Update

main.py

@@ -1048,7 +1048,7 @@ def main():
 
         Cross_val = 'combined',
 
-        sample_no = None,
+        sample_no = 400,
         undersampling = False,
         oversampling = False,
 
@@ -1081,14 +1081,15 @@ def main():
         G_hidden_no = 1,
 
 
-        RB1 = 0.13905385615810364, 
-        RLR = 2.6665447517436225e-05, 
-        R_ac_func = 'leaky', 
-        R_aco_func = 'softmax', 
-        R_hidden = 2719, 
-        R_hidden_no = 3, 
-        R_optim = 'AdamW', 
-        R_tau = 0.606889373892653,
+        RB1 = 0.011464688009852337, 
+        RLR = 0.014508298446114655, 
+        R_ac_func = 'leaky20', 
+        R_hidden = 81, 
+        R_hidden_no = 4, 
+        R_optim = 'Adam',
+
+
+
         ) 
 
 

params.py

@@ -119,7 +119,7 @@ def load_trials(P,name=None):
         'User_U'          : 2,                          # User for the Unlabelled data
         'User_V'          : 3,                          # User for the Validation data
 
-        'FX_sel'          : 'basic',                    # Which features to extract
+        'FX_sel'          : 'all',                      # Which features to extract
         'FX_indeces'      : None,                       # Which features to select after extraction. None = all
         'FX_num'          : None,                       # If given, select the n best features (overwrites 'FX_indeces')
 
@@ -134,7 +134,7 @@ def load_trials(P,name=None):
 
         'sample_no'       : None,                       # Not None: number of samples to reduce/increase all classes to
         'undersampling'   : False,                      # True: undersample all majority classes
-        'oversampling'    : True,                       # True: oversample all minority classes  
+        'oversampling'    : False,                      # True: oversample all minority classes  
         'PCA_n_components': None,                       # Number of components for PCA
 
         'epochs'          : 500,                        # Number of regular training epochs
@@ -143,10 +143,6 @@ def load_trials(P,name=None):
         'save_step'       : 10,                         # Number of epochs after which results are stored
         'batch_size'      : 512,                        # Number of samples per batch
         'noise_shape'     : 100,                        # Size of random noise Z
-
-        'G_no'            : 1,                          # Model number of the new generator
-        'D_no'            : 1,                          # Model number of the new discriminator
-        'C_no'            : 1,                          # Model number of the new classifier
 
         'G_label_sample'  : True,                       # True: randomly sample input labels for G | False: use current sample batch as input
         'G_label_factor'  : 1,                          # Size factor of the input for G in relation to current batch
@@ -253,9 +249,7 @@ def get_dataset_hash_str(self):
         return str(self.get_dataset_hash())
 
     def log(self,txt:str,save:bool=True,error:bool=False,name:str=None):
-        if name is None:
-            name = self.get('log_name')
-        writeLog(txt,save=save,error=error,name=name)
+        writeLog(txt,save=save,error=error,name=(self.get('name')+'_log'))
 
     def save(self):
         make_dir_mod()