Antern ML002

Ayush Singh

October 16, 2021

Introduction

Deep learning is one of the most emerging domains in this era, Learning Deep Learning
is totally worth it! We present ML002 a full deep learning course from scratch covering
mathematics basics to some level in deep learning. We saw many students suffering for
not getting the right resources to learn for free.

We already presented a full-fledged machine learning course ML001 by Antern for

absolutely free, covering in-depth machine learning for beginners. Now Antern Presents
ML002, a full course on Deep Learning.

We also identified some problems that free courses have and feedback from our
previous courses and tried to improve our content based on that. So, Let’s get started!

Why to Learn Deep Learning?

Whether you’re an aspirant of becoming a data scientist or machine learning engineer,

nowadays having a good knowledge of deep learning is a must. It sets you very well and
gives you a new way of thinking about “how machines learn?”.

It also sets you a foundation of learning from images and texts, and In this era for data
scientists and machine learning engineers or AI Engineers, Deep Learning should be in
their skillset.

Why an ML002 Course from Antern ?

Many of the courses online which are free are good but the problem is students are not
able to grasp them, maybe because they are not good at the required mathematics! So
this course ML002 starts from scratch to some level in Deep Learning. We also saw
students have to suffer for finding the right End 2 End projects and Strategy for working
on a project and some general misconception that the students have!
 Objectives

From ML002 we aim to serve students with real world deep learning skills, so after
completing this course they have a decent understanding of deep learning theoretically
and practically.

● Introduce students with the building blocks of mathematics which are required
for deep learning/AI.
● Make the base strong for the students for deep learning.
● Prepare students to work on challenging and complicated projects.
● Prepare students to clear Interviews.
● Make Students familiar with Image and Text processing
● Make students understand the mathematics very easily

Course Syllabus

You can see the complete syllabus by going to this document here. In this syllabus you
will find the topics which we will cover in this ML002, and you may think, it is similar
like other courses, FYI The topic which we will teach covers everything about the topic
with reading the research paper as well! So ML002 is very different!

Module 1: Getting Ready for Deep Learning

Motivation :- Why to learn Deep Learning?

Chapter 1:- Linear Algebra

● Basic properties of a matrix and vectors: scalar multiplication, linear
● transformation, transpose, conjugate, rank, and determinant.
● Inner and outer products, matrix multiplication rule and various algorithms, and
matrix inverse.
● Matrix factorization concept/LU decomposition, Gaussian/Gauss-Jordan
elimination, solving Ax=b linear system of an equation.
● Eigenvalues, eigenvectors, diagonalization, and singular value decomposition.
● Special matrices: square matrix, identity matrix, triangular matrix, the idea about
sparse and dense matrix, unit vectors, symmetric matrix, Hermitian,
skew-Hermitian and unitary matrices.
● Vector space, basis, span, orthogonality, orthonormality, and linear least square.
Chapter 2:- Calculus
● Limits: Introduction, Properties of Limits, Solving Limits, L-Hopital Rule
● Continuity: Introduction, Solving problems, Discontinuities
● Differentiability: Introduction, How does it work? Formal Definition, Mean Value
Theorem, Minima and Maxima, Gradient Descent, derivative, partial derivative,
Jacobian and Hessian Matrices
● Integration : Introduction, Integration by Substitution, Integration by partial
functions, properties of definite integral
● Optimization: Convex Optimization, first and second order optimization
algorithm, Constrained Optimization, Karush–Kuhn–Tucker Approach,
generalized Lagrangian

Chapter 3:- Probability and Information Theory

● Probability:
○ Bayesian Probability
○ Random Variables,
○ Discrete Variables and Probability Mass Function,
○ Continuous Variables and Probability Density Functions
○ ,Marginal Probability
○ ,Conditional Probability,
○ Expectation,Variance and Covariance,
○ Bayes Theorem
○ Probability Distributions,
■ Bernoulli Distribution,
■ Multinoulli Distribution,
■ Gaussian Distribution,
■ Exponential and Laplace Distributions, The Dirac Distribution and
Empirical Distribution,
■ Properties
● Information Theory

Chapter 4:- Machine Learning basics

● Learning Algorithms : Linear Regression, Logistic Regression
● Bias and Variance
● Hyperparameters and Validation Sets
● Bayesian Statistics
● Supervised Learning Algorithms
● Unsupervised Learning
● Challenges Motivating Deep Learning
 Module 2: Deep Learning Fundamentals

Chapter 5:- Deep Learning Fundamentals Part - 1

● What is Deep Learning?

○ Evolution of deep learning
○ Representation learning
○ History of deep learning
○ Formal Definition of deep learning
● An Overview of Neuron of a brain
● Perceptron and How It relates to Neurons?
● Logistic Regression as Neural Network
● Multi-layer Perceptron
● Notations

Chapter 6:- Deep Learning Fundamentals Part - 2

● Perceptron Training
● Multi-layer Perceptron Training
● BackPropagation Training
● Activation Functions and Derivation
 ● Building a simple neural network from scratch In Python
● Building a simple neural network Using Tensorflow and PyTorch
● Designing a Neural Network Using Keras-Tuner

Module 3: Regularization & Optimization

Chapter 7:- Regularization

● Parameter Norm Penalties

● Norm Penalties as Constrained Optimization
● Noise Robustness
● Parameter Tying and Parameter Sharing
● Multi-Task learning
● Semi-Supervised Learning
● Bagging & Boosting
● Dataset Augmentation
● Early Stopping
● Dropout
● Implementation in PyTorch/Tensorflow

Chapter 8:- Optimization

● How does learning differ from pure optimization?
● Challenges
● Batch Gradient Descent
● Stochastic Gradient Descent
● Adam
● RMSProp
● Batch Normalization
● Weights Initialization techniques

Module 4: Convolution Neural Networks

Chapter 9:- Convolutional Neural Nets ( CNN )

● Why do we need CNN over ANN?

● CNN
○ Filters, Feature Maps
○ Convolutions
 ○ Strides
○ Pooling
○ Max-Pooling
○ Average Pooling
○ Padding with different techniques
○ Getting to know the dimensions
● CNN Architectures
○ AlexNet
○ VGG-16
○ VGG-19
○ ResNet
● Designing CNN
● Transfer Learning
● Object Detection
○ Basic object detection model
○ YOLO ( You Only Look Once )
○ R-CNN
○ Fast R-CNN

Chapter 10:- Building Application using ConvNets and Deploying it

● Building cat and dog classifier

● Building Pneumonia Detection using Transfer Learning
● Building Multi-class classifier
● Building Object detector using Coco dataset
● Approaching and Completing a Kaggle Contest

Module 5: Recurrent Neural Network

Chapter 11:- Recurrent Neural Networks

● RNN Applications
● RNN Overview
● Backpropagation In RNN
● LSTM ( Long Short Term Memory )
● Bi-Directional RNN’s
 ● Gated Recurrent Units
● Problems with RNN’s
● Seq2Seq Architectures
● Transformers

Chapter 12:- Building Applications using RNN

● Basic RNN Implementation

● Sentiment Analysis using RNN
● Fake News Detection
● Speech recognition
● Approaching and Completing a Kaggle Contest

Calendar Or Lecture Schedules

● This course will start from 22 October 2021 and a Total of 3-4 Lectures in a week!
● Each Lecture will be of duration 60 - 90 minutes according to the topic!
● If the lecture is delayed, the lecture will be combined with another lecture making
180 minutes of video!

Who’s this course for?

ML002 is for everyone whether someone is student or working professional,

Prerequisite for this course is as follows:-
● Python ( required )
● Machine Learning Fundamentals

Before you start with this course, we ask you to attempt a test before starting with this
course and successfully completing the course!

Lead Instructor : Ayush Singh

Ayush Singh is a Machine Learning Engineer & Data Scientist, He has worked with Data
Scientist Intern @ Artifact and also worked as ML Engineer @ Omdena. He has 2+ YOE
in this field of working in this AI field. He also taught more than 1000s of students,
Ayush Taught ML001 where he helped thousands of students to learn the fundamentals
and he also helped lots of peoples in getting their job through guidance and teaching.
He has also worked in large scale machine learning applications and He founded
Newera where we present End2End courses and also he founded Antern where we
present our live programs. He is currently building MLOps Centric Platform by building
AI Products.

His Machine Learning course is recommended by MIT CS and AI Labs via Twitter and
He is regarded as Teenage Tech RockStar and India’s youngest working data scientist.

Linkedin:- https://www.linkedin.com/in/ayush-singh488/
Twitter:- https://twitter.com/AyushSi69507942
Instagram:- https://www.instagram.com/ai.ayush_singh/

Course Components

1. Lecture ( 3 lectures/week ): At the end of every lecture, you will be given a quiz to
attend in your lms to mark whether you were present or not! The quiz will be
based on lecture and some feedback. We are taking this because we need to
ensure that you’re learning from this course! This will be known as attendance for
this course! Lecture Notes, Slides will be available to you through the resources
tab in your lms as well in the youtube description box.

2. Practical Labs ( 1 lecture/week ) : In the lectures after we have studied the theory,
In the labs we will do the practical implementation of whatever we have studied,
this also marks your attendance, your attendance will be marked after you view
the video and attempt a quiz which will be given at the end of every lecture/labs.

3. Homework Assignment ( 1 HW/Week ) : Every week there will be 1 homework

assignment released, the homework will take around 4-5 hours to complete with
honesty. Homework assignments will be based on lectures and labs. Homework
assignments will have 40-50 questions including subjective and objective type
questions. You need to submit the homework assignment through your lms, all
these details about submission will be given below.

Grades

This course is pass/fail and 60% grade is passing this course and getting your Certificate
of completion letter of this course! If you complete all your homeworks and watch all the
lectures genuinely, you will most likely pass this course.
We take plagiarism very seriously, as our TA’s will manually evaluate your homeworks
every sunday! If we find plagiarism, we will contact you and give you first warning and if
it happens a second time we will remove and ban you from further submitting the
homeworks and you will most likely not get the completion letter from Antern!

If you fail the course, most likely you will be given a second chance to start with this
course!

Components of your final grade

Homework 50%
● Submitted via lms
● Student can submit the homework
assignment at least 2 times per
assignment
● We will drop the students who are
not active every 2 weeks.

Attendance 20%
● We want to ensure you’re watching
the lecture and understanding
everything , so we have an
attendance policy.
● At the middle/end of each lecture
you will be given some quizzes
based on the lecture content.

Final Project 30%

● You will be asked to build a final
project, more details about the
final project and homeworks given
above!

How will we Grade?

● We are AI Geeks, your Problem Sheets will be downloaded and sent to the AI
system and each problem sheet will be similarity checked with our correct answer
and according to the similarity of the documents you will be given marks.
● We will also shuffle the similarity checks with student - student and and here
your similarity check should be low otherwise your solution will be regarded as
copied.
 HW & Final Project

Homework Assignment:-
● HW Assignment will be released at the end of every week which is after the lab
through your lms.
● HW Assignment Solution discussion will be happening once a month which will
be live!
● Homeworks also contains some interview questions and some practical stuff for
you to do!
● HW will also contain some ungraded question and experiment questions.

Final Project:-
● After you complete the course, you will be asked to make a final project. The
project which you will build should be the End2End Project, maybe you can do it
in a group or alone!
● After you complete developing your project you have to write one blog or paper,
showing your approach to how you handled this problem. Below are the more
details on this.

In the proposal, below your project title, include the project category. The category can
be one of:

● Computer Vision
● Natural Language Processing
● Generative Modeling
● Speech Recognition
● Reinforcement Learning
● Healthcare
● Others (Please specify!)

Your project proposal should include the following information:

● What is the problem that you will be investigating? Why is it interesting?

● What are the challenges of this project?
● What dataset are you using? How do you plan to collect it?
● What method or algorithm are you proposing? If there are existing
implementations, will you use them and how? How do you plan to improve or
modify such implementations?
 ● What reading will you examine to provide context and background? If relevant,
what papers do you refer to?
● How will you evaluate your results?

Please Note : Before Starting with the project proposal, be sure to confirm with TA’s
and Instructors about the project which you will work on!

If you’re willing to write a paper or in docs or a blog, you should follow the below
format:-

You have to submit your final project through the form which will be given through your
LMS.

Group Work & Plagiarism Policy

We encourage students to work to learn together and do their work together, however
we don’t encourage students to share their assignments and answers, but we want you to
be as collaborative as you can, please engage with the course online tools like discord to
ask your question and answer peer questions.

Course Technology
For this course, we will be using several technology for smooth learning experience
throughout this course:-

Youtube
We will be using Youtube for streaming of your course lectures and labs, all the video
lectures/labs will be uploaded to the Newera Youtube channel. In the playlist called
ML002, The frequency of the lectures are already told to you!

NEO LMS

We will be using NEO Lms for assignment and attendance policy, so below are the steps
for successfully getting in the learning management system

Following are the steps for your NEO LMS

● Go to https://antern.neolms.com/
● Click on Signup with Access code, and if you’re a student or learner write your
access code to be iamstudent
 ● Click on Sign up! And after that fill in all your details.
● And Click Submit! And Voila You’re in!
● But one more step, Click on classes tab at the left size and click on Enroll and add
your access code to be HDST-TKGM

And you’re set up with your learning, soon more assignments and other tabs will be
shown up!
Important thing to Note here:- We can make only 400 students to enroll because we
don’t have budget for buying the subscription, if you can help this series in buying a
subscription for lms, it would be really helpful.

Discord Server
We will use a discord server to provide you with doubt support and help you if you are
stuck anywhere in the course! There is a category named ML002 and there several
channels are created! Please ask your questions in specific channels for the homework
sets and most importantly have fun.

You can join the discord server:- https://discord.gg/ZmKCjmYnAP

Student Resources

Homework Discussion Session

Each month we will discuss homework through live sessions on youtube and sometimes
1-1 on zoom. This will help students to better approach the problem, we will select some
specific problems to discuss.

Doubt Support

We highly encourage students to ask doubt, no doubts are stupid! If you ask doubts and
learn from our courses it will motivate us to make more amazing content! You can ask
any doubt in our discord server in a specific channel, we will get on live for clearing your
doubt in live for free.

Contact Us!

We have almost listed everything over here, and if you still have any questions, please
ask in youtube comments or contact us!
The easiest way to contact us is via opening a ticket by filling out this form.

Certificate Information
Certificate of completion letter will be only given if you pass this course covering all the
requirements which is already listed in your Grades section above! If you didn’t pass the
course, you can redo this course!

Meet our Team