Narrow Results

A large class of neural network models have their units organized in a lattice with fixed topology or generate their topology during the learning process. These network models can be used as neighborhood preserving map of the input manifold, but such a structure is difficult to manage since these maps are graphs with a number of nodes that is just one or two orders of magnitude less than the number of input points (i.e., the complexity of the map is comparable with the complexity of the manifold) and some hierarchical algorithms were proposed in order to obtain a high-level abstraction of these structures. In this paper a general structure capable to extract high order information from the graph generated by a large class of self–organizing networks is presented. This algorithm will allow to build a two layers hierarchical structure starting from the results obtained by using the suitable neural network for the distribution of the input data. Moreover the proposed algorithm is also capable to build a topology preserving map if it is trained using a graph that is also a topology preserving map.

The graph clustering problem has become highly relevant due to the growing interest of several research communities in social networks and their possible applications. Overlapped graph clustering algorithms try to find subsets of nodes that can belong to different clusters. In social network-based applications it is quite usual for a node of the network to belong to different groups, or communities, in the graph. Therefore, algorithms trying to discover, or analyze, the behavior of these networks needed to handle this feature, detecting and identifying the overlapped nodes. This paper shows a soft clustering approach based on a genetic algorithm where a new encoding is designed to achieve two main goals: first, the automatic adaptation of the number of communities that can be detected and second, the definition of several fitness functions that guide the searching process using some measures extracted from graph theory. Finally, our approach has been experimentally tested using the Eurovision contest dataset, a well-known social-based data network, to show how overlapped communities can be found using our method.

Clustering is one of the most versatile tools for data analysis. In the recent years, clustering that seeks the continuity of data (in opposition to classical centroid-based approaches) has attracted an increasing research interest. It is a challenging problem with a remarkable practical interest. The most popular continuity clustering method is the spectral clustering (SC) algorithm, which is based on graph cut: It initially generates a similarity graph using a distance measure and then studies its graph spectrum to find the best cut. This approach is sensitive to the parameters of the metric, and a correct parameter choice is critical to the quality of the cluster. This work proposes a new algorithm, inspired by SC, that reduces the parameter dependency while maintaining the quality of the solution. The new algorithm, named genetic graph-based clustering (GGC), takes an evolutionary approach introducing a genetic algorithm (GA) to cluster the similarity graph. The experimental validation shows that GGC increases robustness of SC and has competitive performance in comparison with classical clustering methods, at least, in the synthetic and real dataset used in the experiments.

The distance dynamics model was recently proposed to detect the disjoint community of a complex network. To identify the overlapping structure of a network using the distance dynamics model, an overlapping community detection algorithm, called L-Attractor, is proposed in this paper. The process of L-Attractor mainly consists of three phases. In the first phase, L-Attractor transforms the original graph to a link graph (a new edge graph) to assure that one node has multiple distances. In the second phase, using the improved distance dynamics model, a dynamic interaction process is introduced to simulate the distance dynamics (shrink or stretch). Through the dynamic interaction process, all distances converge, and the disjoint community structure of the link graph naturally manifests itself. In the third phase, a recovery method is designed to convert the disjoint community structure of the link graph to the overlapping community structure of the original graph. Extensive experiments are conducted on the LFR benchmark networks as well as real-world networks. Based on the results, our algorithm demonstrates higher accuracy and quality than other state-of-the-art algorithms.

Complex networks depict the individual relationship in a population, which can help to deeply mine the characteristics of complex networks and predict the potential collaboration between individuals by analyzing their interaction within different groups or clusters. However, the existing algorithms are with high complexity, which cost much computational time. In this paper, an efficient graph clustering algorithm based on spectral coarsening is proposed, to deal with the large time complexity of the traditional spectral algorithm. We first find the subset most possibly belonged to the same cluster in the original network, and merge them into a single node. The scale of the network will decrease with the network being coarsened. Then, the spectral clustering algorithm is performed on the coarsened network with the maintained advantages and the improved time efficiency. Finally, the experimental results on the multiple datasets demonstrate that the proposed algorithm, compared with the current state-of-the-art methods, has superior performance.

In today’s world scenario, many of the real-life problems and application data can be represented with the help of the graphs. Nowadays technology grows day by day at a very fast rate; applications generate a vast amount of valuable data, due to which the size of their representation graphs is increased. How to get meaningful information from these data become a hot research topic. Methodical algorithms are required to extract useful information from these raw data. These unstructured graphs are not scattered in nature, but these show some relationships between their basic entities. Identifying communities based on these relationships improves the understanding of the applications represented by graphs. Community detection algorithms are one of the solutions which divide the graph into small size clusters where nodes are densely connected within the cluster and sparsely connected across. During the last decade, there are lots of algorithms proposed which can be categorized into mainly two broad categories; non-overlapping and overlapping community detection algorithm. The goal of this paper is to offer a comparative analysis of the various community detection algorithms. We bring together all the state of art community detection algorithms related to these two classes into a single article with their accessible benchmark data sets. Finally, we represent a comparison of these algorithms concerning two parameters: one is time efficiency, and the other is how accurately the communities are detected.

In this paper, we examine the main advances registered in the last ten years in Pattern Recognition methodologies based on graph matching and related techniques, analyzing more than 180 papers; the aim is to provide a systematic framework presenting the recent history and the current developments. This is made by introducing a categorization of graph-based techniques and reporting, for each class, the main contributions and the most outstanding research results.

Graph clustering is an essential task in data analysis. Recently, there has been a notable trend in the application of deep learning in graph clustering. However, it is worth noting that existing deep graph clustering methods primarily focus on the topological information of nodes while overlooking the use of attribute information in the learned feature representations. In addition, they do not address the consistency of the space distribution in attribute and structure clustering results. To tackle these critical issues, a novel structured deep graph clustering network with consistency constraint (CC-DGC) is proposed. The network first constructs an autoencoder to learn and transmit hierarchical attribute information to the graph autoencoder (GAE). Subsequently, the GAE integrates the received hierarchical attribute information with extracted topological information to generate enhanced clustering representations. Moreover, this paper designs a consistency constraint module to promote consistency between the autoencoder and the GAE by optimizing the cluster space distributions they produce. Finally, the feature extraction and clustering classification processes are synchronized and optimized in a self-supervised manner within a unified framework. Extensive experiments illustrate that the proposed CC-DGC demonstrates superiority over the state-of-the-art deep graph clustering methods on five benchmark datasets.

Software architecture reconstruction plays an important role in software reuse, evolution and maintenance. Clustering is a promising technique for software architecture reconstruction. However, the representation of software, which serves as clustering input, and the clustering algorithm need to be improved in real applications. The representation should contain appropriate and adequate information of software. Furthermore, the clustering algorithm should be adapted to the particular demands of software architecture reconstruction well. In this paper, we first extract Weighted Directed Class Graph (WDCG) to represent object-oriented software. WDCG is a structural and quantitative representation of software, which contains not only the static information of software source code but also the dynamic information of software execution. Then we propose a WDCG-based Clustering Algorithm (WDCG-CA) to reconstruct high-level software architecture. WDCG-CA makes full use of the structural and quantitative information of WDCG, and avoids wrong compositions and arbitrary partitions successfully in the process of reconstructing software architecture. We introduce four metrics to evaluate the performance of WDCG-CA. The results of the comparative experiments show that WDCG-CA outperforms the comparative approaches in most cases in terms of the four metrics.

Any application that represents data as sets of graphs may benefit from the discovery of relationships among those graphs. To do this in an unsupervised fashion requires the ability to find graphs that are similar to one another. That is the purpose of GraphClust. The GraphClust algorithm proceeds in three phases, often building on other tools:

(1) it finds highly connected substructures in each graph;

(2) it uses those substructures to represent each graph as a feature vector; and

(3) it clusters these feature vectors using a standard distance measure. We validate the cluster quality by using the Silhouette method. In addition to clustering graphs, GraphClust uses SVD decomposition to find frequently co-occurring connected substructures. The main novelty of GraphClust compared to previous methods is that it is application-independent and scalable to many large graphs.

Community detection is a pivotal part of network analysis and is classified as an NP-hard problem. In this paper, a novel community detection algorithm is proposed, which probabilistically predicts communities’ diameter using the local information of random seed nodes. The gravitation method is then applied to discover communities surrounding the seed nodes. The individual communities are combined to get the community structure of the whole network. The proposed algorithm, named as Local Gravitational community detection algorithm (LGCDA), can also work with overlapping communities. LGCDA algorithm is evaluated based on quality metrics and ground-truth data by comparing it with some of the widely used community detection algorithms using synthetic and real-world networks.