THE

Dive into this subset of

machine learning and
discover the foundations,
techniques, architectures,

MIT
applications, and benefits
that deep learning can offer
your organization in this
8-week online course.

WAY
DEEP LEARNING:
MASTERING NEURAL
NETWORKS

Copyright © Massachusetts Institute of Technology 2022 1

 THE MIT WAY
Expand your skill set and help your organization make
data-informed predictions.
The AI revolution has brought about extensive
technological advances: conversational systems like Siri
or Alexa, driverless cars, and even automated traders that
compete with their human counterparts on Wall Street.
The subset of machine learning known as deep learning is
the force behind many of these incredible applications.

With the rapid rise of big data across a multitude of

industries, the ability to analyze it in order to detect
trends and make data-driven predictions has become
a necessity for organizations. And having both a solid
theoretical foundation and a practical understanding
of the basic building blocks of deep learning could be
a differentiating factor for professionals in nearly any
sector. This online course provides such expertise, backed
by the reputation of one of the world’s leading research
institutions in AI: MIT.

THIS COURSE IS AIMED AT

Professionals from
Tech enthusiasts who IT professionals looking nearly any industry
wish to learn more to expand their skill set (finance, logistics,
about deep learning with some of the most biotech, manufacturing,
and gain hands-on common and powerful and consulting, among
experience with the vast machine and deep others) that want to
array of techniques and learning algorithms. improve their data
applications it offers. management abilities
to generate valuable
insights.

A basic knowledge of programming languages is a

prerequisite for this program. The course will, however,
explain the necessary foundational concepts of machine
learning and programming, allowing those without previous
experience to fully participate.

Copyright © Massachusetts Institute of Technology 2022 2

FACULTY DE MIT

He is affiliated with the MIT Microsystems Technology

Laboratories, and serves as MTL Associate Director for
Duane S. Boning Computation and CAD. He is also the Engineering Faculty
Clarence J. LeBel Professor Co-Director of the MIT Leaders for Global Operations (LGO)
in Electrical Engineering, program, serving in that role since September 2016. From
and Professor of Electrical 2004 to 2011, he served as Associate Head of the EECS
Engineering and Computer Department at MIT. From 2011 through 2013 he was the
Science in the EECS Department Director/Faculty Lead of the MIT Skoltech Initiative, and
at MIT from 2011 through July 2018, he was the faculty Director of
the MIT/Masdar Institute Cooperative Program.

Copyright © Massachusetts Institute of Technology 2022 3

 Duration Commitment Format CEU

8 weeks 6-8 hours per Online 4.8

week

THROUGHOUT THIS COURSE, YOU WILL:

√ Explore the core mathematical and conceptual ideas

underlying deep neural networks.

√ Experiment with deep learning models and

algorithms using available machine learning toolkits.

Examine application approaches and case studies

√ where deep learning is being used throughout a
variety of industries.

Copyright © Massachusetts Institute of Technology 2022 4

 PROGRAM OUTLINE

MODULE 1 Foundations of Deep Learning

History of Machine Learning

Learning Objectives:
Overview of Key Machine Learning Concepts
• Selected Machine Learning Methods
1. Learn about the history of
• Model Complexity
deep learning, from the
development of single neural
Linear and Logistic Regression Algorithms
networks to deep neural
• Linear Classifiers
networks
• Regression Algorithms
2. Understand the basic building
• Iterative Algorithms
blocks of deep learning
• Revisiting Classification Logistic Regression
3. Use gradient descent to learn
how model coefficients and
Perceptrons and Neural Networks
weights are learned
• The Neuron
• Neural Network

Gradient Descent-Based Learning

• Necessary Steps in Achieving Gradient Descent-Based
Learning

MODULE 2 Basics of Python for Deep Learning

Introduction to Coding Tools

• Google Colab and Python
Learning Objectives:
Basics of Python
1. Learn about Python as a
• Variables, Control, and Functions
programming tool and its
different applications
Overview of Data Science-Related Python Libraries
2. Gain a better understanding of
• What Are Python Libraries?
the Python language through
• Popular Data Science-Related Python Libraries
examples and exercises
3. Learn about the fundamentals
Introduction to Machine Learning using Python
of machine learning models
• Machine Learning With Python Functions
through the basics of Python
• Tiny Implementations of Models and Methods
• Example of Regression and Classification Models
• Training Using Gradient Descent

Python Classes
• Defining and using Python Classes
• Things to Bear in Mind When Learning Python

Copyright © Massachusetts Institute of Technology 2022 5

 MODULE 3 Building and Training Neural Networks

Overfitting vs Generalization
• Regularization
Learning Objectives:
• Applying Regularization in Neural Networks
1. Use data-science Python
Training Curves
libraries to implement larger
• Confusion Matrix
neural networks
• ROC Curves
2. Learn how to define and train
a multilayer perceptron (MLP)
Dropout
neural network
• Generalization Improvement with Dropout
3. Identify the train/validate/
• Dropout Effect on Features
test split paradigm and model
• Dropout in PyTorch
evaluation metrics
• Train/Test/Validate Split
Data Encoding
• Train/Test/Validate Splitting in
• Scaling.
Code

MODULE 4 Deep Networks

Neural Networks: Multilayer Perceptrons (MLPs)

• Activation and Loss Functions
Learning Objectives:
• How Do We Train a Multilayer Perceptron
1. Gain a better understanding of
Autoencoders
complex neural networks and
• Autoencoder Example
how to train them
• Variational Autoencoder
2. Learn about autoencoding as
an architectural method to
Convolutional Neural Networks (CNNs)
customize neural networks
• Max Pooling Layers
and the use of convolutional
• Multiple Filters Across Multiple Channels
neural networks in image
recognition
Deep Neural Networks
3. Analyze basic concepts and
• Challenges of DNNs
challenges of deep neural
• Data Augmentation
networks
• Hyperparameters
• Types of Neural Network Layers in PyTorch

Copyright © Massachusetts Institute of Technology 2022 6

 Deep Neural Network Architectures: CNNs and
MODULE 5
RNNs 

Convolutional Neural Networks (CNNs)

• Transfer learning and fine-tuning, e.g. using VGG16,
Learning Objectives:
Resnet50 models
1. Learn how to adapt existing
Sequence to Sequence Mappings
deep convolution neural
• Word completion, language translation
networks to new applications
using transfer learning
Recurrent Neural Networks (RNNs)
2. Understand sequence to
• Basic recurrent element
sequence mapping as another
• LSTM and GRU
important machine learning
task
3. Gain intuition behind
recurrent neural network
(RNN) structures that learn
from sequence data rather
than from image data

Deep Neural Networks: GANs and

MODULE 6
Transformers

Generative Adversarial Networks (GANs)

• Generative models: Variational Autoencoder to
Learning objectives:
generate new examples
• GANs: Generator and Discriminator in competition
1. Learn how deep neural
networks can generate new
Transformers
examples that look “like”
• Attention networks
training samples
2. Understand generative
Large foundation models: GPT-3
adversarial networks (GANs)
and how they are trained
3. Gain exposure to attention
networks and transformers
as a recent advance in deep
learning

Copyright © Massachusetts Institute of Technology 2022 7

 Building NN-based applications with
MODULE 7
PyTorch II *

Three guided cases studies about Deep Neural Networks

Objetivos de aprendizaje:
Ethical AI
• Bias in data sets, training
1. Gain exposure to modern
• Trends in explainable models, trustworthy-AI
deep neural networks in
action, including CNNs for
image segmentation and
classification, RNNs for text
processing, and transformer
models for wide-ranging
applications
2. Understand the importance of
ethical practices in building
and using deep learning
models

MODULE 8 Capstone Project

• Two week long project

• The students will have to apply a particular deep learning algorithm (or set of algorithms) of their choice to a real
dataset to solve a certain challenge or business problem.
• They will elaborate a report with the results in the form of a presentation or blog post.
• They will be required to develop their project in Google Colab.
• Finally, they will be asked to upload a link to their notebook and their choice of report (uploaded in .pdf) so that it
can be reviewed and corrected by their peers.
• The plan is to use PyTorch with torchvision and torchaudio datasets and pre-trained models.

Copyright © Massachusetts Institute of Technology 2022 8

MIT xPRO AND GLOBAL ALUMNI:
Continuing programs for professionals around the world

About MIT and MIT xPRO Collaboration with Global Alumni

The Massachusetts Insitute of Technology, more Our collaboration with Global Alumni has driven the
commonly referred to as MIT, is a private American launch of MIT xPRO in Portuguese, Spanish, and English.
university founded in 1861. The Institute’s faculty has This EdTech is dedicated to expanding access to the
included 97 Nobel Prize laureates, 58 National Medal knowledge of the world’s top universities, helping break
of Science winners, 58 Rhodes Scholars, and 45 down both geographical and language barriers.
MacArthur Fellows.
Their focus on innovation, technology, and user
MIT xPRO offers online continuing education courses experience, along with their expensive understanding
for professionals at all levels from the fields of science, of our goals as an institution, ensure that quality of
engineering, and technology. All courses are led and our teaching is upheld as we expand our offering to
supervised by MIT faculty. professionals around the globe.

This collaboration offers these professionals the

opportunity to take advantage of the same quality
teaching found on MIT’s campus wherever they are and
in their own language.

USA +1 617 778 2938 | Spain +34 679 492 128

[email protected]

More Information
Copyright © Massachusetts Institute of Technology 2022 9