Assignment 4 CS5370: Deep Icarning for Vision/A15100: Deep Learning/A12100: Decp Learning IIT-Hyderabad Jan-Apr 2021 Max Marks: 50 Due: 30th Mar 2021 11:59 pm Instructions Please use Google Classroom to upload your submission by the deadline mentioned above. Your submission should comprise of a single ZIP file, named Assign4, with all your solutions, including code. For late submissions, 10% is deducted for each day (including weekend) late after an assignment is due. Note that each student begins the course with 10 grace days for late submission of assignments, of which upto 1 grace days can be used for a single assignment. Late submissions will automatically use your grace days balance, if you have any left. You can see your balance on the Marks and Grace Days document, soon to be shared under the course Google drive. You have to use PYTHON for the programming questions. . Please read the Klepartment plagiarism policy. Do not cngage in any form of cheating - strict penalties will be imposed for both givers and takers. Please talk to instructor or TA if you have concerns. 1 Theory (15 marks) You can submit your response as a PDF document, which can be typed out in LaTeX/Worl or handurillen and scanned. I/ handwritten, please ensure legibility of answers. 1. (3 marks) Consider the following time-unrolled RNN: E2 Ez 02 03 W H1 H2 H3 U 11 where Hi: 0; 1; are the hidden states, output states and input states respectively. U.VW are weights of the RNN connecting clifferent modules as shown in the figure. If crrors corresponding to each output O: are E, and E is the total error of the network, compute the following gradients: 1 Assignment 4 CS5370: Deep Icarning for Vision/A15100: Deep Learning/A12100: Decp Learning IIT-Hyderabad Jan-Apr 2021 Max Marks: 50 Due: 30th Mar 2021 11:59 pm Instructions Please use Google Classroom to upload your submission by the deadline mentioned above. Your submission should comprise of a single ZIP file, named Assign4, with all your solutions, including code. For late submissions, 10% is deducted for each day (including weekend) late after an assignment is due. Note that each student begins the course with 10 grace days for late submission of assignments, of which upto 1 grace days can be used for a single assignment. Late submissions will automatically use your grace days balance, if you have any left. You can see your balance on the Marks and Grace Days document, soon to be shared under the course Google drive. You have to use PYTHON for the programming questions. . Please read the Klepartment plagiarism policy. Do not cngage in any form of cheating - strict penalties will be imposed for both givers and takers. Please talk to instructor or TA if you have concerns. 1 Theory (15 marks) You can submit your response as a PDF document, which can be typed out in LaTeX/Worl or handurillen and scanned. I/ handwritten, please ensure legibility of answers. 1. (3 marks) Consider the following time-unrolled RNN: E2 Ez 02 03 W H1 H2 H3 U 11 where Hi: 0; 1; are the hidden states, output states and input states respectively. U.VW are weights of the RNN connecting clifferent modules as shown in the figure. If crrors corresponding to each output O: are E, and E is the total error of the network, compute the following gradients: 1