EE4389: Homework

Professor Cherkassky, University of Minnesota, Fall Semester 2013
  • Guidelines for homework report, pdf file.

  • Homework 5, pdf file, due Thursday 12/05.

    • For Problem 7.7, see example in here.

      • Do NOT use evalsvm() for Problem 7.7 since evalsvm() uses cross-validation on the training data for model selection; however, we have independent validation data set in this problem.

      • Use smo() and svmclass(), instead.

      • To train the SVM classifier with linear kernel, use
        model = smo(data, struct('ker’, 'linear’,'C’, c))
        where data follows the same structure (or format) described in here, and c is the value of the regularization parameter C.

      • To make predictions on the given data, use
        ypred = svmclass(data.X, model)

    • For the 3rd problem, see example in here.

    • Solutions