摘要
为得到可靠的仿真数据,提出一种基于数据符号化表示和云模型的时序数据生成方法。首先用SFVS算法将原数据或是数据的先验知识(很多时候仅能获得相关领域的一些知识而非数据)表达为一个符号矢量,然后用定性定量转换工具——云模型利用符号矢量产生相应的时序数据。仿真实验表明,该方法产生的数据具有与原数据一样的结构特征、知识蕴涵,并具有可控的随机性、复杂性。
In order to generate reliable synthetic datasets, this paper proposed a method based on the symbolic representation of the original data and cloud model. Firstly,the original data or the knowledge about the data（ we often can only get the knowledge of the domain）was symbolic represented with the SFVS algorithm, and then generated synthetic dataset by using the symbolic vector as input by the cloud model. Simulation experiments demonstrate that the characters and the connotative knowledge of the synthetic data set are the same as the original data＇ s, moreover, the randomicity as well as the complexity of the synthetic data set is controllable.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3691-3693,3697,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(60872113)
关键词
时序数据
符号化表示
符号化统计特征矢量
云模型
云发生器
time series data
symbolic representation
statistical feature vector symbolic （ SFVS ）
cloud model
cloud generator
