BCE.hpp

#pragma once
 
#include "Loss.hpp"
#include "utils/Functions.hpp"
 
namespace NeuralNet {
class BCE : public Loss {
 public:
  static double cmpLoss(const Eigen::MatrixXd &o, const Eigen::MatrixXd &y) {
    constexpr double threshold = 1.0e-5;
    Eigen::MatrixXd oThresh = thresh(o, threshold);
    Eigen::MatrixXd yThresh = thresh(y, threshold);
 
    Eigen::MatrixXd loss =
        -(yThresh.array() * oThresh.array().log() +
          (1.0 - yThresh.array()) * (1.0 - oThresh.array()).log());
 
    if (loss.array().isNaN().any())
      throw std::runtime_error(
          "NaN value encountered. Inputs might be too big");
 
    return loss.sum();
  }
 
  static Eigen::MatrixXd cmpLossGrad(const Eigen::MatrixXd &yHat,
                                     const Eigen::MatrixXd &y) {
    constexpr double epsilon = 1.0e-9;
    return (yHat.array() - y.array()) /
           ((yHat.array() * (1.0 - yHat.array())) + epsilon);
  }
};
 
}  // namespace NeuralNet