sigmoid_cross_entropy_error_impl.hpp

Go to the documentation of this file.

#ifndef MLPACK_METHODS_ANN_LOSS_FUNCTION_SIGMOID_CROSS_ENTROPY_ERROR_IMPL_HPP
#define MLPACK_METHODS_ANN_LOSS_FUNCTION_SIGMOID_CROSS_ENTROPY_ERROR_IMPL_HPP

// In case it hasn't yet been included.
#include "sigmoid_cross_entropy_error.hpp"
#include <mlpack/methods/ann/activation_functions/softplus_function.hpp>

namespace mlpack {
namespace ann  {

template<typename InputDataType, typename OutputDataType>
SigmoidCrossEntropyError<InputDataType, OutputDataType>
::SigmoidCrossEntropyError()
{
  // Nothing to do here.
}

template<typename InputDataType, typename OutputDataType>
template<typename PredictionType, typename TargetType>
inline typename PredictionType::elem_type
SigmoidCrossEntropyError<InputDataType, OutputDataType>::Forward(
    const PredictionType& prediction,
    const TargetType& target)
{
  typedef typename PredictionType::elem_type ElemType;
  ElemType maximum = 0;
  for (size_t i = 0; i < prediction.n_elem; ++i)
  {
    maximum += std::max(prediction[i], 0.0) +
        std::log(1 + std::exp(-std::abs(prediction[i])));
  }

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

template<typename InputDataType, typename OutputDataType>
template<typename PredictionType, typename TargetType, typename LossType>
inline void SigmoidCrossEntropyError<InputDataType, OutputDataType>::Backward(
    const PredictionType& prediction,
    const TargetType& target,
    LossType& loss)
{
  loss = 1.0 / (1.0 + arma::exp(-prediction)) - target;
}

template<typename InputDataType, typename OutputDataType>
template<typename Archive>
void SigmoidCrossEntropyError<InputDataType, OutputDataType>::serialize(
    Archive& /* ar */,
    const uint32_t /* version */)
{
  // Nothing to do here
}

} // namespace ann
} // namespace mlpack

#endif