You are on page 1of 10

DECEMBER 2020

QUESTIONS

1.
a) Explain the machine learning concept by taking an
example. Describe the perspective and issues in machine
learning.
b) What is the role of preprocessing of data in machine
learning? Why it is needed? Explain the unsupervised
model of machine learning in detail with an example.

2.
a) Discuss linear regression with an example. Explain the
role of hypothesis function in machine learning models.
b) Explain the concept of perceptron, back propagation and
sigmoid activation function in brief. Differentiate
between classification and regression.

3.
a) What are the different types of Neural networks? Explain
the convolution neural network model in detail.
b) Explain the multilayer perceptron model in detail with
neat diagram.

4.
a) Explain the concept of different layers in Neural network.
What do you mean by the term convolution layer, pooling
layer, loss layer, dense layer? Describe each one in brief.
b) Explain the process of Sub-sampling of input data in neural
network model. Some of the features of Keras framework
for implementing neural networks models.

5.
a) What do you mean by Recurrent neural network? Explain
with the help of a diagram. In which cases this model is
suitable.
b) Explain the Actor critic model. List down what are its
advantages in reinforcement learning.

6.
a) Explain the concept of Reinforcement Learning and its
framework in details.
b) Describe how principle component analysis is carried out
to reduce the dimensionality of data sets.

7.
a) Describe Q-learning in brief. What is SARSA algorithm?
Explain this.
b) Explain the difference between Value iteration and Policy
iteration. What is Markov Decision Process (MDP)?

8.
a) Write short notes on any two:
i) Natural language processing
ii) Application of machine learning in computer vision
iii) Bayesian networks
b) Explain the concept and role of support vector machine
in details. Also, describe its application areas.
ANSWERS
1. a) Machine learning is a subset of artificial intelligence that involves the use of algorithms to learn from data and make predictions
or decisions without being explicitly programmed to do so. For example, a machine learning algorithm could be trained on a dataset
of patient information to predict whether or not a patient has a certain disease. The algorithm would learn the relationship between
the input features (such as age, gender, and medical history) and the output (disease diagnosis) and could then make predictions on
new data.

Machine learning can provide valuable insights and predictions, but there are also challenges and issues to consider. One issue is the
quality of the data used to train the algorithm. If the data is biased or incomplete, the algorithm may make inaccurate predictions.
Another issue is the interpretability of the algorithm. Some machine learning models, such as deep neural networks, can be difficult
to understand and interpret.

b) Preprocessing of data is an important step in machine learning because it can improve the accuracy of the model. Preprocessing
involves cleaning and transforming the data to make it suitable for use in a machine learning algorithm. This can include removing
missing or incorrect values, normalizing or scaling the data, and encoding categorical variables.

Unsupervised learning is a type of machine learning where the algorithm learns from data without explicit labels or outcomes. One
common application of unsupervised learning is clustering, where the algorithm groups similar data points together. For example, a
clustering algorithm could be used to group customers based on their purchasing behavior.

2. a) Linear regression is a type of supervised learning where the algorithm learns a linear relationship between the input features
and the output variable. For example, a linear regression model could be used to predict house prices based on features such as
square footage and number of bedrooms.

The hypothesis function in machine learning represents the relationship between the input features and the output variable. In
linear regression, the hypothesis function is a linear equation that describes how changes in the input features affect changes in the
output variable.

b) A perceptron is a type of artificial neuron that can be used in machine learning models. It takes in multiple inputs and produces a
single output based on a weighted sum of the inputs and an activation function.

Backpropagation is an algorithm used to train neural networks by adjusting the weights of the connections between neurons based
on the error between the predicted output and the actual output.

The sigmoid activation function is a commonly used activation function in neural networks that maps any input value to a value
between 0 and 1.

Classification and regression are two types of supervised learning. Classification involves predicting a categorical outcome, such as
whether an email is spam or not spam. Regression involves predicting a continuous outcome, such as house price.

3. a) There are several different types of neural networks, including feedforward neural networks, recurrent neural networks,
convolutional neural networks (CNNs), and deep neural networks.

A convolutional neural network (CNN) is a type of neural network commonly used in image recognition tasks. It consists of multiple
layers that process image data by applying filters to detect specific features such as edges or corners.

b) A multilayer perceptron (MLP) is a type of feedforward neural network that consists of multiple layers of neurons connected by
weighted connections. The input layer takes in data and passes it through hidden layers where it is processed by neurons with
activation functions before reaching the output layer where predictions are made.

4. a) A neural network consists of multiple layers of neurons connected by weighted connections. The different layers serve different
purposes in processing the data.

A convolution layer applies filters to detect specific features in image data.


A pooling layer reduces the dimensionality of data by combining multiple values into one.
A loss layer calculates the error between predicted outputs and actual outputs.
A dense layer connects every neuron in one layer to every neuron in another layer.
b) Sub-sampling refers to reducing the dimensionality of data by selecting only some values from it. In neural networks, this can be
done using pooling layers that combine multiple values into one.

Keras is an open-source framework for building and training neural networks. Some features include support for multiple backends,
easy model building using pre-built layers, and support for advanced training techniques such as transfer learning.

5. a) A recurrent neural network (RNN) is a type of neural network that can process sequential data by maintaining an internal state
that represents information from previous time steps. RNNs are commonly used in natural language processing tasks such as
language translation or text generation.
A Recurrent Neural Network (RNN) is a type of neural network that is designed to process sequential data. Unlike traditional
feedforward neural networks, RNNs have feedback connections that allow information to persist between time steps. This allows
the network to maintain an internal state that can represent information from previous time steps.

Here is a diagram that illustrates the basic structure of an RNN:

```
+------+ +------+ +------+
| | | | | |
| x1 +--->+ x2 +--->+ x3 |
| | | | | |
+---+--+ +---+--+ +---+--+
^ ^ ^
| | |
h1 h2 h3
```

In this diagram, `x1`, `x2`, and `x3` represent the input data at different time steps, and `h1`, `h2`, and `h3` represent the hidden
state of the network at each time step. The arrows show how information flows through the network.

RNNs are suitable for a wide range of applications that involve sequential data, such as natural language processing, speech
recognition, and time series prediction.

b) The Actor-Critic model is a type of reinforcement learning algorithm that combines the benefits of both value-based and policy-
based methods. In this model, there are two separate components: an actor that selects actions based on the current policy, and a
critic that evaluates the actions taken by the actor and updates the policy accordingly.

Some advantages of the Actor-Critic model include:


- Reduced variance compared to other methods like the REINFORCE algorithm
- The ability to learn both a policy and a value function
- Improved stability and convergence compared to using only a value function or only a policy

6. a) Reinforcement Learning (RL) is a type of machine learning that focuses on training agents to make a sequence of decisions in an
environment in order to maximize a reward signal. In RL, an agent interacts with an environment by taking actions and receiving
feedback in the form of rewards or penalties. The goal of the agent is to learn a policy that maps states to actions in such a way as to
maximize the cumulative reward over time.

The framework of RL consists of several key components:


- An agent that makes decisions
- An environment that the agent interacts with
- A set of actions that the agent can take
- A reward signal that provides feedback to the agent
- A policy that maps states to actions
- A value function that estimates the expected cumulative reward for each state or state-action pair

b) Principal Component Analysis (PCA) is a technique used to reduce the dimensionality of data sets by transforming the data into a
new coordinate system where the axes are ordered by importance. This is done by calculating the eigenvectors and eigenvalues of
the covariance matrix of the data and selecting the eigenvectors with the largest eigenvalues to form a new basis for the data.
7. a) Q-learning is a type of reinforcement learning algorithm that uses a Q-table to store estimates of the expected cumulative
reward for each state-action pair. The algorithm updates these estimates over time based on feedback from the environment in
order to learn an optimal policy.

SARSA (State-Action-Reward-State-Action) is another type of reinforcement learning algorithm that is similar to Q-learning but uses
on-policy updates instead of off-policy updates. This means that SARSA updates its Q-table based on the action actually taken by the
current policy, while Q-learning updates its Q-table based on the action with maximum estimated reward.

b) Value iteration and Policy iteration are two methods used to solve Markov Decision Processes (MDPs). MDPs are mathematical
models used to describe decision-making problems where outcomes are partly random and partly under control of a decision-
maker.

Value iteration involves iteratively updating estimates of the value function until convergence. Policy iteration involves iteratively
evaluating and improving a policy until convergence.

8. a) i) Natural Language Processing (NLP) is a field of artificial intelligence that focuses on enabling computers to understand,
interpret, and generate human language. NLP techniques are used in many applications such as language translation, sentiment
analysis, and text summarization.

ii) Machine learning has many applications in computer vision, including object recognition, image classification, and scene
understanding. Machine learning algorithms can be trained on large datasets of images to learn patterns and features that can be
used to make predictions on new data.

iii) Bayesian networks are graphical models that represent probabilistic relationships between variables. They can be used to model
complex systems where there is uncertainty and to make predictions or inferences about unknown variables based on observed
data. Bayesian networks are commonly used in applications such as medical diagnosis, risk assessment, and decision-making.

b) A Support Vector Machine (SVM) is a type of machine learning algorithm that can be used for classification or regression tasks.
SVMs work by finding the hyperplane that best separates the data into different classes. The hyperplane is chosen to maximize the
margin, which is the distance between the hyperplane and the closest data points from each class.

SVMs have several advantages, including their ability to handle high-dimensional data and their robustness to outliers. They are
commonly used in applications such as image classification, text classification, and bioinformatics.

Some application areas of SVMs include:


- Image classification: SVMs can be trained to recognize objects or scenes in images.
- Text classification: SVMs can be used to classify text documents into different categories based on their content.
- Bioinformatics: SVMs can be used to predict protein structures or to identify disease biomarkers based on genetic data.
AI GENERATED

QUESTIONS

Sure! Here are some additional questions that could be asked about machine learning in a similar format:

9.
a) What is the difference between supervised and unsupervised learning? Give examples of each.
b) Explain the concept of clustering in machine learning. What are some common clustering algorithms?

10.
a) What is overfitting and how can it be avoided in machine learning models?
b) Explain the concept of cross-validation and its importance in model selection.

11.
a) What is a decision tree and how is it used in machine learning?
b) Explain the concept of ensemble learning and its advantages.

12.
a) What is a support vector machine and how does it work?
b) Explain the concept of kernel methods in machine learning.

13.
a) What is deep learning and how does it differ from traditional machine learning?
b) Explain the concept of transfer learning and its applications.

14.
a) What is reinforcement learning and how does it differ from supervised and unsupervised learning?
b) Explain the concept of Q-learning and its applications.

15.
a) What is natural language processing and how is it used in machine learning?
b) Explain the concept of sentiment analysis and its applications.

1. What is the difference between supervised and unsupervised learning?

2. Can you explain the concept of overfitting in machine learning?

3. How does gradient descent work in machine learning?

4. What are some common techniques for feature selection in machine learning?

5. How do decision trees work in machine learning?

6. Can you explain the difference between classification and regression in machine learning?

7. What is the purpose of cross-validation in machine learning?

8. Can you explain the concept of regularization in machine learning?

9. How do support vector machines work in machine learning?

10. What is the role of bias and variance in machine learning

9. a) What is the difference between supervised and unsupervised learning?

b) How can overfitting be prevented in machine learning models?


10. a) What is the curse of dimensionality in machine learning?

b) How can regularization be used to improve the performance of a machine learning model?

11. a) What is deep learning and how does it differ from traditional machine learning?

b) How can hyperparameter tuning be used to optimize the performance of a machine learning model?

12. a) What is transfer learning in machine learning and how is it used?

b) What is the difference between classification and regression tasks in machine learning?

13. a) What is ensemble learning and how does it work?

b) How can feature engineering be used to improve the performance of a machine learning model?

14. a) What is clustering in unsupervised learning and how is it used?

b) What is the difference between batch and online learning in machine learning?

15. a) What is reinforcement learning and what are some applications of this type of machine learning?

b) How can cross-validation be used to evaluate the performance of a machine learning model?
MAY 2022

QUESTIONS

1. a) What are the basic design issues and approaches to machine learning?

b) Define statistical theory and how it is performed in machine learning.

2. a) Differentiate between Training data and Testing data.

b) What do you mean by Gradient Descent ?

3. a) Write the algorithm for back propagation.

b) Explain in detail principal component analysis for dimension reduction.

4. a) What is Reinforcement learning? Explain its detailed concepts.

b) Explain Locally weighted linear regression.

5. a) What is model selection in Machine Learning?

b) Describe the concept of MDP.

6. a) Explain Q learning algorithm assuming deterministic rewards and actions.

b) Explain the concept of Bayesian theorem with an example

7. a) What is Support Vector Machine (SVM ) ? Discuss in detail.

b) Define Bayesian learning and how it impacts in machine learning?

8. Explain the following terms with example:

i) Convex optimization in machine learning

ii) Multilayer network

iii) Attention model

iv) Natural Language Processing

ANSWERS

1. a) Basic design issues in machine learning include selecting appropriate features, choosing an appropriate algorithm, and
evaluating the performance of the model. Approaches to machine learning include supervised learning, unsupervised learning, and
reinforcement learning.

b) Statistical theory involves using statistical methods to analyze data and make predictions. In machine learning, statistical theory is
used to develop models and algorithms that can learn from data.
2. a) Training data is used to train a machine learning model, while testing data is used to evaluate the performance of the model.

b) Gradient descent is an optimization algorithm used in machine learning to minimize the error or cost function of a model by
adjusting the model parameters.

3. a) The backpropagation algorithm is used to train neural networks by iteratively adjusting the weights of the network based on
the error between the predicted output and the actual output.

b) Principal component analysis is a technique used for dimensionality reduction in which the original data is transformed into a new
set of variables that capture the most important information in the data.

4. a) Reinforcement learning is a type of machine learning in which an agent learns to take actions in an environment to maximize a
reward signal. The agent receives feedback in the form of rewards or penalties for its actions and adjusts its behavior accordingly.

b) Locally weighted linear regression is a non-parametric regression method that assigns weights to nearby data points based on
their distance from the point being predicted.

5. a) Model selection involves choosing the best model or algorithm for a particular problem based on its performance on a
validation set.

b) MDP (Markov Decision Process) is a mathematical framework used to model decision-making problems in which the outcomes
depend on both the current state and the actions taken by the decision-maker.

6. a) Q-learning is a reinforcement learning algorithm that learns an optimal policy by estimating the value of each state-action pair
using a Q-value function.

b) Bayesian theorem is a mathematical formula used to update the probability of a hypothesis based on new evidence. For example,
if a coin is flipped and comes up heads, Bayesian theorem can be used to update the probability that the coin is biased.

7. a) Support Vector Machine (SVM) is a supervised learning algorithm used for classification and regression tasks. It works by finding
the hyperplane that best separates the data into different classes.

b) Bayesian learning is a statistical approach to machine learning that involves using Bayes' theorem to update the probability of a
hypothesis based on new evidence.

8. i) Convex optimization in machine learning involves finding the minimum of a convex function, which has only one global
minimum. For example, linear regression can be solved using convex optimization.

ii) A multilayer network is a type of neural network with multiple layers of nodes that can learn complex patterns in data.

iii) Attention model is a neural network architecture that learns to selectively focus on different parts of the input data to improve
performance on specific tasks.

iv) Natural Language Processing (NLP) is a field of study that focuses on developing algorithms and models for processing and
understanding human language. Examples include sentiment analysis, machine translation, and text classification.

You might also like