MEPB321: SUMMER TRAINING

Submitted To:
Dr. Abhishek Mishra Sir

Submitted By:
Name: GIRHEPUJE SAHIL RAHUL
Roll No: 221310023
Branch: ME
Semester: 5th Sem

Department of Mechanical Engineering

NATIONAL INSTITUTE OF TECHNOLOGY

DELHI

2024
Index

1. Abstract
2. Preface
3. Introduction
4. Course Objectives
5. Course Structure and Content
o 6.1 Introduction to Machine Learning
o 6.2 Supervised Learning
 6.2.1 Linear Regression
 6.2.2 Logistic Regression and Regularization
o 6.3 Neural Networks and Deep Learning
o 6.4 Unsupervised Learning
 6.4.1 K-Means Clustering
 6.4.2 Principal Component Analysis (PCA)
o 6.5 Support Vector Machines (SVM)
6. Tools and Technologies Used
7. Challenges Faced
o 8.1 Understanding Mathematical Concepts
o 8.2 Implementing Algorithms from Scratch
o 8.3 Hyperparameter Tuning
8. Key Learning Outcomes
9. Impact on Career Goals
10. Conclusion
11. Appendices
 Abstract

This report presents the details of my summer training in the Machine

Learning course by Andrew Ng, offered through Coursera. The training
provided a comprehensive understanding of fundamental machine learning
concepts, including supervised and unsupervised learning, neural networks, and
support vector machines. Through hands-on projects, I developed practical
skills in implementing and optimizing machine learning algorithms using
Python and MATLAB/Octave.

The training emphasized the importance of data preprocessing, model selection,

and performance evaluation, enabling me to apply machine learning techniques
to real-world problems such as house price prediction, email spam detection,
and customer segmentation. Challenges faced during the training, such as
understanding complex mathematical concepts and tuning hyperparameters,
were overcome through additional learning resources and practical
implementation.
This course has significantly contributed to my knowledge and skills in
Artificial Intelligence and Machine Learning, aligning with my academic and
professional aspirations. The report outlines the course objectives, structure,
projects undertaken, challenges faced, and the learning outcomes achieved
during the training.
 Preface

The field of Machine Learning (ML) is transforming industries and redefining

the future of technology. As a student of Mechanical Engineering with a minor
in Artificial Intelligence and Machine Learning at NIT Delhi, I have always
been fascinated by the potential of intelligent systems to solve complex
problems across diverse domains.
This report is a culmination of my summer training experience in the Machine
Learning course by Andrew Ng on Coursera, which provided me with the
opportunity to delve into the world of data-driven solutions and intelligent
algorithms. The training offered a blend of theoretical knowledge and practical
application, allowing me to understand the fundamentals of machine learning
and their implementation in real-world scenarios.
Through this report, I aim to share my learning journey, including the
challenges I faced, the solutions I implemented, and the insights I gained during
the course. I believe that this experience has equipped me with valuable skills
that will not only enhance my academic pursuits but also prepare me for future
professional opportunities in the field of Artificial Intelligence and Machine
Learning.
I am grateful for the guidance and support of my mentors, peers, and online
communities, who provided invaluable insights and encouragement throughout
this training. I also extend my gratitude to Coursera and Andrew Ng for offering
such a well-structured and insightful course.
 Summer Training Report
Course Title: Machine Learning
Instructor: Andrew Ng
Platform: Coursera
Duration: 8 weeks, e.g., June - August 2024
Institution: National Institute of Technology (NIT) Delhi
Submitted by: Girhepuje Sahil Rahul

1. Introduction

Machine Learning (ML) has emerged as a key component of modern

technology, revolutionizing industries such as healthcare, finance,
manufacturing, and transportation. As part of my summer training, I
completed the Machine Learning course by Andrew Ng on Coursera,
a globally recognized course designed to provide in-depth knowledge
of machine learning fundamentals. The training bridged the gap
between theoretical concepts and their practical applications,
equipping me with the necessary skills to pursue further studies and
projects in Artificial Intelligence (AI) and Machine Learning.
 2. Course Objectives

The course aimed to:

 Introduce the fundamental concepts and techniques of machine
learning.
 Develop hands-on skills for implementing and deploying ML
models.
 Foster an understanding of how machine learning can solve real-
world problems.
 Provide a foundation for advanced topics like deep learning and
reinforcement learning.
 3. Course Structure and Content

The course was structured into multiple modules, each focusing on a

specific area of machine learning. Below is a detailed breakdown:

Module 1: Introduction to Machine Learning

 Definition and importance of machine learning in today’s world.
 Distinction between supervised, unsupervised, and
reinforcement learning.
 Overview of commonly used ML algorithms.
Key Takeaway: Machine learning is a versatile tool with applications
across various domains, making it an essential skill for engineers.

Module 2: Supervised Learning

Linear Regression with One Variable
 Hypothesis function and cost function.
 Gradient descent algorithm for minimizing cost.
Multivariate Linear Regression
 Handling multiple features.
 Feature scaling and polynomial regression.
Project: House Price Prediction
 Built a regression model to predict house prices based on square
footage, number of rooms, and location.

Module 3: Logistic Regression and Regularization

 Binary classification and logistic regression hypothesis.
 Decision boundary and cost function for classification problems.
 Overfitting and regularization using L1 and L2 norms.
Project: Email Spam Detection
 Developed a logistic regression model to classify emails as
spam or non-spam based on word frequencies.

Module 4: Neural Networks and Deep Learning

  Architecture of artificial neural networks.
 Forward and backward propagation.
 Gradient descent optimization in deep networks.
Project: Handwritten Digit Recognition
 Designed and trained a neural network to recognize handwritten
digits from the MNIST dataset, achieving high accuracy.

Module 5: Unsupervised Learning

K-Means Clustering
 Partitioned data into clusters based on feature similarity.
Principal Component Analysis (PCA)
 Reduced dimensionality of high-dimensional datasets.
 Visualized high-dimensional data in 2D and 3D.
Project: Customer Segmentation
 Used K-means clustering to segment customers based on
purchasing behavior.

Module 6: Support Vector Machines (SVM)

 Linear and non-linear SVMs for classification.
 Kernel functions to map data into higher-dimensional space.
Application: Image Classification
 Applied SVMs to classify images of objects into different
categories.

4. Tools and Technologies Used

 Programming Languages: Python, Octave/MATLAB.
 Libraries:
o NumPy: For numerical computations.
o Pandas: For data manipulation and analysis.
o Matplotlib & Seaborn: For data visualization.
o scikit-learn: For implementing machine learning
algorithms.
 Software Platforms: Jupyter Notebooks, Google Colab.

5. Challenges Faced

1. Understanding Mathematical Concepts:

Initially, I found it challenging to grasp some mathematical
concepts, such as gradient descent and backpropagation.
Solution: I referred to additional resources, such as research
papers, textbooks, and video lectures, to strengthen my
understanding.
2. Implementing Algorithms from Scratch:
Implementing algorithms like logistic regression and neural
networks from scratch required a deep understanding of the
underlying principles.
Solution: I practiced coding exercises and participated in online
forums to seek guidance and clarification.
3. Hyperparameter Tuning:
Choosing the right hyperparameters (e.g., learning rate,
regularization term) was difficult, as it significantly impacted
model performance.
Solution: I learned techniques such as cross-validation and grid
search to optimize hyperparameters effectively.
6. Key Learning Outcomes
 1. Theoretical Knowledge: Gained a solid foundation in the core
concepts of machine learning, including supervised and
unsupervised learning, neural networks, and support vector
machines.
2. Practical Skills: Developed proficiency in implementing
machine learning algorithms using Python and MATLAB.
3. Problem-Solving Ability: Learned to apply machine learning
techniques to solve real-world problems, such as price
prediction, classification, and clustering.
4. Data Analysis and Visualization: Enhanced my ability to
analyze and visualize data, making it easier to interpret model
performance and draw insights.

7. Impact on Career Goals

This course has significantly contributed to my career aspirations in

the field of Artificial Intelligence and Machine Learning. It has:
 Provided me with the skills necessary to work on AI-related
projects in my academic and professional life.
 Inspired me to explore advanced topics such as deep learning,
reinforcement learning, and natural language processing.
 Boosted my confidence to participate in hackathons, internships,
and research projects focused on machine learning.

8. Conclusion
The Machine Learning course by Andrew Ng on Coursera was a
transformative learning experience. It not only enhanced my technical
skills but also provided me with a new perspective on how data-
driven solutions can address complex challenges. I am confident that
the knowledge and skills gained from this training will play a crucial
role in my journey as a future engineer and AI enthusiast.

Appendices
Appendix A: Course Certificate

Appendix B: Course Syllabus

A detailed outline of the course modules and topics covered during

the training:

1. Week 1: Introduction to Machine Learning

2. Week 2: Linear Regression with One Variable
3. Week 3: Linear Regression with Multiple Variables
4. Week 4: Logistic Regression
5. Week 5: Regularization
6. Week 6: Neural Networks Representation
7. Week 7: Neural Networks Learning
8. Week 8: Support Vector Machines
9. Week 9: Unsupervised Learning (K-means and PCA)
10. Week 10: Anomaly Detection and Recommender Systems
11. Week 11: Large Scale Machine Learning
12. Week 12: Application Example and Course Conclusion

Appendix C: Python Code Samples

Appendix D: Project Results and Graphs

1. House Price Prediction (Scatter Plot of Predicted vs. Actual

Prices).
2. Logistic Regression Classification Boundary for Spam
Detection.
3. Neural Network Accuracy vs. Epochs for Handwritten Digit
Recognition.