Method train

src/thundersvm/model/svc.cpp:41–145

}

void SVC::train(const DataSet &dataset, SvmParam param) {
    DataSet dataset_ = dataset;
    dataset_.group_classes();
    model_setup(dataset_, param);

    vector<SyncArray<float_type>> alpha(n_binary_models);
    vector<bool> is_sv(dataset_.n_instances(), false);

    int k = 0;
    for (int i = 0; i < n_classes; ++i) {
        for (int j = i + 1; j < n_classes; ++j) {
            train_binary(dataset_, i, j, alpha[k], rho.host_data()[k]);
            vector<int> original_index = dataset_.original_index(i, j);
            CHECK_EQ(original_index.size(), alpha[k].size());
            const float_type *alpha_data = alpha[k].host_data();
            for (int l = 0; l < alpha[k].size(); ++l) {
                is_sv[original_index[l]] = is_sv[original_index[l]] || (alpha_data[l] != 0);
            }
            k++;
        }
    }

    for (int i = 0; i < dataset_.n_classes(); ++i) {
        vector<int> original_index = dataset_.original_index(i);
        DataSet::node2d i_instances = dataset_.instances(i);
        int *n_sv_data = n_sv.host_data();
        for (int j = 0; j < i_instances.size(); ++j) {
            if (is_sv[original_index[j]]) {
                n_sv_data[i]++;
                sv.push_back(i_instances[j]);
                sv_indices.push_back(original_index[j]);
            }
        }
    }

    n_total_sv = sv.size();
    LOG(INFO) << "#total unique sv = " << n_total_sv;
    coef.resize((n_classes - 1) * n_total_sv);

    vector<int> sv_start(1, 0);
    const int *n_sv_data = n_sv.host_data();
    for (int i = 1; i < n_classes; ++i) {
        sv_start.push_back(sv_start[i - 1] + n_sv_data[i - 1]);
    }

    k = 0;
    float_type *coef_data = coef.host_data();
    for (int i = 0; i < n_classes; ++i) {
        for (int j = i + 1; j < n_classes; ++j) {
            const float_type *alpha_data = alpha[k].host_data();
            vector<int> original_index = dataset_.original_index(i, j);
            int ci = dataset_.count()[i];
            int cj = dataset_.count()[j];
            int m = sv_start[i];
            for (int l = 0; l < ci; ++l) {
                if (is_sv[original_index[l]]) {
                    coef_data[(j - 1) * n_total_sv + m++] = alpha_data[l];
                }