Research On Density Peak Clustering Algorithm Based On Fuzzy Set

There are a lot of fuzzy phenomena in the real world,which can be represented by fuzzy sets.In the era of big data,the scale of data is getting bigger and bigger.Cluster analysis has become a very important task in data mining.Density peak clustering algorithm is a clustering algorithm that can quickly identify clustering centers and distribute data.It has a simple principle and is easy to implement.It has good clustering effect on large-scale data sets,so it is widely used in various fields.At present,the traditional methods with high complexity are used for fuzzy set clustering,which is difficult to realize the clustering analysis of large-scale fuzzy dataset.At the same time,there are some defects in the noise detection strategy of density peak clustering algorithm,which reduces the clustering accuracy.Based on the above problems,this paper studies in the following aspects:In the intuitionistic fuzzy environment,an improved weighted intuitionistic fuzzy distance operator based on Canberra distance is proposed,which overcomes the problem that traditional distance operators are sensitive to large singular values.Then the operator is used to calculate the weight of each attribute,which reduces the subjectivity of determining the attribute weight manually.Finally,the density peak clustering algorithm is used to cluster,which reduces the complexity of the algorithm.The experimental results show that the algorithm reduces the running time of intuitionistic fuzzy clustering and improves the clustering accuracy.In the hesitant fuzzy environment,an improved weighted hesitant fuzzy distance operator is proposed based on Euclidean distance and Hamming distance.It does not need to complete and order the membership degree elements during distance calculation,which reduces the subjectivity of artificial complement elements and the complexity of calculation.Then the weight calculation formula is used to calculate the weight,and finally the density peak clustering algorithm is used to cluster.Experimental results show that the algorithm reduces the running time of the hesitant fuzzy clustering algorithm and improves the clustering accuracy.Aiming at the defects that the noise detection strategy of density peak clustering algorithm cannot accurately identify some sample points and is highly sensitive to parameter selection,a density peak clustering algorithm based on sinusoidal fuzzy entropy was proposed.Firstly,a sinusoidal fuzzy entropy is constructed,which overcomes the counterintuitive problem of traditional fuzzy entropy.Then,it is used to determine the classification criteria of three-way clustering.Finally,the optimal thresholds are determined by genetic algorithm,and the core region,fringe region and trivial region are divided under the three-way clustering framework.The points in trivial region are recognized as noise points.Experimental results show that the algorithm can identify noise points more accurately,improve the clustering accuracy,and reduce the sensitivity of algorithm results to parameter.The above researches explore the application of density peak clustering algorithm in fuzzy environment,and analyze the improvement of density peak clustering algorithm by using fuzzy theory,which better solves the problems of low precision and high complexity of fuzzy clustering,and improves the noise detection ability of the algorithm.Experimental results verify the effectiveness and feasibility of the proposed algorithm,which has certain research significance.