Minor refactor

README.md

@@ -15,7 +15,6 @@ pip install -r requirements.txt
 
 Perform experiment by running the following lines:
 ```
-cd cogito-nlp-workshop-spring-2020/
 python train.py
 ```
 

metrics.py

@@ -0,0 +1,21 @@
+from keras import backend as K
+
+
+def recall_m(y_true, y_pred):
+        true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))
+        possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))
+        recall = true_positives / (possible_positives + K.epsilon())
+        return recall
+
+
+def precision_m(y_true, y_pred):
+        true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))
+        predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))
+        precision = true_positives / (predicted_positives + K.epsilon())
+        return precision
+
+
+def f1_m(y_true, y_pred):
+    precision = precision_m(y_true, y_pred)
+    recall = recall_m(y_true, y_pred)
+    return 2 * ((precision * recall) / (precision + recall + K.epsilon()))

models.py

@@ -0,0 +1,21 @@
+from keras.layers import Input, Dense, Dropout
+from keras.models import Model
+
+
+def get_model(num_feats, class_nbs):
+	x = Input(shape=(num_feats, ))
+	shared = Dense(64, activation="relu")(x) # x
+	sub1 = Dense(32, activation="relu")(shared)
+	#sub1 = Dropout(0.5)(sub1)
+
+	sub2 = Dense(32, activation="relu")(shared)
+	#sub2 = Dropout(0.5)(sub2)
+
+	sub3 = Dense(32, activation="relu")(shared)
+	#sub3 = Dropout(0.5)(sub3)
+
+	out1 = Dense(class_nbs[0], activation="softmax")(sub1)
+	out2 = Dense(class_nbs[1], activation="softmax")(sub2)
+	out3 = Dense(class_nbs[2], activation="softmax")(sub3)
+
+	model = Model(inputs=x, outputs=[out1, out2, out3])

notes

@@ -23,9 +23,3 @@ NLP:
 - Main challenge, natural language is ambigeous and complex -> need to kind of guess a lot
 - split sentence in sentences -> split on periods ex.
 - Classify text if positive or negative -> IMDB dataset
-
-
-
-
-
-