Question

Q8: (8 marks) [basic design problem] Given below is a single node in a neural network. Supposing that d is 4, x={4,2,5,2), and w={0.2,0.3.0.4,0.1), b=0.1, and that the activation function is a standard ReLU, that is =max(0,x), where x is the input to the activation function. b 1 W1 WN X 1 Id (a) What is the output of this node? [2 marks] (b) Describe the difference between, recognition and detection in terms of how you would use a Deep Convolutional Network to solve the problem? [2 marks] (e) In the VGG convolutional neural network, the first layer (CONV1) adopts a set of 3x3 filters for convolution, followed by a ReLU, then the second layer (CONV2) similarly adopts 3x3 Convolution filters, followed again by a ReLU. For the output of a single location (node) in CONV2, how many pixels of the input image would impact on the result? (i.e., if you changed the pixel values). [2 marks] (d) You could remove one of these layers from the network, perhaps modifying the other. Would this have an impact on the quality of the learning? Explain your answer in at most a few sentences. (Note that multiple answers are possible to this question, state any assumptions you make). [2 marks]

can someone answer the part b

Answer 1

Answering the part b as specified in the question

8(b).

The difference between the Recognition and Detection in terms of Deep Convolutional Neural Network is given below :

Object Recognition	Object Detection
In terms of Deep Convolutional Neural Network, recognition is the task of assigning the label with the appreciate message based on the type of object.	In terms of Deep Convolutional Neural Network, detection is the task of detecting the object in the image.
Deep Convolutional Neural Network carries out this task using the retained CNN models on the Imagenet dataset.	Deep Convolutional Neural Network carries out this task using the custom CNN model which is trained on a dataset from scratch.
The object recognition requires more training time as compared to object detection.	The object detection requires less training time as compared to object recognition.
The object recognition task also required the help of a classical neural network to assign labels.	The object detection does not require the help of classical neutral network.