Multi-label feature selection based on fuzzy neighborhood rough sets

Abstract Multi-label feature selection, a crucial preprocessing step for multi-label classification, has been widely applied to data mining, artificial intelligence and other fields. However, most of the existing selection methods dealing with mixed have following problems: (1) These rarely consider importance features from multiple perspectives, which analyzes not comprehensive enough. (2) select subsets according positive region, while ignoring uncertainty implied by upper approximation. To address these problems, method based on fuzzy neighborhood rough set is developed in this article. First, approximation accuracy decision are defined model, new conditional entropy designed. Second, measure proposed combining information view approximate algebra view, evaluate different views. Finally, forward algorithm removing redundant decrease complexity classification. The experimental results illustrate validity stability systems, when compared related ten datasets.