吉林大学学报(信息科学版) ›› 2020, Vol. 38 ›› Issue (4): 457-466.

• • 上一篇    下一篇

量子计算在增量式大数据并行挖掘中的应用

李晓峰1,王妍玮2,李东3   

  1. 1. 黑龙江外国语学院信息工程系,哈尔滨150025; 2. 普度大学机械工程系,印第安纳州西拉法叶市IN47906; 3. 哈尔滨工业大学计算机科学与技术学院,哈尔滨150001
  • 收稿日期:2019-10-17 出版日期:2020-07-24 发布日期:2020-08-13
  • 作者简介:李晓峰( 1978— ) ,男,哈尔滨人,黑龙江外国语学院教授,博士,主要从事人工智能、机器学习和智慧医疗等研究,( Tel) 86-451-88121567( E-mail) lixiaofeng@ hiu. net. cn。
  • 基金资助:
    国家自然科学基金资助项目( 61803117) ; 教育部科技发展中心产学研创新基金资助项目( 2018A01002) ; 国家科技部创新方法专项基金资助项目( 2017IM010500)

Application of Quantum Computing in Incremental Parallel Mining of Large Data

LI Xiaofeng1,WANG Yanwei2,LI Dong3   

  1. 1. Department of Information Engineering,Heilongjiang International University,Harbin 150025,China;
    2. Department of Mechanical Engineering,Purdue University,West Lafayette,Indianan IN47906,US;
    3. School of Computer Science and Technology,Harbin Institute of Technology,Harbin 150001,China
  • Received:2019-10-17 Online:2020-07-24 Published:2020-08-13

摘要: 针对传统大数据并行挖掘方法是一次性对所有数据进行挖掘,导致挖掘时间较长,挖掘精度较低等问
题,采用量子计算对增量式大数据并行挖掘方法进行优化设计。首先,按照数据挖掘的基本流程搭建并行数据
挖掘模型; 然后分别通过定义量子比特、量子搜索算法、量子神经网络处理以及量子映射变换4 个步骤,实现
增量式数据的预处理,利用矩阵向量相乘分解得到过滤权重组合,通过该组合实现预处理结果的并行协同过
滤; 最后通过量子模糊聚类得出增量式大数据并行挖掘结果。实验结果表明,应用量子计算的增量式大数据并
行挖掘方法的平均召回率为97. 25%,并行挖掘时间在2. 1 ~ 3. 2 s 的范围内浮动,准确率超过95%,且该方法
的收敛性最好,寻优能力强。

关键词: 量子计算, 增量式, 大数据, 并行挖掘

Abstract: In view of the problem that the traditional big data parallel mining method always mines all the data at
one time,resulting in a long mining time and low mining accuracy,the quantum computing is adopted to
optimize the incremental big data parallel mining method. Firstly,the parallel data mining model is built
according to the basic process of data mining. Then on the mining model respectively by defining a quantum bit,
quantum search algorithm,quantum neural network processing and mapping transformation,the incremental data
preprocessing,filtering weights are obtained by decomposition of matrix-vector multiplication,preprocessing
results by using the combination of parallel collaborative filtering. Finally,by quantum fuzzy clustering,large
incremental data parallel mining results are obtained. The experimental results show that the average recall rate
of the incremental big data parallel mining method using quantum computing is 97. 25%,the parallel mining
time is within the range of 2. 1 ~ 3. 2 s,and the accuracy rate is always above 95%. And this method has the
best convergence and strong optimization ability.

Key words: quantum computing, incremental, big data, parallel mining

中图分类号: 

  • TP391