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
|
%--------------------------------------------------------------------------
% This file is part of the ASTRA Toolbox
%
% Copyright: 2010-2016, iMinds-Vision Lab, University of Antwerp
% 2014-2016, CWI, Amsterdam
% License: Open Source under GPLv3
% Contact: astra@uantwerpen.be
% Website: http://www.astra-toolbox.com/
%--------------------------------------------------------------------------
classdef DARToptimizer < handle
%----------------------------------------------------------------------
properties (SetAccess=public,GetAccess=public)
% optimization options
max_evals = 100; % SETTING: Maximum number of function evaluations during optimization.
tolerance = 0.1; % SETTING: Minimum tolerance to achieve.
display = 'off'; % SETTING: Optimization output. {'off','on','iter'}
verbose = 'yes'; % SETTING: verbose? {'yes','no}
metric = ProjDiffOptimFunc(); % SETTING: Optimization object. Default: ProjDiffOptimFunc.
% DART options
DART_iterations = 20; % SETTING: number of DART iterations in each evaluation.
D_base = [];
end
%----------------------------------------------------------------------
properties (SetAccess=private,GetAccess=public)
stats = Statistics();
end
%----------------------------------------------------------------------
methods (Access=public)
%------------------------------------------------------------------
% Constructor
function this = DARToptimizer(D_base)
this.D_base = D_base;
% statistics
this.stats = Statistics();
this.stats.register('params');
this.stats.register('values');
this.stats.register('score');
end
%------------------------------------------------------------------
function opt_values = run(this, params, initial_values)
if nargin < 3
for i = 1:numel(params)
initial_values(i) = eval(['this.D_base.' params{i} ';']);
end
end
% fminsearch
options = optimset('display', this.display, 'MaxFunEvals', this.max_evals, 'TolX', this.tolerance);
opt_values = fminsearch(@this.optim_func, initial_values, options, params);
% save to D_base
for i = 1:numel(params)
eval(sprintf('this.D_base.%s = %d;',params{i}, opt_values(i)));
end
end
%------------------------------------------------------------------
end
%----------------------------------------------------------------------
methods (Access=protected)
%------------------------------------------------------------------
function score = optim_func(this, values, params)
% copy DART
D = this.D_base.deepcopy();
% set parameters
for i = 1:numel(params)
eval(sprintf('D.%s = %d;',params{i}, values(i)));
D.output.pre = [D.output.pre num2str(values(i)) '_'];
end
% evaluate
if D.initialized == 0
D.initialize();
end
rng('default');
D.iterate(this.DART_iterations);
% compute score
score = this.metric.calculate(D, this);
% statistics
this.stats.add('params',params);
this.stats.add('values',values);
this.stats.add('score',score);
% output
if strcmp(this.verbose,'yes')
disp([num2str(values) ': ' num2str(score)]);
end
end
%------------------------------------------------------------------
end
end
|
