Abstract | ||
---|---|---|
We investigate practical selection of hyper-parameters for support vector machines (SVM) regression (that is, epsilon-insensitive zone and regularization parameter C). The proposed methodology advocates analytic parameter selection directly from the training data, rather than re-sampling approaches commonly used in SVM applications. In particular, we describe a new analytical prescription for setting the value of insensitive zone epsilon, as a function of training sample size. Good generalization performance of the proposed parameter selection is demonstrated empirically using several low- and high-dimensional regression problems. Further, we point out the importance of Vapnik's epsilon-insensitive loss for regression problems with finite samples. To this end, we compare generalization performance of SVM regression (using proposed selection of epsilon-values) with regression using 'least-modulus' loss (epsilon=0) and standard squared loss. These comparisons indicate superior generalization performance of SVM regression under sparse sample settings, for various types of additive noise. |
Year | DOI | Venue |
---|---|---|
2004 | 10.1016/S0893-6080(03)00169-2 | Neural Networks |
Keywords | Field | DocType |
generalization performance,regression problem,prediction accuracy,vc theory,proposed selection,noise estimation,support vector machine,loss function,proposed parameter selection,analytic parameter selection,parameter selection,svm parameter,high-dimensional regression problem,support vector machine regression,practical selection,svm regression,insensitive zone,svm application,complexity control | Square (algebra),Pattern recognition,Regression,Regression analysis,Support vector machine,Regularization (mathematics),Artificial intelligence,Estimation theory,Artificial neural network,Mathematics,Sample size determination,Machine learning | Journal |
Volume | Issue | ISSN |
17 | 1 | 0893-6080 |
Citations | PageRank | References |
390 | 24.38 | 6 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Vladimir Cherkassky | 1 | 1064 | 126.66 |
Yunqian Ma | 2 | 533 | 44.21 |