Research on Unbalanced Data Classification Algorithm for Complex Network Data Streams

Yingxue Cai,  Jia Chen, Song Hu, Hui Hu, Sibo Huang, Zhaoquan Cai*

Huizhou University, Huizhou, 516000, China

Abstract: In order to improve the ability of fast processing and recognition of unbalanced data in complex network data flow, it needs to carry out fast classification and analysis of data. A fast classification algorithm based on association mining and fuzzy C-means clustering is proposed for unbalanced data flow in complex networks. Non-equilibrium data flow model of complex network data flow is constructed by nonlinear time series analysis method, and the delay scale characteristic parameters of non-equilibrium data flow in complex network data flow are extracted. Taking the extracted association rules as the feature quantity, the data clustering is processed, the fuzzy C-means classification algorithm is used to realize the optimal classification of the non-equilibrium data in the complex network data flow, and the convergence control of the classification center is carried out with the differential evolution method. The global convergence of the classification process is improved. The simulation results show that the proposed method has good accuracy and low error rate in the classification of unbalanced data in complex network data streams.

Keywords: Complex network; Data; Fuzzy C-means; Clustering; Classification