Marking Rubric
Question 1
• All libraries are imported in this cell [10 marks].
• Incomplete libraries are imported in this cell [5 marks].
• No libraries are imported in this cell [0 mark].
Question 2
• Fashion MNIST dataset is downloaded from tf.keras.datasets and pre-processed
properly [10 marks].
• Fashion MNIST dataset is downloaded from tf.keras.datasets but no preprocessing
codes are given [5 marks].
• Fashion MNIST dataset is not downloaded from tf.keras.datasets [0 marks].
Question 3
• Some samples are plotted in the grid-shape format where class names are correctly
identified [10 marks].
• Only one sample is plotted where class names are correctly identified [5 marks].
• No images are plotted [0 mark].
Question 4
• Data samples are normalized properly in the range of [0,1] [5 marks].
• Data samples are normalized in the wrong range [2 marks].
• Data samples are not normalized [0 marks].
Question 5
• Neural networks are properly built from tf.keras.sequential with correct parameters:
hidden layers, hidden nodes, input size [10 marks].
• Neural networks are built from tf.keras,sequential with incorrect parameters [5 marks].
• Neural networks are not properly built [0 mark].
Question 6
• Model is successfully compiled with correct loss function and optimizer [5 marks].
• Model is successfully compiled with incorrect loss function and optimizer [2 marks].
• Model is not successfully compiled [0 mark].
Question 7
• Model is successfully trained with correct hyper-parameters [10 marks].
• Model is successfully trained with incorrect hyper-parameters [5 marks].
• Model is not successfully trained [0 mark].
Question 8
• Model is evaluated properly with the testing set and the classification rate is displayed
[5 mark].
• Model is evaluated properly with the testing set but no classification rate is displayed
[2 mark].
• Model is not evaluated properly with the testing set [0 mark].
Question 9
• Images are successfully rescaled to the size of 32*32 [10 marks].
• Images are not rescaled or there is some error in rescaling the images [0 mark].
Question 10
• ResNet50 is successfully built from tensorflow.keras.applications and model is
summarized using model.summary() [10 marks].
• ResNet5- is successfully built from tensorflow.keras.applications but model is not
summarized [5 marks].
• No model is created [0 marks].
Question 11
• Model is compiled properly with correct loss function and optimizer [5 marks].
• Model is compiled properly with incorrect loss function and optimizer [2 marks].
• No model is compiled [0 mark].
Question 12
• Model is trained with correct epoch, batch size and validation split [10 marks]
• Model is trained successfully with incorrect epoch, batch size and validation split [5
marks].
• Model is not trained [0 mark].
Question 13
• Model is evaluated properly with the testing set where the classification rate is
displayed [5 marks].
• Model is evaluated properly with the testing set but no classification rate is displayed
[2 marks].
• Model is not evaluated properly with the testing set [0 mark].
Question 14
• Model is successfully changed to VGG16 with correct training configurations [20
marks].
• Model is successfully changes to VGG16 with incorrect training configurations [15
marks].
• No model is changed [0 mark].
Question 15
• Model is evaluated properly with the testing set where the classification rate is
displayed [5 marks].
• Model is evaluated properly with the testing set but no classification rate is displayed
[2 marks].
• Model is not evaluated properly with the testing set [0 mark].
Question 16
• The reason is correctly described and supported by solid evidence: plots, numerical
results, counter examples, etc [15 marks].
• The reason is correctly described but not supplemented by solid evidence [8 marks].
• The reason is incorrectly described [0 mark].
Theoretical Questions
Question 1
• Successfully explain non-optimality of DFS with a supporting example [10 marks]
• Successfully explain non-optimality of DFS without a supporting example [5 marks]
• Incorrectly explain non-optimality of DFS [0 mark]
Question 2
• Correct heuristic function is chosen with a supporting example [10 marks]
• Correct heuristic function is chosen without a supporting example [5 marks]
• Incorrect heuristic function is chosen [0 mark]
Question 3
• Successfully explain the ordered crossover with a supporting example [10 marks]
• Successfully explain the ordered crossover without a supporting example [5 marks]
• Incorrectly explain the ordered crossover [0 mark]
Question 4
• Correct design of neural networks with numerical examples [20 marks]
• Correct designs of neural networks without numerical examples [10 marks]
• Incorrect designs of neural networks [0 mark]
