Support vector machine

A support vector machine (SVM) can be used both for classification and for regression. The SVM is a frequently used tool in particular with regard to classification tasks.

The fundamental goal of an SVM is to find a hyperplane in an N-dimensional space, wherein the distance between the closest data point and the plane is maximized. A hyperplane can only separate the space linearly (also called linear SVM). A non-linear separation is also possible by means of a so-called kernel trick (also called kernel SVM). The N-dimensional space is transformed into a higher-dimensional space here. A linear separation with a hyperplane is possible in an accordingly higher-dimensional space.

If a distinction needs to be made between several classes, several support vector machines are generated internally and classification takes place by means of comparisons. A one-class SVM can also be trained and used for anomaly detection.

Supported properties

ONNX support

The following ONNX operators are supported:

• SVMRegressor
• SVMClassifier

Supported kernel functions are listed in the following table:

Kernel function	Description
Linear
Radial Basis Function (RBF)
Sigmoid
Polynomial

For samples of the export of SVMs as ONNX, see ONNX export of an SVM.

Classification limitation With classification models, only the output of the labels is mapped in the PLC. The scores/probabilities are not available in the PLC.