1
$\begingroup$

Is it correct to say that, a difference between decision trees and linear SVMs is that the hyperplanes used by the decision trees are perpendicular to axis?

$\endgroup$
1
  • 1
    $\begingroup$ Is this question homework? Consider the tag self-study, and read its description and wiki. $\endgroup$
    – Firebug
    May 22, 2018 at 12:50

1 Answer 1

1
$\begingroup$

Assuming you meant the independent variables (features) axes, the phrasing is a bit off. The SVM (linear or otherwise) uses a single decision hyperplane. The decision trees, however, are not bound to a single hyperplane: they use multiple decision rules. Some tree architectures use oblique decisions as well.

An ordinary stump (one-level decision tree) uses a single hyperplane (like a SVM) orthogonal to a feature.

$\endgroup$
1
  • 1
    $\begingroup$ They also learn the hyperplanes by very different means, and the hyperplanes in a tree are NOT necessarily decision boundaries. $\endgroup$ May 22, 2018 at 14:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.