Computers & Mathematics with Applications
| Hi-index | 0.01 |
Classification experiments are made with neural network algorithm and support vector machine method separately. The samples are divided into three groups and two kinds of support vector machines based on polynomial kernel and radial basis function are applied by changing the parameter values. The simulated results show that, as for the dataset with less training samples, using simple structure learning function will avoid the over fitting problem. In contrast, the learning function with slightly simple structure will reduce the generalization ability. In the experiment, the Penalty factor C is introduced in order to allow the training samples to be classified wrongly. Increasing the value of C , generalization ability of the learning machine can be improved. Using cross-validation method to choose parameter values can improve the classification accuracy. The experimental results show that the support vector machine method is superior to the neural network algorithm.