Problem & Convolutional Neural Networks
Consider the use of the following convolutional neural network (CNN) for image understanding applications. The image has an input layer with a 5 x 5 resolution, and the network includes a convolution layer (CI), a Max-Pooling layer (PI), and a fully-connected layer.
Filter size: 2 x 2
Max-Pooling Layer
Fully Connected Layer
Convolutional Layer
Input Image
(#)_ Briefly answer the following two questions on the structure and functionality of using the CNN for image understanding. Restrict your answer to about 10 words per question.
1. What is the purpose of the convolution and pooling layers? (2)
2. What is the purpose of the fully connected layer?
(b) Given the input binary image, solve the following problems:
1. Using a 2 x 2 filter as a convolution kernel, determine the number of neurons needed in the convolutional layer. (2)
2. Compute the output feature maps of the convolutional layer with the given 2 x 2 filter in Part (1). All intermediate computation steps must be shown explicitly. (points)
3. Use a 2 x 2 max-pooling layer with a stride distance of 2 to avoid overlapping. Compute the output feature map of the max-pooling layer.