Total No. of Questions : 8] SEAT No.

8
23
PB3799 [6262]-58 [Total No. of Pages :2

ic-
tat
T.E. (Artificial Intelligence and Data Science)

1s
ARTIFICIAL NEURAL NETWORK

6:3
02 91
(2019 Pattern) (Semester- II) (317531)

9:4
0
40
Time : 2½ Hours ] 1/0 13 [Max. Marks : 70
0
Instructions to the candidates:
5/2
.23 GP

1) Solve questions Q.1 or Q.2, Q.3 or Q.4, Q.5 or Q.6, Q.7 or Q.8.
2) Neat diagrams must be drawn wherever necessary.
E
82

8
3) Figures to the right indicate full marks.
C

23
ic-
4) Assume suitable data if necessary.
16

tat
8.2

1s
Q1) a) How does Hopefield network work and state its limitations. [6]
.24

6:3
91
b) Exemplify stimulated annealing with its advantages and disadvantages.[6]
49

9:4
30

c) Define: [6]
40
01
02

i) Pattern association
5/2
GP

ii) Pattern classification

1/0

iii) Pattern mapping tasks

CE
82

8
23
.23

OR
ic-
16

tat
Q2) a) Explain in detail stochastic gradient approach. [6]
8.2

1s

b) State basic functional units of ANN for pattern recognition tasks. [6]
.24

6:3
91
49

c) What is catastrophic forgetting in neural network? [6]

9:4
30
40
01
02

Q3) a) Why Kohonens network are called self organizing maps? [6]
5/2
GP

b) What is Adaptive Resonance Theory and its applications? [6]

1/0

c) Define following: [6]

CE
82

i) Learning vector quantization

.23

ii) Adaptive pattern classification

16
8.2

OR
.24
49

P.T.O.
Q4) a) How to recognize character using ART network? [6]

8
23
b) What is competitive learning in neural network and its limitations? [6]

ic-
tat
c) Explain SOM architecture and its uses. [6]

1s
6:3
02 91
9:4
Q5) a) Why do we prefer Convolution Neural Networks(CNN) Over Artificial

0
Neural Networks(ANN) for image data as input? [6]

40
1/0 13
b) Write short note on: [6]
0
5/2
.23 GP

i) AlexNET
ii) VGG-16
E
82

8
C

23
iii) Residual networks

ic-
c) Explain the role of the flattening layer in CNN. [5]
16

tat
8.2

1s
OR
.24

6:3
91
49

9:4
Q6) a) What exactly is a CNN and how does it work? [6]
30
40

b) Define bias and variance. What is bias-variance trade-off? [6]

01
02

c) What do we use a pooling layer in a CNN? [5]

5/2
GP
1/0
CE
82

8
23
Q7) a) Explain automatic language translation with its three basic rules. [6]
.23

b) Exemplify recognition of Olympic Games symbols. ic-[6]

16

tat
8.2

c) What is NET talk? [5]

1s
.24

6:3
91

OR
49

9:4
30
40

Q8) a) Exemplify pattern classification? [6]

01
02

b) Write a short note on: [6]

5/2
GP

i) Texture classification
1/0

ii) Texture segmentation

CE
82

c) Illustrate about Neocognitron? [5]

.23
16
8.2


.24
49

[6262]-58 2