Department of CSE(DS)
B. Tech. Mid Question Bank (R20 Regulation)
Academic Year: 2024-25 Semester: VII
Subject Name: Deep Learning
Faculty Name: Dr. K. Murali, Mr M Ravindran
PART-A
Q.No Questions Marks BL CO Unit No
1 What is Deep Feed Forward networks? 2M BL1 CO1 I
2 List reasonably common hidden unit types. 2M BL1 CO1 I
3 List out different set of layers in Feed forward 2M BL1 CO1 I
networks.
4 Describe Gradient Descent. 2M BL2 CO1 I
5 Give reason for the term “feed forward” used in the 2M BL2 CO1 I
feed forward networks.
6 State Universal Approximation Theorem. 2M BL2 CO1 I
7 Describe Regularization for deep learning. 2M BL2 CO2 II
8 List the common types of parameter norm 2M BL1 CO2 II
penalties.
9 Define dataset augmentation. Why dataset 2M BL2 CO2 II
augmentation is important?
10 Illustrate Semi supervised learning. 2M BL3 CO2 II
11 Why are parameter tying and sharing used? 2M CO2 II
12 Define Dropout. How does dropout prevent 2M BL2 CO2 II
overfitting?
13 How do objectives differ in Learning vs. 2M BL2 CO3 III
Optimization?
14 List the Parameter initialization strategies. 2M BL2 CO3 III
15 Mention the different challenges in optimization. 2M BL3 CO3 III
16 Discuss about Adam optimizer. 2M BL2 CO3 III
17 Write short notes on Quasi-Newton methods. 2M BL2 CO3 III
18 Define Meta-Algorithms? List some common Meta- 2M BL2 CO3 III
Algorithms.
19 What is meant by convolution? 2M BL1 CO4 IV
20 How pooling handles inputs of varying size. 2M BL4 CO4 IV
21 List three stages of a convolutional network. 2M BL1 CO4 IV
22 Explain feature map. 2M BL2 CO4 IV
23 Define unshared convolution. 2M BL1 CO4 IV
24 Write short notes on parameter sharing. 2M BL2 CO4 IV
25 Define Model compression. 2M BL1 CO5 V
26 Write short notes on Dataset augmentation. 2M BL2 CO5 V
27 What is Natural Language Processing? 2M BL2 CO5 V
28 List the main components of a Speech recognition 2M BL2 CO5 V
system.
29 Define Computer vision? List the applications of 2M BL1 CO5 V
computer vision.
30 Define Hierarchy of words. 2M BL1 CO5 V
PART-B
Q.No Questions Marks BL CO Unit No
1 Describe Deep feed forward networks. 3M BL1 CO1 I
2 Discuss about Hidden Units. 3M BL2 CO1 I
3 Explain Cost function in Gradient based learning. 3M BL3 CO1 I
4 Describe Back Propagation algorithm. 3M BL1 CO1 I
5 Write short notes on Architecture Design. 3M BL4 CO1 I
6 List and explain the historical trends in Deep 3M BL1 CO1 I
Learning.
7 Give an example of learning XOR function to explain 6M BL2 CO1 I
a fully functioning feed forward network.
8 Explain in detail about the concept of Gradient 6M BL2 CO1 I
based learning.
9 Develop a Deep Feed forward network and explain. 6M BL6 CO1 I
10 Explain Regularization for deep learning? 3M BL2 CO2 II
11 Define Norm Penalties. How do you formulate an 3M BL2 CO2 II
optimization problem with a norm penalty as a
constrained optimization problem?
12 Explain Multi-Task Learning. 3M BL2 CO2 II
13 Illustrate Meta-algorithm used for early stopping. 3M BL3 CO2 II
14 Explain Adversarial Training. 3M BL2 CO2 II
15 Discuss the steps involved in computing Tangent 3M BL3 CO2 II
distance.
16 What are techniques used for regularization? 6M BL3 CO2 II
Explain.
17 Analyse and write short notes on Dataset 6M BL4 CO2 II
Augmentation.
18 Explain popular ensemble methods besides 6M BL3 CO2 II
Bagging.
19 Explain the role of Optimization in Deep Learning. 3M BL3 CO3 III
20 What is the difference between Learning and Pure 3M BL3 CO3 III
Optimization? Explain.
21 Discuss the challenges in Neural Network 3M BL2 CO3 III
Optimization.
22 Explain the basic Optimization algorithms. 3M BL3 CO3 III
23 List various algorithms with adaptive learning rates. 3M Bl2 CO3 III
Explain.
24 Discuss about supervised Pre-training. 3M BL2 CO3 III
25 Give an example for convolution. 3M BL2 CO4 IV
26 Describe Pooling with suitable example. 3M BL2 CO4 IV
27 Explain the key similarities between CNNs and the 3M BL2 CO4 IV
visual cortex.
28 Discuss in detail the variants of the Basic 3M BL2 CO4 IV
Convolution Function.
29 Explain about parameter sharing. 3M BL2 CO4 IV
30 Illustrate Equivariant representation. 3M BL3 CO4 IV
31 Differentiate locally connected layers, tiled 6M BL4 CO4 IV
convolution and standard convolution with suitable
examples and diagram.
32 Explain the following with suitable diagram. 6M BL4 CO4 IV
i. Sparse interactions.
ii. Parameter sharing.
33 Evaluate variants of the basic convolution function? 6M BL5 CO4 IV
34 Discuss about Large scale distributed 3M BL2 CO5 V
implementations.
35 Explain about Contrast normalization in computer 3M BL3 CO5 V
vision.
36 Illustrate N-Gram models. 3M BL3 CO5 V
37 What are the common types of Speech recognition 3M BL3 CO5 V
systems? Explain.
38 Describe Noise-Contrastive Estimation. 3M BL2 CO5 V
39 How to speeding-up Gradient descent using 3M BL4 CO5 V
sampling? Explain.
40 Explain the strategies used for efficient large-scale 6M BL2 CO5 V
model training.
41 How do you calculate probabilities in N-gram 6M BL4 CO5 V
models?
42 Discuss various methods used for high-dimensional 6M BL2 CO5 V
outputs.
Note
Mid Question paper contains Part-A and Part-B. Each question in Part-A
carries 2 marks. Prepare a list of 6 short answer questions from each unit
as per the format shown above for PART-A.
For PART-B, prepare a list of 6 questions carrying 3 marks each from 3 rd
Unit. Prepare a list of 6 questions carrying 3 marks each and 3 questions
carrying 6 marks each from 1st, 2nd, 4th and 5th Units as per the format shown
above for PART-B.
