一种贝叶斯证据框架下支持向量机建模方法的研究_阎威武.pdfVIP

第19 卷第5 期 　控　制　与　决　策 2004 年5 月 Vol. 19 No . 5 　Control and Decision　 　 May 2004 : 1001-0920(2004) 05-0525-04 阎威武, 常俊林, 邵惠鹤 ( , 200030) : 研究贝叶斯证据框架下标准支持向量机和最小二乘支持向量机的估计 法和建模方法, 给出一种高斯核支 持向量机估计 法的参数选择和调整方法. 将贝叶斯证据框架下支持向量机的建模方法应用于非线性系统的辨识, 仿真结果表明, 该方法对于工业过程建模是十分有效的. : 支持向量机; 贝叶斯; 建模; 证据框架 : T P274: A Modeling method based on support vector machines within the Bayesian evidence framework - , - , - YA N Wei wu CH AN G Jun lin SHA O H ui he ( Departme t of Automatio , Sha ghai Jiaoto g U iversity , Sha ghai 200030, Chi a. Correspo de t: YAN Wei- , - : @ . ) wu E mail ya ww sjtu hotmail com Abstract: Regressio algorithms a d modeli g met hod based o st a dard support vector machi e ( Sta dard SVM ) a d least squares support vector machi e ( LS SVM ) w ithi the Bayesia evide ce framew ork are discussed. A parameter selecti g a d tu i g method withi t he Bayesia evide ce framework is proposed for SVM regressio w ith Gauss ker el. SVM modeli g m ethod w it hi t he Bayesia evide ce framework is applied to o li ear system ide tificatio , a d effective results are obtai ed. : ; ; , Key words support vector machi e bayesia modeli g evide ce framew ork 1引　　言 [ 1] [ 2] Vap ik . , M acky[ 3] , Kow k[ 4] ( SVM) . . , , , ( Sta dard SVM ) ( LS SVM ) , . , , , . , . .