提出了基于结构风险上界的SVM参数选择方法。首先，从理论上分析了SVM结构风险上界的计算方法，给出了结构风险上界的算法步骤；其次，以结构风险上界作为SVM泛化性评价准则对5个UCI公开数据库和经过实测建立的两个特征库（包括二类和多类数据）进行了参数选择仿真实验，并与5-折交叉验证的实验结果进行了比较，结果表明，基于结构风险上界的SVM参数选择方法有效、省时。
Support Vector Machine (SVM) is an intelligent technology for classification problems. Because of its flexibility, computational efficiency and capacity to handle high dimensional data, SVM has become a popular research issue in recent years. Selection of optimal parameters is important for an SVM. The traditional methods, such as the k-fold cross validation, can select optimal parameters, but would take too much time. In this paper, a method of SVM parameter selection based on the bound of structure risk is proposed. First, the bound of the structure risk is theoretically analyzed. Then, the simulated experiments with several datasets are designed. Comparisons are made between the proposed method and the method based on the 5-folds cross validation. The results show that the proposed method is effective and takes less time, and it would be very suitable for target recognition problems.
