Abstract: In the neighborhood rough set model,with the increasing of the size of information granules,the majority vo-ting rule based neighborhood classifier (NC) is easy to misjudge the classes of unknown samples.To remedy this deficiency,based on the idea of collaborative representation based classification (CRC),we proposed a neighborhood colla-borative representation based classification method,namely,the neighborhood collaborative classifier (NCC).NCC firstly performs feature selection in the classification learning task with neighborhood rough set model,and then finds the neighborhood space of unknown sample under selected features.Finally,instead of the majority voting rule in the neighborhood space,NCC judges the class of unknown sample with the collaborative representation,which considers the class with the minimal reconstruction error for unknown sample as the predicted category.Experimental results on 4 UCI data sets show that compared with NC,the proposed NCC achieves satisfactory performance in larger information granules and compared with CRC,and the proposed NCC greatly reduces the size of the dictionary while maintaining good classification accuracy,and improves the efficiency of classification.

Key words: Classification,Collaborative representation,Feature selection,Neighborhood,Rough set