SVM parameters optimization based on artificial bee colony algorithm and its application in handwriting verification

In order to overcome the defect of falling into local optimal solution which all the common SVM parameters optimization methods had in different degree, a new SVM parameters optimization method based on artificial bee colony algorithm was proposed and applied to handwriting verification. Penalty factor C and kernel function parameter of SVM were taken as the optimization object, and classification accuracy of SVM was used as fitness value. Then the artificial bee colony algorithm was adopted in this work to achieve the global optimal solution of parameter C and . The proposed method was tested on four UCI standard datasets and compared with genetic algorithm and other conventional optimization algorithms. It was indicated from the result that the proposed method overcame the local optimal solution problem and acquire higher classification accuracy. The cost time of searching optimized parameters of small number classification problem was also reduced. Then the proposed method was applied to handwriting verification. Mean and variance of high frequency wavelet coefficient matrixes of handwriting images were taken as the classification feature. At last, the proposed method was used in handwriting verification and high classification precision was acquired.