关于大数据网络中数据分类优化识别研究

Research on Data Classification Optimization and Recognition in Big Data Network

对大数据网络中数据进行分类优化识别,能提升数据管理的效率。数据处理中,难以保证数据样本集的纯度,降低了数据分类的准确率,需要对大数据网络中数据分类优化识别进行研究。传统方法建立增量式的分类模型对具有不确定性的概念漂移数据分类问题进行处理,具有较高的执行效率,但存在数据分类识别的准确率较差的问题。提出基于改进蛙跳的数据分类优化识别方法。采用基于流形的半监督算法,选择出有效数据特征子集,利用改进蛙跳数据分类方法对选择出的子集进行分类优化识别,通过混沌搜索数据分类的初始化解,进行变异操作生成新的数据个体,依据新的个体对青蛙位置进行更新,设计新的寻找方法,提高了算法的寻优能力,完成对大数据网络中数据分类优化识别。实验结果表明,利用所提方法能有效提高数据分类优化识别的准确率。

The classification optimization recognition for data in big data network can improve the efficiency of data management. In data processing, it is difficult to ensure the purity of data sample set, which reduces the accuracy of data classification. The incremental classification model established by traditional method has high execution effi- ciency, but it has the poor accuracy of data classification and recognition. Therefore, a method of data classification optimization recognition based on improved frog leaping was presented. This research used the semi - supervised al- gorithm based on manifold to select the effective data feature subset. Then, the research used the improved leapfrog data classification to classify, optimize and recognize selected subsets. Through the initial solution of chaotic search data classification, new data individuals were generated by mutation operation. Moreover, the frog position was updated based on new individuals. Finally, our search designed new search method to improve the optimizing ability of algorithm. Thus, the data classification and recognition in big data network was completed. Simulation results show that the proposed method can effectively improve the accuracy of classification optimization recognition of data.