基于模糊熵的支撑矢量预选取方法

在基于支撑矢量机的分类器学习算法中，预先选择支撑矢量是非常重要的．依据模糊熵理论，提出一种启发式的支撑矢量预选取方法——模糊熵方法．该方法针对支撑矢量数目较小的情况，可以有效地预选取出包含支撑矢量的边界集．利用边界集作为训练集可以大大简化支撑矢量机的训练而不影响分类性能．与其它方法相比，该方法的主要优点是不需要参数来确定边界集的阈值．仿真实验结果表明该方法是有效和可行的．

The Pre-selecting Method of Support Vectors Based on Fuzzy Entropy

For the support vector machine based learning algorithm of classifier, it is very importance for the support vector to be pre-selected. Based on fuzzy entropy theory, it proposes a new heuristic method for pre-selecting support vector. Under the circumstances that there are a little support vectors in training set, the new method can effectively pre-select the boundary subset which contain overwhelming majority support vectors. To substitute the boundary subset for training set, our method greatly reduces the training samples, while the ability of support vector machine to classification is unaffected. Comparing with other analogous methods, the merit of our method is that there are no parameters for determining the border of subset. The simulate results indicate that our approach is efficient and practical.