摘要
针对支持向量机(SVM)本身抗噪声能力低和训练数据类别不均匀会造成分类结果偏向数目较大一类的倾向性等问题,本文提出了去噪声的加权SVM分类方法。在该方法中,通过引入主成分分析方法来降维去除噪声,再通过引入加权系数的方式,补偿了上述倾向性造成的不利影响,提高了预测分类精度。对污水处理过程运行状态的分类实验表明该方法的有效性。
A new classification algorithm based on support vector machine (SVM) theory and principal component analysis (PCA) techniques is presented. Noise is eliminated by PCA to increases the predicted classification accuracy. When training sets with uneven class sizes are used, the result is undesirably biased towards the larger class. The cause of this effect and the compensation method are proposed in this paper. Numerical experiments for classifying operation state of wastewater treatment processes show that the proposed algorithm can be used to obtain better classification accuracy than the original SVM.
出处
《电路与系统学报》
CSCD
2004年第4期97-102,共6页
Journal of Circuits and Systems
基金
国家863基金资助项目2002AA412010)
关键词
支持向量机
主成分分析
分类精度
污水处理过程
support vector machine
principal component analysis
classification accuracy
wastewater treatment processes
