Deep Learning-Based Density Clustering Simulation of Unstructured Big Data
Conventionally,traditional methods are time-consuming and prone to incorrect data density allocation,which affects the data clustering accuracy.Therefore,this paper proposed a fast density clustering method for non-structural big data based on deep learning.Firstly,the data density function was used to calculate all density values of unstructured big data.Secondly,the proximity search technology was adopted to find the best center of each cluster.Then,the Alex Net network was used to construct a learning framework for data clustering.Meanwhile,data feature vectors were extracted by mapping.Thirdly,pseudo labels were obtained by loss function as a basis for backpropaga-tion.In order to improve the clustering speed and accuracy of the model,small-lot gradient descent was used to opti-mize the model parameter,thus achieving the non-structural big data density clustering.Experimental results show that the proposed method can make the data with similar density integrate more closely with each other and make the data with large density differences sparse,so it has good data density clustering effect.
Deep learningNon-structural big dataData densityPseudo label
NETL
NSTL
万方数据
