ada_delta_update.hpp

Go to the documentation of this file.

#ifndef MLPACK_CORE_OPTIMIZERS_ADA_DELTA_UPDATE_HPP
#define MLPACK_CORE_OPTIMIZERS_ADA_DELTA_UPDATE_HPP

#include <mlpack/prereqs.hpp>

namespace mlpack {
namespace optimization {

class AdaDeltaUpdate
{
 public:
  AdaDeltaUpdate(const double rho = 0.95, const double epsilon = 1e-6) :
      rho(rho),
      epsilon(epsilon)
  {
    // Nothing to do.
  }

  void Initialize(const size_t rows, const size_t cols)
  {
    // Initialize empty matrices for mean sum of squares of parameter gradient.
    meanSquaredGradient = arma::zeros<arma::mat>(rows, cols);
    meanSquaredGradientDx = arma::zeros<arma::mat>(rows, cols);
  }

  void Update(arma::mat& iterate,
              const double stepSize,
              const arma::mat& gradient)
  {
    // Accumulate gradient.
    meanSquaredGradient *= rho;
    meanSquaredGradient += (1 - rho) * (gradient % gradient);
    arma::mat dx = arma::sqrt((meanSquaredGradientDx + epsilon) /
        (meanSquaredGradient + epsilon)) % gradient;

    // Accumulate updates.
    meanSquaredGradientDx *= rho;
    meanSquaredGradientDx += (1 - rho) * (dx % dx);

    // Apply update.
    iterate -= (stepSize * dx);
  }

  double Rho() const { return rho; }
  double& Rho() { return rho; }

  double Epsilon() const { return epsilon; }
  double& Epsilon() { return epsilon; }

 private:
  // The smoothing parameter.
  double rho;

  // The epsilon value used to initialise the mean squared gradient parameter.
  double epsilon;

  // The mean squared gradient matrix.
  arma::mat meanSquaredGradient;

  // The delta mean squared gradient matrix.
  arma::mat meanSquaredGradientDx;
};

} // namespace optimization
} // namespace mlpack

#endif