| 2. | This thesis aims to discuss the clustering techniques with the background of large - scale nuclear physics science data mining . first , we introduce the key techniques and the main task in data mining , then we analyze the data preprocessing techniques and clustering techniques combine data mining techniques with science data . from data preprocessing aspect , we propose some methods of segmenting , denoising , integrating and transforming , and we use “ truncation method ” and “ successive difference method ” in data reduction , at last we extract information from the science data
论文基于大规模核物理科学数据挖掘的背景,全面介绍了数据挖掘的关键技术和主要任务,从理论、算法和应用三个层次,结合科学数据的特点来分析预处理技术和聚类方法,提出了很多实用的预处理方法:对hdf5科学数据进行分块、除噪、集成、变换等,同时对它使用“截断法”和“逐层求差法”进行规约,并对数据进行信息提取。 |
