Function sigmoid_train

ml/src/svm.cpp:1970–2100

Platt's binary SVM Probabilistic Output: an improvement from Lin et al.

// Platt's binary SVM Probabilistic Output: an improvement from Lin et al.
static void
sigmoid_train(
    int l, const double* dec_values, const double* labels, double& A, double& B)
{
  double prior1 = 0, prior0 = 0;

  for (int i = 0; i < l; i++)
    if (labels[i] > 0)
      prior1 += 1;
    else
      prior0 += 1;

  const int max_iter = 100; // Maximal number of iterations

  const double min_step = 1e-10; // Minimal step taken in line search

  const double sigma = 1e-12; // For numerically strict PD of Hessian

  const double eps = 1e-5;

  const double hiTarget = (prior1 + 1.0) / (prior1 + 2.0);

  const double loTarget = 1 / (prior0 + 2.0);

  double* t = Malloc(double, l);

  // Initial Point and Initial Fun Value
  A = 0.0;

  B = std::log((prior0 + 1.0) / (prior1 + 1.0));

  double fval = 0.0;

  for (int i = 0; i < l; i++) {
    if (labels[i] > 0)
      t[i] = hiTarget;
    else
      t[i] = loTarget;

    double fApB = dec_values[i] * A + B;

    if (fApB >= 0)
      fval += t[i] * fApB + std::log(1 + std::exp(-fApB));
    else
      fval += (t[i] - 1) * fApB + std::log(1 + std::exp(fApB));
  }

  int iter = 0;
  for (; iter < max_iter; iter++) {
    // Update Gradient and Hessian (use H' = H + sigma I)
    double h11 = sigma; // numerically ensures strict PD
    double h22 = sigma;
    double h21 = 0.0;
    double g1 = 0.0;
    double g2 = 0.0;

    for (int i = 0; i < l; i++) {
      double fApB = dec_values[i] * A + B;