多标签特征选择研究进展

特征选择一直是机器学习和数据挖掘中的一个重要问题。在多标签学习任务中，数据集中的每个样本都与多个标签相关联，标签与标签之间通常也是相关的。在多标签高维数据分析中，为降低特征维数和提高分类性能，研究者们提出了多标签特征选择方法。系统综述了多标签特征选择的研究进展。在介绍多标签分类以及评价准则之后，详细分析了多标签特征选择的三类方法，即过滤式算法、包裹式算法和嵌入式算法，对多标签特征选择未来的研究提出展望。

Research Progress of Multi-Label Feature Selection

Feature selection has always been an important issue in machine learning and data mining. In multi-label learning tasks, each sample in the multi-label dataset is associated with multiple labels and different labels are also usually related. In multi-label high-dimensional data analysis, multi-label feature selection methods are proposed to reduce feature dimension and improve classification performance. This paper reviews the research progress of multi-label feature selection. After introducing multi-label classification and evaluation criteria, three kinds of multi-label feature selection approaches are analyzed in detail, namely, filter, wrapper, and embedded algorithms. Finally, the future research of multi-label feature selection is prospected.