Abstract: From the point of view of pattern recognition, the paper analyzes t he outlier detection method and points out that the method may become have to re alize with the increase ment of samples. In order to reduce the size of optimiza tion, the approach of outlier detection based on support objects is proposed. By the new method, the operation size can be reduced greatly, and the demand of r eal-time can be met.

摘要： 从模式识别的角度出发,分析了基于核的野点检测方法,指出随着样本数目的增 加,该方法会因为过大的优化规模而无法实际操作,为此提出了基于支持对象的野点检测方 法,该方法大大降低了运算规模和内存需求,保证了野点检测实时性的要求?

关键词: 野点检测, 支持对象, 核方法

CLC Number: 

• TP311