Parallelization Algorithm for the Calculation of Typical Testors Based on YYC

In the present work, a new method is proposed to find typical testors, which helps reduce the number of features needed to carry out classification processes. Using the strategy of divide and conquer, we divide a basic matrix into blocks, then we find the typical testors of the defined blocks. To find the typical testors of the complete basic matrix, unions between the elements of these sets obtained from the blocks are tested. A criterion is developed to determine when the unions of typical testors of blocks form typical testors of the complete matrix. The performance of the method is evaluated using synthetic matrices. The execution time of the method in parallel and sequential versions was compared and contrasted with the YYC algorithm used for the complete basic matrix. Finally, its performance is analyzed in a real database obtained from the UCI Repository.