【作者】 张健沛； 程丽丽； 杨静； 马骏；

【Author】 Zhang Jianpei,Cheng Lili,Yang Jing,and Ma Jun (College of Computer Science and Technology,Harbin Engineering University,Harbin 150001)

【机构】 哈尔滨工程大学计算机科学与技术学院；

【摘要】 提出了一种基于人工鱼群优化算法的支持向量机集成模型(AFSASVM).在独立训练出一批个体支持向量机后,利用人工鱼群优化算法对集成中个体支持向量机的权值进行优化,选择权值大于某一阈值的部分个体支持向量机参与集成,实现一种基于选择性集成思想的支持向量机集成模型.在标准UCI数据集合和StatLog项目集合上的仿真对比实验表明,该方法可以得到更好的集成性能,显示了AFSA在多分类器集成权值优化方面的有效性,同时在运行效率上AFSA也具有明显的优势.更多还原

【Abstract】 Ensemble learning has become a hot topic in the machine learning recently.The generalization performance of ensemble classification systems has been improved dramatically by training and combining some accurate and diverse classifiers.The model of support vector machine (SVM) ensemble based on artificial fish_swarm algorithm(AFSA) is proposed after an analyzing the drawbacks of the known algorithms such as GASEN and CLU_ENN.The AFSA is used to optimize the ensemble weights of base SVMs for SVM ensemble after an individual SVM of ensemble is trained.Those SVMs with weights larger than a given threshold value are ensembled.The method of selective ensemble is achieved to obtain better performance than traditional ones that ensemble all of the base SVMs.The simulated experiment results on UCI and StatLog show that the proposed method has better performance and the AFSA has its superiority in optimizing weights of SVM ensembles,and also in operation efficiency.更多还原