一种基于局部线性嵌入的SVM增量学习方法
|
摘要
由于SVM对高维数据分类的耗时较长,计算复杂度较高,而PCA-SVM对高维数据分类的准确率相对较低,提出了利用LLE-ISVM方法对高维数据降维后采用SVM方法进行分类,利用LLE降维对新增样本进行约减后,淘汰新增样本中的非支持向量用于简化运算,实现了基于局部线性嵌入(LLE)的SVM增量学习过程(LLE-ISVM)。并将该算法用于MNIST数据库和瓷片表面缺陷检测分类过程。实验结果表明:该算法对高维数据的运算速度与精度都有所提高,能实现完整增量学习过程,较为准确快速地实现磁片表面的缺陷检测分类过程。
Due to the long time and high computational complexity taken by SVM to classify high-dimensional data, and the PCA-SVM for high dimensional data classification hasrelatively low accuracy rate. A new method is proposed to use the LLE-ISVM classifier for reducing dimension of the high dimension data,then the SVM is used for classification. The LLE-dimensionality reduction is used to reduce the new sample, the non-support vector in the new sample is eliminated in order to simplify the operation.The SVM incremental learning process(LLE-ISVM) based on local linear embedding(LLE) is implemented. The algorithm is used in MNIST dataset and porcelain surface defect detection and classification.The experimental results show that the algorithm improves the operation speed and accuracy for high dimensional data.
Due to the long time and high computational complexity taken by SVM to classify high-dimensional data, and the PCA-SVM for high dimensional data classification hasrelatively low accuracy rate. A new method is proposed to use the LLE-ISVM classifier for reducing dimension of the high dimension data,then the SVM is used for classification. The LLE-dimensionality reduction is used to reduce the new sample, the non-support vector in the new sample is eliminated in order to simplify the operation.The SVM incremental learning process(LLE-ISVM) based on local linear embedding(LLE) is implemented. The algorithm is used in MNIST dataset and porcelain surface defect detection and classification.The experimental results show that the algorithm improves the operation speed and accuracy for high dimensional data.
引文
[1] 周蓉.支持张量机的在线学习算法研究[D].广州:华南理工大学,2014.
[2] LUH G C,LIN C Y.PCA based immune networks for human face recognition[J].Applied soft computing,2011,11(2):1743-1752.
[3] ROBINSON SL,BENNETTRJ.A typology of deviant workplace behaviors:a multidimensional scaling study[J].Academy of management journal,1995,38(2):555-572.
[4] ZHANG X X,JIA Y D.Symmetrical null space LDA for face and ear recognition[J].Elsevier science publishers,2007,70(4/5/6):842-848.
[5] TENENBAUM J B,DE SILVA V,LANGFORD J C.A global geometric framework for nonlinear dimensionality reduction[J].Science,2000,290(5500):2319-2323.
[6] ROWEIS S T,SAUL L K.Nonlinear dimensionality reduction by locally linear embedding[J].Science,2000,290(5500):2323-2326.
[7] ZHANG Z Y,ZHA H Y.Principal manifolds and nonlinear dimensionality reduction via tangent space alignment[J].Journal of Shanghai unicersity,2004,8(4):406-424.
[8] DONOHO D L,GRIMES C.Hessian eigenmaps:locally linear embedding techniques for high-dimensional data[J].Praceedings of the national academy of arts and sciences of the United States of America,2003,100(10):5591-5596.
[9] WANG L,ZHANG Y,FENG J.On the Euclidean distance of images[J].IEEE transactions on pattern analysis & machine intelligence,2005,27(8):1334-1339.
[10] 郑莉莉,黄鲜萍,梁荣华.基于支持向量机的人体姿态识别[J].浙江工业大学学报,2012,40(6):670-675.
[11] 兰秀菊,,张丽霞,鲁建厦,等.基于小波分析和PSO-SVM的控制图混合模式识别[J].浙江工业大学学报,2012,40(5):532-536.
[12] 叶永伟,任设东,陆俊杰,等.基于SVM的汽车涂装线设备故障诊断[J].浙江工业大学学报,2015,43(6):670-675.
[13] 王晓晔,王正欧.K-最近邻分类技术的改进算法[J].电子与信息学报,2005,27(3):487-491.
[2] LUH G C,LIN C Y.PCA based immune networks for human face recognition[J].Applied soft computing,2011,11(2):1743-1752.
[3] ROBINSON SL,BENNETTRJ.A typology of deviant workplace behaviors:a multidimensional scaling study[J].Academy of management journal,1995,38(2):555-572.
[4] ZHANG X X,JIA Y D.Symmetrical null space LDA for face and ear recognition[J].Elsevier science publishers,2007,70(4/5/6):842-848.
[5] TENENBAUM J B,DE SILVA V,LANGFORD J C.A global geometric framework for nonlinear dimensionality reduction[J].Science,2000,290(5500):2319-2323.
[6] ROWEIS S T,SAUL L K.Nonlinear dimensionality reduction by locally linear embedding[J].Science,2000,290(5500):2323-2326.
[7] ZHANG Z Y,ZHA H Y.Principal manifolds and nonlinear dimensionality reduction via tangent space alignment[J].Journal of Shanghai unicersity,2004,8(4):406-424.
[8] DONOHO D L,GRIMES C.Hessian eigenmaps:locally linear embedding techniques for high-dimensional data[J].Praceedings of the national academy of arts and sciences of the United States of America,2003,100(10):5591-5596.
[9] WANG L,ZHANG Y,FENG J.On the Euclidean distance of images[J].IEEE transactions on pattern analysis & machine intelligence,2005,27(8):1334-1339.
[10] 郑莉莉,黄鲜萍,梁荣华.基于支持向量机的人体姿态识别[J].浙江工业大学学报,2012,40(6):670-675.
[11] 兰秀菊,,张丽霞,鲁建厦,等.基于小波分析和PSO-SVM的控制图混合模式识别[J].浙江工业大学学报,2012,40(5):532-536.
[12] 叶永伟,任设东,陆俊杰,等.基于SVM的汽车涂装线设备故障诊断[J].浙江工业大学学报,2015,43(6):670-675.
[13] 王晓晔,王正欧.K-最近邻分类技术的改进算法[J].电子与信息学报,2005,27(3):487-491.