Answer the following questions on Convolutional Neural Network (CNN)

(10+10=20 marks) Question 3 Answer the following questions on Convolutional Neural Network (CNN) 1. Figure 1 is a diagram of a small convolutional neural network that converts a 1616 image into 6 output values. The network has the following layers/operations from input to output: convolution with 4 filters, max pooling, ReLu, and finally a fully- connected layer. For this network we will not be using any bias/offset parameters. Please answer the following questions about this network. 4@12x12 1616 4@6x6 1 Convolution fully- max pooling 4 filters 5x5 connected 2x2 Stride 1 Stride 2 Figure 1: A small CNN network. (a) How many weights in the convolutional layer do we need to learn? [2 marks] (b) How many ReLu operations are performed on the forward pass? [2 marks] (c) How many weights do we need to learn for the entire network? [2 marks] (d) True or false: A fully-connected neural network with the same size layers as the above network (16 16 4 12 12 4 x6 6 6 1) can represent any classifier that the above convolutional network can represent. [2 marks] (e) What is the disadvantage of a fully-connected neural network compared to a convolutional neural network with the same size layers? [2 marks] 2. Consider the following issue in problem solving: A R&D team in a company is faced with solving a pattern classification problem, and has decided to use an artificial neural network model. However, the researchers cannot decide between using a multilayer perceptron (MLP) and a Convolutional Neural Network (CNN). In your capacity as a consultant to this team, discuss the pros and cons of both types of the models, in the context of various factors that the team must consider before a fi- nal decision about network type can be made. Illustrate your answer with examples. [10 marks]