基于加权深度森林算法的公安敏感数据流动态挖掘研究
Research on Dynamic Mining of Public Security Sensitive Data Flow Based on Weighted Deep Forest
陈予雯1
作者信息
- 1. 河南警察学院 网络安全系,河南 郑州 450046
- 折叠
摘要
公安敏感数据流隐蔽性与复杂性较高,数据挖掘过程中难以对这些数据进行分类,导致数据挖掘质量下降.为解决这一问题,提出基于加权深度森林算法的公安敏感数据流动态挖掘方法.该方法利用本地化差分隐私技术,采集公安部门终端应用的敏感数据流,再根据最大类间散度获取敏感数据流向量,通过计算最佳散度提取公安敏感数据流可挖掘特征,然后结合可挖掘特征使用加权深度森林算法求解敏感数据流类别密度,引入孤立因子对数据流进行动态分类,实现公安敏感数据流动态挖掘,实际应用效果较好.
Abstract
The public security sensitive data flow exhibit high levels of concealment and complexity,making it dif-ficult to classify these data in the data mining process,resulting in the decline of data mining quality.Therefore,a dynamic mining method of public security sensitive data flow based on weighted depth forest is proposed.The local-ized differential privacy technology is employed to collect the privacy sensitive data flow of the public security depart-ment applications.After the sensitive data flow vector is obtained according to the maximum inter-class divergence,the mining features of public security sensitive data flow are extracted by calculating the best divergence.The dynamic mining of public security sensitive data flow is realized by calculating the category density of data flow with the mining features and the weighted depth forest combined,and dynamically classifying the data flow based on isolation factors.The proposed method proves to be effective in practical application.
关键词
加权深度森林/敏感数据流/动态挖掘/差分隐私/可挖掘特征/孤立因子Key words
weighted depth forest/sensitive data flow/dynamic mining/differential privacy/mining features/isolation factors引用本文复制引用
基金项目
河南省研究生教育改革与质量提升工程项目(YJS2023AL099)
出版年
2024
NETL
NSTL
万方数据