- Replace magic number '1' (class column) with Config.classColumn var…

…iable.

- Comments
- Keep only inner speech condition in LSTM and EEGNet

Python_Processing/Utilitys.py

@@ -145,6 +145,9 @@ def getFullNameFromSbjNumber(sbjNumber):
 
 
 def filterCondition(X, Y, conditionID, discardNonEssentialCols=True):
+    # If discardNonEssentialCols is set to True, it will only return the Class and Session# columns and discard the other 2
+    # This is needed in saveSubjectDataToMat.py because we need only those two columns.
+
     # Keep only trials for given condition
     idxInnerSpeechTrials = (Y[:, Config.conditionColumn] == conditionID)
 
@@ -153,7 +156,7 @@ def filterCondition(X, Y, conditionID, discardNonEssentialCols=True):
 
     if discardNonEssentialCols:
         # Keep only essential columns
-        columnsToKeep = [Config.classColumn, Config.sessionColumn]  # Class and Session# columns
+        columnsToKeep = [Config.classColumn, Config.sessionColumn]
         Y = Y[:, columnsToKeep]
 
     return X, Y

testLSTM.py

@@ -24,14 +24,14 @@
 
 # Load all trials for a single subject
 X, Y = Extract_data_from_subject(Config.datasetDir, N_S, datatype)
-#X, Y = filterCondition(X, Y, Config.idInnerCondition, discardNonEssentialCols=False)
+X, Y = filterCondition(X, Y, Config.idInnerCondition, discardNonEssentialCols=False)
 
 
 electrodes = ["D5","D6","D7","D8","D9","D10","D11","D12","D13","D14","D15","D16","D17","D18","D19","D20","D21","D22","D23","D24","D25","D26","D27","D28","D29","D30","D31","D32"]
 #electrodes = ["A5","A6","A7","A8","A9","A10","A11","A14","A16","A17","A18","A19","A21","A24","A25","A28","A29","B3","B5","B6","B7","B8","B9","B10","B11","B12","B13","B16","B21","B26","C10","C13","C15","C16","C18","C28","C29","D5","D17","D18","D19","D20","D25","D27","D28","D29","D30"]
 #X = selectElectrodes(X, electrodes)
 X = Select_time_window(X)  # Keep only action interval
-y = Y[:, 1]
+y = Y[:, Config.classColumn]
 
 useSpectralDomainData = False
 if useSpectralDomainData:
@@ -79,7 +79,7 @@
 sgd = SGD(learning_rate=0.001)
 adam = Adam(learning_rate=0.0001)
 model.compile(optimizer=adam, loss='sparse_categorical_crossentropy', metrics=['accuracy'])
-model.fit(X_train, y_train, epochs=40, batch_size=64, validation_split=0.25)
+model.fit(X_train, y_train, epochs=40, batch_size=16, validation_split=0.25)
 
 
 #print(model.predict(X_test))

testSVM.py

@@ -30,7 +30,7 @@
 electrodes = ["A5","A6","A7","A8","A9","A10","A11","A14","A16","A17","A18","A19","A21","A24","A25","A28","A29","B3","B5","B6","B7","B8","B9","B10","B11","B12","B13","B16","B21","B26","C10","C13","C15","C16","C18","C28","C29","D5","D17","D18","D19","D20","D25","D27","D28","D29","D30"]
 #X = selectElectrodes(X, electrodes)
 X = Select_time_window(X)  # Keep only action interval
-y = Y[:, 1]
+y = Y[:, Config.classColumn]
 
 psds, freqs = psd_array_welch(X, 256, fmin=8, fmax=100, verbose=False)
 X = psds.reshape(psds.shape[0], -1)  # Flatten data into 2D

testSVΜ_AllSubjects.py

@@ -27,9 +27,9 @@
 
     electrodes = ["D5","D6","D7","D8","D9","D10","D11","D12","D13","D14","D15","D16","D17","D18","D19","D20","D21","D22","D23","D24","D25","D26","D27","D28","D29","D30","D31","D32"]
     #electrodes = ["A5","A6","A7","A8","A9","A10","A11","A14","A16","A17","A18","A19","A21","A24","A25","A28","A29","B3","B5","B6","B7","B8","B9","B10","B11","B12","B13","B16","B21","B26","C10","C13","C15","C16","C18","C28","C29","D5","D17","D18","D19","D20","D25","D27","D28","D29","D30"]
-    X = selectElectrodes(X, electrodes)
+    #X = selectElectrodes(X, electrodes)
     X = Select_time_window(X)  # Keep only action interval
-    y = Y[:, 1]  # Keep only labels
+    y = Y[:, Config.classColumn]  # Keep only labels
 
     psds, freqs = psd_array_welch(X, 256, fmin=8, fmax=100, verbose=False)
     X = psds.reshape(psds.shape[0], -1)  # Flatten data into 2D