital news is very rich in term of content and link to other news/sources but lack of category information. This make the user could not easily identify or grouping all the news that they read into set of groups. Naturally, digital news are linked data because every digital new has relation/connection with other digital news/resources. The most appropriate model for linked data is graph model. Graph model is suitable for this purpose due its flexibility in describing relation and its easy-to-understand visualization. To handle the grouping issue, we use graph clustering approach. There are many graph clustering algorithm available, such as MST Clustering, Chameleon, Makarov Clustering and Star Clustering. From all of these options, we choose Star Clustering because this algorithm is more easy-to-understand, more accurate, efficient and guarantee the quality of clusters results. In this research, we investigate the accuracy of the cluster results by comparing it with expert judgement. We got quite high accuracy level, which is 80.98% and for the cluster quality, we got promising result which is 62.87%.

Evolutionary Graph Clustering for Protein Complex Identification.

This paper presents a graph clustering algorithm, called EGCPI, to discover protein complexes in protein-protein interaction (PPI) networks. In performing its task, EGCPI takes into consideration both network topologies and attributes of interacting proteins, both of which have been shown to be important for protein complex discovery. EGCPI formulates the problem as an optimization problem and tackles it with evolutionary clustering. Given a PPI network, EGCPI first annotates each protein with corresponding attributes that are provided in Gene Ontology database. It then adopts a similarity measure to evaluate how similar the connected proteins are taking into consideration the network topology. Given this measure, EGCPI then discovers a number of graph clusters within which proteins are densely connected, based on an evolutionary strategy. At last, EGCPI identifies protein complexes in each discovered cluster based on the homogeneity of attributes performed by pairwise proteins. EGCPI has been tested with several real data sets and the experimental results show EGCPI is very effective on protein complex discovery, and the evolutionary clustering is helpful to identify protein complexes in PPI networks. The software of EGCPI can be downloaded via: https://github.com/hetiantian1985/EGCPI.

Approximate normalized cuts without Eigen-decomposition

Most traditional weighted graph clustering algorithms are solved by spectral method, which is only suitable for small scale datasets because of the high space and time complexity. How to reduce the computational complexity of graph cut clustering to process the massive complex data has become a challenging problem. To overcome this problem, we design an approximate normalized cuts algorithm without eigen-decomposition for large scale clustering. On the one hand, the space requirement of normalized cut is decreased by sampling a few data points to infer the global features of dataset instead of using the whole affinity matrix; on the other hand, the graph cut clustering procedure is accelerated in an iterative way that using the approximate weighted kernel k-means to optimize the objective function of normalized cut, which avoids the direct eigen-decomposition of Laplacian matrix. We also analyze the approximation error of the proposed algorithm and compare it with other state-of-the-arts clustering algorithms on several benchmark datasets. The experimental results demonstrate that our method can efficiently do the clustering when the number of data objects exceeds tens of thousands.

An SVD based Real Coded Genetic Algorithm for Graph Clustering

This paper proposes a novel graph clustering model based on genetic algorithm using a random point bipartite graph. The model uses random points distributed uniformly in the data space and the measurement of distance from these points to the test points have been considered as proximity. Random points and test points create an adjacency matrix. To create a similarity matrix, correlation coefficients are computed from the given bipartite graph. The eigenvectors of the singular value decomposition of the weighted similarity matrix are considered and the same are passed to an elitist GA model for identifying the cluster centers. The model has been tasted with the standard datasets and the performance has been compared with existing standard algorithms.

Parallel local graph clustering

Graph clustering has many important applications in computing, but due to growing sizes of graph, even traditionally fast clustering methods such as spectral partitioning can be computationally expensive for real-world graphs of interest. Motivated partly by this, so-called local algorithms for graph clustering have received significant interest due to the fact that they can find good clusters in a graph with work proportional to the size of the cluster rather than that of the entire graph. This feature has proven to be crucial in making such graph clustering and many of its downstream applications efficient in practice. While local clustering algorithms are already faster than traditional algorithms that touch the entire graph, they are sequential and there is an opportunity to make them even more efficient via parallelization. In this paper, we show how to parallelize many of these algorithms in the shared-memory multicore setting, and we analyze the parallel complexity of these algorithms. We present comprehensive experiments on large-scale graphs showing that our parallel algorithms achieve good parallel speedups on a modern multicore machine, thus significantly speeding up the analysis of local graph clusters in the very large-scale setting.

Spatio-temporal graph clustering algorithm based on attribute and structural similarity

Spatio-temporal graph clustering algorithm based on attribute and structural similarity

Discovering top-k non-redundant clusterings in attributed graphs

Many graph clustering algorithms focus on producing a single partition of the vertices in the input graph. Nevertheless, a single partition may not provide sufficient insight about the underlying data. In this context, it would be interesting to explore alternative clustering solutions. Many areas, such as social media marketing demand exploring multiple clustering solutions in social networks to allow for behavior analysis to find, for example, potential customers or influential members according to different perspectives. Additionally, it would be desirable to provide not only multiple clustering solutions, but also to present multiple non-redundant ones, in order to unleash the possible many facets from the underlying dataset. In this paper, we propose RM-CRAG, a novel algorithm to discover the top-k non-redundant clustering solutions in attributed graphs, i.e., a ranking of clusterings that share the least amount of information, in the information theoretic sense. We also propose MVNMI, an evaluation criterion to assess the quality of a set of clusterings. Experimental results using different datasets show the effectiveness of the proposed algorithm.

Alignment of PPI Networks Using Semantic Similarity for Conserved Protein Complex Prediction.

Network alignment is a computational technique to identify topological similarity of graph data by mapping link patterns. In bioinformatics, network alignment algorithms have been applied to protein-protein interaction (PPI) networks to discover evolutionarily conserved substructures at the system level. In particular, local network alignment of PPI networks searches for conserved functional components between species and predicts unknown protein complexes and signaling pathways. In this article, we present a novel approach of local network alignment by semantic mapping. While most previous methods find protein matches between species by sequence homology, our approach uses semantic similarity. Given Gene Ontology (GO) and its annotation data, we estimate functional closeness between two proteins by measuring their semantic similarity. We adopted a new semantic similarity measure, simVICD, which has the best performance for PPI validation and functional match. We tested alignment between the PPI networks of well-studied yeast protein complexes and the genome-wide PPI network of human in order to predict human protein complexes. The experimental results demonstrate that our approach has higher accuracy in protein complex prediction than graph clustering algorithms, and higher efficiency than previous network alignment algorithms.

Graph-based multimodal clustering for social multimedia

Real world datasets often consist of data expressed through multiple modalities. Clustering such datasets is in most cases a challenging task as the involved modalities are often heterogeneous. In this paper we propose a graph-based multimodal clustering approach. The proposed approach utilizes an example relevant clustering in order to learn a model of the "same cluster" relationship between a pair of items. This model is subsequently used in order to organize the items of the collection to be clustered in a graph, where the nodes represent the items and a link between a pair of nodes exists if the model predicted that the corresponding pair of items belong to the same cluster. Eventually, a graph clustering algorithm is applied on the graph in order to produce the final clustering. The proposed approach is applied on two problems that are typically treated using clustering techniques; in particular, it is applied on the problem of detecting social events and to the problem of discovering different landmark views in collections of social multimedia.

Min-Hash를 이용한 효율적인 대용량 그래프 클러스터링 기법

Graph clustering is widely used to analyze a graph and identify the properties of a graph by generating clusters consisting of similar vertices. Recently, large graph data is generated in diverse applications such as Social Network Services (SNS), the World Wide Web (WWW), and telephone networks. Therefore, the importance of graph clustering algorithms that process large graph data efficiently becomes increased. In this paper, we propose an effective clustering algorithm which generates clusters for large graph data efficiently. Our proposed algorithm effectively estimates similarities between clusters in graph data using Min-Hash and constructs clusters according to the computed similarities. In our experiment with real-world data sets, we demonstrate the efficiency of our proposed algorithm by comparing with existing algorithms.

A consensus graph clustering algorithm for directed networks

Finding groups of highly related vertices in undirected graphs has been widely investigated. Nevertheless, a very few strategies are specially designed for dealing with directed networks. In particular, strategies based on the maximization of the modularity adjusted to overcome the resolution limit for directed networks have not been developed. The analysis of the characteristics of the clusters produced by these approaches is highly important since among the most used strategies for detecting communities in directed networks are the modularity maximization-based algorithms for undirected graphs. Towards these remarks, in this paper we propose a consensus-based strategy, named ConClus, for providing partitions for directed networks guided by the adjusted modularity measure. In the computational experiments, we compared ConClus with benchmark strategies, including Infomap and OSLOM, by using hundreds of LFR networks. ConClus outperformed Infomap and was competitive with OSLOM even for graphs with high mixture index and small-sized clusters, to which modularity-based algorithms have limitations. ConClus outperformed all algorithms when considering the networks with the highest average and maximum in-degrees among the networks used in the experiments.

Extracting Train Driver’s Eye-Gaze Patterns Using Graph Clustering

The present paper investigates train drivers’ eye-gaze data to find important features to explain behavioral differences between experienced and inexperienced drivers. The obtained eye-gaze data contain too complex transition structure to find any meaningful patterns that might be common across or differentiate drivers. The Markov Cluster Algorithm, which is an unsupervised algorithm for graph clustering, is therefore utilized to divide such a structure into clusters that represent constituent eye-gaze patterns of frequent occurrence. As a result, a common eye-gaze pattern was identified to represent a perception tactic that drivers would repetitively move their gaze ahead soon after looking at other specific areas. Comparing cluster structures extracted with different clustering parameter settings clarified that all of the drivers implement this tactic more or less, but that they are different in that the experienced drivers can consistently follow it while the inexperienced drivers can not.