The Research on Model of Group Behavior Based on Mobile Network Mining and High-Speed Data Streams
High-speed data stream is a data flow velocity exceeds the processing power of integrated classifier; integrated classifier training can not reach all the most recent data to update the classification model. To this end, this chapter introduces the optimal Bayesian classification theory, and its integration on the basis of analysis of the expected classification error of the bias variance decomposition, and finally presents a sampling bias based on an integrated high-speed data stream classification algorithm (Ensemble Classifiers Algorithm for Classify High Speed Data Stream based of Biased Sample, CDSBS), theoretical analysis is the experimental verification show that the algorithm can effectively reduce the integrated classifier training update at the same time, the classification remains a high classification performance.
KeywordsGroup Behavior Mobile Network Mining Data Streams
Unable to display preview. Download preview PDF.
- 2.Vitter J S,Wang M, Lyer B.Data cube approximation and Histograms via wavelets. Proceeding of CIKM, 1998 Google Scholar
- 3.Vitter, J.S., Wang, M.: Approxlinate computation of multidimensional aggregates of sparse data using wavelets. In: Proceeding of the 2002 ACM-SIGMOD international conference Management of Data (1999)Google Scholar
- 5.Schapire, R.E.: The strength of weak learnability. Machine Learning 5(2), 197–227 (1990)Google Scholar
- 10.Wang, L., Sugiyama, M., Yang, C., et al.: On the margin explanation of boosting algorithms. In: 21st Annual Conference on Learning Theory(COLT) (2008)Google Scholar
- 11.Opitz, D.W.: Feature selection for ensembles. In: Proceedings of the sixteenth international conference on Artificial intelligence and the eleventh Innovative applications of artificial intelligence conference innovative applications of artificial intelligence, pp. 379–384 (1999)Google Scholar