基于支持向量机的无穷维AdaBoost算法及其应用.pdfVIP

学兔兔 第31卷 第4期 仪 器 仪 表 学 报 Vo1．31 No．4 2010年4月 Chinese Joumal of Scientific Instrument Apr．2010 基于支持向量机的无穷维AdaBoost算法及其应用：I= 刘 冲，张均东，曾 鸿，任 光，纪玉龙 (大连海事大学 大连 116026) 摘 要：AdaBoost算法能够集成比随机猜测略好的弱分类器，输出较高分类精度的强分类器。为了进一步提高AdaBoost算法 的分类精度，建立了一种基于支持向量机的无穷维AdaBoost算法，实现无穷维AdaBoost算法的关键是建立一个新的支持向量 机核函数，使此核函数集成无穷多个AdaBoost算法弱分类器。将无穷维AdaBoost算法用于模拟电路故障诊断，故障诊断结果 表明：无穷维AdaBoost算法分类精度优于有限维AdaBoost算法，提高了AdaBoost算法的分类精度。 关键词：AdaBoost算法；支持向量机；核函数；超平面；分类精度 中图分类号：TP181 文献标识码：A 国家标准学科分类代码：520．2040 Infinite dimension AdaBoost algorithm based on support vector machine and its application Liu Chong，Zhang Jundong，Zeng Hong，Ren Guang，Ji Yulong (Marine Engineering College，Dalian Maritime University，Dalian 116026，China) Abstract：AdaBoost algorithm can achieve better classification ability than that of weak classifier．An infinite Ada- Boost algorithm based on support vector machine is proposed to apply to the classification problem，and the classifi— cation ability of the algorithm is further improved．The key improvement of the algorithm is that an infinite number of hypotheses are embedded into a new kernel function of suppo~vector machine．Infinite AdaBoost weak classifiers are integrated into the new kernel function．The new algorithm was applied to the fault diagnosis problem of analog cir- cuits during designing stage．Experiments were conducted using the algorithm．Experiment results show that the infi— nite AdaBoost algorithm based on suppo~vector machine is superior to finite AdaBoost algorithm．Th e classification accuracy of AdaBoost algorithm is improved． Key words：AdaBoost