Label Distribution Entropy Regularized Fuzzy C-means Algorithm For Balanced Clustering And Its Application In Wireless Sensor Network

With the rapid development of science and technology,massive data information is produced in our daily life,which poses a great challenge to the field of pattern recognition.Cluster analysis,as a classical unsupervised learning method,is a hot topic in the field of pattern recognition.Its purpose is to group similar samples into the same category according to certain criteria and find potential values and rules from massive data.However,the application scenarios of Wireless Sensor Network(WSN)and population distribution need to fully consider the equilibrium problem,and the traditional clustering algorithm cannot solve this problem.Therefore,balanced clustering has attracted more and more attention.As an important branch in the field of pattern recognition,it can satisfy both high-quality cluster and balanced cluster.For the soft equilibrium clustering problem,the main research and innovation work of this paper is as follows:1.In-depth study of load balancing in wireless sensor networks clustering protocols is a soft balance clustering problem.Firstly,the analysis of wireless sensor network clustering routing is essentially a clustering problem.Secondly,according to the load problem existing in WSN,the balanced clustering is divided into hard balanced clustering and soft balanced clustering.Finally,a soft equilibrium clustering model is designed and applied to wireless sensor networks.2.A Label Distribution Entropy Regularized Fuzzy C-means Algorithm for Balanced Clustering(FCM_LDE)model is established and optimization algorithm are given.Firstly,the hidden relationship between Fuzzy membership matrix and label matrix is studied,and a square loss term for regularizing Fuzzy C-means(FCM)is designed.Inspired by the strategy of balanced cluster method of label distribution design,the clustering distribution information revealed by label matrix is used.A label distribution entropy is designed to balance FCM.Secondly,the traditional FCM model,ordinary loss term and label distribution entropy are combined to build a soft equilibrium clustering model.Finally,Augmented Lagrange Multipliers(ALMs)method is used to propose an alternate update optimization method for solving the model,and the algorithm flow and complexity analysis are given.3.In order to verify the effectiveness of FCM_LDE,six real data sets and a simulation environment were used to evaluate balanced clustering.First,the Locality Preserving Projection(LPP)method is used to de-noise the original data.Secondly,six balanced data sets are compared with other algorithms,and the graphs obtained by some evaluation indexes are used.The results show that the proposed algorithm has good advantages in clustering performance and balancing performance.Finally,for wireless sensor networks,the simulation environment is set up by setting network operation parameters.Simulation analysis shows that the balance performance of the proposed algorithm has certain advantages in network life cycle,number of dead nodes and network energy consumption.