classify_class_svm — Classify a feature vector by a support vector machine.
classify_class_svm computes the best Num classes of the feature vector Features with the SVM SVMHandle and returns them in Class. If the classifier was created in the Mode = 'one-versus-one', the classes are ordered by the number of votes of the sub-classifiers. If Mode = 'one-versus-all' was used, the classes are ordered by the value of each sub-classifier (see create_class_svm for more details). If the classifier was created in the Mode = 'novelty-detection', it determines whether the feature vector belongs to the same class as the training data (Class = 1) or is regarded as outlier (Class = 0). In this case Num must be set to 1 as the classifier only determines membership.
Before calling classify_class_svm, the SVM must be trained with train_class_svm.
Number of best classes to determine.
Default value: 1
Suggested values: 1, 2, 3, 4, 5
Result of classifying the feature vector with the SVM.
If the parameters are valid the operator classify_class_svm returns the value 2 (H_MSG_TRUE). If necessary, an exception is raised.
John Shawe-Taylor, Nello Cristianini: “Kernel Methods for Pattern
Analysis”; Cambridge University Press, Cambridge; 2004.
Bernhard Schölkopf, Alexander J.Smola: “Learning with Kernels”; MIT Press, London; 1999.