基于最小包含球的领域自适应算法

Field Adaptive Algorithm based on Minimum Enclosing Ball

相同应用领域因不同时间、地点或设备,检测到的数据域可能会出现不完全一致的现象,从而可能导致机器学习效率降低。为有效地进行数据域间知识传递,在原有支持向量域描述(SVDD)算法的基础上,提出一种全新的数据域中心点校正的领域自适应算法,并使用人造数据集和KDD CUP99大数据集验证算法。实验证明,该领域自适应算法效果较好,将其应用于大数据集可减少核心集元素个数,提高运算效率。

Due to different time, places or equipments, data domain detected by the same application field may not be consistent completely, which may lead to lowering in machine learning efficiency. To effectively transmit knowledge between data domains, a new field adaptive algorithm of date center correction is presented based on the original support vector domain description （SVDD） algorithm and the artificial data sets and KDD CUP99 large data collection validation algorithm are applied. It is shown that its application in the large data collection can reduce the number of elements in the core set and improve the efficiency.