A soft neighborhood rough set model and its applications

Neighborhood rough set theory is widely used to measure the uncertainty of data in machine learning and data mining. However, the neighborhood radius has a significant influence on the effectiveness and robustness of the models and algorithms based on this theory. To address this problem, the soft-margin theory is introduced into neighborhood rough sets to define a soft neighborhood rough set model that can adaptively determine the appropriate neighborhood radius for each sample. The model effectively reduces the influence of the neighborhood radius on the uncertainty measure. Specific properties and theoretical analysis of the new model are presented. Based on soft neighborhood rough sets, this study presents a feature selection algorithm and classifier. The experimental results demonstrate that the designed algorithms exhibit acceptable performance, confirming that the soft neighborhood rough set model is feasible and robust.