A Multi-Objicetive Optimization Memetic Algorithm For Community Detection In Complex Networks

With the rapid development of technology,complex networks have unwittingly affected people's lives.For example,from tangible transportation networks,communication networks,power networks to intangible economic networks,information networks,social networks,and the like.These networks can be abstracted as a form representation of a graph,with nodes representing objects,and nodes-to-node connections representing a relationship that exists between objects.The community structure is an important attribute of a complex network.It has the characteristics of tight internal connections and sparse external connections.Discovering the characteristics of complex network community structure helps to analyze network behavior and reveal potential laws in the network.In recent years,researchers have proposed a series of algorithms to discover the community structure of complex networks.Commonly used in community testing is the optimization method,which is to transform the community detection problem into the target optimization problem.Due to the complexity of the network in real life,the structure is numerous,and the way of optimizing multiple objective functions in community detection can better discover the network community structure.Therefore,this paper adopts multi-objective optimization method and local search operator to construct a new multi-objective crypto-parent algorithm.The main work and innovations of this paper are as follows:1.Research on complex network community structure and multi-objective optimization algorithm.For the firm and sparse judgment of the community,the researchers defined the same metrics for the community according to different criteria,such as module degree,module density,and community score.Combining different targets and optimizing at the same time helps to comprehensively consider the multiple characteristics of the community.This paper focuses on multi-objective algorithms and analyzes the characteristics of different algorithms.2.A community detection algorithm based on multi-objective cryptosystem is proposed.The algorithm uses the community score and modularity as the optimization objective function,and uses the uniform crossover and point mutation operation to evolve.The coding scheme adopts the coding method based on the neighboring point.The coding method does not need to know the number of communities in advance,and is convenient for processing most of the actual.A network that does not know the number of associations.3.The population initialization strategy using random walks.The initialization method of random walk is to ensure that each individual generated is a safe individual compared with the traditional random initialization.Taking the Markov transition probability as the node walk standard not only ensures the individual's effectiveness,but also maintains the diversity of the population.4.Simulated annealing operator is introduced as a local search strategy.Simulated annealing algorithm is a heuristic algorithm with good searching ability.In this paper,we make some improvements to the simulated annealing algorithm.Such adjustments are good for searching for good individuals.The Simulation experiments are carried out on the network generated by artificial synthetic network platform and the real network in the real world,and comparative analysis is carried out with other algorithms.The experimental results verify the effectiveness of community detection proposed in this paper.