Skip to content

GitLab

Commit 827f9253 authored by Carlos Mejia's avatar Carlos Mejia
Browse files

Regroupement des extractions des variables depuis VARARGIN dans Code-2S-SOM/learn_2s_som.m

parent 9334adea
Hide whitespace changes
Inline Side-by-side
Showing with 245 additions and 199 deletions
Code-2S-SOM/learn_2s_som.m
View file @ 827f9253
......@@ -47,13 +47,21 @@ function [sMap sMap_denorm Result] = learn_2s_som(A,nb_neurone,varargin)
% Detailed explanation goes here
% Valeurts par defaut
tracking = 0;
init = 'lininit';
lattice = 'rect';
bool_norm = 0;
% Valeurs par defaut
tracking = 0;
init = 'lininit';
lattice = 'rect';
Result = [];
bool_verbose = 0;
bool_norm = 0;
bool_rad = 0;
bool_trainlen = 0;
bool_2ssom = 0;
bool_DimData = 0;
bool_lambda = 0;
bool_eta = 0;
Result = [];
init_with_make = 1;
pre_train = 1;
......@@ -76,26 +84,68 @@ function [sMap sMap_denorm Result] = learn_2s_som(A,nb_neurone,varargin)
while (i<=length(varargin))
if ischar(varargin{i})
switch varargin{i},
case { 'verbose', '-verbose' },
bool_verbose = 1;
case { 'data_name' },
data_casename = varargin{i+1};
i=i+1;
data_casename = varargin{i+1}; i=i+1;
case { 'comp_names' },
data.colheaders = varargin{i+1};
i=i+1;
data.colheaders = varargin{i+1}; i=i+1;
case { 'norm' },
bool_norm = 1;
type_norm = varargin{i+1}; i=i+1;
case { 'init' },
init = varargin{i+1}; i=i+1;
case { 'tracking' },
tracking = varargin{i+1}; i=i+1;
case { 'lattice' },
lattice = varargin{i+1}; i=i+1;
case 'radius'
bool_rad = 1;
rad = varargin{i+1}; i=i+1;
case 'trainlen'
bool_trainlen = 1;
trlen = varargin{i+1}; i=i+1;
case 'S2-SOM'
disp('** S2-SOM Active **');
bool_2ssom = 1;
case 'DimData'
DimData = varargin{i+1}; i=i+1;
for di=1:length(DimData)
DimBloc(di).Dim = DimData(di);
end
bool_DimData = 1;
case 'lambda'
lambda=varargin{i+1}; i=i+1;
if length(lambda) < 1
error('lambda est de longueur nulle ! Il doit y avoir au moins une valeur')
end
bool_lambda = 1;
case 'eta'
eta = varargin{i+1}; i=i+1;
if length(eta) < 1
error('eta est de longueur nulle ! Il doit y avoir au moins une valeur')
end
bool_eta = 1;
otherwise
error(sprintf(' *** %s error: argument(%d) ''%s'' inconnu ***\n', ...
mfilename, i, varargin{i}));
end
else
error(sprintf(' *** %s error: argument non-string inattendu (en %d-iemme position) ***\n', ...
mfilename, i));
end
i=i+1;
end
sD = som_data_struct(data.data,'name', data_casename,'comp_names', upper(ListVar));
i=1;
while (i<=length(varargin) && bool_norm==0)
if strcmp(varargin{i},'norm')
bool_norm=1;
type_norm=varargin{i+1};
end
i=i+1;
end
% i=1;
% while (i<=length(varargin) && bool_norm==0)
% if strcmp(varargin{i},'norm')
% bool_norm=1;
% type_norm=varargin{i+1};
% end
% i=i+1;
% end
%normalisation des donnees
if bool_norm
......@@ -111,21 +161,21 @@ function [sMap sMap_denorm Result] = learn_2s_som(A,nb_neurone,varargin)
end
if ~isempty(varargin)
i=1;
while i<=length(varargin)
if strcmp(varargin{i},'init')
init=varargin{i+1};
end
if strcmp(varargin{i},'tracking')
tracking=varargin{i+1};
end
if strcmp(varargin{i},'lattice')
lattice=varargin{i+1};
end
i=i+1;
end
end
% if ~isempty(varargin)
% i=1;
% while i<=length(varargin)
% if strcmp(varargin{i},'init')
% init=varargin{i+1};
% end
% if strcmp(varargin{i},'tracking')
% tracking=varargin{i+1};
% end
% if strcmp(varargin{i},'lattice')
% lattice=varargin{i+1};
% end
% i=i+1;
% end
% end
fprintf(1,[ '\n-- ------------------------------------------------------------------\n', ...
'-- New 2S-SOMTraining function:\n', ...
......@@ -165,193 +215,189 @@ function [sMap sMap_denorm Result] = learn_2s_som(A,nb_neurone,varargin)
fprintf(1,' <som init END>.\n')
end
bool_rad=0;
bool_trainlen=0;
if ~isempty(varargin)
% bool_rad=0;
% bool_trainlen=0;
% if ~isempty(varargin)
%
% i=1;
% while i<=length(varargin)
% if ischar(varargin{i})
% switch varargin{i}
% case 'radius'
% bool_rad=1;
% loc_rad=i;
% rad=varargin{loc_rad+1};
% i=i+1;
% case 'trainlen'
% bool_trainlen=1;
% loc_trainlen=i;
% trlen=varargin{loc_trainlen+1};
% i=i+1;
% otherwise
% i=i+1;
% end
% else
% i=i+1;
% end
% end
if pre_train
tracking_ini = tracking;
%tracking_ini = 1;
i=1;
while i<=length(varargin)
if ischar(varargin{i})
switch varargin{i}
case 'radius'
bool_rad=1;
loc_rad=i;
rad=varargin{loc_rad+1};
i=i+1;
case 'trainlen'
bool_trainlen=1;
loc_trainlen=i;
trlen=varargin{loc_trainlen+1};
i=i+1;
otherwise
i=i+1;
end
else
i=i+1;
% batchtrain avec radius ...
if (bool_rad && ~bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec radius ... ')
if tracking_ini, fprintf(1,'\n'); end
j=1;
while j<length(rad)
sMap=som_batchtrain(sMap,sD_norm.data,'radius',[rad(j) rad(j+1)],'tracking',tracking_ini);
j=j+1;
end
end
if pre_train
tracking_ini = tracking;
%tracking_ini = 1;
% batchtrain avec radius ...
if (bool_rad && ~bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec radius ... ')
if tracking_ini, fprintf(1,'\n'); end
%rad=varargin{loc_rad+1};
% batchtrain avec trainlen ...
if (~bool_rad && bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec trainlen ... ')
if tracking_ini, fprintf(1,'\n'); end
j=1;
while j<=length(trlen)
sMap=som_batchtrain(sMap,sD_norm.data,'trainlen',trlen(j),'tracking',tracking_ini);
j=j+1;
end
end
% batchtrain avec radius et trainlen
if (bool_rad && bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec radius et trainlen ... \n')
if tracking_ini, fprintf(1,'\n'); end
if length(rad)==length(trlen)+1
j=1;
while j<length(rad)
sMap=som_batchtrain(sMap,sD_norm.data,'radius',[rad(j) rad(j+1)],'tracking',tracking_ini);
sMap=som_batchtrain(sMap,sD_norm.data,'radius',[rad(j) rad(j+1)],'trainlen',trlen(j),'tracking',tracking_ini);
j=j+1;
end
else
error('vecteur radius doit avoir un element en plus que le vecteur trainlen ')
end
% batchtrain avec trainlen ...
if (~bool_rad && bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec trainlen ... ')
if tracking_ini, fprintf(1,'\n'); end
%trlen=varargin{loc_trainlen+1};
j=1;
while j<=length(trlen)
sMap=som_batchtrain(sMap,sD_norm.data,'trainlen',trlen(j),'tracking',tracking_ini);
j=j+1;
end
current_perf = som_distortion(sMap,sD_norm);
fprintf(1,'--> som_distortion apres entrainement initiale = %s\n', num2str(current_perf));
else
fprintf(1,'** batchtrain initial non active **\n')
end
% end
% %S2-SOM
% bool_2ssom=0;
% bool_DimData=0;
% bool_lambda=0;
% bool_eta=0;
%
% if ~isempty(varargin)
% i=1;
% while i<=length(varargin)
% if ischar(varargin{i})
% switch varargin{i}
%
% case 'S2-SOM'
% disp('** S2-SOM Active **');
% bool_2ssom=1;
% i=i+1;
% %mettre en bloc
% case 'DimData'
% i=i+1;
% DimData=varargin{i};
% for di=1:length(DimData)
% DimBloc(di).Dim=DimData(di);
% end
% bool_DimData=1;
% case 'lambda'
% i=i+1;
% lambda=varargin{i};
% if length(lambda) < 1
% error('lambda est de longueur nulle ! Il doit y avoir au moins une valeur')
% end
% bool_lambda=1;
% case 'eta'
% i=i+1; eta=varargin{i};
% if length(eta) < 1
% error('eta est de longueur nulle ! Il doit y avoir au moins une valeur')
% end
% bool_eta=1;
% otherwise
% i=i+1;
%
% end
% else
% i=i+1;
% end
% end
if (bool_2ssom)
if(bool_lambda && bool_eta && bool_DimData)
best_i = 0;
best_j = 0;
bestperf = inf;
i_train = 1;
n_train = length(lambda)*length(eta);
for i=1:length(lambda)
for j=1:length(eta)
fprintf(1,'\n-- batchtrainRTOM (%d/%d) with lambda=%s and eta=%s ... ',i_train, ...
n_train, num2str(lambda(i)),num2str(eta(j)));
if tracking, fprintf(1,'\n'); end
end
end
% batchtrain avec radius et trainlen
if (bool_rad && bool_trainlen)
fprintf(1,'\n-- BATCHTRAIN initial avec radius et trainlen ... \n')
if tracking_ini, fprintf(1,'\n'); end
%rad=varargin{loc_rad+1};
%trlen=varargin{loc_trainlen+1};
if length(rad)==length(trlen)+1
[Result(i,j).sMap Result(i,j).bmus Result(i,j).Alpha Result(i,j).Beta] = som_batchtrainRTOM( ...
sMap, sD_norm, ...
'TypeAlgo','2SSOM', ...
'DimData',DimData, ...
'DimBloc',DimBloc, ...
'lambda', lambda(i), ...
'eta',eta(j), ...
'radius',[rad(round(length(rad)/2)) ...
rad((round(length(rad)/2))+1)], ...
'trainlen',trlen(round(length(trlen)/2)), ...
'tracking',tracking);
j=1;
while j<length(rad)
sMap=som_batchtrain(sMap,sD_norm.data,'radius',[rad(j) rad(j+1)],'trainlen',trlen(j),'tracking',tracking_ini);
j=j+1;
current_perf = som_distortion(Result(i,j).sMap,sD_norm);
fprintf(1,' --> som_distortion=%s\n', num2str(current_perf));
% end
%end
% best_i=0;
% best_j=0;
% bestperf=inf;
% for i=1:length(lambda)
% for j=1:length(eta)
%
Result(i,j).Perf = current_perf;
if Result(i,j).Perf < bestperf
best_i = i;
best_j = j;
end
else
error('vecteur radius doit avoir un element en plus que le vecteur trainlen ')
i_train = i_train + 1;
end
end
current_perf = som_distortion(sMap,sD_norm);
fprintf(1,'--> som_distortion apres entrainement initiale = %s\n', num2str(current_perf));
sMap = Result(best_i,best_j).sMap;
else
fprintf(1,'** batchtrain initial non active **\n')
error('manque de parametre')
end
elseif (bool_lambda || bool_eta || bool_DimData)
error('mentionnez si vous voulez S2-SOM')
end
%S2-SOM
bool_2ssom=0;
bool_DimData=0;
bool_lambda=0;
bool_eta=0;
if ~isempty(varargin)
i=1;
while i<=length(varargin)
if ischar(varargin{i})
switch varargin{i}
case 'S2-SOM'
disp('** S2-SOM Active **');
bool_2ssom=1;
i=i+1;
%mettre en bloc
case 'DimData'
i=i+1;
DimData=varargin{i};
for di=1:length(DimData)
DimBloc(di).Dim=DimData(di);
end
bool_DimData=1;
case 'lambda'
i=i+1;
lambda=varargin{i};
if length(lambda) < 1
error('lambda est de longueur nulle ! Il doit y avoir au moins une valeur')
end
bool_lambda=1;
case 'eta'
i=i+1; eta=varargin{i};
if length(eta) < 1
error('eta est de longueur nulle ! Il doit y avoir au moins une valeur')
end
bool_eta=1;
otherwise
i=i+1;
end
else
i=i+1;
end
end
if (bool_2ssom)
if(bool_lambda && bool_eta && bool_DimData)
best_i = 0;
best_j = 0;
bestperf = inf;
i_train = 1;
n_train = length(lambda)*length(eta);
for i=1:length(lambda)
for j=1:length(eta)
fprintf(1,'\n-- batchtrainRTOM (%d/%d) with lambda=%s and eta=%s ... ',i_train, ...
n_train, num2str(lambda(i)),num2str(eta(j)));
if tracking, fprintf(1,'\n'); end
[Result(i,j).sMap Result(i,j).bmus Result(i,j).Alpha Result(i,j).Beta] = som_batchtrainRTOM( ...
sMap, sD_norm, ...
'TypeAlgo','2SSOM', ...
'DimData',DimData, ...
'DimBloc',DimBloc, ...
'lambda', lambda(i), ...
'eta',eta(j), ...
'radius',[rad(round(length(rad)/2)) ...
rad((round(length(rad)/2))+1)], ...
'trainlen',trlen(round(length(trlen)/2)), ...
'tracking',tracking);
current_perf = som_distortion(Result(i,j).sMap,sD_norm);
fprintf(1,' --> som_distortion=%s\n', num2str(current_perf));
% end
%end
% best_i=0;
% best_j=0;
% bestperf=inf;
% for i=1:length(lambda)
% for j=1:length(eta)
%
Result(i,j).Perf = current_perf;
if Result(i,j).Perf < bestperf
best_i = i;
best_j = j;
end
i_train = i_train + 1;
end
end
sMap = Result(best_i,best_j).sMap;
else
error('manque de parametre')
end
elseif (bool_lambda || bool_eta || bool_DimData)
error('mentionnez si vous voulez S2-SOM')
end
end
% end
% denormalisation de la Map
if bool_norm
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment