Total No. of Questions : 8] SEAT No.

8
23
P-559 [Total No. of Pages : 2

ic-
tat
[6004]-494

4s
0:4
B.E. (Computer Engineering)

02 91
4:2
DEEP LEARNING

0
31
9/0 13
(2019 Pattern) (Semester - VIII) (410251)
0
6/2
Time : 2½ Hours] [Max. Marks : 70
.23 GP

Instructions to the candidates :

E
80

1) Solve Q.1 or Q.2, Q.3 or Q.4, Q.5 or Q.6, Q.7or Q.8

8
C

23
2) Figures to the right indicate full marks.

ic-
3) Neat diagrams must be drawn whenever necessary.
16

tat
4) Make suitable assumption whenever necessary.
8.2

4s
.24

0:4
91
Q1) a) Explain Pooling Layer with its need and different types. [6]
49

4:2
30

b) Draw and explain CNN (Convolution Neural Network) architecture in

31

detail. [6]
01
02
6/2
GP

c) Explain ReLU Layer in detail. What are the advantages of ReLU over
9/0

Sigmoid? [6]
CE
80

8
OR

23
.23

ic-
Q2) a) Explain all the features of pooling layer. [6]
16

tat
b) Explain Dropout Layer in Convolutional Neural Network. [6]
8.2

4s
.24

0:4

c) Explain working of Convolution Layer with its features. [6]

91
49

4:2
30
31

Q3) a) What is RNN? What is need of RNN? Explain in brief about working of
01
02

RNN (Recurrent Neural Network). [6]

6/2
GP
9/0

b) How LSTM and Bidirectional LSTM works. [6]

CE
80

c) Explain Unfolding computational graphs with example. [5]

.23

OR
16
8.2

P.T.O.
.24
49
Q4) a) What are types of RNN (Recurrent Neural Network)? How to train RNN

8
23
explain in brief. [6]

ic-
b) Explain Encoder-Decoder Sequence to Sequence architecture with its

tat
application. [6]

4s
0:4
c) Differentiate between Recurrent and Recursive Neural Network. [5]

02 91
4:2
0
31
9/0 13
Q5) a) Explain Boltzmann machine in details. [6]
0
6/2
b) Explain GAN (Generative Adversarial Network) architecture with an
.23 GP

example. [6]
E
80

8
c) Do GANs (Generative Adversarial Network) find real or fake images? If
C

23
yes explain it in detail. [6]

ic-
16

OR

tat
8.2

4s
Q6) a) Differentiate generative and discriminative models in GAN (Generative
.24

0:4
Adversarial Network). [6]
91
49

4:2
30

b) What are applications of GAN (Generative Adversarial Network)? Explain

31

any four in detail. [6]

01
02
6/2

c) Write Short Note on Deep generative model and Deep Belief Networks.[6]
GP
9/0
CE
80

8
Q7) a) Explain Markov Decision Process with Markov property. [6]

23
.23

b) Explain in detail Dynamic programming algorithms for reinforcement

ic-
16

tat
learning. [6]
8.2

4s
.24

c) Explain Simple reinforcement learning for Tic-Tac-Toe. [5]

0:4
91
49

4:2

OR
30
31

Q8) a) Write Short Note on Q Learning and Deep Q-Networks. [6]

01
02

b) What are the challenges of reinforcement learning? Explain any four in

6/2
GP

detail. [6]
9/0
CE
80

c) What is deep reinforcement learning? Explain in detail. [5]

.23
16


8.2
.24

[6004]-494 2
49