Skip to content
GitLab
Select Git revision
  • zonalstats_histogram default
Find file BlameHistoryPermalink
AutoencoderModel.h 5.28 KiB
1 
#ifndef AutoencoderModel_h
2 
#define AutoencoderModel_h
3
4 
#include "otbMachineLearningModelTraits.h"
5 
#include "otbMachineLearningModel.h"
6 
#include <fstream>
7 
#include <shark/Algorithms/StoppingCriteria/AbstractStoppingCriterion.h>
8
9 
#include <shark/Models/FFNet.h>
10 
#include <shark/Models/Autoencoder.h>
11 
namespace otb
12 
{
13 
template <class TInputValue, class NeuronType>
14 
class ITK_EXPORT AutoencoderModel: public  MachineLearningModel<itk::VariableLengthVector< TInputValue> , itk::VariableLengthVector< TInputValue>>
15 
{
16
17 
public:
18
19 
	typedef AutoencoderModel Self;
20 
	typedef MachineLearningModel<itk::VariableLengthVector< TInputValue> , itk::VariableLengthVector< TInputValue>> Superclass;
21 
	typedef itk::SmartPointer<Self> Pointer;
22 
	typedef itk::SmartPointer<const Self> ConstPointer;
23
24 
	typedef typename Superclass::InputValueType 			InputValueType;
25 
	typedef typename Superclass::InputSampleType 			InputSampleType;
26 
	typedef typename Superclass::InputListSampleType 		InputListSampleType;
27 
	typedef typename InputListSampleType::Pointer 			ListSamplePointerType;
28 
	typedef typename Superclass::TargetValueType 			TargetValueType;
29 
	typedef typename Superclass::TargetSampleType 			TargetSampleType;
30 
	typedef typename Superclass::TargetListSampleType 		TargetListSampleType;
31
32 
	/// Confidence map related typedefs
33
34 
	typedef typename Superclass::ConfidenceValueType  				ConfidenceValueType;
35 
	typedef typename Superclass::ConfidenceSampleType 				ConfidenceSampleType;
36 
	typedef typename Superclass::ConfidenceListSampleType      		ConfidenceListSampleType;
37
38 
	/// Neural network related typedefs
39 
	typedef shark::Autoencoder<NeuronType,shark::LinearNeuron> OutAutoencoderType;
40 
	typedef shark::Autoencoder<NeuronType,NeuronType> AutoencoderType;
41 
	typedef shark::FFNet<NeuronType,shark::LinearNeuron> NetworkType;
42
43
44 
	itkNewMacro(Self);
45 
	itkTypeMacro(AutoencoderModel, DimensionalityReductionModel);
46
47 
	//unsigned int GetDimension() {return m_NumberOfHiddenNeurons[m_net.size()-1];};  // Override the Dimensionality Reduction model method, it is used in the dimensionality reduction filter to set the output image size
48 
	itkGetMacro(NumberOfHiddenNeurons,itk::Array<unsigned int>);
49 
	itkSetMacro(NumberOfHiddenNeurons,itk::Array<unsigned int>);
50
51 
	itkGetMacro(NumberOfIterations,unsigned int);
52 
	itkSetMacro(NumberOfIterations,unsigned int);
53
54 
	itkGetMacro(Epsilon,double);
55 
	itkSetMacro(Epsilon,double);
56
57 
	itkGetMacro(InitFactor,double);
58 
	itkSetMacro(InitFactor,double);
59
60 
	itkGetMacro(Regularization,itk::Array<double>);
61 
	itkSetMacro(Regularization,itk::Array<double>);
62
63 
	itkGetMacro(Noise,itk::Array<double>);
64 
	itkSetMacro(Noise,itk::Array<double>);
65
66 
	itkGetMacro(Rho,itk::Array<double>);
67 
	itkSetMacro(Rho,itk::Array<double>);
68
69 
	itkGetMacro(Beta,itk::Array<double>);
70 
	itkSetMacro(Beta,itk::Array<double>);
7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135
itkGetMacro(WriteLearningCurve,bool); itkSetMacro(WriteLearningCurve,bool); itkSetMacro(WriteWeights, bool); itkGetMacro(WriteWeights, bool); itkGetMacro(LearningCurveFileName,std::string); itkSetMacro(LearningCurveFileName,std::string); bool CanReadFile(const std::string & filename); bool CanWriteFile(const std::string & filename); void Save(const std::string & filename, const std::string & name="") ITK_OVERRIDE; void Load(const std::string & filename, const std::string & name="") ITK_OVERRIDE; void Train() ITK_OVERRIDE; template <class T, class Autoencoder> void TrainOneLayer(shark::AbstractStoppingCriterion<T> & criterion,Autoencoder &,unsigned int, unsigned int,double, double, shark::Data<shark::RealVector> &, std::ostream&); template <class T, class Autoencoder> void TrainOneSparseLayer(shark::AbstractStoppingCriterion<T> & criterion,Autoencoder &, unsigned int, unsigned int,double, double,double, shark::Data<shark::RealVector> &, std::ostream&); template <class T> void TrainNetwork(shark::AbstractStoppingCriterion<T> & criterion,double, double,double, shark::Data<shark::RealVector> &, std::ostream&); protected: AutoencoderModel(); ~AutoencoderModel() ITK_OVERRIDE; virtual TargetSampleType DoPredict(const InputSampleType& input, ConfidenceValueType * quality = ITK_NULLPTR) const; virtual void DoPredictBatch(const InputListSampleType *, const unsigned int & startIndex, const unsigned int & size, TargetListSampleType *, ConfidenceListSampleType * quality = ITK_NULLPTR) const; private: /** Network attributes */ //std::vector<AutoencoderType> m_net; NetworkType m_net; itk::Array<unsigned int> m_NumberOfHiddenNeurons; /** Training parameters */ unsigned int m_NumberOfIterations; // stop the training after a fixed number of iterations double m_Epsilon; // Stops the training when the training error seems to converge itk::Array<double> m_Regularization; // L2 Regularization parameter itk::Array<double> m_Noise; // probability for an input to be set to 0 (denosing autoencoder) itk::Array<double> m_Rho; // Sparsity parameter itk::Array<double> m_Beta; // Sparsity regularization parameter double m_InitFactor; // Weight initialization factor (the weights are intialized at m_initfactor/sqrt(inputDimension) ) bool m_WriteLearningCurve; // Flag for writting the learning curve into a txt file std::string m_LearningCurveFileName; // Name of the output learning curve printed after training bool m_WriteWeights; }; } // end namespace otb #ifndef OTB_MANUAL_INSTANTIATION #include "AutoencoderModel.txx" #endif #endif