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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
|
/*=========================================================================
Program: Insight Segmentation & Registration Toolkit
Module: itkMultilayerNeuralNetworkBase.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 __itkMultilayerNeuralNetworkBase_txx
#define __itkMultilayerNeuralNetworkBase_txx
#include "itkMultilayerNeuralNetworkBase.h"
#include "itkErrorBackPropagationLearningFunctionBase.h"
#include "itkErrorBackPropagationLearningWithMomentum.h"
#include "itkQuickPropLearningRule.h"
namespace itk
{
namespace Statistics
{
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::MultilayerNeuralNetworkBase()
{
typedef ErrorBackPropagationLearningWithMomentum<TLearningLayer,TTargetVector> DefaultLearningFunctionType;
m_LearningFunction = DefaultLearningFunctionType::New();
m_LearningRate = 0.001;
//#define __USE_OLD_INTERFACE Comment out to ensure that new interface works
#ifdef __USE_OLD_INTERFACE
m_NumOfLayers = 0;
m_NumOfWeightSets=0;
#endif
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::SetLearningRate(ValueType lr)
{
m_LearningRate=lr;
this->Modified();
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::SetLearningFunction(LearningFunctionInterfaceType* f)
{
m_LearningFunction=f;
this->Modified();
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::~MultilayerNeuralNetworkBase()
{
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::AddLayer(LayerInterfaceType* layer)
{
//Automatically set the layer Id based on position in the layer vector.
layer->SetLayerId(m_Layers.size());
m_Layers.push_back(layer);
//#define __USE_OLD_INTERFACE Comment out to ensure that new interface works
#ifdef __USE_OLD_INTERFACE
m_NumOfLayers++;
#endif
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
typename MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>::LayerInterfaceType*
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::GetLayer(int layer_id)
{
return m_Layers[layer_id].GetPointer();
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
const typename MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>::LayerInterfaceType*
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::GetLayer(int layer_id) const
{
return m_Layers[layer_id].GetPointer();
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
typename MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>::NetworkOutputType
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::GenerateOutput(TMeasurementVector samplevector)
{
this->m_Layers[0]->ForwardPropagate(samplevector);
unsigned int i;
for (i = 0; i < this->m_Layers.size() && i < this->m_Weights.size(); i++)
{
this->m_Weights[i]->ForwardPropagate(
this->m_Layers[i]->GetOutputVector() );
this->m_Layers[i + 1]->ForwardPropagate();
}
NetworkOutputType temp_output;
temp_output.SetSize(this->m_Layers[i]->GetNumberOfNodes());
for(unsigned int k=0; k<temp_output.Size(); k++)
{
temp_output[k]=this->m_Layers[i]->GetOutputVector()[k];
}
return temp_output;
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::BackwardPropagate(NetworkOutputType errors)
{
unsigned int i = this->m_Layers.size();
i--;
this->m_Layers[i]->BackwardPropagate(errors);
i--;
while (i > 0)
{
this->m_Layers[i]->BackwardPropagate();
i--;
}
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::InitializeWeights()
{
unsigned int num_wts = this->m_Weights.size();
for(unsigned int i=0; i<num_wts; i++)
{
this->m_Weights[i]->InitializeWeights();
}
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::UpdateWeights(ValueType itkNotUsed(lr))
{
unsigned int i = this->m_Layers.size();
while(i>1)
{
i--;
m_LearningFunction->Learn(this->m_Layers[i],m_LearningRate);
this->m_Layers[i]->GetInputWeightSet()->UpdateWeights(m_LearningRate);
}
}
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::AddWeightSet(typename LayerInterfaceType::WeightSetInterfaceType* weightset)
{
weightset->SetWeightSetId(m_Weights.size());
m_Weights.push_back(weightset);
//#define __USE_OLD_INTERFACE Comment out to ensure that new interface works
#ifdef __USE_OLD_INTERFACE
m_NumOfWeightSets++;
#endif
}
#ifdef __USE_OLD_INTERFACE
//Moved definition to header in attempt to fix compiler issues on MS Express 5.0 compiler.
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
typename MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>::LayerInterfaceType::WeightSetInterfaceType*
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::GetWeightSet(unsigned int id)
{
return m_Weights[id].GetPointer();
}
#endif
#ifdef __USE_OLD_INTERFACE
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
const typename MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>::LayerInterfaceType::WeightSetInterfaceType*
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::GetWeightSet(unsigned int id) const
{
return m_Weights[id].GetPointer();
}
#endif
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::SetLearningRule(LearningFunctionInterfaceType* l)
{
m_LearningFunction = l;
this->Modified();
}
/** Print the object */
template<class TMeasurementVector, class TTargetVector,class TLearningLayer>
void
MultilayerNeuralNetworkBase<TMeasurementVector,TTargetVector,TLearningLayer>
::PrintSelf( std::ostream& os, Indent indent ) const
{
os << indent << "MultilayerNeuralNetworkBase(" << this << ")" << std::endl;
Superclass::PrintSelf( os, indent );
//os << indent << "m_Layers = " << m_Layers << std::endl;
//os << indent << "m_Weights = " << m_Weights << std::endl;
if(m_LearningFunction)
{
os << indent << "m_LearningFunction = " << m_LearningFunction << std::endl;
}
os << indent << "m_LearningRate = " << m_LearningRate << std::endl;
os << indent << "NumOfLayers = " << m_Layers.size() << std::endl;
os << indent << "NumOfWeightSets = " << m_Weights.size() << std::endl;
}
} // end namespace Statistics
} // end namespace itk
#endif
|
