1
$\begingroup$

From Wiki:

Ensemble methods use multiple learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms

Multiple kernel learning refers to a set of machine learning methods that use a predefined set of kernels and learn an optimal linear or non-linear combination of kernels as part of the algorithm.

What is the difference?

$\endgroup$
1
$\begingroup$

Multiple kernel learning is restricted to the kernel methods. Ensemble learning can collect together any classification methods -- kernel SVM and Random Forest and logistic regression could all appear in the same ensemble.

$\endgroup$
3
  • $\begingroup$ So if I combine lets say 3 predictions of logistic regression by another logistic regression (linear combination of 3 linear kernels) is it then both MKL and ensemble? $\endgroup$ – rep_ho Jul 15 '15 at 14:19
  • $\begingroup$ @user2173836 Logistic regression as I understand it is not a kernel method. $\endgroup$ – Sycorax Jul 15 '15 at 14:50
  • $\begingroup$ you can use a kernel trick $\endgroup$ – rep_ho Jul 16 '15 at 8:27
1
$\begingroup$

You can find a useful explanation on that topic in Sun et a. (2003), particularly in Section 3:

It reaches a conclusion that multiple kernel learning (MKL) is a special instance of ensemble learning (EL). For example, EL does not suffer from the limitation of MKL that the classifiers have the sub-kernels of the same size and the same support vector coefficients (for SVMs).

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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