Reg.

No:

SET A
DEPARTMENT OF ARTIFICIAL INTELLIGENCE AND DATA SCIENCE
CAT II Examination - NOV 2024
AD3501-DEEP LEARNING
Year/Sem/Sec : III/V Date : 04.11.2024
Department : AI&DS Duration : 3 hours
Faculty Name : ATHIRAYAN.S Max. Marks : 100
PART A (10 x 2= 20)
CO
1. What is meant by scalar? R
3
CO
2. What is meant by Probability Distribution? R
3
CO
3. Write about Gradient descent U
3
CO
4. Write the concept about hyperparameter U
3
What are some of the most used applications of Deep learning? CO
5. U
3
What is meant by convolution? CO
6. R
4
CO
7. What is meant by parameter sharing? R
4
CO
8. What is meant by pooling? R
4
CO
9. What is tiled convolution? R
4
CO
10. What are the Non-Linear Activation Functions? U
4
PART B (5X13=65)
11(A i) Describe deep learning and machine learning. (7) R CO3
) ii) Give the various concepts of probability. (6) (OR)
11 (i) Discuss the bias variance trade off (7) R CO3
(B) ii) Discuss over fitting and under fitting (6)
Assess the following.
12(A i) Expectation. (5) U CO3
) ii) Variance. (4)
iii) Covariance. (4) (OR)
i) Illustrate the importance of principal components analysis. (6) U
12(B) CO3
ii) Explain support vector machines in detail. (7)
13(A Explain the operations of Deep Feed forward network (13) (OR) U CO3
 )
Explain Probability Mass function and Probability Density function.
13(B) U CO3
(13)
14(A Construct an architecture that show complex layer terminology and U CO4
) Simple layer terminology in convolutional neural network (13) (OR)
Describe in detail about the following.
14(B) i) Parameter Sharing. (7) U CO4
ii) Equivariant representation. (6)
15(A Differentiate locally connected layers, tiled convolution and standard U CO4
) convolution with suitable examples and diagram. (13) (OR)
Write short notes
15(B) Max Pooling. (6) U CO4
Explain Pooling with down sampling. (7)

PART C (1 x15=15)

16( State and explain the concept of Convolution neural network learning.
U CO4
A) (OR)
16( State and Explain optimizers and its types (10)
U CO4
B) Describe Gradient Computation (5)
 Reg.No:

SET B
DEPARTMENT OF ARTIFICIAL INTELLIGENCE AND DATA SCIENCE
CAT II Examination - NOV 2024
AD3501-DEEP LEARNING
Year/Sem/Sec : III/V Date : 04.11.2024
Department : AI&DS Duration : 3 hours
Faculty Name : ATHIRAYAN.S Max. Marks : 100
PART A (10 x 2= 20)
1. What is meant by Vector? R CO3
2. What is meant by Tensors? R CO3
3. Define Stochastic Gradient Descent R CO3
4. illustrate the concept of Over fitting and under fitting U CO3
5. How is Deep learning better than machine learning? U CO3
6. What is meant by sparse interaction in CNN? R CO4
7. What do you mean by Translational equivariance? U CO4
8. What are the advantages of pooling layer? U CO4
9. What is meant by transposed Convolution? R CO4
Why do we use convolutions for images instead of using fully
10. U CO4
connected layers?
PART B (5X13=65)
i) What is Deep Learning? (3)
11(A ii) Describe how different parts of an Artificial Intelligence system R CO3
) relate to each other within different AI disciplines in detail with
diagram. (10) (OR)
11 i. Discuss about scalars.(7) R CO3
(B) ii. Give detail description of Matrices. (6)
12(A Extrapolate conditional probability and develop a summary of various U CO3
) common probability distribution. (13) (OR)
12(B) Explain supervised learning and unsupervised learning algorithm U CO3
13(A
Describe Stochastic Gradient Descent. (13) (OR) U CO3
)
13(B) Explain Principal Components Analysis. (13) U CO3
14(A Discuss in detail the variants of the Basic Convolution Function. (13) U CO4
) (OR)
Write an example of function for Convolution operation and describe U
14(B) CO4
briefly (13)
15(A Explain the following with suitable diagram.
i) Sparse interactions. (6) U CO4
)
Parameter sharing. (7) (OR)
15(B) Describe Pooling with suitable example. (13) U CO4

PART C (1 x15=15)

16(
Explain loss functions and its types with suitable Diagrams. (15) (OR) U CO4
A)
16(
Describe the Concept of Convolution Variants. (15) U CO4
B)