[Math] How to solve the dual problem of SVM

convex optimizationoptimization

By solving the primal form of SVM (support vector machine), we can get the dual form of this problem. The more details are shown in wiki of SVM.
enter image description here

Given this dual problem, how can I solve the maximization problem ? Thanks !

Best Answer

Being a concave quadratic optimization problem, you can in principle solve it using any QP solver. For instance you can use MOSEK, CPLEX or Gurobi. All of them come with free trial or academic license.

Due to its typical dimension, and the peculiar structure, there are some first-order gradient based algorithms usually used by specialized packages. I suggest you to look at the LIBSVM package. The source code is available, plenty of documentation, examples and specialized versions.

Don't re-invent the wheel unless you really need a special one.

Related Question