Support taille arbitraire de sets dans Dataugv6

higher/dataug.py

@@ -709,17 +709,38 @@ class Data_augV6(nn.Module): #Optimisation sequentielle
         self._fixed_mag = fixed_mag
         
         self._TF_set_size=3
-        #if self._TF_set_size>self._nb_tf:
-        #    print("Warning : TF sets size higher than number of TF. Reducing set size to %d"%self._nb_tf)
-        #    self._TF_set_size=self._nb_tf
-        assert self._nb_tf>=self._TF_set_size
-        self._TF_sets=[]
-        for i in range(1,self._nb_tf):
-            for j in range(i,self._nb_tf):
-                if i!=j:
-                    self._TF_sets+=[torch.tensor([0, i, j])]
-        #print(self._TF_sets)
-        #self._TF_sets=[torch.tensor([0, i, j]) for i in range(1,self._nb_tf)] #All VS Identity
+        self._fixed_TF=[0]
+        assert self._TF_set_size>=len(self._fixed_TF)
+
+        if self._TF_set_size>self._nb_tf:
+            print("Warning : TF sets size higher than number of TF. Reducing set size to %d"%self._nb_tf)
+            self._TF_set_size=self._nb_tf
+        
+        ## Genenerate TF sets ##
+        if self._TF_set_size==len(self._fixed_TF): 
+            print("Warning : using only fixed set of TF : ", self._fixed_TF)
+            self._TF_sets=torch.tensor([self._fixed_TF])
+        else: 
+            def generate_TF_sets(n_TF, set_size, idx_prefix=[]):
+                TF_sets=[]
+                print(set_size, idx_prefix)
+                if len(idx_prefix)!=0:
+                    if set_size>2:
+                        for i in range(idx_prefix[-1]+1, n_TF):
+                            TF_sets += generate_TF_sets(n_TF=n_TF, set_size=set_size-1, idx_prefix=idx_prefix+[i])
+                    else:
+                        #if i not in idx_prefix:
+                        TF_sets+=[torch.tensor(idx_prefix+[i]) for i in range(idx_prefix[-1]+1, n_TF)]
+                elif set_size>1:
+                     for i in range(0, n_TF):
+                        TF_sets += generate_TF_sets(n_TF=n_TF, set_size=set_size, idx_prefix=[i])
+                else:
+                    TF_sets+=[torch.tensor([i]) for i in range(0, n_TF)]
+                return TF_sets
+
+            self._TF_sets=generate_TF_sets(self._nb_tf, self._TF_set_size, self._fixed_TF)
+        
+        ## Plan TF learning schedule ##
         self._TF_schedule = [list(range(len(self._TF_sets))) for _ in range(self._N_seqTF)]
         for n_tf in range(self._N_seqTF) :
             TF.random.shuffle(self._TF_schedule[n_tf])

higher/test_dataug.py

@@ -10,17 +10,17 @@ tf_names = [
     'FlipLR',
     'Rotate',
     'TranslateX',
-    'TranslateY',
-    'ShearX',
-    'ShearY',
+    #'TranslateY',
+    #'ShearX',
+    #'ShearY',
 
     ## Color TF (Expect image in the range of [0, 1]) ##
-    'Contrast',
-    'Color',
-    'Brightness',
-    'Sharpness',
-    'Posterize',
-    'Solarize', #=>Image entre [0,1] #Pas opti pour des batch
+    #'Contrast',
+    #'Color',
+    #'Brightness',
+    #'Sharpness',
+    #'Posterize',
+    #'Solarize', #=>Image entre [0,1] #Pas opti pour des batch
 
     #Color TF (Common mag scale)
     #'+Contrast',
@@ -44,10 +44,10 @@ tf_names = [
     #'BadTranslateY',
     #'BadTranslateY_neg',
 
-    'BadColor',
-    'BadSharpness',
-    'BadContrast',
-    'BadBrightness',
+    #'BadColor',
+    #'BadSharpness',
+    #'BadContrast',
+    #'BadBrightness',
 
     #Non fonctionnel
     #'Auto_Contrast', #Pas opti pour des batch (Super lent)
@@ -91,11 +91,11 @@ if __name__ == "__main__":
     print('-'*9)
     '''
     #### Augmented Model ####
-    '''
+    #'''
     t0 = time.process_time()
     tf_dict = {k: TF.TF_dict[k] for k in tf_names}
     #tf_dict = TF.TF_dict
-    aug_model = Augmented_model(Data_augV6(TF_dict=tf_dict, N_TF=3, mix_dist=0.0, fixed_prob=False, fixed_mag=True, shared_mag=True), LeNet(3,10)).to(device)
+    aug_model = Augmented_model(Data_augV6(TF_dict=tf_dict, N_TF=1, mix_dist=0.0, fixed_prob=False, fixed_mag=True, shared_mag=True), LeNet(3,10)).to(device)
     #aug_model = Augmented_model(Data_augV5(TF_dict=tf_dict, N_TF=2, mix_dist=0.5, fixed_mag=True, shared_mag=True), WideResNet(num_classes=10, wrn_size=160)).to(device)
     #aug_model = Augmented_model(RandAug(TF_dict=tf_dict, N_TF=2), LeNet(3,10)).to(device)
     print(str(aug_model), 'on', device_name)
@@ -116,9 +116,9 @@ if __name__ == "__main__":
 
     print('Execution Time : %.00f '%(time.process_time() - t0))
     print('-'*9)
-    '''
-    #### TF tests ####
     #'''
+    #### TF tests ####
+    '''
     res_folder="res/brutus-tests/"
     epochs= 150
     inner_its = [1]
@@ -168,4 +168,4 @@ if __name__ == "__main__":
                             #plot_resV2(log, fig_name=res_folder+filename, param_names=tf_names)
                             print('-'*9)
 
-    #'''    
+    '''