论文部分内容阅读
支持向量机中的参数直接影响其推广能力,针对参数选取的主观性,提出基于改进的遗传算法优化其参数,并将其应用于银行个人信用的五等级分类问题中,针对多分类问题,设计了3个二值分类器,不同分类的参数不同,通过实验证实可以达到更精细的分类效果.
According to the subjectivity of parameter selection, a new genetic algorithm based on improved genetic algorithm is proposed to optimize its parameters, which is applied to the five-level classification of bank personal credit. For the multi-classification problem, the design of the support vector machine The three binary classifiers, different parameters of different categories, confirmed by experiments can achieve more detailed classification results.