add_sample_class_svmT_add_sample_class_svmAddSampleClassSvmAddSampleClassSvmadd_sample_class_svm (Operator)

Name

add_sample_class_svmT_add_sample_class_svmAddSampleClassSvmAddSampleClassSvmadd_sample_class_svm — Add a training sample to the training data of a support vector machine.

Signature

add_sample_class_svm( : : SVMHandle, Features, Class : )

Description

add_sample_class_svmadd_sample_class_svmAddSampleClassSvmAddSampleClassSvmadd_sample_class_svm adds a training sample to the support vector machine (SVM) given by SVMHandleSVMHandleSVMHandleSVMHandlesvmhandle. The training sample is given by FeaturesFeaturesFeaturesfeaturesfeatures and ClassClassClassclassValclass. FeaturesFeaturesFeaturesfeaturesfeatures is the feature vector of the sample, and consequently must be a real vector of length NumFeaturesNumFeaturesNumFeaturesnumFeaturesnum_features, as specified in create_class_svmcreate_class_svmCreateClassSvmCreateClassSvmcreate_class_svm. ClassClassClassclassValclass is the target of the sample, which must be in the range of 0 to NumClassesNumClassesNumClassesnumClassesnum_classes-1 (see create_class_svmcreate_class_svmCreateClassSvmCreateClassSvmcreate_class_svm). In the special case of 'novelty-detection'"novelty-detection""novelty-detection""novelty-detection""novelty-detection" the class is to be set to 0 as only one class is assumed. Before the SVM can be trained with train_class_svmtrain_class_svmTrainClassSvmTrainClassSvmtrain_class_svm, training samples must be added to the SVM with add_sample_class_svmadd_sample_class_svmAddSampleClassSvmAddSampleClassSvmadd_sample_class_svm. The usage of support vectors of an already trained SVM as training samples is described in train_class_svmtrain_class_svmTrainClassSvmTrainClassSvmtrain_class_svm.

The number of currently stored training samples can be queried with get_sample_num_class_svmget_sample_num_class_svmGetSampleNumClassSvmGetSampleNumClassSvmget_sample_num_class_svm. Stored training samples can be read out again with get_sample_class_svmget_sample_class_svmGetSampleClassSvmGetSampleClassSvmget_sample_class_svm.

Normally, it is useful to save the training samples in a file with write_samples_class_svmwrite_samples_class_svmWriteSamplesClassSvmWriteSamplesClassSvmwrite_samples_class_svm to facilitate reusing the samples and to facilitate that, if necessary, new training samples can be added to the data set, and hence to facilitate that a newly created SVM can be trained with the extended data set.

Execution Information

Multithreading type: reentrant (runs in parallel with non-exclusive operators).
Multithreading scope: global (may be called from any thread).
Processed without parallelization.

This operator modifies the state of the following input parameter:

SVMHandleSVMHandleSVMHandleSVMHandlesvmhandle

During execution of this operator, access to the value of this parameter must be synchronized if it is used across multiple threads.

Parameters

SVMHandleSVMHandleSVMHandleSVMHandlesvmhandle (input_control, state is modified) class_svm → (handle)

SVM handle.

FeaturesFeaturesFeaturesfeaturesfeatures (input_control) real-array → (real)

Feature vector of the training sample to be stored.

ClassClassClassclassValclass (input_control) number → (integer / real)

Class of the training sample to be stored.

Result

If the parameters are valid the operator add_sample_class_svmadd_sample_class_svmAddSampleClassSvmAddSampleClassSvmadd_sample_class_svm returns the value 2 ( H_MSG_TRUE) . If necessary, an exception is raised.