Method ComputePrimalLoss

tensorflow/core/kernels/hinge-loss.h:89–93

Hinge loss for binary classification for a single example. Hinge loss equals max(0, 1 - y * wx) (see https://en.wikipedia.org/wiki/Hinge_loss). For weighted instances loss should be multiplied by the instance weight.

  // equals max(0, 1 - y * wx) (see https://en.wikipedia.org/wiki/Hinge_loss).
  // For weighted instances loss should be multiplied by the instance weight.
  double ComputePrimalLoss(const double wx, const double example_label,
                           const double example_weight) const final {
    const double y_wx = example_label * wx;
    return std::max(0.0, 1 - y_wx) * example_weight;
  }

  double PrimalLossDerivative(const double wx, const double label,
                              const double example_weight) const final {