1
$\begingroup$

I am trying to find the dual optimization problem for the following adaptive SVM. $\mathbf{w}_t$ are the weights of the target SVM. $C$ and $B$ are regularization parameters (constants), $\mathbf{w}_s$ are the weights of the source SVM (constants) and $\mathbf{\xi}$ is a slack variable.

enter image description here

enter image description here

$\endgroup$
  • $\begingroup$ Since $w_{S}$ is a constant, you can drop the last two terms. $\endgroup$ – jpmuc Mar 29 '17 at 5:49
1
$\begingroup$

I can't see any error in your derivation. Perhaps the form of the dual you were given has that term dropped, since it doesn't depend on the variables you are optimising over it can safely be ignored without affecting the solution. You could check this by taking any solution and comparing the function value for the primal and dual equations.

$\endgroup$

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.