You want to create a classification application using a convolutional neural network to sort different qualities of cucumbers at a farm in Cameron Highlands. For the dataset, you have collected thousands of pictures annotated with labels of high quality, medium quality, low quality and damaged cucumbers. (a) Discuss TWO (2) advantages of using convolutional layers over fully connected layers. [4 marks] (b) Explain the importance of using non-linearities between the layers of neural networks? [2 marks] (c) As you train your model, you realize that you do not have enough data to classify for damaged or low quality cucumbers. Suggest THREE (3) data augmentation techniques that can be used to increase the amount of training data. [3 marks] (d) Briefly explain why dividing a dataset into train, validation and test sets can assist to identify overfitting? [3 marks] (e) Your training and validation accuracies are as in the graph below. What can you do to improve the results? [3 marks] Accuracy Train Validation epoch