IT 323

ML: Machine
Learning
Conducted by Ruchika Pharswan
at Delhi Technological University

8 Aug 2024
Brief About Me
Email : ruchikapharswan2024@gmail.com

B.Tech (CSE) M.Tech (CSE) || Ph.D. coursework graduated

Teaching Assistant IIT Delhi, Assistant Prof. GGSIPU, TA IIHT

Journal, Book Chapters, Conferences papers || Active Reviewer

Governance of AI ,Technology Adoption, Social Media Analysis, User

Behavior,
Lecture Presentation | Ruchika Pharswan @Delhi Technological University 2
Jens Martensson
 INSTRUCTIONS
NO entry in class after 15 mins , once class the started.

NO proxies attendance. NO back attendances will be given.

If there will be mass bunk, then for next lecture, few of you will be randomly selected to present assigned topics.

Topic presentation (10), Assignments (5) will hold weightage of CWS component.

Also feel free to ask any doubts in class or even after class you can drop your query in AIES
WhatsApp group. Kindly send your rest of the requests and queries to CR , refrain individual
messages unless important. Thanks.
DO NOT UPLOAD THIS PPT. ANYWHERE !

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
By the end of this course you can assure following :

• Understand the fundamental concepts and algorithms of machine learning.

• Develop a comprehensive understanding of fundamental machine learning concepts, algorithms, and
techniques, including supervised and unsupervised learning, classification, regression, clustering, and
dimensionality reduction.
• Apply principles and algorithms to evaluate models generated from data.
• Learn to critically evaluate the performance of machine learning models using appropriate metrics.
• Develop the ability to identify and formulate problems suitable for machine learning solutions, design
appropriate models, and interpret results in practical applications.
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
Course Outline

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
UNIT 1
Introduction to machine learning

Sources : Books, Internet, Self

Referring to Prof. Dinesh Kumar Vishwakarma Lecture Notes.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
Machine learning is an application of artificial
intelligence (AI) that provides systems the ability to
automatically learn and improve from experience
without being explicitly programmed. Machine
learning focuses on the development of computer
programs that can access data and use it learn for
themselves.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
Any hunches ?
 Foundation of AI : Turing Test

Alan Turing

The Turing Test is a method of

inquiry in artificial intelligence (AI)
for determining whether or not a
computer is capable of thinking like
a human being.

Fool the tester

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

The Turing Test was the technique of exploration in AI for deciding whether or not a computer is competent of thinking
like a human being.

Limitation of Turing test : Nature of question was limited

Computer score high only if the questions were formulated in the queries, that can be answered either in "Yes" or "No“
or related to a narrow field of knowledge.

Whereas when questions were open-ended and needed conversational answers, computer scored less.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
PEAS Representation

 PEAS stands for PERFORMANCE , ENVIRONMENT, ACTUATORS, SENSORS.

PEAS is a type of model on which an AI agent works upon. When we define an AI agent or rational agent,
then we can group its properties under PEAS representation model. Here performance measure is the
objective for the success of an agent's behavior.

Let's suppose a self-driving car then PEAS representation will be:

Performance: Safety, time, legal drive, comfort

Environment: Roads, other vehicles, road signs,

pedestrian

Actuators: Steering, accelerator, brake, signal, horn

Sensors: Camera, sonar.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
History of Machine learning

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 Requires
Knowledge Intelligence

Complicated
Voluminous
Dynamic
Challenging to identify accurately

Intelligence requires knowledge. Knowledge is collection of facts.

ML technique is a method that achieves knowledge.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 No Human Expertise, Unknown Challenges, data driven learning

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Jens Martensson
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Jens Martensson
 Machine Learning
System
Inputs: Data + Output
Data: Similar to the traditional
system, this is the information you
want to work with.

However, in a machine learning

context, this data is typically much
larger and more complex.

Output: This is the desired result

(Labels) (Model) that you want the system to produce
based on the data.

For example, if you want a machine

TRAINING learning model to recognize cats in
images, the output could be labels
indicating whether a cat is present in
each image.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Output: Program (Model)
1. Representation of Some Phenomenon
2. Non-Maths /Stats Model

A MODEL is a mathematical or computational representation that

encapsulates patterns, relationships, or rules that have been learned from
data.

A model in machine learning is a learned representation that can make

predictions or decisions based on input data. It’s the core of what makes
machine learning systems powerful and adaptable.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

How a Model Works
1.Training:
1. During the training phase, a machine learning algorithm processes a dataset that includes both input
data and the corresponding desired outputs (also known as labels or targets).
2. The algorithm adjusts its internal parameters in such a way that it can map the input data to the
correct output. These internal parameters form the model.

2.Inference:
1. After training, the model can take new, unseen input data and apply the learned rules or patterns to
predict an output.

After training, the model can make predictions, classify data, or even generate new
Model Output content based on what it has learned.

TASK OF MACHINE LEARNING

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Types of Models
•Supervised Learning Models: Trained on labeled data, where the correct output is provided during
training.
•Unsupervised Learning Models: Learn from unlabeled data, identifying patterns or structures in the data
without explicit outputs.
•Reinforcement Learning Models: Learn by interacting with an environment, receiving feedback in the
form of rewards or penalties.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 Steps in ML
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Supervised Learning

Supervised learning involves learning a function that maps input data to output labels, based on a
set of input-output pairs. The model is trained on labeled data, where the correct output is provided
for each input in the training set.

•Classification:

• Objective: Predict a categorical label for given input data.

• Example: Spam detection in emails (spam or not spam), image recognition (cat or dog).
• Algorithms: Logistic Regression, Decision Trees, Support Vector Machines (SVM), k-
Nearest Neighbors (k-NN), Neural Networks.
•Regression:

• Objective: Predict a continuous value for given input data.

• Example: Predicting house prices, forecasting stock prices.
• Algorithms: Linear Regression, Ridge Regression, Lasso Regression, Polynomial
Regression, Neural Networks.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

The iris dataset contains three classes of flowers, Versicolor, Setosa, Virginica, and each class contains 4
features, ‘Sepal length’, ‘Sepal width’, ‘Petal length’, ‘Petal width’. The aim of the iris flower classification
is to predict flowers based on their specific features.
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Unsupervised Learning
Unsupervised learning involves finding hidden patterns or intrinsic structures in input data that is not labeled. The model
tries to learn the structure or distribution of the data without any supervision.

•Clustering:
• Objective: Group similar data points together based on some notion of similarity.
• Example: Customer segmentation, grouping similar documents.
• Algorithms: k-Means, Hierarchical Clustering, DBSCAN, Gaussian Mixture Models (GMM).

Anomaly Detection:
• Objective: Identify rare items, events, or observations that do not conform to the general distribution of the data.
• Example: Fraud detection, network security.
• Algorithms: Isolation Forest, One-Class SVM, Auto encoders.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 Diversity of data -refers to the variety or heterogeneity of the data used to train
Data Diversity and evaluate models.

Importance of Data Diversity

1.Improves Generalization:
A diverse dataset helps the model learn patterns that generalize well to new, unseen data. If the training
data is diverse, the model is less likely to overfit to specific patterns that don't hold across all data points.

2.Avoids Bias:
Lack of diversity can lead to biased models.
For example, if a facial recognition system is trained only on images of people with a certain skin tone, it
may perform poorly on images of people with different skin tones.

3.Captures Edge Cases:

Diverse data includes edge cases and rare situations, which are important for building robust models that
perform well across all possible inputs, not just the most common ones.

4.Reflects Real-World Complexity:

Real-world problems often involve complex and varied scenarios.
Diverse data ensures that the model can handle this complexity and doesn't just perform well on a narrow
subset of possible situations.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 Guess the distribution ?

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Normal Distribution: More common in real-life data due to natural variations, the central limit theorem, and the
way multiple factors interact in complex systems.

A normal distribution, also known as a Gaussian distribution, is characterized by a symmetric, bell-shaped

curve where most data points cluster around a central mean, with fewer points appearing as you move further
from the mean.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Feature Space

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 Inductive Learning
Inductive learning is a key concept in machine learning where a model learns general rules from specific
examples or data points. In this learning approach, the machine is provided with labeled training data, and
from this, it attempts to infer a generalized function that can make predictions on unseen data

Key Idea of Inductive Learning

•Specific to General: The model starts with specific instances (training data) and generalizes from these
examples to learn a broader pattern or rule.

•Learning from Observations: Inductive learning relies on learning patterns from observed data without assuming
the data fits into a pre-defined theory or structure.

How It Works

1.Training Data
2. Generalization
3.Prediction (tasks)

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
EXAMPLE : Lets take an example of loan approval system: learner uses examples from historical data to learn the
function F(X), then applies it to new data for making predictions. (income, credit score, and employment history.)

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Hyper parameter vs Parameter / features

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Types Of Classification
 Which type of
classification ?

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Linear Regression

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Linear
Strong Relationship Weak Relationship

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Another formula

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Practice Question

X Y
2 3
4 7
6 5
8 10

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 Decide –X and Y

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Previous year question

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Goodness of Fit

Goodness of Fit refers to how well a model fits the observed data. It measures the discrepancy between the
observed data and the values predicted by the model. In simple terms, it helps us assess how well the chosen
model explains the variability in the data.

Why is Goodness of Fit Important?

When building a model, especially in machine learning or statistical analysis, it's crucial to know whether the
model you’ve created accurately represents the data or if it's underfitting or overfitting. A model with a good fit
will make accurate predictions on both training data and unseen data (generalization).

SST: The total variability in the dependent variable.

SSR: The variability explained by the model.
SSE: The remaining variability not explained by the
model.

Together, these terms help evaluate the performance of

a regression model and indicate how well it fits the data.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Coefficient of Correlation

X Y
1 1
2 1
3 2
4 2
5 4

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
ANOTHER FORMULA

Coefficient of Determination
Coefficient of Correlation

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 A correlation coefficient of 0.956 indicates a very strong
positive relationship between the two variables.

This means that 91.4% of the variability in the dependent variable 𝑌 can be explained by the independent variable 𝑋.
The remaining 8.6% is due to factors not captured by the model.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Coefficient of Determination

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Residual Analysis (also called errors)

A residual is the individual difference between the observed value and the predicted value for each data point
in a regression model. Residuals show how much each data point deviates from the regression line.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Standard Error of Estimate (SEE) (s) (σ ^2)

(s) Or

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Practice Questions

The SEE of 0.60 means that, on average, the actual values of the dependent variable Y (observed values) differ
from the predicted values 𝑌^ by about 0.60 units.

In other words, the typical prediction error or residual is 0.60 units away from the actual data points.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Test of Slope Coefficient

Formula for the t-statistic: estimated slope

standard error of the slope

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
You’re a marketing analyst for any Toys. You find β0 = –0.1, β1 = 0.7 and s = 0.6055
Is the relationship significant the .05 level of significance?

X Y
1 2
2 3
3 5
4 4
5 6

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Step 2: Calculate the Standard Error of the Slope

Given : s = 0.6055

t- calculated value

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Another Formula

t- calculated value

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

At the 0.05 significance level (two-tailed test), we look up the critical t-value for 3 degrees of freedom.
Using a t-distribution table, the critical value for 𝛼=0.05α=0.05 and 𝑑𝑓=3df=3 is approximately 3.182.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 The calculated t-statistic is 3.654.
The critical t-value is 3.182.

Decision: we reject the null hypothesis.

Since ∣3.654∣>3.182

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 This topic is part of optimization

Linear regression with gradient Descent

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
 EXPAND Function

Partial differentiation W.r.t

“C”

Putting old value of “c”

and “m ”

Gradient Value to determine first

iteration of “c” value i.e. new value

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 New “c” Value

Likewise :

Partial differentiation W.r.t “m”

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 Putting old value of “c” and “m ”

Gradient Value to determine first

iteration of “m” value i.e. new value

New “m” Value

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
In Batch Gradient Descent we were considering all the examples for every step of Gradient Descent. But what if our
dataset is very huge. Deep learning models crave for data. The more the data the more chances of a model to be good.
Suppose our dataset has 5 million examples, then just to take one step the model will have to calculate the gradients of
all the 5 million examples. This does not seem an efficient way.

It requires to calculate the gradients for the whole data set to perform just one update.

BGD can be very slow and is intractable for datasets that don't fit In memory ,it also doesn't allow us to update the
model online i.e BGD isn’t performed on data set that update continuously.

To tackle this problem we have Stochastic Gradient Descent.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

In Stochastic Gradient Descent (SGD), we consider just one example at a time to take a single step.
Stochastic gradient descent(SGD) performs a parameter update for each training example 𝑥(i) and label y(i).

Lecture Presentation | Ruchika Pharswan @ Delhi Technological University

While SGD’s fluctuation, on the one hand, enables it
to jump to new and potentially better local minima
On the other hand, this ultimately complicates
convergence to the exact minimum, as SGD will
keep overshooting
To tackle this problem we have Mini-Batch Gradient Descent.

Lecture Presentation | Ruchika Pharswan @ Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
LECTURE PRESENTATION | RUCHIKA PHARSWAN @ DELHI TECHNOLOGICAL UNIVERSITY
 What are Bias and Variance?
Bias
It can be defined as an inability of machine
learning algorithms such as Linear
Regression to capture the true relationship
between the training data points
.
High bias, also known as underfitting,
means the model did not learn enough
from the dataset.

LECTURE PRESENTATION | RUCHIKA PHARSWAN @ DELHI TECHNOLOGICAL UNIVERSITY

 What are Bias and Variance?
Variance

variance implies the error value that occurs when

we try to make predictions by using data that is not
previously seen by the model

High variance, also known as overfitting, means the

model focuses too much on specific patterns in the
training dataset and does not generalize well on
unseen data.

LECTURE PRESENTATION | RUCHIKA PHARSWAN @ DELHI TECHNOLOGICAL UNIVERSITY

Performance Evaluation : Confusion matrix
The Confusion Matrix is a tool used to evaluate the performance of classification models in Machine Learning
(ML). It is a square matrix that compares the predicted labels (output) with the actual labels (ground truth) of the
data, giving a comprehensive view of how well the model is performing.

For a binary classification problem, the confusion matrix is a 2x2 matrix. It consists of four components:

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

 True Positive
Interpretation: You predicted positive and it’s true (Actual -positive).

You predicted that a woman is pregnant and she actually is.

True Negative:

Interpretation: You predicted negative and it’s true ( Actual-nve).

You predicted that a man is not pregnant and he actually is not.

False Positive: (Type 1 Error)

Interpretation: You predicted positive and it’s false (actual –nve)

You predicted that a man is pregnant but he actually is not.

False Negative: (Type 2 Error)

Interpretation: You predicted negative and it’s false (actual –positive).

You predicted that a woman is not pregnant but she actually is.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Several important metrics can be calculated using the values from the confusion matrix:

The above equation can be explained by saying, from all the classes we have predicted as positive, how many
are actually positive.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
The above equation can be explained by saying, from all the positive
classes, how many we predicted correctly.

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Steps to calculate AUC

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Another formula for AUC Calculation

Lecture Presentation | Ruchika Pharswan @Delhi Technological University

Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
Lecture Presentation | Ruchika Pharswan @Delhi Technological University
All the Best, Prepare well

LECTURE PRESENTATION | RUCHIKA PHARSWAN @ DELHI TECHNOLOGICAL UNIVERSITY