Hands-On Machine Learning with C++

With SVMs

The Shogun library also has an implementation of the multi-class SVM algorithm in the CMulticlassLibSVM class. The instances of this class can be configured with a parameter named C, which is a measure of the allowance of misclassification with a kernel object. In the following example, we use an instance of the CGaussianKernel class for the kernel object. This object also has parameters for configuration, but we used only one named the combined_kernel_weight parameter because it gave the most reasonable configuration for our model after a series of experiments. Let's look at the code in the following block:

Some <CDenseFeatures<DataType>> features;Some<CMulticlassLabels> labels;Some<CDenseFeatures<DataType>> test_features;Some<CMulticlassLabels> ...