| Multi-objective evolutionary algorithm has been widely used in recent years,and it has been widely used in industry,scheduling,automation and other fields.With the rapid increase of the amount of information,the generation of a large number of information will certainly be a large number of demand for the increase in the number of goals to optimize the rapid increase.Up to now,most of the multi-objective optimization algorithms are based on the two target,and there is no advantage in the Pareto front when there are more than three targets.Therefore,it is necessary to do a great deal of research on the application of multi-objective and the research of algorithm itself.In the case of high dimension,the number of non dominated solutions will decrease sharply,leading to the discontinuity of the Pareto front surface.Most of the existing algorithms,such as MOEA/D and NSGA-III,make up the Pareto optimal plane by dividing the target space into several sub problems and trying to find the optimal solution.But in the process of studying MOEA/D,we find that not all sub problems can find the unique optimal solution,that is to say,there are many sub problems corresponding to an optimal solution.So,this thesis proposes the MOEA/DM algorithm,the Pareto algorithm to solve the final frontier of non dominated solution insufficient number of problems by reducing the number of the number of target space neutron problems increase each sub problem corresponding to the optimal solution.Through the experiment we found that it can show superiority in the high dimensional case.The research of complex network clustering has been widely used in recent years.It has been widely used in the research of social networks,biological networks and the world wide web.The network cluster structure is one of the most common and important attributes in the research of complex network topology and implicit patterns.The network cluster has the characteristics that the same kind of nodes are very close to each other,and the connection between different classes is sparse.Today,the social network,the relationship network is rapidly increasing and becoming more complex,it has a very valuable research significance.Through the analysis of the characteristics of complex network clustering,it is found that the intra class links are very close in the network.This fits the characteristics of multi-objective optimization.In this thesis,we try to apply the MOEA/DM algorithm to the problem,and we find that it has some advantages over existing algorithms. |
