大数据子集特征遗忘启发价值信息提取方法  

Method of Value Information Extraction Based on Big Data Subset Features Forgotten and Inspired

作　　者：向春枝[1] 

机构地区：[1]河南广播电视大学开放教育教务处,郑州450008

出　　处：《科技通报》2014年第10期160-162,共3页Bulletin of Science and Technology

基　　金：河南省教育厅2012年度河南省高等学校青年骨干教师资助计划项目(教高〔2012〕626号);河南省教育厅2013年度科学技术研究重点项目资助计划项目(13A520208;教科技〔2013〕68号)

摘　　要：传统的大数据中价值信息提取方法采用基于模糊学习理论的数据融合处理方法,将预定学习序列输入神经网络,通过模糊启发,对预定序列进行多模型映射,此方法模型复杂,且启发率低。提出一种大数据子集特征遗忘启发的价值信息提取方法,对大数据进行非线性映射归一化,使每个子集实现并行运算,通过混沌方法提取子集特征,并建立混沌模型下的子集特征遗忘启发链,针对不同子集中的价值信息,依据遗忘启发链实现启发,提取价值信息。采用一组大数据下的伪随机价值信息进行提取测试,仿真实验表明,本文价值信息提取方法的提取率达到了98%,对于大数据下的价值信息提取具有很好的指导意义。The fuzzy approach learning theory data fusion method is used in the traditional big data value information extrac-tion, the method is based on a predetermined sequence of inputs to the neural network learning, the predetermined se-quence for multi-model mapping is inspired by fuzzy method, this model is complex and the inspiring rate is low. A value information extraction method of big data is proposed based on subset features forgotten and inspired, each subset is done with parallel computing, a subset of feature extraction method is used with chaos, sub-forgotten inspired collection features the chain is established under chaos model, for different subset of the value of information, with forgetting inspired chain, the value of information is extracted. A large data with pseudo-random distribution is used for test, the simulation experi-ments show that the with the proposed method, the value of information extraction rate reaches 98%, it has good guidance for the value of information extracting under big data.

关 键 词：大数据 子集特征 遗忘启发 价值信息提取 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]