LDRP Institute of Technology and Research, Gandhinagar

Department of Computer Engineering and Information Technology

QUESTION BANK
Subject Name & Code: Neural Network and Deep Learning (CT803B-N)

UNIT-1
Introduction to machine learning and neural Networks
SR
QUESTIONS
NO.
1 What is Machine Learning? Explain its different types.
2 Comparison between biological Neuron and artificial Neuron.
3 Compared Computers Traditionally work and Neural Networks.
4 What is need of Artificial Neural Network in today’s era?
5 Explain the equation formulation of ANN.
6 Define Neural Network and What are the different component of its.
7 What is Activation function? Enlist different types of activation function.
8 Write short notes on:
a) Linear Activation Function
b) The Sigmoid(Logistic) function
c) The ReLu(Rectified Linear Unit) function
d) McCulloch–Pitts unit-the first computation unit
9 What are the limitation of McCulloch–Pitts unit neuron?
10 What is perceptron. Briefly explain with its formulation and diagram of neurons.
11 List out and explain the basic element of the perceptron?
12 What do you understand by Perceptron Learning algorithm? Justify in your own words.
13 Describe the Convergence theorem for Perceptron Learning Algorithm.
14 Derive the proof of the convergence theorem for Perceptron Learning algorithm.
15 Difference between Soft computing and hard computing.
16 Explain neural network architecture with different layers.
17 Explain liner separability in terms of neural network.
18 Explain characteristics of neural network.
19 Explain Auto Associative Memory in term of ANN.
20 Explain biological structure of human brain.
21 Explain applications of Neural Network.
22 What linear separability in neural network?
23 Find out Error of the given network (in figure 1), where two inputs and one output,
the values lies between -1 to 1, hence there is no need to normalize the value. Assume
two neurons in the hidden layer. Page no 57, Ref Book: S.rajasekaran
24 Consider the unit shown on Figure 1.

Figure 1: Artificial Neural Network unit

Suppose that the weights corresponding to the three inputs have the following
values: w1=5, w2= -6, w3=2 and the activation of the unit is given by step function:

Calculate what will be the output value y of the unit for each of the following
input patterns:
 UNIT-2
Feed Forward Networks
SR
QUESTIONS
NO.
1 What is a Multi-layer Perceptron (MLP)?
2 What is Data Normalization, and Why Do We Need It?
3 What is Gradient Descent Rule? Write down the Gradient Descent algorithm steps.
4 Explain Back propagation with its algorithm.
5 What is the limitation of Back Propagation learning? Explain in details.
6 Differentiate between Gradient Descent and Perceptron training rule.
7 Write a short note on
a) Variational auto encoder
b) Regularized auto encoder

UNIT-3
Convolutional Neural Networks

SR
QUESTIONS
NO.
1 What do you mean by CNN? Write down the advantages of CNN.
2 Draw and explain the architecture of convolutional network.
3 Draw and explain the LeNet architecture.
4 Which architecture network that Challenged CNNs? Given a briefly reason of it.
5 Define Overfitting Problem with respect to AlexNet Architecture.

UNIT-4
Recurrent Neural Networks

SR
QUESTIONS
NO.
1 What is RNN. Explain briefly.
2 Explain LSTM (Long Short-Term Memory). Also design and explain it components the
basicstructure of LSTM.
3 Draw and explain the RNN architecture.
4 List out the pros and cons of a typical RNN architecture.
 UNIT-5
Deep Neural Networks

SR
QUESTIONS
NO.
1 What is the difference between Machine Learning and Deep Learning?
2 What are some of the most used applications of Deep Learning?
3 What are some of the Deep Learning frameworks or tools that you have used?
4 What are autoencoders?
5 What is data normalization in Deep Learning?
6 What are hyperparameters in Deep Learning?
7 How can hyperparameters be trained in neural networks?
8 What is the meaning of dropout in Deep Learning?
9 What are some of the advantages of using TensorFlow?
10 What are some of the applications of transfer learning in Deep Learning?

UNIT-6
Generative Models

SR
QUESTIONS
NO.
1 What are Boltzmann Machines? And What is a Restricted Boltzmann Machine?
2 What are generative adversarial networks (GANs)? Why are generative adversarial
networks (GANs) so popular?
3 How do Restricted Boltzmann Machines work?
4 Applications of Restricted Boltzmann Machine.
5 Explain Gibbs Sampling in details
6 What are Autoencoders? How do Autoencoders Work?
7 What are the Different Layers of Autoencoders? Explain with diagram.
8 What is the Difference Between GANs and Autoencoders?
9 How Variational Autoencoders Differ From Other Autoencoders?
10 Explain Generative Adversarial Network.
11 Why are Generative Adversarial Networks classed as unsupervised?
12 How can GAN be used to create images solely from text?
13 How many non-image data be used to improve GAN performance?