Outputting the decision boundary function for one-class SVM with fitcsvm?

Say I am performing outlier detection, or performing a one-class SVM classification using fitcsvm. I am only interested in doing so to obtain the function that separates the decision boundaries. How do I obtain this?

If the decision boundary is when , how do I obtain ?

Many thanks!

Answer 1

[Aditya Patil]

You can find the equation for decision boundary explicitly only if linear kernel is used. In other cases, the score function is not calculated in terms of x, hence it's not possible to derive equation for when . You can however plot the decision boundary by finding points for which .

See this answer for more details.