Deep Learning Handbook

The Deep Learning Handbook is a project in progress to help study the Deep Learning book by Goodfellow et al.

Goodfellow’s masterpiece is a vibrant and precious resource to introduce the booming topic of deep learning. However, many found the accompanying video lectures, slides, and exercises not pedagogic enough for a fresh starter.

I used the ‘dlbook’ as the primary reference for the machine learning class that I have taught in Spring 2019/2020 at the computer science department, American University of Beirut.

I would like to share my experience by publishing the slides and some of the assignments on this page. The project may be developed further into a full handbook and guide accompanying the fascinating text of Goodfellow et al.

The target audience comprises:

• undergraduate and graduate students willing to know more about machine and deep learning, be able to read research papers or start a graduation project or a master thesis in the same context,
• developers and practitioners aspiring to a bit more math and philosophy,
• or mathematicians liking to have some hands-on and a bit more coding experience,
• any other bored or sleepless person.

Currently, only part I and part II are covered.

Slides

Part I

• Introduction

• Linear Algebra

• Probability

• Numerical Analysis

• Machine Learning Basics

Part II

• Multi-Layer Perceptrons

• Regularization

• Optimization

• Convolutional Neural Networks

• Recurrent Neural Networks

Exercises

Assignment 1 - Linear Algebra

Linear Algebra Exercises

Assignment 2 - Probability

Probability Exercises

Assignment 3 - Numerical Computations

Numerical Computations Exercises

Notebook

Assignment 4 - Machine Learning Basics

Machine Learning Basics Exercises

Notebook

Assignment 5 - Multi-layer Perceptrons

Multi-layer perceptrons

Notebook

Assignment 6: Regularization

Regularization

Notebook

Assignment 7: Optimization

Coming soon

Assignment 8: Convolutional Neural Networks (CNN)

Coming soon

Assignment 9: Recurrent Neural Networks (RNN)

Coming soon

Multi-choice questions: Test your deep learning knowledge

Test Your Skills

Competition

In the context of the Spring 2020 class offering, we have organized an in-class Kaggle competition under the theme: Android Malware Classification.

You can still have a late submission

Winners solutions:

1. Houdeib & Haidar
2. Tachanka

Anonymous Student testimony

I think this is one of the best graduate courses I’ve taken during my Master’s degree. The special thing about this course is the parallelism between the theoretical parts and the practical assignments that we were solving. So, it developed both my theoretical and practical skills. Also, making the project as a Kaggle competition stimulated me to work more on it and to compete with other teams in my class.

Contributions

If you would like to contribute to this guide/handbook, please contact the author at mn115 at aub.edu.lb

Citation

To cite this handbook, please use this bibtex entry:

@book{nassar-dlh-2020,
    title={Deep Learning Handbook},
    author={Mohamed Nassar},
    publisher={Zenodo},
    doi={10.5281/zenodo.3881558},
    note={\url{http://mnassar.github.io/deeplearninghandbook}},
    year={2020}
}

Author

scholar researchgate