Need help with the assignment

1.(1 point) In the Convolutional Neural Networks, the input image is given as 553. The convolutional layer has a filter as 333 as below. The pad is 1 (so the image becomes 7 73 ). The stride is 2 . The bias is 1 . After applying this filter, what is the output volume? Show the output volume of this along with the computation process. 2. (2 points) For edge detection, we are going to try the following two filters (Filter-A and FilterB) for the operation of Convolutional layer in CNN. The pad is 0 (No padding). The stride is 1 . The bias is 0 . ( Filter A) ( Filter- , \begin{tabular}{|l|l|l|} \hline 1 & 2 & 1 \\ \hline 0 & 0 & 0 \\ \hline1 & 2 & 1 \\ \hline \end{tabular} We have gray scale images, typically zero is taken to be black, and 255 is taken to be white. (2-1) We want to apply Filter-A for this Image-1. Show the output volume (66) of this input Image-1 along with the computation process. (22) We want to apply Filter-B for this Image-1. Show the output volume (66) of this input Image- 1 along with the computation process. (2-3) Which filter, Filter-A or Filter-B, is acting as a horizontal detector? (3). (2 points) An excellent overview paper on Deep Learning was published in the Nature. The authors, Yann LeCun, Yoshua Bengio, and Geoffrey Hinton, are pioneers and leading scientists in Deep Learning field. - Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. "Deep learning.", Nature 521.7553, 436444.(2015) You can access the paper through http://www.nature.comature/journal/v5217553/fullature 14539.html or download it from Module 4. Read the paper and answer the following questions. Do not just write down (copy and paste) the contents of the paper. You should write in your own understanding. You can focus on sections of "Convolutional neural networks" and "Image understanding with deep convolutional networks" of the paper. 1. What is the role of Convolutional layer? 2. Why do we need more than one filter? 3. What is the role of pooling layer? 4. How local combinations of edges, motifs, parts, objects are related in a hierarchy of features? 5. What are the major successful application areas of CNN ? 6. Why CNN is so successful