一种改进的局部支持向量机算法

An improved algorithm of local support vector machine

局部支持向量机是一种用途广泛的分类器,无论在理论研究还是实际应用方面,局部支持向量机都受到越来越多的关注。目前,许多传统的局部支持向量机算法都存在一个问题,即模型中样本比例失衡,导致无法提高分类精度。在加权支持向量机的启发下,提出了将加权思想应用在局部支持向量机Falk-SVM中的WFalk-SVM算法,并通过实验分析验证了WFalk-SVM的可行性及其有效性,最后对WFalk-SVM算法进行分析总结。

Local support vector machine is a widely used classifier. It has been attracting more and more attention both on its theoretical research and practical applications. Nowadays, there exists a problem in many traditional local support vector machine algorithms： the imbalance of the number of the samples leads to the difficulty in improving the classification accuracy. In this paper, firstly, under the inspiration of weighted support vector machine, the algorithm named WFalkSVM is proposed, which uses weight in FalkSVM. Secondly, the experiments demonstrates the feasibility and effectiveness. At last, we conclude the weighted FalkSVM.