通过采用Spearman相关系数矩阵取代临时分类标记来构造标签相关性模块,提出一种改进的带Spearman相关性的多标签高斯随机域(MLQ-GRF)算法,以减少临时分类标记的不确定性.实验对比所得结果表明,文中构造的改进的MLQ-GRF算法对于扰动和带误差的临时分类标记有更好的稳定性,能提高分类的精确度.
An improved multi-label Gaussian random field algorithm is proposed to reduce the uncertainty of temporary labels.The spearman correlation matrix is used to build a label-relevant module instead of temporary labels.The results of comparative experiments show that the proposed algorithm is stable for temporary labels with tolerance and disturbance and it increases the accuracy of classification.