基于大数据的时序数据分类模型鲁棒性验证
Robustness Verification of Temporal Data Classification Model Based on Big Data
李艳1
作者信息
- 1. 朔州师范高等专科学校,山西 朔州 036000
- 折叠
摘要
不可逆的时序数据是物联网产业的重要组成部分.但如何识别来源复杂且多样的时序数据,并对其进行分类,保障计算模型的顺畅运转是其中的难点.针对此问题,研究提出了基于大数据的时序数据分类模型,采用局部投影算法增强并识别获取的时序数据关键特征,使用动态主成分分析方法,联合麻雀搜索算法构建数据分类模型.研究结果表明:模型的FPR值平均比其他算法高出14.38％,在多种噪声干扰下,模型的FpR值下降幅度较其他算法更小,鲁棒性明显优于其他算法.
Abstract
Irreversible timing data is an important part of the Internet of Things industry.How-ever,it is difficult to identify and classify the time series data from complex and diverse sources to ensure the smooth operation of the calculation model.To solve this problem,a time-series data classification model based on big data is proposed.The local projection algorithm is used to en-hance and identify the key features of the time-series data obtained.The dynamic principal com-ponent analysis method and the sparrow search algorithm are used to construct the data classifica-tion model.The results show that the FPR value of the model is 14.38％higher than that of other algorithms on average.Under various noise interference,the FPR value of the model de-creases less than that of other algorithms,and its robustness is obviously better than that of other algorithms.
关键词
大数据/时间序列数据/局部投影法/动态主成分分析/麻雀搜索算法Key words
Big data/Time series data/Local projection method/Dynamic principal component analysis/Sparrow search algorithm引用本文复制引用
出版年
2024
NETL
NSTL
万方数据