www.Notesfree.

in

CS3491 ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

COURSE OBJECTIVES:
The main objectives of this course are to:
 Study about uninformed and Heuristic search techniques.
 Learn techniques for reasoning under uncertainty
 Introduce Machine Learning and supervised learning algorithms
 Study about ensembling and unsupervised learning algorithms
 Learn the basics of deep learning using neural networks

n
UNIT I PROBLEM SOLVING

e.i
Introduction to AI - AI Applications - Problem solving agents – search algorithms –
uninformed search strategies – Heuristic search strategies – Local search and optimization
problems – adversarial search – constraint satisfaction problems (CSP)

UNIT II PROBABILISTIC REASONING

e
Acting under uncertainty – Bayesian inference – naïve bayes models. Probabilistic
reasoning – Bayesian networks – exact inference in BN – approximate inference in BN –
sfr
causal networks.

UNIT III SUPERVISED LEARNING

Introduction to machine learning – Linear Regression Models: Least squares, single
& multiple variables, Bayesian linear regression, gradient descent, Linear Classification
ote

Models: Discriminant function – Probabilistic discriminative model - Logistic regression,

Probabilistic generative model – Naive Bayes, Maximum margin classifier – Support vector
machine, Decision Tree, Random forests

UNIT IV ENSEMBLE TECHNIQUES AND UNSUPERVISED LEARNING

Combining multiple learners: Model combination schemes, Voting, Ensemble
.N

Learning - bagging, boosting, stacking, Unsupervised learning: K-means, Instance Based

Learning: KNN, Gaussian mixture models and Expectation maximization

UNIT V NEURAL NETWORKS

Perceptron - Multilayer perceptron, activation functions, network training – gradient
w

descent optimization – stochastic gradient descent, error backpropagation, from shallow

networks to deep networks –Unit saturation (aka the vanishing gradient problem) – ReLU,
hyperparameter tuning, batch normalization, regularization, dropout.
ww

PRACTICAL EXERCISES:
1. Implementation of Uninformed search algorithms (BFS, DFS)
2. Implementation of Informed search algorithms (A*, memory-bounded A*)
3. Implement naïve Bayes models
4. Implement Bayesian Networks
5. Build Regression models
6. Build decision trees and random forests
7. Build SVM models
8. Implement ensembling techniques
9. Implement clustering algorithms

www.Notesfree.in
www.Notesfree.in

10. Implement EM for Bayesian networks

11. Build simple NN models
12. Build deep learning NN models

COURSE OUTCOMES:
At the end of this course, the students will be able to:
CO1: Use appropriate search algorithms for problem solving
CO2: Apply reasoning under uncertainty
CO3: Build supervised learning models
CO4: Build ensembling and unsupervised models

n
CO5: Build deep learning neural network models

e.i
TEXT BOOKS:
1. Stuart Russell and Peter Norvig, “Artificial Intelligence – A Modern Approach”, Fourth
Edition, Pearson Education, 2021.
2. Ethem Alpaydin, “Introduction to Machine Learning”, MIT Press, Fourth Edition, 2020.

REFERENCES:
e
sfr
1. Dan W. Patterson, “Introduction to Artificial Intelligence and Expert Systems”, Pearson
Education,2007
2. Kevin Night, Elaine Rich, and Nair B., “Artificial Intelligence”, McGraw Hill, 2008
3. Patrick H. Winston, "Artificial Intelligence", Third Edition, Pearson Education, 2006
4. Deepak Khemani, “Artificial Intelligence”, Tata McGraw Hill Education, 2013
ote

(http://nptel.ac.in/)
5. Christopher M. Bishop, “Pattern Recognition and Machine Learning”, Springer, 2006.
6. Tom Mitchell, “Machine Learning”, McGraw Hill, 3rd Edition,1997.
7. Charu C. Aggarwal, “Data Classification Algorithms and Applications”, CRC Press, 2014
8. Mehryar Mohri, Afshin Rostamizadeh, Ameet Talwalkar, “Foundations of Machine
Learning”, MIT Press, 2012.
.N

9. Ian Goodfellow, Yoshua Bengio, Aaron Courville, “Deep Learning”, MIT Press, 2016
w
ww

www.Notesfree.in
www.Notesfree.in

UNIT I PROBLEM SOLVING

Introduction to AI - AI Applications - Problem solving agents – search algorithms –
uninformed search strategies – Heuristic search strategies – Local search and optimization
problems – adversarial search – constraint satisfaction problems (CSP)

Part A

1. What is Artificial Intelligence?

Artificial Intelligence is the study of how to make computers do things which
at the moment people do better.

n
2. What is an agent?

e.i
An agent is anything that can be viewed as perceiving its environment through
sensors and acting upon that environment through actuators.

3. What are the different types of agents?

e
A human agent has eyes, ears, and other organs for sensors and hands, legs,
mouth, and other body parts for actuators.
sfr
A robotic agent might have cameras and infrared range finders for sensors
and various motors for actuators.
A software agent receives keystrokes, file contents, and network packets as
sensory inputs and acts on the environment by displaying on the screen, writing files,
ote

and sending network packets.

Generic agent - A general structure of an agent who interacts with the
environment.

4. Define rational agent.

.N

For each possible percept sequence, a rational agent should select an action that is
expected to maximize its performance measure, given the evidence provided by the
percept sequence and whatever built-in knowledge the agent has. A rational agent
should be autonomous.
w

5. List down the characteristics of intelligent agent.

Internal characteristics are
Learning/reasoning: an agent has the ability to learn from previous
ww

experience and to successively adapt its own behaviour to the environment.

Reactivity: an agent must be capable of reacting appropriately to influences or
information from its environment.
Autonomy: an agent must have both control over its actions and internal
states. The degree of the agent’s autonomy can be specified. There may need
intervention from the user only for important decisions.
Goal-oriented: an agent has well-defined goals and gradually influence its
environment and so achieve its own goals.
External characteristics are

www.Notesfree.in
www.Notesfree.in

Communication: an agent often requires an interaction with its environment

to fulfil its tasks, such as human, other agents, and arbitrary information sources.
Cooperation: cooperation of several agents permits faster and better solutions
for complex tasks that exceed the capabilities of a single agent.
Mobility: an agent may navigate within electronic communication networks.
Character: like human, an agent may demonstrate an external behaviour with
many human characters as possible.

6. What are various applications of AI? or What can AI do today?

n
 Robotic vehicles
 Speech recognition

e.i
 Autonomous planning and scheduling
 Game playing
 Spam fighting
 Logistics planning
 Robotics
 Machine Translation
e
sfr
7. Are reflex actions (such as flinching from a hot stove) rational? Are they
intelligent?

Reflex actions can be considered rational. If the body is performing the action,
ote

then it can be argued that reflex actions are rational because of evolutionary
adaptation. Flinching from a hot stove is a normal reaction, because the body wants to
keep itself out of danger and getting away from something hot is a way to do that.

Reflex actions are also intelligent. Intelligence suggests that there is reasoning
.N

and logic involved in the action itself.

8. Is AI a science, or is it engineering? Or neither or both? Explain.

w

AI is both science and engineering. Observing and experimenting, which are

at the core of any science, allows us to study artificial intelligence. From what we
learn by observation and experimentation, we are able to engineer new systems that
encompass what we learn and that may even be capable of learning themselves.
ww

9. What are the various agent programs in intelligent systems?

Simple reflex agents
Model-based reflex agents
Goal-based agents
Utility-based agents

www.Notesfree.in
www.Notesfree.in

10. Define the problem solving agent.

A Problem solving agent is a goal-based agent. It decides what to do by
finding sequence of actions that lead to desirable states. The agent can adopt a goal
and aim at satisfying it. Goal formulation is the first step in problem solving.

11. Define the terms goal formulation and problem formulation.

Goal formulation based on the current situation and the agent’s performance
measure is the first step in problem solving. The agent’s task is to find out which
sequence of actions will get to a goal state.

n
Problem formulation is the process of deciding what actions and states to
consider given a goal.

e.i
12. List the steps involved in simple problem solving agent.
(i) Goal formulation
(ii) Problem formulation
(iii) Search

e
(iv) Search Algorithm
(v) Execution phase
sfr
13. What are the components of well-defined problems? (or)
What are the four components to define a problem? Define them?
The four components to define a problem are,
1) Initial state – it is the state in which agent starts in.
ote

2) A description of possible actions – it is the description of possible actions

which are available to the agent.
3) The goal test – it is the test that determines whether a given state is goal
(final) state.
4) A path cost function – it is the function that assigns a numeric cost (value)
.N

to each path.
The problem-solving agent is expected to choose a cost function that reflects
its own performance measure.
w

14. Differentiate toy problems and real world problems?

A toy problem is intended to illustrate various problem solving methods. It can
be easily used by different researchers to compare the performance of algorithms. A
ww

real world problem is one whose solutions people actually care about.

15. Give example for real world end toy problems.

Real world problem examples:
 Airline travel problem.
 Touring problem.
 Traveling salesman problem
 VLSI Layout problem
 Robot navigation

www.Notesfree.in
www.Notesfree.in

 Automatic Assembly
 Internet searching
Toy problem Examples:
 8 – Queen problem
 8 – Puzzle problem
 Vacuum world problem

16. How will you measure the problem-solving performance?

We can evaluate an algorithm’s performance in four ways:

n
Completeness: Is the algorithm guaranteed to find a solution when there is
one?
Optimality: Does the strategy find the optimal solution?

e.i
Time complexity: How long does it take to find a solution?
Space complexity: How much memory is needed to perform the search?

17. What is the application of BFS?

e
It is simple search strategy, which is complete i.e. it surely gives solution if
solution exists. If the depth of search tree is small then BFS is the best choice. It is
sfr
useful in tree as well as in graph search.

18. State on which basis search algorithms are chosen?

Search algorithms are chosen depending on two components.
ote

1) How is the state space – That is, state space is tree structured or graph?
Critical factor for state space is what is branching factor and depth level of that tree or
graph.
2) What is the performance of the search strategy? A complete, optimal search
strategy with better time and space requirement is critical factor in performance of
.N

search strategy.

19. Evaluate performance of problem-solving method based on depth-first search

algorithm?
w

DFS algorithm performance measurement is done with four ways –

1) Completeness – It is complete (guarantees solution)
2) Optimality – it is not optimal.
ww

3) Time complexity – It’s time complexity is O (b).

4) Space complexity – its space complexity is O (b d+1).

20. List some of the uninformed search techniques.

The uninformed search strategies are those that do not take into account the
location of the goal. That is these algorithms ignore where they are going until they
find a goal and report success. The various uninformed search strategies are
 Breadth-first search
 Uniform-cost search
 Depth-first search

www.Notesfree.in
www.Notesfree.in

 Depth-limited search
 Iterative deepening depth-first search
 Bidirectional search

21. What is the power of heuristic search? (or) Why does one go for heuristics
search?
Heuristic search uses problem specific knowledge while searching in state
space. This helps to improve average search performance. They use evaluation
functions which denote relative desirability (goodness) of a expanding node set. This
makes the search more efficient and faster. One should go for heuristic search because

n
it has power to solve large, hard problems in affordable times.

e.i
22. What are the advantages of heuristic function?
Heuristics function ranks alternative paths in various search algorithms, at
each branching step, based on the available information, so that a better path is
chosen. The main advantage of heuristic function is that it guides for which state to

e
explore now, while searching. It makes use of problem specific knowledge like
constraints to check the goodness of a state to be explained. This drastically reduces
sfr
the required searching time.

23. State the reason when hill climbing often gets stuck?
Local maxima are the state where hill climbing algorithm is sure to get struck.
ote

Local maxima are the peak that is higher than each of its neighbour states, but lower
than the global maximum. So we have missed the better state here. All the search
procedure turns out to be wasted here. It is like a dead end.

24. When a heuristic function h is said to be admissible? Give an admissible

.N

heuristic function for TSP?

Admissible heuristic function is that function which never over estimates the
cost to reach the goal state. It means that h(n) gives true cost to reach the goal state
‘n’. The admissible heuristic for TSP is
w

a. Minimum spanning tree.

b. Minimum assignment problem
ww

25. What do you mean by local maxima with respect to search technique?
Local maximum is the peak that is higher than each of its neighbour states, but
lowers than the global maximum i.e. a local maximum is a tiny hill on the surface
whose peak is not as high as the main peak (which is a optimal solution). Hill
climbing fails to find optimum solution when it encounters local maxima. Any small
move, from here also makes things worse (temporarily). At local maxima all the
search procedure turns out to be wasted here. It is like a dead end.

26. How can we avoid ridge and plateau in hill climbing?

www.Notesfree.in
www.Notesfree.in

Ridge and plateau in hill climbing can be avoided using methods like
backtracking, making big jumps. Backtracking and making big jumps help to avoid
plateau, whereas, application of multiple rules helps to avoid the problem of ridges.

27. Differentiate Blind Search and Heuristic Search.

Parameters Blind search Heuristic search
It is also known Uninformed It is also known Informed
Known as
Search Search
Using It doesn’t use knowledge for It uses knowledge for the
Knowledge the searching process. searching process.

n
It finds solution slow as It finds a solution more quickly.
Performance compared to an informed

e.i
search.
Completion It is always complete. It may or may not be complete.
Cost Factor Cost is high. Cost is low.
It consumes moderate time It consumes less time because of
Time
because of slow searching. quick searching.
Direction
e
No suggestion is given
regarding the solution in it.
There is a direction given about
the solution.
sfr
It is lengthier while It is less lengthy while
Implementation
implemented. implemented.
It is comparatively less It is more efficient as efficiency
efficient as incurred cost is takes into account cost and
Efficiency more and the speed of finding performance. The incurred cost
ote

the Breadth-First solution is is less and speed of finding

slow. solutions is quick.
Computational Comparatively higher Computational requirements are
requirements computational requirements. lessened.
Size of search Solving a massive search task Having a wide scope in terms of
problems is challenging. handling large search problems.
.N

Example Example
a) Breadth first search a) Best first search
b) Uniform cost search b) Greedy search
Examples of
c) Depth first Search c) A* search
Algorithms
w

d) Depth limited search d) AO* Search

e) Iterative deepening search e)Hill Climbing Algorithm
f) Bi – Directional Search
ww

28. What is CSP?

CSP are problems whose state and goal test conform to a standard structure
and very simple representation. CSPs are defined using set of variables and a set of
constraints on those variables. The variables have some allowed values from specified
domain. For example – Graph coloring problem.

29. How can minimax also be extended for game of chance?

www.Notesfree.in
www.Notesfree.in

In a game of chance, we can add extra level of chance nodes in game search tree.
These nodes have successors which are the outcomes of random element. The
minimax algorithm uses probability P attached with chance node di based on this
value. Successor function S(N,di) give moves from position N for outcome di

Part B
1. Enumerate Classical “Water jug Problem”. Describe the state space for this
problem and also give the solution.
2. How to define a problem as state space search? Discuss it with the help of an

n
example
3. Solve the given problem. Describe the operators involved in it.

e.i
Consider a Water Jug Problem : You are given two jugs, a 4-gallon one and
a 3-gallon one. Neither has any measuring markers on it. There is a pump
that can be used to fill the jugs with water. How can you get exactly 2 gallons
of water into the 4-gallon jug ? Explicit Assumptions: A jug can be filled

e
from the pump, water can be poured out of a jug onto the ground, water can
be poured from one jug to another and that there are no other measuring
devices available.
sfr
4. Define the following problems. What types of control strategy is used in the
following problem.
i.The Tower of Hanoi
ii.Crypto-arithmetic
ote

iii.The Missionaries and cannibals problems

iv.8-puzzle problem
5. Discuss uninformed search methods with examples.
6. Give an example of a problem for which breadth first search would work
better than depth first search.
.N

7. Explain the algorithm for steepest hill climbing

8. Explain the A* search and give the proof of optimality of A*
9. Explain AO* algorithm with a suitable example. State the limitations in the
algorithm?
w

10. Explain the nature of heuristics with example. What is the effect of heuristics
accuracy?
11. Explain the various types of hill climbing search techniques.
ww

12. Discuss about constraint satisfaction problem with a algorithm for solving a
crypt arithmetic Problem.
13. Solve the following Crypt arithmetic problem using constraints satisfaction
search procedure.
CROSS
+ROADS
------------
DANGER
----------------

www.Notesfree.in
www.Notesfree.in

14. Explain alpha-beta pruning algorithm and the Minmax game playing
algorithm with example?
15. Solve the given problem. Describe the operators involved in it.
Consider a water jug problem: You are given two jugs, a 4-gallon one and a
3-gallon one. Neither have any measuring Markers on it. There is a pump
that can be used to fill the jug with water. How can you get exactly 2 gallons
of water into the 4 gallon jug? Explicit Assumptions: A jug can be filled from
the pump, water can be poured out of a jug onto the ground, water can be
poured from one jug to another and that there are no other measuring

n
devices available.

UNIT II PROBABILISTIC REASONING

e.i
Acting under uncertainty – Bayesian inference – naïve bayes models. Probabilistic
reasoning – Bayesian networks – exact inference in BN – approximate inference in BN –
causal networks.

e
Part A
1. Why does uncertainty arise?
Agents almost never have access to the whole truth about their environment.
sfr
Uncertainty arises because of both laziness and ignorance. It is inescapable in
complex, nondeterministic, or partially observable environments
Agents cannot find a categorical answer.
 Uncertainty can also arise because of incompleteness, incorrectness in agents
understanding of properties of environment.
ote

2. Differentiate uncertainty with ignorance.

A key condition that differentiates ignorance from uncertainty is the absence
of knowledge about the factors that influence the issues

3. What is the need for probability theory in uncertainty?

.N

Probability provides the way of summarizing the uncertainty that comes from
our laziness and ignorance. Probability statements do not have quite the same kind
of semantics known as evidences.

4. What is the need for utility theory in uncertainty?

w

Utility theory says that every state has a degree of usefulness, or utility to in
agent, and that the agent will prefer states with higher utility.

5. Define principle of maximum expected utility (MEU)?

ww

`The fundamental idea of decision theory is that an agent is rational if and only if
it chooses the action that yields the highest expected utility, averaged over all the
possible outcomes of the action. This is called the principle of maximum expected
utility (MEU).

6. Mention the needs of probabilistic reasoning in AI.

 When there are unpredictable outcomes.
 When specifications or possibilities of predicates becomes too large to
handle.
 When an unknown error occurs during an experiment.

www.Notesfree.in
www.Notesfree.in

7. What does the full joint probability distribution specify?

The full joint probability distribution specifies the probability of each
complete assignment of values to random variables. It is usually too large to create
or use in its explicit form, but when it is available it can be used to answer queries
simply by adding up entries for the possible worlds corresponding to the query
propositions.

8. State Bayes' Theorem in Artificial Intelligence.

Bayes' theorem is also known as Bayes' rule, Bayes' law, or Bayesian
reasoning, which determines the probability of an event with uncertain
knowledge. It is a way to calculate the value of P(B|A) with the knowledge of

n
P(A|B). Bayes' theorem allows updating the probability prediction of an event by
observing new information of the real world.

e.i
Example: If cancer corresponds to one's age then by using Bayes' theorem,
we can determine the probability of cancer more accurately with the help of age.

P(A/B)=[P(A)*P(B/A)]/P(B)

e
9. Given that P(A)=0.3,P(A|B)=0.4 and P(B)=0.5, Compute P(B|A).
sfr
0.4 = (0.3*P(B/A))/0.5
ote

P(B/A) = 0.66

10. What is Bayesian Belief Network?

A Bayesian network is a probabilistic graphical model which represents a set
of variables and their conditional dependencies using a directed acyclic graph. It is
.N

also called a Bayes network, belief network, decision network, or Bayesian model.
Bayesian networks are probabilistic, because these networks are built from a
probability distribution, and also use probability theory for prediction and
anomaly detection.
A Bayesian network is a directed graph in which each node is annotated with
w

quantitative probability information. The full specification is as follows:

1. Each node corresponds to a random variable, which may be discrete or
continuous.
2. A set of directed links or arrows connects pairs of nodes. If there is an arrow
ww

from node X to node Y , X is said to be a parent of Y. The graph has no directed

cycles (and hence is a directed acyclic graph, or DAG).
3. Each nodeXi has a conditional probability distribution P(Xi |Parents(Xi))
that quantifies the effect of the parents on the node.

Part B

1. How to get the exact inference form Bayesian network?

www.Notesfree.in
www.Notesfree.in

2. Explain variable elimination algorithm for answering queries on Bayesian

networks?
3. Define uncertain knowledge, prior probability and conditional probability.
State the Bayes’ theorem. How it is useful for decision making under
uncertainty? Explain belief networks briefly?
4. Explain the method of handling approximate inference in Bayesian networks.
5. What is Bayes’ rule? Explain how Bayes’ rule can be applied to tackle
uncertain Knowledge.
6. Discuss about Bayesian Theory and Bayesian network.
7. Explain how does Bayesian statistics provide reasoning under various kinds
of uncertainty?

n
8. How to get the approximate inference from Bayesian network.
9. Construct a Bayesian Network and define the necessary CPTs for the given

e.i
scenario. We have a bag of three biased coins a,b and c with probabilities of
coming up heads of 20%, 60% and 80% respectively. One coin is drawn
randomly from the bag (with equal likelihood of drawing each of the three
coins) and then the coin is flipped three times to generate the outcomes X1,
X2 and X3.

e
a. Draw a Bayesian network corresponding to this setup and define the
relevant CPTs.
sfr
b. Calculate which coin is most likely to have been drawn if the flips come
up HHT
10. Consider the following set of propositions
Patient has spots
ote

Patient has measles

Patient has high fever
Patient has Rocky mountain spotted fever.
Patient has previously been inoculated against measles.
Patient was recently bitten by a tick
.N

Patient has an allergy.

a) Create a network that defines the casual connections among these nodes.
b) Make it a Bayesian network by constructing the necessary conditional
probability matrix.
w
ww

www.Notesfree.in
www.Notesfree.in

UNIT III SUPERVISED LEARNING

Introduction to machine learning – Linear Regression Models: Least squares, single
& multiple variables, Bayesian linear regression, gradient descent, Linear Classification
Models: Discriminant function – Probabilistic discriminative model - Logistic regression,
Probabilistic generative model – Naive Bayes, Maximum margin classifier – Support vector
machine, Decision Tree, Random forests.

PART - A

1. What is Machine Learning?

Machine learning is a branch of computer science which deals with system

n
programming in order to automatically learn and improve with experience. For
example: Robots are programed so that they can perform the task based on data

e.i
they gather from sensors. It automatically learns programs from data.

2. Mention the difference between Data Mining and Machine learning?

Machine learning relates with the study, design and development of the
algorithms that give computers the capability to learn without being explicitly

e
programmed. While, data mining can be defined as the process in which the
unstructured data tries to extract knowledge or unknown interesting patterns.
sfr
3. What is ‘Overfitting’ in Machine learning?
In machine learning, when a statistical model describes random error or noise
instead of underlying relationship ‘overfitting’ occurs. When a model is
excessively complex, overfitting is normally observed, because of having too
ote

many parameters with respect to the number of training data types. The model
exhibits poor performance which has been overfit.

4. Why overfitting happens?

The possibility of overfitting exists as the criteria used for training the model
is not the same as the criteria used to judge the efficacy of a model.
.N

5. How can you avoid overfitting?

By using a lot of data overfitting can be avoided, overfitting happens relatively
as you have a small dataset, and you try to learn from it. But if you have a small
w

database and you are forced to come with a model based on that. In such situation,
you can use a technique known as cross validation. In this method the dataset
splits into two section, testing and training datasets, the testing dataset will only
test the model while, in training dataset, the datapoints will come up with the
ww

model. In this technique, a model is usually given a dataset of a known data on

which training (training data set) is run and a dataset of unknown data against
which the model is tested. The idea of cross validation is to define a dataset to
“test” the model in the training phase.

6. What are the five popular algorithms of Machine Learning?

 Decision Trees
 Neural Networks (back propagation)
 Probabilistic networks
 Nearest Neighbor
 Support vector machines

www.Notesfree.in
www.Notesfree.in

7. What are the different Algorithm techniques in Machine Learning?

The different types of techniques in Machine Learning are

 Supervised Learning
 Unsupervised Learning
 Semi-supervised Learning
 Reinforcement Learning
 Transduction
 Learning to Learn

n
8. What are the three stages to build the hypotheses or model in machine
learning?

e.i
 Model building
 Model testing
 Applying the model

e
9. What is the standard approach to supervised learning?
The standard approach to supervised learning is to split the set of example into
the training set and the test.
sfr
10. What is ‘Training set’ and ‘Test set’?
In various areas of information science like machine learning, a set of data is
used to discover the potentially predictive relationship known as ‘Training Set’.
ote

Training set is an examples given to the learner, while Test set is used to test the
accuracy of the hypotheses generated by the learner, and it is the set of example
held back from the learner. Training set are distinct from Test set.

11. What is the difference between artificial learning and machine learning?
Designing and developing algorithms according to the behaviours based on
.N

empirical data are known as Machine Learning. While artificial intelligence in

addition to machine learning, it also covers other aspects like knowledge
representation, natural language processing, planning, robotics etc.

12. What are the advantages of Naive Bayes?

w

In Naïve Bayes classifier will converge quicker than discriminative models

like logistic regression, so you need less training data. The main advantage is that
it can’t learn interactions between features.
ww

13. What is the main key difference between supervised and unsupervised
machine learning?
supervised learning Unsupervised learning
The supervised learning technique needs Unsupervised learning does not
labelled data to train the model. For need any labelled dataset. This is
example, to solve a classification problem the main key difference between
(a supervised learning task), you need to supervised learning and
have label data to train the model and to unsupervised learning.
classify the data into your labelled groups.

www.Notesfree.in
www.Notesfree.in

14. What is a Linear Regression?

In simple terms, linear regression is adopting a linear approach to modeling
the relationship between a dependent variable (scalar response) and one or more
independent variables (explanatory variables). In case you have one explanatory
variable, you call it a simple linear regression. In case you have more than one
independent variable, you refer to the process as multiple linear regressions.

15. What are the disadvantages of the linear regression model?

One of the most significant demerits of the linear model is that it is sensitive
and dependent on the outliers. It can affect the overall result. Another notable

n
demerit of the linear model is overfitting. Similarly, underfitting is also a
significant disadvantage of the linear model.

e.i
16. What is the difference between classification and regression?
Classification is used to produce discrete results; classification is used to
classify data into some specific categories. For example, classifying emails into
spam and non-spam categories. Whereas, we use regression analysis when we are

e
dealing with continuous data, for example predicting stock prices at a certain point
in time.
sfr
17. What is the difference between stochastic gradient descent (SGD) and
gradient descent (GD)?
Both algorithms are methods for finding a set of parameters that minimize a
loss function by evaluating parameters against data and then making adjustments.
ote

In standard gradient descent, you'll evaluate all training samples for each set of
parameters. This is akin to taking big, slow steps toward the solution. In stochastic
gradient descent, you'll evaluate only 1 training sample for the set of parameters
before updating them. This is akin to taking small, quick steps toward the solution.

18. What are the different types of least squares?

.N

Least squares problems fall into two categories: linear or ordinary least
squares and nonlinear least squares, depending on whether or not the residuals are
linear in all unknowns. The linear least-squares problem occurs in statistical
regression analysis; it has a closed-form solution.
w

19. What is the difference between least squares regression and multiple
regression?
The goal of multiple linear regression is to model the linear relationship
ww

between the explanatory (independent) variables and response (dependent)

variables. In essence, multiple regression is the extension of ordinary least-squares
(OLS) regression because it involves more than one explanatory variable.

20. What is the principle of least squares?

Principle of Least Squares" states that the most probable values of a system of
unknown quantities upon which observations have been made, are obtained by
making the sum of the squares of the errors a minimum.

21. What are some advantages to using Bayesian linear regression?

www.Notesfree.in
www.Notesfree.in

Doing Bayesian regression is not an algorithm but a different approach to

statistical inference. The major advantage is that, by this Bayesian processing, you
recover the whole range of inferential solutions, rather than a point estimate and a
confidence interval as in classical regression.

22. What Is Bayesian Linear Regression?

In Bayesian linear regression, the mean of one parameter is characterized by a
weighted sum of other variables. This type of conditional modeling aims to
determine the prior distribution of the regressors as well as other variables
describing the allocation of the regress and eventually permits the out-of-sample
forecasting of the regress and conditional on observations of the regression

n
coefficients.

e.i
23. What are the advantages of Bayesian Regression?
 Extremely efficient when the dataset is tiny.
 Particularly well-suited for online learning as opposed to batch learning,
when we know the complete dataset before we begin training the model.
This is so that Bayesian Regression can be used without having to save
data.

e
 The Bayesian technique has been successfully applied and is quite strong
sfr
mathematically. Therefore, using this requires no additional prior
knowledge of the dataset.

24. What are the disadvantages of Bayesian Regression?

 The model's inference process can take some time.
ote

 The Bayesian strategy is not worthwhile if there is a lot of data accessible

for our dataset, and the regular probability approach does the task more
effectively.

25. What are types of classification models?

 Logistic Regression
.N

 Naive Bayes
 K-Nearest Neighbors
 Decision Tree
 Support Vector Machines
w

26. Why is random forest better than SVM?

Random Forest is intrinsically suited for multiclass problems, while SVM is
ww

intrinsically two-class. For multiclass problem you will need to reduce it into
multiple binary classification problems. Random Forest works well with a mixture
of numerical and categorical features.

27. Which is better linear regression or random forest?

Multiple linear regression is often used for prediction in neuroscience.
Random forest regression is an alternative form of regression. It does not make the
assumptions of linear regression. We show that linear regression can be superior
to random forest regression.

28. Which is better linear or tree based models?

www.Notesfree.in
www.Notesfree.in

If there is a high non-linearity & complex relationship between dependent &

independent variables, a tree model will outperform a classical regression method.
If you need to build a model which is easy to explain to people, a decision tree
model will always do better than a linear model.

29. Is linear discriminant analysis classification or regression?

Linear Discriminant Analysis is a simple and effective method for
classification. Because it is simple and so well understood, there are many
extensions and variations to the method.

30. What is probabilistic discriminative model?

n
Discriminative models are a class of supervised machine learning models
which make predictions by estimating conditional probability P(y|x). In order to

e.i
use a generative model, more unknowns should be solved: one has to estimate
probability of each class and probability of observation given class.

31. What is SVM?

It is a supervised learning algorithm used both for classification and regression

e
problems. A type of discriminative modelling, support vector machine (SVM)
creates a decision boundary to segregate n-dimensional space into classes. The
sfr
best decision boundary is called a hyperplane created by choosing the extreme
points called the support vectors.

32. What is Decision tree?

A type of supervised machine learning model where data is continuously split
ote

according to certain parameters. It has two main entities–decision nodes and

leaves. While leaves are the final outcomes or decisions, nodes are the points
where data is split.

33. What is Random forest?

It is a flexible and easy-to-use machine learning algorithm that gives great
.N

results without even using hyper-parameter tuning. Because of its simplicity and
diversity, it is one of the most used algorithms for both classification and
regression tasks.
34. What is Decision Tree Classification?
w

A decision tree builds classification (or regression) models as a tree structure,

with datasets broken up into ever-smaller subsets while developing the decision
tree, literally in a tree-like way with branches and nodes. Decision trees can
handle both categorical and numerical data.
ww

35. What Is Pruning in Decision Trees, and How Is It Done?

Pruning is a technique in machine learning that reduces the size of decision
trees. It reduces the complexity of the final classifier, and hence improves
predictive accuracy by the reduction of overfitting.
Pruning can occur in:
● Top-down fashion. It will traverse nodes and trim subtrees starting at the
root
● Bottom-up fashion. It will begin at the leaf nodes
There is a popular pruning algorithm called reduced error pruning, in which:
● Starting at the leaves, each node is replaced with its most popular class

www.Notesfree.in
www.Notesfree.in

● If the prediction accuracy is not affected, the change is kept

● There is an advantage of simplicity and speed

36. Do you think 50 small decision trees are better than a large one? Why?
Yes. Because a random forest is an ensemble method that takes many weak
decision trees to make a strong learner. Random forests are more accurate, more
robust, and less prone to overfitting.

37. You’ve built a random forest model with 10000 trees. You got delighted after
getting training error as 0.00. But, the validation error is 34.23. What is going
on? Haven’t you trained your model perfectly?

n
The model has overfitted. Training error 0.00 means the classifier has
mimicked the training data patterns to an extent, that they are not available in the

e.i
unseen data. Hence, when this classifier was run on an unseen sample, it couldn’t
find those patterns and returned predictions with higher error. In a random forest,
it happens when we use a larger number of trees than necessary. Hence, to avoid
this situation, we should tune the number of trees using cross-validation.

e
38. When would you use random forests vs SVM and why?
There are a couple of reasons why a random forest is a better choice of the
sfr
model than asupport vector machine:
● Random forests allow you to determine the feature importance. SVM’s can’t
do this.
● Random forests are much quicker and simpler to build than an SVM.
● For multi-class classification problems, SVMs require a one-vs-rest method,
ote

which is less scalable and more memory intensive.

Part – B
1. Assume a disease so rare that it is seen in only one person out of every
million. Assume also that we have a test that is effective in that if a person has
the disease, there is a 99 percent chance that the test result will be positive;
.N

however, the test is not perfect, and there is a one in a thousand chance that
the test result will be positive on a healthy person. Assume that a new patient
arrives and the test result is positive. What is the probability that the patient
has the disease?
w

2. Explain Naïve Bayes Classifier with an Example.

3. Explain SVM Algorithm in Detail.
4. Explain Decision Tree Classification.
5. Explain the principle of the gradient descent algorithm. Accompany your
ww

explanation with a diagram. Explain the use of all the terms and constants
that you introduce and comment on the range of values that they can take.
6. Explain the following
a) Linear regression
b) Logistic Regression

www.Notesfree.in
www.Notesfree.in

UNIT IV ENSEMBLE TECHNIQUES AND UNSUPERVISED LEARNING

Combining multiple learners: Model combination schemes, Voting, Ensemble Learning -
bagging, boosting, stacking, Unsupervised learning: K-means, Instance Based Learning:
KNN, Gaussian mixture models and Expectation maximization

PART - A
1. What is bagging and boosting in ensemble learning?
Bagging is a way to decrease the variance in the prediction by generating additional
data for training from dataset using combinations with repetitions to produce multi-sets of the
original data. Boosting is an iterative technique which adjusts the weight of an observation

n
based on the last classification.

e.i
2. What is stacking in ensemble learning?
Stacking is one of the most popular ensemble machine learning techniques used to
predict multiple nodes to build a new model and improve model performance. Stacking
enables us to train multiple models to solve similar problems, and based on their combined
output, it builds a new model with improved performance.

e
3. Which are the three types of ensemble learning?
sfr
The three main classes of ensemble learning methods are bagging, stacking, and
boosting, and it is important to both have a detailed understanding of each method and to
consider them on your predictive modeling project.

4. Why ensemble methods are used?

ote

There are two main reasons to use an ensemble over a single model, and they are
related; they are: Performance: An ensemble can make better predictions and achieve better
performance than any single contributing model. Robustness: An ensemble reduces the
spread or dispersion of the predictions and model performance.

5. What is a voting classifier?

.N

A voting classifier is a machine learning estimator that trains various base models or
estimators and predicts on the basis of aggregating the findings of each base estimator. The
aggregating criteria can be combined decision of voting for each estimator output
w

6. What type of classifiers are used in weighted voting method?

The performance-weighted-voting model integrates five classifiers including logistic
regression, SVM, random forest, XGBoost and neural networks. We first used cross-
validation to get the predicted results for the five classifiers.
ww

7. What is difference between K means and Gaussian mixture?

K-Means is a simple and fast clustering method, but it may not truly capture
heterogeneity inherent in Cloud workloads. Gaussian Mixture Models can discover
complex patterns and group them into cohesive, homogeneous components that are
close representatives of real patterns within the data set.

8. What are Gaussian mixture models How is expectation maximization used in it?
Expectation maximization provides an iterative solution to maximum
likelihood estimation with latent variables. Gaussian mixture models are an approach

www.Notesfree.in
www.Notesfree.in

to density estimation where the parameters of the distributions are fit using the
expectation-maximization algorithm.
9. What is k-means unsupervised learning?
K-Means clustering is an unsupervised learning algorithm. There is no labeled
data for this clustering, unlike in supervised learning. K-Means performs the division
of objects into clusters that share similarities and are dissimilar to the objects
belonging to another cluster. The term 'K' is a number.

10. What is the difference between K-means and KNN?

KNN is a supervised learning algorithm mainly used for classification
problems, whereas K-Means (aka K-means clustering) is an unsupervised learning

n
algorithm. K in K-Means refers to the number of clusters, whereas K in KNN is the
number of nearest neighbors (based on the chosen distance metric).

e.i
11. What is expectation maximization algorithm used for?
` The EM algorithm is used to find (local) maximum likelihood parameters of a
statistical model in cases where the equations cannot be solved directly. Typically
these models involve latent variables in addition to unknown parameters and known
data observations.

e
sfr
12. What is the advantage of Gaussian process?
Gaussian processes are a powerful algorithm for both regression and
classification. Their greatest practical advantage is that they can give a reliable
estimate of their own uncertainty.
ote

13. What are examples of unsupervised learning?

Some examples of unsupervised learning algorithms include K-Means
Clustering, Principal Component Analysis and Hierarchical Clustering.

14. How do you implement expectation maximization algorithm?

The two steps of the EM algorithm are:
.N

E-step: perform probabilistic assignments of each data point to some class

based on the current hypothesis h for the distributional class parameters;
M-step: update the hypothesis h for the distributional class parameters based
on the new data assignments.
w

15. What is the principle of maximum likelihood?

The principle of maximum likelihood is a method of obtaining the optimum
values of the parameters that define a model. And while doing so, you increase the
ww

likelihood of your model reaching the “true” model.

Part – B
1. Explain briefly about unsupervised learning structure?
2. Explain various learning techniques involved in unsupervised learning?
3. What is Gaussian process? And explain in detail of Gaussian parameter
estimates with suitable examples.
4. Explain the concepts of clustering approaches. How it differ from classification.
5. List the applications of clustering and identify advantages and disadvantages of
clustering algorithm.
6. Explain about EM algorithm.

www.Notesfree.in
www.Notesfree.in

7. List non-parametric techniques and Explain K-nearest neighbour estimation.

UNIT V NEURAL NETWORKS

Perceptron - Multilayer perceptron, activation functions, network training – gradient
descent optimization – stochastic gradient descent, error backpropagation, from shallow
networks to deep networks –Unit saturation (aka the vanishing gradient problem) – ReLU,
hyperparameter tuning, batch normalization, regularization, dropout.

1. What is perceptron and its types?

A Perceptron is an Artificial Neuron. It is the simplest possible Neural Network. Neural
Networks are the building blocks of Machine Learning.

n
e.i
2. Which activation function is used in multilayer perceptron?
Image result for Perceptron - Multilayer perceptron, activation functions
The Sigmoid Activation Function: Activation in Multilayer Perceptron Neural Networks.

3. What are the activation functions of MLP?

e
In MLP and CNN neural network models, ReLU is the default activation function for
hidden layers. In RNN neural network models, we use the sigmoid or tanh function for
sfr
hidden layers. The tanh function has better performance. Only the identity activation function
is considered linear.

4. Does MLP have activation function?

Multilayer perceptrons (MLP) has been proven to be very successful in many applications
ote

including classification. The activation function is the source of the MLP power. Careful
selection of the activation function has a huge impact on the network performance.

5. What is the difference between a perceptron and a MLP?

The Perceptron was only capable of handling linearly separable data hence the multi-layer
perception was introduced to overcome this limitation. An MLP is a neural network capable
.N

of handling both linearly separable and non-linearly separable data.

6. What are the types of activation function?

Popular types of activation functions and when to use them
w

 Binary Step Function

 Linear Function
 Sigmoid
 Tanh
ww

 ReLU
 Leaky ReLU
 Parameterised ReLU
 Exponential Linear Unit

7. What is MLP and how does it work?

A multilayer perceptron (MLP) is a feedforward artificial neural network that generates a
set of outputs from a set of inputs. An MLP is characterized by several layers of input nodes
connected as a directed graph between the input and output layers. MLP uses
backpropogation for training the network.

www.Notesfree.in
www.Notesfree.in

8. Why do you require Multilayer Perceptron?

MLPs are useful in research for their ability to solve problems stochastically, which often
allows approximate solutions for extremely complex problems like fitness approximation.

9. What are the advantages of Multilayer Perceptron?

Advantages of Multi-Layer Perceptron:
A multi-layered perceptron model can be used to solve complex non-linear problems.
It works well with both small and large input data.
It helps us to obtain quick predictions after the training.
It helps to obtain the same accuracy ratio with large as well as small data.

n
10. What do you mean by activation function?

e.i
An activation function is a function used in artificial neural networks which outputs a
small value for small inputs, and a larger value if its inputs exceed a threshold. If the inputs
are large enough, the activation function "fires", otherwise it does nothing.

11. What are the limitations of perceptron?

e
Perceptron networks have several limitations. First, the output values of a perceptron can
take on only one of two values (0 or 1) because of the hard-limit transfer function. Second,
sfr
perceptrons can only classify linearly separable sets of vectors.

12. How many layers are there in perceptron?

This is known as a two-layer perceptron. It consists of two layers of neurons. The first
layer is known as hidden layer, and the second layer, known as the output layer, consists of a
ote

single neuron.

13. is stochastic gradient descent same as gradient descent?

Compared to Gradient Descent, Stochastic Gradient Descent is much faster, and more
suitable to large-scale datasets. But since the gradient it's not computed for the entire dataset,
and only for one random point on each iteration, the updates have a higher variance.
.N

14. How is stochastic gradient descent used as an optimization technique?

Stochastic gradient descent is an optimization algorithm often used in machine learning
applications to find the model parameters that correspond to the best fit between predicted
w

and actual outputs. It's an inexact but powerful technique. Stochastic gradient descent is
widely used in machine learning applications.

15. Does stochastic gradient descent lead to faster training?

ww

Gradient Descent is the most common optimization algorithm and the foundation of how
we train an ML model. But it can be really slow for large datasets. That's why we use a
variant of this algorithm known as Stochastic Gradient Descent to make our model learn a lot
faster.

16. What is stochastic gradient descent and why is it used in the training of neural
networks?
Stochastic Gradient Descent is an optimization algorithm that can be used to train neural
network models. The Stochastic Gradient Descent algorithm requires gradients to be
calculated for each variable in the model so that new values for the variables can be
calculated.

www.Notesfree.in
www.Notesfree.in

17. What are the three main types gradient descent algorithm?
There are three types of gradient descent learning algorithms: batch gradient descent,
stochastic gradient descent and mini-batch gradient descent.

18. What are the disadvantages of stochastic gradient descent?

SGD is much faster but the convergence path of SGD is noisier than that of original
gradient descent. This is because in each step it is not calculating the actual gradient but an
approximation. So we see a lot of fluctuations in the cost.

19. How do you solve the vanishing gradient problem within a deep neural network?

n
The vanishing gradient problem is caused by the derivative of the activation function used
to create the neural network. The simplest solution to the problem is to replace the activation

e.i
function of the network. Instead of sigmoid, use an activation function such as ReLU

20. What is the problem with ReLU?

Key among the limitations of ReLU is the case where large weight updates can mean that
the summed input to the activation function is always negative, regardless of the input to the

e
network. This means that a node with this problem will forever output an activation value of
0.0. This is referred to as a “dying ReLU“
sfr
21. Why is ReLU used in deep learning?
The ReLU function is another non-linear activation function that has gained popularity in
the deep learning domain. ReLU stands for Rectified Linear Unit. The main advantage of
using the ReLU function over other activation functions is that it does not activate all the
ote

neurons at the same time.

22. Why is ReLU better than Softmax?

As per our business requirement, we can choose our required activation function.
Generally , we use ReLU in hidden layer to avoid vanishing gradient problem and better
computation performance , and Softmax function use in last output layer .
.N

Part – B

1. Draw the architecture of a single layer perceptron (SLP) and explain its
w

operation. Mention its advantages and disadvantages.

2. Draw the architecture of a Multilayer perceptron (MLP) and explain its
operation. Mention its advantages and disadvantages.
3. Explain the stochastic optimization methods for weight determination.
ww

4. Describe back propagation and features of back propagation.

5. Write the flowchart of error back-propagation training algorithm.
6. Develop a Back propagation algorithm for Multilayer Feed forward neural
network consisting of one input layer, one hidden layer and output layer from
first principles.
7. List the factors that affect the performance of multilayer feed-forward neural
network.
8. Difference between a Shallow Net & Deep Learning Net.
9. How do you tune hyperparameters for better neural network performance?
Explain in detail.

www.Notesfree.in