Research On Improved Swarm Intelligence Algorithms And Their Applications In Clustering

Swarm intelligent(SI)algorithms are highly efficient and widely used.The metaheuristic algorithms based on SI do not seek to establish and fit an accurate mathematical model of the optimized problem,and have the ability to solve problems efficiently and effectively that are difficult or even impossible to be solved by other optimization methods.They have obvious advantages in universality and robustness.For SI algorithms,how to further accelerate the convergence speed of natural element heuristic algorithm becomes the key problem to improve the performance of swarm intelligence optimization algorithm.Clustering algorithm is a kind of classical unsupervised learning method in the field of data mining,which can find out the potential structural relations between data and aggregate the individuals with similarity into the same cluster.It is a kind of efficient data aggregation technology.At present,most clustering algorithms are difficult to obtain accurate clustering results in the face of high-dimensional and complex data,so how to improve the accuracy of clustering has become a key problem to improve the performance of clustering analysis algorithm.In this paper,starting from such problems,using the characteristics of SI algorithms,the SI algorithms are improved to enhance the optimization ability and efficiency of the algorithm,so as to form a new SI algorithm,and it is applied to the fuzzy C-means clustering algorithm.The main points and creative ideas are shown below:(1)In this paper,the characteristics of existing swarm intelligence optimization algorithms are summarized and analyzed,and a backbone optimization framework(BOFCS)based on cross-stage evolution is proposed.This framework can be applied to all swarm intelligence optimization algorithms and has universality and universality.It is also shown that the framework converges to the global optimal solution set with probability 1.(2)On the basis of the backbone optimization model based on cross-stage evolution,whale optimization algorithm(WOA)is selected to form the backbone whale optimization algorithm based on cross-stage evolution(BWOACS)to verify the universality and effectiveness of the model.(3)Fuzzy clustering algorithm is one of the common algorithms in clustering.This algorithm is an algorithm to determine the clustering degree of each data point by membership degree.In this paper,the improved BWOACS is combined with the fuzzy Cmeans clustering algorithm(FCM)to form the backbone whale optimized fuzzy clustering model based on the cross-stage evolution(BWOACS-FCM).The algorithm can improve the clustering accuracy and improve the defect that individuals are prone to fall into the local optimal value.