Deep Learning If4071
Uploaded by
rejirejisha02Deep Learning If4071
Uploaded by
rejirejisha02Question Paper Code : 20687
MEMTech DEGREE EXAMINATIONS, NOVEMBEVDECEMBER
2024.
Elective
Applied Eleetronica
IE 4071 DEEP LEARNING
Common to M E. Comnputer Scicnce and Engineering/M.E. Computer Seience and
NotN
Ne
Engineering (With Specinlization in Artificinl Intelligence and Machine
Computer Seience and Engineering (With Specialization
E Mobile and Pervasive Computing/M.E. Multimedia T'echnology/
in
ME.Software Engineering/ME. VLSI and Embedded Systen
M
Tech. Information Technology)
(Regulations 2021)
Tirme Three bours
:
Maxunum: 100 marks
Answer ALL questions.
PARTA-(10 x2= 20 marks)
How is lesrning rate is defined in machine learning?
2 What is Deep Learning?
What are the disadvantages of Gradient
descent method?
State the common data pre-processing techniques used for
neutral networks.
Give some applications of Convolutional Neural Networks?
6 What is Residual network?
7. Define Word Embedding.
8 List the lunitations of Backpropagation through Time.
9 Compare deep learning and reinforcement learning.
10 What is the primary objective of Deep Network (DQN) in
Q reinforcement
learning?
PART B (6 13 = 65 mnarks)
(a) Describe the
linear approach uscd for binary
modelng classification
Or
(b)
Describe the
percoptronmodel with
example
12 (a)
Explan about feature
netwarks. engineering and its
importance in neural
Or
(b) Discuss about the
categories ofhyperparameters,
13 (a) Explain the implementation of Transfer Jearning with Image Data n
Deep Learning algorithm
Or
(b) Describe about an implication of deep neural networks like VGGTe on
computer vision tasks.
14. (a) Describe the architecture of a Bi-directional Long Short Term Memory
and also mention about its advantages and
disadvantages.
Or
(b) Descnbe the architecture of sequence to sequence (seq 2 seq) models
including roles of the encodes and decodes RNN.
(a) llustrate about Q-table and how is it used in Q-learning
Or
(b) Design and propose a Convolutional Auto Encoder architecture for a any
application
PART C (1* 15 = 15 marks)
16 (a) Deseribe the process of back propagation in deep neural networks with
example.
Or
Networks architecture for Image
(b) Design a Convolutional Neural
Classification.
20687
