用于多标记学习的局部顺序分类器链算法

标记间的相关性在分类问题中具有重要作用,目前有研究将标记相关性引入多标记学习,通过分类器链的形式将标记结果引入属性空间,为学习其他标记提供有用信息。分类器链中标记的预测顺序具有随机性,分类结果存在着很大的不确定性与不稳定性,且容易造成错误信息的传播。为此充分考虑标记的局部分布特性,提出了一种局部顺序分类器链算法,解决分类器链中分类器顺序问题。实验表明,该算法性能优于其他常用多标记学习算法。

Locally ordinal classifier chains for multi-label learning

The correlation among different labels plays an important role in classification problems, and recent studies have taken into account label correlation during multi-label learning. The label information is marked into the attribute space through the classifier chains and provides useful information for the other labels during the classification process. The classification results are indeterminate and instable because of the random classifier order in the classifier chain. Besides, it may cause to propagate the error label information. This paper fully considerd the local distribution of instance labels, and proposed a locally ordinal classifier chain algorithm. Experimental results show that, the new algorithm outperforms the other commonly used multi-label algorithms most of the time.