Calculation Method for Semi-Monolayer Covering Approximation Sets Fushing GPU
A semi-monolayer covering rough set is a high-quality and efficient rough-set model for matching set-valued information systems.There are a large number of computationally intensive and simple logical operations in the calculation process of a semi-monolayer covering approximation set.Therefore,this study proposes a matrix-based algorithm for semi-monolayer covering approximation sets to utilize the powerful computing performance of Graphics Processing Unit(GPU)to accelerate the calculation process.In order to achieve this goal,Boolean matrices are used to represent the elements in the semi-monolayer covering approximate sets.Boolean matrix operators corresponding to set operations are introduced,and a matrix representation of the semi-monolayer covering rough approximation set(DE,DA,DE0,and DA0)is proposed.A matrix based semi-monolayer covering approximation set algorithm(M_SMC)is designed.The corresponding theorems prove the equivalence between the matrix representation and the set representation of a semi-monolayer covering approximate sets.However,there is an issue with excessive memory consumption in the matrix storage and calculation steps during the operation of M_SMC.To deploy the algorithm on GPU with limited graphics memory and to optimize the matrix storage and calculation steps,a batch processing matrixed semi-monolayer covering approximation set algorithm(BM_SMC)is proposed.The experimental results on 10 datasets show that the BM_SMC algorithm fused with the GPU improves the computational efficiency by 2.16-11.3 times compared with the BM_SMC algorithm using only the CPU.The BM_SMC algorithm can fully utilize the GPU under limited storage-space conditions and effectively improve the computational efficiency of semi-monolayer coverage approximation sets.
semi-monolayer covering approximation setsSet-Valued Information Systems(SVIS)matrizationGraphics Processing Unit(GPU)accelerationbatch processing
国家科技期刊平台
NSTL
万方数据
