In SVM algorithm

when training sets with uneven class sizes are used

the prediction result of classifier is undesirably biased towards the class with more samples in the training set.That is to say

the larger the sample size

the smaller the classification error

whereas the smaller the sample size

the larger the classification error.Aiming at this orientation problem and with the analysis of the cause of it

an improved method based on genetic crossover operator was proposed

for the training set with small size generate new samples by using crossover operation

thereby compensates for the unfavorable impact caused by the bias of the training data class size.Simulation experiment results on UCI stander data shows that the proposed method has better classification accurate compared with stander support vector machine method.