Function test_normal_force_last_weight

dlib/test/svm_c_linear_dcd.cpp:247–337

// ----------------------------------------------------------------------------------------

    void test_normal_force_last_weight(bool have_bias, bool force_weight)
    {
        typedef matrix<double,10,1> sample_type;
        dlog << LINFO << "have_bias: "<< have_bias << "   force_weight: "<< force_weight;


        typedef linear_kernel<sample_type> kernel_type;


        svm_c_linear_trainer<kernel_type> linear_trainer_cpa;

        svm_c_linear_dcd_trainer<kernel_type> linear_trainer;

        svm_c_linear_dcd_trainer<kernel_type>::optimizer_state state;

        const double C = 1;
        linear_trainer.set_epsilon(1e-10);
        linear_trainer_cpa.set_epsilon(1e-11);
        linear_trainer_cpa.set_relative_epsilon(1e-11);

        linear_trainer_cpa.force_last_weight_to_1(force_weight);

        linear_trainer.force_last_weight_to_1(force_weight);
        linear_trainer.include_bias(have_bias);

        std::vector<sample_type> samples;
        std::vector<double> labels;

        // make an instance of a sample vector so we can use it below
        sample_type sample;

        decision_function<kernel_type> df, df2;

        running_stats<double> rs;

        dlib::rand rnd;
        // Now lets go into a loop and randomly generate 10000 samples.
        double label = +1;
        for (int i = 0; i < 40; ++i)
        {
            // flip this flag
            label *= -1;

            sample = 0;

            // now make a random sparse sample with at most 10 non-zero elements
            for (int j = 0; j < 5; ++j)
            {
                int idx = rnd.get_random_32bit_number()%9;
                double value = rnd.get_random_double();

                sample(idx) = label*value + label;
            }

            sample(9) = 4;

            // Also save the samples we are generating so we can let the svm_c_linear_trainer
            // learn from them below.