关键词: 粗糙集,邻域,属性约简,块集,快速算法

Abstract: Calculating each record’s δ-neighborhood elements is the most frequent and complex step in neighborhood rough set model.In this paper,we proposed the concept of the block sets according to the distribution of records in the space,then proved that each record’s δ-neighborhood elements can only be contained in its own block set and its adjacent block sets.Based on the block-set-neighborhood theory,we presented a quick attribute reduct algorithm on neighborhood rough set,which can reduce the complexity of calculating each record’s δ-neighborhood elements.Moreover,the algorithm’s validity was verified by several data sets from UCI.Expermental results show that our algorithm is effective and feasible.

Key words: Rough set,Neighborhood,Attribute reduct,Block set,Efficient algorithm