Chuan Shi

A GENETIC ALGORITHM FOR DETECTING COMMUNITIES IN LARGE-SCALE COMPLEX NETWORKS

Network model recently becomes a popular tool for studying complex systems. Detecting meaningful communities in complex networks, as an important task in network modeling and analysis, has attracted great interests in various research areas. This paper proposes a genetic algorithm with a special encoding schema for community detection in complex networks. The algorithm employs a metric, named modularity Q as the fitness function and applies a special locus-based adjacency encoding schema to represent the community partitions. The encoding schema enables the algorithm to determine the number of communities adaptively and automatically, which provides great flexibility to the detection process. In addition, the schema also significantly reduces the search space. Extensive experiments demonstrate the effectiveness of the proposed algorithm.

A New Genetic Algorithm for Community Detection

With the rapidly grown evidence that various systems in nature and society can be modeled as complex networks, community detection in networks becomes a hot research topic in many research fields. This paper proposes a new genetic algorithm for community detection. The algorithm uses the fundamental measure criterion modularity Q as the fitness function. A special locus-based adjacency encoding scheme is applied to represent the community partition. The encoding scheme is suitable for the community detection based on the reason that it determines the community number automatically and reduces the search space distinctly. In addition, the corresponding crossover and mutation operators are designed. The experiments in three aspects show that the algorithm is effective, efficient and steady.

On selection of objective functions in multi-objective community detection

There is a surge of community detection of complex networks in recent years. Different from conventional single-objective community detection, this paper formulates community detection as a multi-objective optimization problem and proposes a general algorithm NSGA-Net based on evolutionary multi-objective optimization. Interested in the effect of optimization objectives on the performance of the multi-objective community detection, we further study the correlations (i.e., positively correlated, independent, or negatively correlated) of 11 objective functions that have been used or can potentially be used for community detection. Our experiments show that NSGA-Net optimizing over a pair of negatively correlated objectives usually performs better than the single-objective algorithm optimizing over either of the original objectives, and even better than other well-established community detection approaches.

A novel genetic algorithm for overlapping community detection

There is a surge of community detection on complex network analysis in recent years, since communities often play special roles in the network systems. However, many community structures are overlapping in real word. For example, a professor collaborates with researchers in different fields. In this paper, we propose a novel algorithm to discover overlapping communities. Different from conventional algorithms based on node clustering, our algorithm is based on edge clustering. Since edges usually represent unique relations among nodes, edge clustering will discover groups of edges that have the same characteristics. Thus nodes naturally belong to multiple communities. The proposed algorithm apply a novel genetic algorithm to cluster on edges. A scalable encoding schema is designed and the number of communities can be automatically determined. Experiments on both artificial networks and real networks validate the effectiveness and efficiency of the algorithm.

Ranking-based Clustering on General Heterogeneous Information Networks by Network Projection

Recently there is an increasing attention in heterogeneous information network analysis, which models networked data as networks including different types of objects and relations. Many data mining tasks have been exploited in heterogeneous networks, among which clustering and ranking are two basic tasks. These two tasks are usually done separately, whereas recent researches show that they can mutually enhance each other. Unfortunately, these works are limited to heterogeneous networks with special structures (e.g. bipartite or star-schema network). However, real data are more complex and irregular, so it is desirable to design a general method to manage objects and relations in heterogeneous networks with arbitrary schema. In this paper, we study the ranking-based clustering problem in a general heterogeneous information network and propose a novel solution HeProjI. HeProjI projects a general heterogeneous network into a sequence of sub-networks and an information transfer mechanism is designed to keep the consistency among sub-networks. For each sub-network, a path-based random walk model is built to estimate the reachable probability of objects which can be used for clustering and ranking analysis. Iteratively analyzing each sub-network leads to effective ranking-based clustering. Extensive experiments on three real datasets illustrate that HeProjI can achieve better clustering and ranking performances compared to other well-established algorithms.

A multi-objective approach for community detection in complex network

Detecting community structure is crucial for uncovering the links between structures and functions in complex networks. Most contemporary community detection algorithms employ single optimization criteria (e.g., modularity), which may have fundamental disadvantages. This paper considers the community detection process as a Multi-Objective optimization Problem (MOP). Correspondingly, a special Multi-Objective Evolutionary Algorithm (MOEA) is designed to solve the MOP and two model selection methods are proposed. The experiments in artificial and real networks show that the multi-objective community detection algorithm is able to discover more accurate community structures.

An estimation of distribution algorithm based on nonparametric density estimation

Probabilistic models play a key role in an estimation of distribution algorithm(EDA). Generally, the form of a probabilistic model has to be chosen before executing an EDA. In each generation, the probabilistic model parameters will be estimated by training the model on a set of selected individuals and new individuals are then sampled from the probabilistic model. In this paper, we propose to use probabilistic models in a different way: firstly generate a set of candidate points, then find some as offspring solutions by a filter which is based on a nonparametric density estimation method. Based on this idea, we propose a nonparametric estimation of distribution algorithm (nEDA) for global optimization. The major differences between nEDA and traditional EDAs are (1) nEDA uses a generating-filtering strategy to create new solutions while traditional EDAs use a model building-sampling strategy to generate solutions, and (2) nEDA utilizes a nonparametric density model with traditional EDAs usually utilize parametric density models. nEDA is compared with a traditional EDA which is based on Gaussian model on a set of benchmark problems. The preliminary experimental results show that nEDA is promising for dealing with global optimization problems.

Multi-objective decisionmaking in the detection of comprehensive community structures

Community detection in complex networks has attracted a lot of attentions in recent years. Compared with the traditional single-objective community detection approaches, the multi-objective approaches based on evolutionary computation can provide a decision maker with more flexible and promising solutions. How to make effective use of the optimal solution set returned by the multi-objective community detection approaches is an important yet unsolved issue. Through leveraging an existing multi-objective community detection algorithm, this paper proposes four model selection methods to aid the decision makers to select the preferable community structures. The experiments with three synthetic and real social networks illustrate that the proposed method can discover more authentic and comprehensive community structures than traditional single-objective approaches.

A novel algorithm for hierarchical community structure detection in complex networks

Networks have in recent years emerged as an invaluable tool for describing and quantifying complex systems in many branches of science. Recent studies suggest that network often exhibit hierarchical organization, where vertices divide into groups that further subdivided into groups of groups, and so forth over multiple scales. In this paper, we introduce a novel algorithm that searches for the hierarchical structure. The method iteratively combines the similar communities with the elaborate design of community similarity and combination threshold. The experiments on artificial and real networks show that the method is able to obtain reasonable hierarchical structure solutions.

A decomposition based multiobjective evolutionary algorithm with semi-supervised classification

In multiobjective evolutionary algorithms, how to select the optimal solutions from the offspring candidate set significantly affects the optimization process. Usually, the selection process is largely based on the real objective values or surrogate model estimating objective values. However, these selection processes are very time consuming sometimes, especially for some real optimization problems. Recently, some researches began to employ supervised classification to assist offspring selection, but these works are difficult to prepare the exact positive and negative samples or time consuming of parameter tuning problems. In order to solve these disadvantages, we propose a decomposition based multiobjective evolutionary algorithm with semi-supervised classification. This approach using random sampling and non-dominated sorting to construct semi supervised classifier. In each generation, a set of candidate solutions are generated for each subproblem and only good solutions are reserved by classifier. If there is more than one good solutions, we calculate each of good solutions by real objective function and choose the best one as the offspring solution. Based on the typical decomposition based multiobjective evolutionary algorithm MOEA/D, we design algorithm framework through integrating the novel offspring selection process based on semi-supervised classification. Experiments show that the proposed algorithm performs best in most test cases and improves the performance of MOEA/D.