Study On Sample Selection By Using SFL Algorithm

SFL is one of the classical algorithms. In this paper, a new Instances Selection algorithm based on shuffled frog leaping shortly named SFLSS-1 and SFLSS-2 are presented. In this algorithm, each sample can be regard as a gene of a frog and the condensed subset can be regard as a frog. The frog was constantly being updated in each iteration. Finally, the maximum value of the frog which is obtained by the SFL is the best condensed subset. Compared to traditional methods, such as MCS and GA Sample Selection, the condensed sets obtained by SFLSS-1 and SFLSS-2 are superior in storage and classification accuracy.