1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
|
/*=========================================================================
Program: Insight Segmentation & Registration Toolkit
Module: itkTwoHiddenLayerBackPropagationNeuralNetwork.txx
Language: C++
Date: $Date$
Version: $Revision$
Copyright (c) Insight Software Consortium. All rights reserved.
See ITKCopyright.txt or http://www.itk.org/HTML/Copyright.htm for details.
This software is distributed WITHOUT ANY WARRANTY; without even
the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
PURPOSE. See the above copyright notices for more information.
=========================================================================*/
#ifndef __itkTwoHiddenLayerBackPropagationNeuralNetwork_txx
#define __itkTwoHiddenLayerBackPropagationNeuralNetwork_txx
#include "itkTwoHiddenLayerBackPropagationNeuralNetwork.h"
namespace itk
{
namespace Statistics
{
/** Constructor */
template<class TMeasurementVector, class TTargetVector>
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::TwoHiddenLayerBackPropagationNeuralNetwork()
{
typedef IdentityTransferFunction<ValueType> tfType1;
m_InputTransferFunction = tfType1::New();
typedef TanSigmoidTransferFunction<ValueType> tfType2;
m_FirstHiddenTransferFunction = tfType2::New();
m_SecondHiddenTransferFunction = tfType2::New();
typedef TanSigmoidTransferFunction<ValueType> tfType3;
m_OutputTransferFunction= tfType3::New();
typedef SumInputFunction<ValueType*, ValueType> InputFcnType;
m_InputFunction=InputFcnType::New();
m_NumOfInputNodes = 0;
m_NumOfFirstHiddenNodes = 0;
m_NumOfSecondHiddenNodes = 0;
m_NumOfOutputNodes = 0;
m_FirstHiddenLayerBias = 1.0;
m_SecondHiddenLayerBias = 1.0;
m_OutputLayerBias = 1.0;
}
/** Intialize */
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::Initialize()
{
if(m_NumOfInputNodes == 0 )
{
itkExceptionMacro("ERROR: Number of Input Nodes must be greater than 0!");
}
if(m_NumOfFirstHiddenNodes == 0 )
{
itkExceptionMacro("ERROR: Number of Hidden Layer 1 Nodes must be greater than 0!");
}
if(m_NumOfSecondHiddenNodes == 0 )
{
itkExceptionMacro("ERROR: Number of Hidden Layer 2 Nodes must be greater than 0!");
}
if(m_NumOfOutputNodes == 0 )
{
itkExceptionMacro("ERROR: Number of Output Nodes must be greater than 0!");
}
//Define weights of Nodes
typename LearningLayerType::WeightSetType::Pointer InputLayerOutputWeights = LearningLayerType::WeightSetType::New();
InputLayerOutputWeights->SetNumberOfInputNodes(m_NumOfInputNodes);
InputLayerOutputWeights->SetNumberOfOutputNodes(m_NumOfFirstHiddenNodes);
InputLayerOutputWeights->SetCompleteConnectivity();
InputLayerOutputWeights->SetBias(m_FirstHiddenLayerBias);
InputLayerOutputWeights->SetRange(1.0); //0.5
InputLayerOutputWeights->Initialize();
typename LearningLayerType::WeightSetType::Pointer HiddenLayer1OutputWeights = LearningLayerType::WeightSetType::New();
HiddenLayer1OutputWeights->SetNumberOfInputNodes(m_NumOfFirstHiddenNodes);
HiddenLayer1OutputWeights->SetNumberOfOutputNodes(m_NumOfSecondHiddenNodes);
HiddenLayer1OutputWeights->SetCompleteConnectivity();
HiddenLayer1OutputWeights->SetBias(m_SecondHiddenLayerBias);
HiddenLayer1OutputWeights->SetRange(1.0); //0.5
HiddenLayer1OutputWeights->Initialize();
typename LearningLayerType::WeightSetType::Pointer HiddenLayer2OutputWeights = LearningLayerType::WeightSetType::New();
HiddenLayer2OutputWeights->SetNumberOfInputNodes(m_NumOfSecondHiddenNodes);
HiddenLayer2OutputWeights->SetNumberOfOutputNodes(m_NumOfOutputNodes);
HiddenLayer2OutputWeights->SetCompleteConnectivity();
HiddenLayer2OutputWeights->SetBias(m_OutputLayerBias);
HiddenLayer2OutputWeights->SetRange(1.0); //0.5
HiddenLayer2OutputWeights->Initialize();
//Define layers
typename LearningLayerType::Pointer inputlayer = LearningLayerType::New();
inputlayer->SetLayerTypeCode(LearningLayerType::INPUTLAYER);
inputlayer->SetNumberOfNodes(m_NumOfInputNodes);
inputlayer->SetTransferFunction(m_InputTransferFunction);
inputlayer->SetNodeInputFunction(m_InputFunction);
typename LearningLayerType::Pointer hiddenlayer1 = LearningLayerType::New();
hiddenlayer1->SetLayerTypeCode(LearningLayerType::HIDDENLAYER);
hiddenlayer1->SetNumberOfNodes(m_NumOfFirstHiddenNodes);
hiddenlayer1->SetTransferFunction(m_FirstHiddenTransferFunction);
hiddenlayer1->SetNodeInputFunction(m_InputFunction);
typename LearningLayerType::Pointer hiddenlayer2 = LearningLayerType::New();
hiddenlayer2->SetLayerTypeCode(LearningLayerType::HIDDENLAYER);
hiddenlayer2->SetNumberOfNodes(m_NumOfSecondHiddenNodes);
hiddenlayer2->SetTransferFunction(m_SecondHiddenTransferFunction);
hiddenlayer2->SetNodeInputFunction(m_InputFunction);
typename LearningLayerType::Pointer outputlayer = LearningLayerType::New();
outputlayer->SetLayerTypeCode(LearningLayerType::OUTPUTLAYER);
outputlayer->SetNumberOfNodes(m_NumOfOutputNodes);
outputlayer->SetTransferFunction(m_OutputTransferFunction);
outputlayer->SetNodeInputFunction(m_InputFunction);
Superclass::AddLayer(inputlayer);
Superclass::AddLayer(hiddenlayer1);
Superclass::AddLayer(hiddenlayer2);
Superclass::AddLayer(outputlayer);
Superclass::AddWeightSet(InputLayerOutputWeights);
Superclass::AddWeightSet(HiddenLayer1OutputWeights);
Superclass::AddWeightSet(HiddenLayer2OutputWeights);
//HACK: NOTE: You can not set the WeightSets until after the layers are added to the network because
// the LayerId's must have been set prior to the Weights being added to the layers.
// The ordering of putting together the networks is crucial. Layers must be added to network
// prior to weights being added to layers.
inputlayer->SetOutputWeightSet(InputLayerOutputWeights);
hiddenlayer1->SetInputWeightSet(InputLayerOutputWeights);
hiddenlayer1->SetOutputWeightSet(HiddenLayer1OutputWeights);
hiddenlayer2->SetInputWeightSet(HiddenLayer1OutputWeights);
hiddenlayer2->SetOutputWeightSet(HiddenLayer2OutputWeights);
outputlayer->SetInputWeightSet(HiddenLayer2OutputWeights);
}
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::SetInputTransferFunction(TransferFunctionInterfaceType* f)
{
m_InputTransferFunction=f;
}
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::SetFirstHiddenTransferFunction(TransferFunctionInterfaceType* f)
{
m_FirstHiddenTransferFunction=f;
}
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::SetOutputTransferFunction(TransferFunctionInterfaceType* f)
{
m_OutputTransferFunction=f;
}
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::SetInputFunction(InputFunctionInterfaceType* f)
{
m_InputFunction=f;
}
/** Generate output */
template<class TMeasurementVector, class TTargetVector>
typename TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector, TTargetVector>::NetworkOutputType
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::GenerateOutput(TMeasurementVector samplevector)
{
return Superclass::GenerateOutput(samplevector);
}
/** Print the object */
template<class TMeasurementVector, class TTargetVector>
void
TwoHiddenLayerBackPropagationNeuralNetwork<TMeasurementVector,TTargetVector>
::PrintSelf( std::ostream& os, Indent indent ) const
{
os << indent << "TwoHiddenLayerBackPropagationNeuralNetwork(" << this << ")" << std::endl;
os << indent << "m_NumOfInputNodes = " << m_NumOfInputNodes << std::endl;
os << indent << "m_NumOfFirstHiddenNodes = " << m_NumOfFirstHiddenNodes << std::endl;
os << indent << "m_NumOfSecondHiddenNodes = " << m_NumOfSecondHiddenNodes << std::endl;
os << indent << "m_NumOfOutputNodes = " << m_NumOfOutputNodes << std::endl;
os << indent << "m_FirstHiddenLayerBias = " << m_FirstHiddenLayerBias << std::endl;
os << indent << "m_OutputLayerBias = " << m_OutputLayerBias << std::endl;
os << indent << "m_InputFunction = " << m_InputFunction << std::endl;
os << indent << "m_InputTransferFunction = " << m_InputTransferFunction << std::endl;
os << indent << "m_FirstHiddenTransferFunction = " << m_FirstHiddenTransferFunction << std::endl;
os << indent << "m_OutputTransferFunction = " << m_OutputTransferFunction << std::endl;
Superclass::PrintSelf( os, indent );
}
} // end namespace Statistics
} // end namespace itk
#endif
|
