摘要
训练数据集的可分性判别是机器学习领域的一个重要研究内容。本文针对该问题,提出了一种基于线性判别分析的数据集可分性判别算法。该算法首先对待判断的训练数据集进行线性判别分析,求得最佳投影直线,然后将原始样本投影到最佳投影直线上,根据投影点的位置关系判断训练数据集的可分性。在人造数据集实例上的检测结果,充分证明了本文判定算法的正确性。
The separability discrimination of training data set is an important research in machine learning field.For this question,in this paper,we proposed a data set separability discriminant algorithm based on linear discriminant analysis.The algorithm firstly performed linear discriminant analysis on training data set,and obtained the best projection line.Then project the original samples onto the best projection line,and judge the separability of the training set according to the distribution of projections.The test results of artifical data set examples fully verify the correctness of the proposed algorithm.
出处
《科技通报》
北大核心
2013年第4期31-32,35,共3页
Bulletin of Science and Technology
关键词
可分性
判别算法
线性判别分析
投影直线
separability
discriminant algorithm
linear discriminant analysis
projection line
