Multi-Improvement On Density-Based Clustering Algorithm And Its Applications

With the advent of the era of big data and the gradual maturity of artificial intelligence,the way to analyze and utilize large-scale raw data and to extract value information from it is the focus of academic research.Clustering technology plays an important role in data mining.The potential internal structure of data from massive data is the key research task in the field of artificial intelligence.Up to now,the unsupervised learning field mainly includes two types of most competitive clustering techniques,one is Clustering by Fast Search and Find of Density Peaks(DPC)algorithm,and the other is based on density with noise.Density-Based Spatial Clustering of Applications with Noise(DBSCAN)algorithm.However,the above algorithms still have some shortcomings:(1)When the density peak clustering algorithm is adopted,the truncation distance parameter must be manually set;(2)the density peak clustering algorithm needs to manually circle the cluster center point;(3)based on density The spatial clustering algorithm of noise has the problem that the global parameter clustering radius Eps needs to be manually set.Combined with the above issues,put forward targeted rectification opinions:(1)When adopting the spatial clustering algorithm based on density and noise,the global parameter clustering radius Eps must be obtained by manual setting.A parameter adaptive density-based noise-based spatial clustering is proposed for this problem(Spatial)The Clustering with Density and Noise Based on Parameter Adaptation(CS-DBSCAN)algorithm uses the cuckoo search algorithm to quickly solve the global optimization problem and improve the clustering performance of the algorithm.(2)When adopting the density peak clustering algorithm,the truncation distance parameter must be set manually.At this time,the Density Peak Clustering Based on Maximum Density(Max-DPC)algorithm can be used.The algorithm introduces the improved idea of the density point from the sample point to the density point of the sample point as the cutoff value,avoiding the artificial setting of the cutoff distance and improving the performance of the clustering algorithm.(3)When adopting the density peak clustering algorithm,it must be done manually by setting the truncation distance parameter,and also including the clustering center point,and propose a density peak cluster based on the truncation distance and the cluster center point automatic selection strategy.Density Peak Clustering Algorithm Based on Choosing Automatically for Cut-off Distance and Cluster Center(CSA-DPC).Based on the improvement scheme(2),the cluster center point is determined according to the similarity change between the cluster center points,which also makes the clustering result more accurate.(4)For the spatial clustering algorithm based on density with noise,the clustering radius Eps needs to be manually set and the density peak clustering algorithm needs to manually circle the clustering center point,and proposes bat-based optimization clustering.Re-clustering Algorithm Based on Bat Optimized Clustering and Its Applications(BA-DPC).By introducing the bat optimization algorithm,the improved DBSCAN clustering algorithm first obtains the initial clustering result,and then automatically selects the clustering center of the DPC algorithm according to the initial clustering result to avoid the artificial participation in the clustering center selection and better generation.Clustering results.