【摘要】 针对局部和全局一致性的分类算法LGC未考虑标签之间的相关性,提出了一种基于局部和全局一致性的多标签分类(MLGC)算法。该方法新增加了一个标签与标签之间的约束,在分类时考虑了标签之间的相关性,再取出1/10的数据集使用该算法,求出每个标签的自适应阈值,利用阈值对整个数据集进行预测。实验结果表明,所提出算法在Emotion和Yeast数据集上均优于原来算法,将此算法应用于区域医疗大数据的项目中,也取得了良好的分类结果。更多还原

【Abstract】 The multi-label classification didn’t take the relationship of the labels into consideration,so this paper improved LGC and proposed a multi-label classification based on the local and global consistency( MLGC). The method added a new constraint between labels,which consider the correlation between tags. Besides,this paper put forward a suitable threshold. Experimental results based on the emotion and yeast data sets shows this method is superior to LGC. Then the algorithm is applied to the project of medical,which also achieved great result.更多还原