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
|
/*=========================================================================
*
* Copyright UMC Utrecht and contributors
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0.txt
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
*=========================================================================*/
#include "itkAdaptiveStepsizeOptimizer.h"
#include <vnl/vnl_math.h>
#include "itkSigmoidImageFilter.h"
namespace itk
{
/**
* ************************* Constructor ************************
*/
AdaptiveStepsizeOptimizer::AdaptiveStepsizeOptimizer() = default;
/**
* ************************** UpdateCurrentTime ********************
*/
void
AdaptiveStepsizeOptimizer::UpdateCurrentTime()
{
using SigmoidType = itk::Functor::Sigmoid<double, double>;
if (this->m_StepSizeStrategy == "Adaptive")
{
if (this->GetCurrentIteration() > 0)
{
/** Make sigmoid function
* Compute beta such that sigmoid(0)=0
* We assume Max>0, min<0 */
SigmoidType sigmoid;
sigmoid.SetOutputMaximum(this->GetSigmoidMax());
sigmoid.SetOutputMinimum(this->GetSigmoidMin());
sigmoid.SetAlpha(this->GetSigmoidScale());
const double beta = this->GetSigmoidScale() * std::log(-this->GetSigmoidMax() / this->GetSigmoidMin());
sigmoid.SetBeta(beta);
/** Formula (2) in Cruz */
const double inprod = inner_product(this->m_PreviousSearchDirection, this->GetGradient());
this->m_CurrentTime += sigmoid(-inprod);
this->m_CurrentTime = std::max(0.0, this->m_CurrentTime);
}
/** Save for next iteration */
this->m_PreviousSearchDirection = this->GetSearchDirection();
}
/** Decaying or constant step size. */
else if (this->m_StepSizeStrategy == "Decaying")
{
/** Almost Robbins-Monro: time = time + E_0.
* If you want the parameter estimation but no adaptive stuff,
* this may be use useful: */
// this->m_CurrentTime += ( this->GetSigmoidMax() + this->GetSigmoidMin() ) / 2.0;
this->m_CurrentTime += 1.0;
}
else if (this->m_StepSizeStrategy == "Constant")
{
this->m_CurrentTime = 0.0;
}
} // end UpdateCurrentTime()
} // end namespace itk
|
