In order to improve precision of data classification of kernel-based extreme learning machine（ KELM）

propose KELM classification parameter optimization method

GA-KELM

which combines K-fold cross-validation（K-CV） and genetic algorithms（GA）

the average precision of multiple models of resulting of CV training as GA fitness evaluation function value

provide evaluation criteria for parameter optimization of KELM

and then the KELM algorithm is used to get the optimization parameters of GA for data classification. Using UCI dataset for simulation

results show that the proposed method is superior to GA-SVM and GA-BP algorithm on the overall performance

with a higher classification precision.