III Year/V Semester

AD 8552 - Machine Learning

Course WorkBook
 Index Sheet - Coursework Book

SlNo Content Pages

1 Course Notes 1-33

2 Content Beyond the Syllabus 34-36

3 Assignment 37

4 Resources 37

5 Question Bank(Part-A,B,C) 38-40

6 Multiple Choice Questions 41-44

 Index Sheet - Activities

SlNo Activity Teacher’s Initial

1 Activity Corner-1 Project Based Learning

2 History Corner

3 Activity Corner-2 Linear & Nonlinear

Model

4 Activity Corner-3 Linear & Nonlinear

Model

5 Programming Corner-1

6 Programming Corner-2

7 Research Corner - Expert Systems

8 Activity Corner-3 Data Sets

9 Problem Solving WorkSpace-1 PCA

10 Problem Solving WorkSpace-2 LDA

11 Research Corner - PCA & LDA

12 Activity Corner - Content Beyond the

Syllabus

13 Unit-1 Assignment

14 MCQ Corner
 UNIT I MACHINE LEARNING BASICS

Unit Syllabus Content

Introduction to Machine Learning (ML) - Essential concepts of ML – Types of learning –

Machine learning methods based on Time – Dimensionality – Linearity and Non linearity –
Early trends in Machine learning – Data Understanding Representation and visualization

1.Introduction to Machine Learning

Machine learning is a branch of artificial intelligence (AI) and computer science that
focuses on using data and algorithms to mimic how humans learn and improve accuracy
over time.
Over the past few decades, technological advances in storage and processing power have
enabled several innovative products powered by machine learning. B. Netflix's
recommendation engine and self-driving cars.
Machine learning is an important part of the growing field of data science. By using
statistical techniques, algorithms are trained to classify or predict and uncover key insights
in data mining projects. These insights drive decisions within applications and
organizations, ideally influencing key growth indicators. As big data continues to expand
and grow, the market demand for data scientists will increase. It should help us to identify
the most relevant business questions and the data to answer them.

Source : CBINSIGHTS Figure 1: Promising AI/ML/DL StartUps in 2021-22

1.1 ML vs DL vs ANN
Deep learning and machine learning are usually used interchangeably, so it's important to
be aware of the nuances of the two. Machine learning, deep learning, and neural networks
are sub-areas of artificial intelligence. However, neural networks are actually a subfield of
machine learning, and deep learning is a subfield of neural networks.

Deep learning and machine learning differ in how each algorithm learns. In "deep"
machine learning, labeled datasets, also known as supervised learning, can be used to
inform algorithms, but labeled datasets are not always required. Deep learning can take
unstructured data in its raw form (such as text or images) and automatically determine a
set of characteristics that distinguish different categories of data. This eliminates some of
the required human intervention and allows for larger data sets.

Classical or "non-deep" machine learning relies heavily on human intervention to learn.

Human experts determine a set of features to understand the differences between data
inputs, but they typically need to learn from more structured data. A neural network, or
artificial neural network (ANN), consists of node layers, including an input layer, one or
more hidden layers, and an output layer. Each node or artificial neuron is connected to
another node and has associated weights and thresholds. When an individual node's output
exceeds a specified threshold, that node wakes up and sends data to the next layer of the
network. Otherwise, no data will be passed from this node to the next layer of the network.
"Deep" in deep learning simply refers to the number of layers in the neural network. A
neural network consisting of three or more layers with inputs and outputs can be
considered a deep learning algorithm or deep neural network. A neural network with only
three layers is just a basic neural network. Deep learning and neural networks are believed
to be accelerating progress in areas such as computer vision, natural language processing,
and speech recognition.

2.Essential concepts of ML

UC Berkeley divides the learning system of machine learning algorithms into three main
parts.

1. Decision-making process: Machine learning algorithms are commonly used

for prediction or classification. Based on some input data, which may or may
not be labeled, the algorithm makes inferences about patterns in the data.
2. Error function: The error function evaluates the model's predictions. If you
have a known example, you can use the error function to make comparisons
and determine the accuracy of your model.
 3. Model optimization process: If the model fits the data points in the training
set better, the weights are adjusted to reduce the discrepancy between the
known examples and the model's estimates. The algorithm repeats this
"evaluation and optimization" process, updating the weights autonomously
until an accuracy threshold is reached.

3.Types of learning

Machine learning models fall into four main categories.

3.1 Supervised Machine Learning

Supervised learning is defined by using a labeled dataset to train an algorithm to classify
data or accurately predict outcomes . As input data is fed into the model, the model adjusts
the weights until it fits well. This is done as part of the cross-validation process and ensures
that the model avoids overfitting or underfitting. Supervised learning helps organizations
solve a variety of real-world problems at scale, including: B. Put spam in a separate folder
from your inbox. Methods used in supervised learning include neural networks, naive
Bayes, linear regression, logistic regression, random forests, and support vector machines
(SVM).

3.2 Unsupervised Machine Learning

Unsupervised learning uses machine learning algorithms to analyze and group unlabeled
datasets. These algorithms discover hidden patterns and groups in data without the need
for human intervention. The method's ability to discover similarities and differences in
information makes it ideal for exploratory data analysis, cross-selling strategies, customer
segmentation, and image and pattern recognition. It is also used to reduce the number of
features in the model through the process of dimensionality reduction. Principal
Component Analysis (PCA) and Singular Value Decomposition (SVD) are two popular
approaches to do this. Other algorithms used in unsupervised learning include neural
networks, k-means clustering, and probabilistic clustering techniques.

3.3 Semi-supervised Learning

Semi-supervised learning offers a middle ground between supervised and unsupervised
learning. During training, a small labeled dataset is used for classification and feature
extraction from a large unlabeled dataset. Semi-supervised learning can solve the problem
of insufficient labeled data for supervised learning algorithms. It's also useful when tagging
enough data is too costly.
3.4 Reinforcement Learning
Reinforcement machine learning is a machine learning model similar to supervised
learning, but the algorithm is not trained on sample data. This model learns over time
through trial and error. A set of successful results is enriched to develop the best
recommendations or policies for a particular problem.

Figure 2 : Classification of ML Methods

Activity Corner- Project Based Learning

Login into the following link using our college email

https://teachablemachine.withgoogle.com/

Create Models for the following applications aforementioned in the above link.

● Bananameter
● Snap Clap Whistle
● Head Tilt

Screen record your executions and store them in your github repository AD8552_ML.
4.Machine learning methods based on Time

4.1 Static Learning

Static learning refers to learning from data captured as a single snapshot
Data properties remain constant over time. Once the model is trained
Data (using either supervised or unsupervised learning) can be a trained model
applied to similar data at any time in the future, the model remains valid,
works as expected.

4.2 Dynamic Learning

It is time sensitive and changes over time. So the training of the model is
Static process, but the model must be trained continuously (or after each)
reasonable time frame) remains valid. A typical example of such a problem
Weather forecasts and stock market forecasts. Model trained one year ago
Predicting tomorrow's weather and prices is utterly useless for
each share tomorrow.

5.Dimensionality
Dimensionality in measurements alludes to the number of properties a dataset has. For
instance, medical services information is famous for having immense measures of factors
(for example circulatory strain, weight, cholesterol level). In an ideal world, this
information could be addressed in a calculation sheet, with one segment addressing each
aspect. Practically speaking, this is challenging to do, to some degree in light of the fact that
numerous factors are between related (like weight and circulatory strain).

Note: Dimensionality implies something else in different areas of arithmetic and science.
For instance, in material science, dimensionality can as a rule be communicated as far as
basic aspects like mass, time, or length. In framework variable based math, two units of
measure have a similar dimensionality on the off chance that the two proclamations are
valid:
1.A capability exists that maps one variable onto another variable.
2.The opposite of the capability in (1) does the converse.

High Dimensional implies that the quantity of aspects are astoundingly high — so high that
estimations become very troublesome. With high layered information, the quantity of
elements can surpass the quantity of perceptions. For instance, microarrays, which
measure quality articulation, can contain several many examples. Each example can contain
a huge number of genes.One individual (for example one perception) has a huge number of
conceivable quality blends. Different regions where elements surpass perceptions
incorporate money, high goal imaging, and site examination (for example publicizing,
slithering, or positioning).

Decrease of dimensionality means to improve on comprehension of information, either

mathematically or outwardly. Information trustworthiness is kept up with. To lessen
dimensionality, you could join related information into bunches involving a device like
complex scaling to recognize similitudes in information. We could likewise utilize bunching
to gather things together.

5.1 Curse of Dimensionality

The scourge of dimensionality typically alludes to what happens when we add an ever
increasing number of factors to a multivariate model. The more aspects you add to an
informational collection, the more troublesome it becomes to foresee specific amounts. You
would imagine that more is better. Notwithstanding, with regards to adding factors, the
inverse is valid. Each additional variable outcomes in a remarkable diminishing in prescient
power.

As a straightforward model, suppose we are utilizing a model to foresee the area of a huge
microscopic organisms in a 25cm2 petri dish. The model may be genuinely exact at nailing
the molecule down to the closest square cm. Notwithstanding, suppose we add only
another aspect: Instead of a 2D petri dish you utilize a 3D recepticle . The prescient space
increments dramatically, from 25 cm2 to 125 cm3. At the point when we add more aspects,
it's a good idea that the computational weight likewise increments. It wouldn't be difficult
to pinpoint where microorganisms may be in a 3D model. Be that as it may, it's a seriously
difficult undertaking.

The measurable revile of dimensionality alludes to a connected truth: an expected example

size n will develop dramatically with information that has d aspects. In basic terms, adding
more aspects could imply that the example size we really want immediately become
unmanageable.

History Corner

Most referred research article for Dimensionality of Data which was published in 1977 for
your reference.

Finney, D.J. (1977). “Dimensions of Stat” Journal of the Royal Stat. Society. Series C (Applied
Stat). 26, No.3, p.285-289.
6.Linearity & Non Linearity

Figure 3 : Illustration of Linear & Non Linear Data

Linear data is data that can be represented on a line graph. This means that there is a clear
relationship between the variables and that the graph will be a straight line.

Activity Corner

Write the inference about the aforementioned graph.

1.
2.

3.

4.

Non-linear data, on the other hand, cannot be represented on a line graph. This is because
there is no clear relationship between the variables and the graph will be curved.

Activity Corner

Write the inference about the aforementioned graph.

1.

2.
3.

4.

6.1 Guidelines for Choosing Linear & Non Linear Models

The basic principle is to utilize direct relapse first to decide if it can fit the specific kind of
bend in our information. In the event that we can't acquire a sufficient fit utilizing straight
relapse, that is the point at which we could have to pick nonlinear relapse.

Direct relapse is simpler to utilize, less difficult to decipher, and we acquire more insights
that assist us with surveying the model. While straight relapse can display bends, it is
moderately confined looking like the bends that it can fit. Now and again it can't fit the
particular bend in our information.

Nonlinear relapse can fit a lot more kinds of bends, yet it can require more exertion both to
view as the best fit and to decipher the job of the free factors. Also, R-squared isn't
legitimate for nonlinear relapse, and it is difficult to compute p-values for the boundary
gauges.

Programming Corner

Execute the following Python Program in Colab/Jupyter Notebook and draw the output
graph in the following workspace.

import pandas as pd
import numpy as np
from sklearn import datasets
import matplotlib.pyplot as plt

# Load the IRIS Dataset

#
iris = datasets.load_iris()
X = iris.data
y = iris.target

# Create a scatter plot

#
plt.scatter(X[:50, 0], X[:50, 1], color='green', marker='o', label='setosa')
plt.scatter(X[50:100, 0], X[50:100, 1], color='black', marker='x', label='versicolor')
plt.xlabel('sepal length [cm]')
plt.ylabel('petal length [cm]')
plt.legend(loc='upper left')
plt.show()

Output:

Inference:
 Programming Corner

Execute the following Python Program in Colab/Jupyter Notebook and draw the output
graph in the following workspace.

import pandas as pd
import numpy as np
from sklearn import datasets
import matplotlib.pyplot as plt
# Load the IRIS Dataset
#
iris = datasets.load_iris()
X = iris.data
y = iris.target
# Create a scatter plot
#
plt.scatter(X[50:100, 0], X[50:100, 1], color='black', marker='x', label='versicolor')
plt.scatter(X[100:150, 0], X[100:150, 1], color='red', marker='+', label='verginica')
plt.xlabel('sepal length [cm]')
plt.ylabel('petal length [cm]')
plt.legend(loc='upper left')
plt.show()

Output:
Inference:

7.Occam's Razor

Occam's (or Ockham's) razor is a principle attributed to the 14th century logician and
Franciscan friar William of Ockham. Ockham was the village in the English county of Surrey
where he was born.

The principle states that "Entities should not be multiplied unnecessarily."

8.No Free Lunch Theorem

The theorem states that all optimization algorithms perform equally well when their
performance is averaged across all possible problems.

9.Law of Diminishing Returns

The law of diminishing returns is an economic theory predicting that after an optimal level
of capacity, adding additional factors of production decreases the output.

10.Early trends in Machine learning

10.1 Expert System

An Expert System (ES) is an information based framework that utilizes information about
its application space and uses an inferencing (reason) system to take care of issues that
would somehow require human skill or mastery. The force of master frameworks stems
fundamentally from the particular information about a tight space put away in the master
framework's information base.
10.2 Categories of Expert System Applications

Classification - identify an object based on stated characteristics

Diagnosis Systems - infer malfunction or disease from observable data
Monitoring - compare data from a continually observed system to prescribe behavior
Process Control - control a physical process based on monitoring
Design - configure a system according to specifications
Scheduling & Planning - develop or modify a plan of action
Generation of Options - generate alternative solutions to a problem

10.3 Benefits of Expert Systems

1. An Es can finish its important for the errands a lot quicker than a human master.
2. The blunder pace of fruitful frameworks is low, at times a lot of lower than the human
mistake rate for a similar errand.
3. ESs make reliable proposals
4. ESs are a helpful vehicle for bringing to the mark of utilization hard to-utilize wellsprings
of information.
5. ESs can catch the scant skill of an exceptionally qualified master.
6. ESs can turn into a vehicle for developing hierarchical information, rather than the
information on people in the association.
7. At the point when use as preparing vehicles, ESs bring about a quicker expectation to
learn and adapt for fledglings.
8. The organization can work an ES in conditions dangerous for people.

10.4 Limitations of Expert Systems

1. Constraints of the innovation
2. Issues with information obtaining
3. Functional spaces as the main area of ES application
4. Keeping up with human aptitude in associations

Research Corner

Using the following link

https://www.sciencedirect.com/journal/expert-systems-with-applications/issues

Identify a research article which was published between 2021 to 2022 of your choice.
1.Write the title of the article.
2.Read and write the glimpse of the complete paper in the following workspace within 50
words/8 Sentences.

11.Understanding the data

Figure 4: Understanding entities,attributes,data types

 Figure 5: Advanced Understanding of entities,attributes,data types

There are two key stages of Data Understanding: a Data Assessment and Data Exploration.

11.1 Answer the following questions for Data Assessment

● What data is available?

● How much data is available?
● Do you have access to the ground truth, the values you’re trying to predict?
● What format will the data be in?
● Where does it reside?
● How can the data be accessed?
● Which fields are most important?
● How do the multiple data sources get joined?
● What important metrics are reported using this data?
● If applicable, how does the data map to the current method of completing the task
today?
11.2 Activities for Data Exploration

Count the number of records — is this what we expected?

What are the datatypes? Will we need to change these for a machine learning model?
Look for missing values — how should we deal with these?
Verify the distribution of each column — are they matching the distribution we expect (e.g.
normally distributed)?
Search for outliers — are there anomalies in our data? Are all values valid (e.g. no ages less
than 0)?
Validated if wer data is balanced — are different groups represented in wer data? Are there
enough examples of each class we wish to predict?
Is there bias in our data — are subgroups in our data treated more favorable than others?

Activity Corner

Use the following github link https://github.com/awesomedata/awesome-public-datasets

1. Identify a Domain/Area of your choice

2. Choose a dataset of your choice
3. Answer the questions in Data Assessment
4. Perform the data exploration
11.3 Representation & Visualization of Data

11.4 Principal Components Analysis

Principal Component Analysis is an unsupervised learning algorithm that is utilized for the
dimensionality decrease in AI. A measurable cycle changes over the perceptions of related
highlights into a bunch of directly uncorrelated elements with the assistance of
symmetrical change. These new changed highlights are known as the Principal
Components. One of the famous apparatuses is utilized for exploratory information
examination and prescient demonstration. It is a method to draw solid examples from the
given dataset by lessening the fluctuations.

Figure 6: Visualization of PCA

PCA by and large attempts to track down the lower-layered surface to project the
high-layered information.PCA works by considering the change of each property on the
grounds that the high trait shows the great split between the classes, and thus it lessens the
dimensionality. A few certifiable uses of PCA are picture handling, film suggestion
framework, improving the power portion in different correspondence channels. It is an
element extraction method, so it contains the significant factors and drops the most
non-significant variable.

Problem

Given the data in Table, reduce the dimension from 2 to 1 using the Principal Component
Analysis (PCA) algorithm.
 Feature Example-1 Example-2 Example-3 Example-4

X1 4 8 13 7

X2 11 4 5 14

Step 1: Calculate Mean

The figure shows the scatter plot of the given data points.

Calculate the mean of X1 and X2 as shown below.

Step 2: Calculation of the covariance matrix.

The covariances are calculated as follows:

The covariance matrix is,

Step 3: Eigenvalues of the covariance matrix

The characteristic equation of the covariance matrix is,

Solving the characteristic equation we get,

Step 4: Computation of the eigenvectors

To find the first principal components, we need only compute the eigenvector
corresponding to the largest eigenvalue. In the present example, the largest eigenvalue is λ1
and so we compute the eigenvector corresponding to λ1.

The eigenvector corresponding to λ = λ1 is a vector

satisfying the following equation:

This is equivalent to the following two equations:

Using the theory of systems of linear equations, we note that these equations are not
independent and solutions are given by,
that is,

where t is any real number.

Taking t = 1, we get an eigenvector corresponding to λ1 as

To find a unit eigenvector, we compute the length of X1 which is given by,

Therefore, a unit eigenvector corresponding to λ1 is

By carrying out similar computations, the unit eigenvector e2 corresponding to the

eigenvalue λ= λ2 can be shown to be,

Step 5: Computation of first principal components

let,

be the kth sample in the above Table (dataset). The first principal component of this
example is given by (here “T” denotes the transpose of the matrix)

For example, the first principal component corresponding to the first example

is calculated as follows:

The results of the calculations are summarized in the below Table.

X1 4 8 13 7

X2 11 4 5 14

PC -4.3052 3.7361 5.6928 -5.1238

Step 6: Geometrical meaning of first principal components

First, we shift the origin to the “center”

and then change the directions of coordinate axes to the directions of the eigenvectors e1
and e2.

The coordinate system for principal components

Next, we drop perpendiculars from the given data points to the e1-axis (see below Figure).

Projections of data points on the axis of the first principal component

The first principal components are the e1-coordinates of the feet of perpendiculars, that is,
the projections on the e1-axis. The projections of the data points on the e1-axis may be
taken as approximations of the given data points hence we may replace the given data set
with these points.

Now, each of these approximations can be unambiguously specified by a single number,

namely, the e1-coordinate of approximation. Thus the two-dimensional data set can be
represented approximately by the following one-dimensional data set.

Geometrical representation of one-dimensional approximation to the data set

Problem Solving Workspace

Compute the principal component of following data

CLASS 1 X = 2 , 3 , 4 Y = 1 , 5 , 3

CLASS 2 X = 5 , 6 , 7 Y = 6 , 7 , 8
11.5 Linear Discriminant Analysis

Linear Discriminant Analysis (LDA) is a speculation of Fisher's straight discriminant, a

strategy utilized in Statistics, design acknowledgment and AI to find a straight mix of
highlights that portrays or isolates at least two classes of items or occasions. This strategy
projects a dataset onto a lower-layered space with great class-distinguishableness to stay
away from overfitting ("revile of dimensionality"), and to lessen computational expenses.
The subsequent mix might be utilized as a direct classifier or, all the more generally, for
dimensionality decrease before resulting grouping.

The first Linear discriminant was depicted for a 2-class issue, and it was afterwards
summed up as "multi-class Linear Discriminant Analysis" or "Various Discriminant
Analysis" by C. R. Rao in 1948 (The usage of various estimations in issues of organic
classification).In a nutshell, the objective of a LDA is frequently to project an element space
(a dataset n-layered examples) into a more modest subspace

k(where k≤n−1), while keeping up with the class-unfair data. As a general rule,
dimensionality decrease doesn't just assist with diminishing computational expenses for a
given order task, yet it can likewise be useful to stay away from overfitting by limiting the
blunder in boundary assessment.
Problem
 Problem Solving Workspace

Factory "ABC" produces very expensive and high quality chip rings that their qualities are
measured in terms of curvature and diameter. Result of quality control by experts is given
in the table below.

Curvature Diameter Quality Control Result

2.95 6.63 Passed

2.53 7.79 Passed

3.57 5.65 Passed

3.16 5.47 Passed

2.58 4.46 Not Passed

2.16 6.22 Not Passed

3.27 3.52 Not Passed

As a consultant to the factory, you get a task to set up the criteria for automatic quality
control. Then, the manager of the factory also wants to test your criteria upon new type of
chip rings that even the human experts are argued to each other. The new chip rings have
curvature 2.81 and diameter 5.46. Can you solve this problem by employing Discriminant
Analysis?
 Research Corner

Download the following research article from the following link and answer the following
questions.

Link : https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3972280

Questions

a.List the variables used in the dataset in Page 12.

b.List the learning models used in the paper.

c.Perform PCA and LDA using the data set in Page 12.

d.List the findings related to feature selection in Page 6 & 7.

e.Identify a research paper in the literature review section from Page 3 to 7 and check the
references for the full citation using the number.Download it.

f.List the merits and demerits discussed in e.

 Content Beyond the Syllabus - Machine Intelligence(MI)

Machine intelligence is advanced computing that enables a technology (a machine, device,

or algorithm) to interact with its environment intelligently, meaning it can take actions to
maximize its chance of successfully achieving its goals.

The concept of machine intelligence highlights the intersection of machine learning and
artificial intelligence, as well as the broad spectrum of opportunities and approaches in the
field.

Research of MI in Google

Google is at the forefront of innovation in Machine Intelligence, with active research

exploring virtually all aspects of machine learning, including deep learning and more
classical algorithms. Exploring theory as well as application, much of our work on language,
speech, translation, visual processing, ranking and prediction relies on Machine
Intelligence. In all of those tasks and many others, we gather large volumes of direct or
indirect evidence of relationships of interest, applying learning algorithms to understand
and generalize.

Machine Intelligence at Google raises deep scientific and engineering challenges, allowing
us to contribute to the broader academic research community through technical talks and
publications in major conferences and journals. Contrary to much of current theory and
practice, the statistics of the data we observe shifts rapidly, the features of interest change
as well, and the volume of data often requires enormous computation capacity. When
learning systems are placed at the core of interactive services in a fast changing and
sometimes adversarial environment, combinations of techniques including deep learning
and statistical models need to be combined with ideas from control and game theory.

Research of MI in Microsoft

Intelligent machines and intelligent software rely on algorithms that can reason about
observed data to make predictions or decisions that are useful. Such systems rely on
machine learning and artificial intelligence, combining computation, data, models, and
algorithms. Microsoft mission, in the Machine Intelligence theme at Microsoft Research
Cambridge, is to expand the reach and efficiency of machine intelligence technology.
Microsoft research how to incorporate structured input data such as code and molecules
effectively into deep learning models. Microsoft invent new methods so models can
accurately quantify their uncertainty when making predictions. Microsoft build models
that learn from small data that is corrupted or only partially observed. Microsoft develop
deep learning algorithms that apply to interactive settings in gaming and in decision
making task, where model predictions have consequences on future inputs.

Improving the performance of machine learning methods demands an ever-increasing scale

in computation while retaining flexibility to develop new models. Microsoft research new
AI compiler technology that will make it easier to express rich algorithms while effectively
utilizing modern accelerators.

Recent MI Research Article

Nair, Vineet, et al. "ADVISER: AI-Driven Vaccination Intervention Optimiser for Increasing
Vaccine Uptake in Nigeria." arXiv preprint arXiv:2204.13663 (2022).

Activity Corner

Identify a recent research article related to Machine Intelligence and write its citation in
IEEE format.
 Unit-1 Assignment

Identify a Society related problem in your area/native and share possible solutions
strategies through Machine Learning.

Resources
1.https://doi.org/10.1007/978-3-030-26622-6
2.Andrew NG Youtube Tutorial - Machine Learning
3.Introduction to Machine Learning,IITM,NPTEL/Swayam Course.
 AD8552- Machine Learning - Question Bank
Unit-1: Introduction to Machine Learning
Part A
Sl.no Questions Blooms
CO
Level*
1 Define Big Data.Give Examples. CO1 B1
2 Relate BigData with Artificial Intelligence and Machine Learning. CO1 B3
3 “Storage and Computational Resources from Cloud will impact B5
the performance of Machine Learning Algorithms by the role of CO1
Big Data” - Justify.
4 List the types of learning algorithms in Machine Learning. CO1 B1
5 Compare the types of learning algorithms in Machine Learning. CO1 B5
6 List the Machine Learning methods based on time. CO1 B1
7 Identify the metric to quote the property of Orthogonality in a B1
CO1
data set.
8 Define Curse of Dimensionality. CO1 B1
9 Relate data density with model accuracy. CO1 B3
10 “Image classification of different birds is a valid static B5
CO1
learning”-Justify
11 “Prediction of Car Engine Performance is a valid dynamic B5
CO1
learning”-Justify
12 Compare linear data and non linear data. CO1 B5
13 Compare linear model and non linear model. CO1 B5
14 Test that y=3ex is producing a non linear data. CO1 B6
15 Transform y=7ex into a linear data modeling equation. CO1 B2
16 Write the algorithms for linear model. CO1 B3
17 Write the algorithms for non linear model. CO1 B3
18 State the Principle of Parsimony. CO1 B1
19 State the NFL theorem. CO1 B1
20 State the Law of Diminishing Returns. CO1 B1
21 Define Expert Systems.Give Examples. CO1 B1
22 List the steps in data preprocessing. CO1 B1
23 Write the steps to understand the nature of data. CO1 B3
24 Write the examples of entities in a data set. CO1 B3
25 Write the examples of attributes in a data set. CO1 B3
26 Write the examples of datatypes in a dataset. CO1 B3
27 Relate Data Visualization with Data Dimensionality. CO1 B2
28 Define Principal Component Analysis(PCA). CO1 B1
29 Write the algorithm of PCA. CO1 B3
30 Define Linear Discriminant Analysis(LCA). CO1 B1
31 Write the algorithm of LCA. CO1 B3
32 Compare PCA and LDA. CO1 B4
Part B
Sl.no Questions Blooms
CO
Level*

1 List and explain the types of learning methods in Machine CO1 B3

Learning.
2 List and explain the types of learning methods in Machine CO1 B3
Learning based on time.
3 Illustrate linear and nonlinear separation between the classes in CO1 B4
a dataset.
4 Compare and Contrast Piecewise linear regression with non CO1 B5
linear regression.
5 Summarize the concept of PCA with a help of data set.Argue by CO1 B5
transforming the data set into a matrix and share the findings.
6 Summarize the concept of LDA with a help of data set.Argue by CO1 B5
transforming the data set into a matrix and share the findings.
Part C

Sl.no Questions CO Bloom

s
Level*

1 “Traffic Alert using Google Maps”.Write your findings related to CO1 B6

learning methods in Machine Learning.

2 Propose a specific problem in health care application and CO1 B6

formulate your way of solving the problem using Machine
Learning.

3 Criticize the performance of PCA and LDA using your own data CO1 B6
set with three variables.
 NAME :

CLASS :
Machine Learning
DATE :
30 Questions

1. 1.Real-Time decisions, Game AI, Learning Tasks, Skill Acquisition, and Robot Navigation are applications of which of the
following

A Unsupervised Learning: Clustering B Reinforcement Learning

C Supervised Learning: Classification D Unsupervised Learning: Regression

2. 2. What is Machine Learning?

· Artificial Intelligence
· Deep Learning
· Offline Learning

A (i) and (ii) B None

C All D Only (i)

3. 3. Which of the following is not type of learning?

A Semi-unsupervised Learning B Unsupervised Learning

C Reinforcement Learning D Supervised Learning

4. 4. Targeted marketing, Recommended Systems, and Customer Segmentation are applications in which of the following.

A Supervised Learning: Classification B Reinforcement Learning

C Unsupervised Learning: Regression D Unsupervised Learning: Clustering

5. 5. Fraud Detection, Image Classification, Diagnostic, and Customer Retention are applications in which of the following

A Reinforcement Learning B Supervised Learning: Classification

C Unsupervised Learning: Regression D Unsupervised Learning: Clustering

6. 6. What is Machine Learning (ML)?

The autonomous acquisition of knowledge through the The selective acquisition of knowledge through the use
A B
use of manual programs of manual programs

The selective acquisition of knowledge through the use The autonomous acquisition of knowledge through the
C D
of computer programs use of computer programs

7.
7. Which of the following are common classes of problems in machine learning?

A Clustering B Classification

C All of the above D Regression

 8. 8. Father of Machine Learning (ML)

A Geoffrey Everest Hinton B Geoffrey Chaucer

C Geoffrey Hill D None

9. 9. Which of the following is not a supervised learning?

A PCA B Naive Bayesian

C Decision Tree Answer D Linear Regression

10. 10. Different learning methods does not include?

A Deduction B Analogy

C Introduction D Memorization

11. 11. Linear Regression is a supervised machine learning algorithm.

A False B True

12. 12. Machine learning is a sub field of ___

A App Development B Web Development

C Game Development D Artificial Intelligence

13. 13. Which is the best language for machine learning ___

A Java B JavaScript

C Python D C++

14.
14. Machine learning model which is build on sample data is known as ___

A Transfer Data B None

C Training Data D Clean Data

15. 15. Application of machine learning methods to large databases is called ___

A Data Mining B Block Chain

C Web Development D App Development

16. 16. Output of training process in machine learning is called ___

A Accuracy B Outcome

C None of the above D Machine Learning Model

17. 17. What is unsupervised learning?

It is a diagram or chart that helps determine a course of It is used to train algorithms that classify data or predict
A B
action or show a statistical probability. outcomes accurately by using labeled datasets.

It is a type of algorithm that learns patterns from

C D None of the above
untagged data and predicts a outcome.
18. 18. What is Supervised learning?

It is used to train algorithms that classify data or predict

A None of the above B
outcomes accurately by using labeled datasets.

It is a diagram or chart that helps determine a course of It is a type of algorithm that learns patterns from
C D
action or show a statistical probability. untagged data and predicts a outcome.

19. 19. Which of the following is the best machine learning method?

A Accuracy B Scalable

C Fast D All of the above

20.
20. Which of the following is a widely used and effective machine learning algorithm based on the idea of bagging?

A Classification B Decision Tree

C Regression D Random Forest

21. 21. When performing regression or classification, which of the following is the correct way to preprocess the data?

Normalize the data → PCA → normalize PCA output →

A B PCA → normalize PCA output → training
training

C Normalize the data → PCA → training D None of the above

22. 22. A Machine Learning technique that helps in detecting the outliers in data

A Anomaly Detection B Clustering

C Classification D All of the above

23. 23. ………………. algorithms enable the computers to learn from data, and even improve themselves, without being
explicitly programmed.

A ArtificialIntelligence B None of the above

C Machine Learning D Deep Learning

24. 24. If machine learning model output involves target variable then that model is called as

A Reinforcement Learning B All of the above

C Descriptive Model D Predictive Model

25. 25. Data used to build a data mining model

A Training Data B Hidden Data

C Validation Data D Test Data

26. 26. The problem of finding hidden structure in unlabeled data is called…

A Supervised Learning B None of the above

C Reinforcement Learning D Unsupervised Learning

27. 27. Supervised learning and unsupervised clustering both require at least one

A Categorical attribute B Hidden Attribute

C Input attribute D Output Attribute

28. 28. ML is a field of AI consisting of learning algorithms that?

A At executing some task B Over time with experience

C Improve their performance D All of the above

29. 29. What is the application of machine learning methods to a large database called?

A Internet of things B Data Mining

C Artificial Intelligence D Big data computing

30. 30. Which of the following are common classes of problems in machine learning?

A All of the above B Classification

C Regression D Clustering