resnet.hpp

#include <mlpack/core.hpp>
#include <mlpack/core/data/split_data.hpp>

#include <mlpack/core/optimizers/sgd/sgd.hpp>
#include <mlpack/core/optimizers/adam/adam_update.hpp>

#include <mlpack/methods/ann/layer/layer.hpp>
#include <"ffn_hack.hpp">

#include "utils.hpp"

using namespace mlpack;
using namespace mlpack::ann;
using namespace mlpack::optimization;

using namespace arma;
using namespace std;

template<typename OutputLayer, typename initialization, typename optimizerType>
class ResNet{
private:
  double RATIO;
  int ITERATIONS_PER_CYCLE;
  int CYCLES;
  double STEP_SIZE;
  int BATCH_SIZE;
  FFN<OutputLayer, initialization> model;
  optimizerType optimizer;
  mat tempDataset;
  mat train, valid;
  int iter=0;
  
  
  
public:
  mat output;
  Resnet(double RATIO,int ITERATIONS_PER_CYCLE,int CYCLES,double STEP_SIZE,int BATCH_SIZE);
  
  mat LoadData(const char* path);

  template<typename layer,typename... Args>
  void  conv_forward(mat input,Args... args){
	model.Add<layer >(...args);
	prevInput=input;
  //we are going just one step further.
	model.Forward(input,output,iter,iter);
	iter=iter+1;
	return output;
  }
  
  void  train(double stepSize,mat Target){
      
      mat params = this->Parameters();
      optimizer.Initialize(params.n_rows,params.n_cols);
      optimizer.Update(this->output,stepSize,Params);
  }


  
};