l1_loss_impl.hpp

Go to the documentation of this file.

#ifndef MLPACK_METHODS_ANN_LOSS_FUNCTION_L1_LOSS_IMPL_HPP
#define MLPACK_METHODS_ANN_LOSS_FUNCTION_L1_LOSS_IMPL_HPP

// In case it hasn't yet been included.
#include "l1_loss.hpp"

namespace mlpack {
namespace ann  {

template<typename InputDataType, typename OutputDataType>
L1Loss<InputDataType, OutputDataType>::L1Loss(const bool mean):
  mean(mean)
{
  // Nothing to do here.
}

template<typename InputDataType, typename OutputDataType>
template<typename PredictionType, typename TargetType>
typename PredictionType::elem_type
L1Loss<InputDataType, OutputDataType>::Forward(
    const PredictionType& prediction,
    const TargetType& target)
{
  if (mean)
    return arma::accu(arma::mean(prediction - target));

  return arma::accu(prediction - target);
}

template<typename InputDataType, typename OutputDataType>
template<typename PredictionType, typename TargetType, typename LossType>
void L1Loss<InputDataType, OutputDataType>::Backward(
    const PredictionType& prediction,
    const TargetType& target,
    LossType& loss)
{
  loss = arma::sign(prediction - target);
}

template<typename InputDataType, typename OutputDataType>
template<typename Archive>
void L1Loss<InputDataType, OutputDataType>::serialize(
    Archive& ar,
    const uint32_t /* version */)
{
  ar(CEREAL_NVP(mean));
}

} // namespace ann
} // namespace mlpack

#endif