基于熵和蚁群聚类算法的模糊支持向量机

摘 要: 模糊支持向量机(FSVM)对传统支持向量机(SVM)在对外围点和噪声数据敏感的缺陷做了重要改进。选取合适的聚类中心计算符合数据本身特征分布的隶属度，能使分类更加准确，提高测试精度。论文基于模糊支持向量机思想，提出一种新的模糊聚类模型—基于熵和蚁群聚类算法的模糊支持向量机(EAFSVM)，为聚类中心和隶属度的计算提出了新方法。实验对比传统SVM和FSVM，结果表明EAFSVM测试精度较高，尤其对多类数据、大规模数据具有较好的分类能力。

Fuzzy SVM with entropy membership and ant colony optimization

Abstract: Compared with traditional support vector machine (SVM), Fuzzy support vector machine (FSVM) made a great improvement in sensitive to outliers or noises. By selecting a proper clustering center to calculate membership, FSVM could obtain higher precision optimization hyperplane. Based on FSVM, through improving clustering center and membership, we propose a new fuzzy clustering method—Fuzzy SVM with Entropy Membership and Ant Colony Optimization(EAFSVM). Experiments show EAFSVM has a better precision and classification performance, especially to multi-class and large scale dates.