Neighborhood rough sets based matrix approach for calculation of the approximations

Approximations of a concept in neighborhood rough sets (NRS) are used to induce rules and extract feature subset from hybrid information systems. To compute the approximations conveniently, this paper presents a new matrix view of NRS and the matrix representation of the lower and upper approximations in NRS. To be robust for tolerating the noisy samples in the data, a probabilistic neighborhood rough sets (PNRS) is proposed. According to the matrix view, a new method as well as its algorithm is presented for calculation of the lower and upper approximations under NRS and PNRS.