Skip to content
GitLab
Select Git revision
  • zonalstats_histogram default
Find file BlameHistoryPermalink
cbTrainAutoencoder.txx 6.63 KiB
1
2 
#ifndef cbTrainAutoencoder_txx
3 
#define cbTrainAutoencoder_txx
4
5 
#include "cbLearningApplicationBaseDR.h"
6
7 
namespace otb
8 
{
9 
namespace Wrapper
10 
{
11
12 
template <class TInputValue, class TOutputValue>
13 
void
14 
cbLearningApplicationBaseDR<TInputValue,TOutputValue>
15 
::InitAutoencoderParams()
16 
{
17
18
19 
  AddChoice("model.tiedautoencoder", "Shark Tied Autoencoder");
20 
  AddChoice("model.autoencoder", "Shark Autoencoder");
21 
  SetParameterDescription("model.autoencoder",
22 
                          "This group of parameters allows setting Shark autoencoder parameters. "
23 
                          );
24
25
26 
  //Tied Autoencoder
27 
  AddParameter(ParameterType_Choice, "model.autoencoder.istied",
28 
               "tied weighth <tied/untied>");
29 
  SetParameterDescription(
30 
    "model.autoencoder.istied",
31 
    "Parameter that determine if the weights are tied or not <tied/untied>");
32
33
34 
  AddChoice("model.autoencoder.istied.yes","Tied weigths");
35 
  AddChoice("model.autoencoder.istied.no","Untied weights");
36
37
38
39 
  //Number Of Iterations
40 
  AddParameter(ParameterType_Int, "model.autoencoder.nbiter",
41 
               "Maximum number of iterations during training");
42 
  SetParameterInt("model.autoencoder.nbiter",100, false);
43 
  SetParameterDescription(
44 
    "model.autoencoder.nbiter",
45 
    "The maximum number of iterations used during training.");
46
47 
  AddParameter(ParameterType_Float, "model.autoencoder.epsilon",
48 
               " ");
49 
  SetParameterFloat("model.autoencoder.epsilon",0, false);
50 
  SetParameterDescription(
51 
    "model.autoencoder.epsilon",
52 
    " ");
53
54
55 
  AddParameter(ParameterType_Float, "model.autoencoder.initfactor",
56 
               " ");
57 
  SetParameterFloat("model.autoencoder.initfactor",1, false);
58 
  SetParameterDescription(
59 
    "model.autoencoder.initfactor", "parameter that control the weight initialization of the autoencoder");
60
61 
   //Number Of Hidden Neurons
62 
  AddParameter(ParameterType_StringList ,  "model.autoencoder.nbneuron",   "Size");
63 
  /*AddParameter(ParameterType_Int, "model.autoencoder.nbneuron",
64 
               "Number of neurons in the hidden layer");
65 
  SetParameterInt("model.autoencoder.nbneuron",10, false);*/
66 
  SetParameterDescription(
67 
    "model.autoencoder.nbneuron",
68 
    "The number of neurons in each hidden layer.");
69
70 
  //Regularization
7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
AddParameter(ParameterType_StringList, "model.autoencoder.regularization", "Strength of the regularization"); SetParameterDescription("model.autoencoder.regularization", "Strength of the L2 regularization used during training"); //Noise strength AddParameter(ParameterType_StringList, "model.autoencoder.noise", "Strength of the noise"); SetParameterDescription("model.autoencoder.noise", "Strength of the noise"); // Sparsity parameter AddParameter(ParameterType_StringList, "model.autoencoder.rho", "Sparsity parameter"); SetParameterDescription("model.autoencoder.rho", "Sparsity parameter"); // Sparsity regularization strength AddParameter(ParameterType_StringList, "model.autoencoder.beta", "Sparsity regularization strength"); SetParameterDescription("model.autoencoder.beta", "Sparsity regularization strength"); AddParameter(ParameterType_OutputFilename, "model.autoencoder.learningcurve", "Learning curve"); SetParameterDescription("model.autoencoder.learningcurve", "Learning error values"); MandatoryOff("model.autoencoder.learningcurve"); } template <class TInputValue, class TOutputValue> void cbLearningApplicationBaseDR<TInputValue,TOutputValue> ::BeforeTrainAutoencoder(typename ListSampleType::Pointer trainingListSample, std::string modelPath) { std::string TiedWeigth = GetParameterString("model.autoencoder.istied"); std::cout << TiedWeigth << std::endl; if(TiedWeigth == "no") { TrainAutoencoder<AutoencoderModelType>(trainingListSample,modelPath); } /* if(TiedWeigth == "yes") { TrainAutoencoder<TiedAutoencoderModelType>(trainingListSample,modelPath); } */ if(TiedWeigth != "yes" && TiedWeigth != "no") { std::cerr << "istied : invalid choice <yes/no>" << std::endl; } } template <class TInputValue, class TOutputValue> template <typename autoencoderchoice> void cbLearningApplicationBaseDR<TInputValue,TOutputValue> ::TrainAutoencoder(typename ListSampleType::Pointer trainingListSample,std::string modelPath) { typename autoencoderchoice::Pointer dimredTrainer = autoencoderchoice::New(); itk::Array<unsigned int> nb_neuron; itk::Array<float> noise; itk::Array<float> regularization; itk::Array<float> rho; itk::Array<float> beta; std::vector<std::basic_string<char>> s_nbneuron= GetParameterStringList("model.autoencoder.nbneuron"); std::vector<std::basic_string<char>> s_noise= GetParameterStringList("model.autoencoder.noise"); std::vector<std::basic_string<char>> s_regularization= GetParameterStringList("model.autoencoder.regularization"); std::vector<std::basic_string<char>> s_rho= GetParameterStringList("model.autoencoder.rho"); std::vector<std::basic_string<char>> s_beta= GetParameterStringList("model.autoencoder.beta"); nb_neuron.SetSize(s_nbneuron.size());
141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182
noise.SetSize(s_nbneuron.size()); regularization.SetSize(s_nbneuron.size()); rho.SetSize(s_nbneuron.size()); beta.SetSize(s_nbneuron.size()); for (int i=0; i<s_nbneuron.size(); i++){ nb_neuron[i]=std::stoi(s_nbneuron[i]); noise[i]=std::stof(s_noise[i]); regularization[i]=std::stof(s_regularization[i]); rho[i]=std::stof(s_rho[i]); beta[i]=std::stof(s_beta[i]); } dimredTrainer->SetNumberOfHiddenNeurons(nb_neuron); dimredTrainer->SetNumberOfIterations(GetParameterInt("model.autoencoder.nbiter")); dimredTrainer->SetEpsilon(GetParameterFloat("model.autoencoder.epsilon")); dimredTrainer->SetInitFactor(GetParameterFloat("model.autoencoder.initfactor")); dimredTrainer->SetRegularization(regularization); dimredTrainer->SetNoise(noise); dimredTrainer->SetRho(rho); dimredTrainer->SetBeta(beta); dimredTrainer->SetWriteWeights(true); if (HasValue("model.autoencoder.learningcurve") && IsParameterEnabled("model.autoencoder.learningcurve")) { std::cout << "yo" << std::endl; dimredTrainer->SetWriteLearningCurve(true); dimredTrainer->SetLearningCurveFileName(GetParameterString("model.autoencoder.learningcurve")); } dimredTrainer->SetInputListSample(trainingListSample); std::cout << "before train" << std::endl; dimredTrainer->Train(); std::cout << "after train" << std::endl; dimredTrainer->Save(modelPath); } } //end namespace wrapper } //end namespace otb #endif