December 3, 2020

3 Must-Take Deep Learning Courses

best deep learning courses in 2020

Will AI replace your job? Probably so.

75 million jobs will be replaced by AI and automation by 2022 according to PwC.

But should you be worried? Not necessary. 133 million more jobs will emerge to offset that replacement and create 58 million additional jobs, all because of the use of AI.

Change is inevitable and change is the only constant. But you can ride on this wave of changes by learning AI now. Here are 3 must-take deep learning courses to get started in AI.

Disclosure: Some of the links below are affiliate links, meaning, at no additional cost to you, Self Learn Data Science will earn a commission if you click through and make a purchase.

Deep Learning Courses

1. Deep Learning Specialization []

Comprehensive contents
Real life programming assignments
Uses tensorflow, a popular deep learning framework
Lacks mentorship

Reviews: Another excellent course by Andrew Ng, a well-known figure in the AI space. The Deep Learning Specialization consists of a series of 5 courses, namely;

  1. Neural Networks and Deep Learning
  2. Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
  3. Structuring Machine Learning Projects
  4. Convolutional Neural Networks
  5. Sequence Models

This is the most complete deep learning MOOC that brings learners through the whole AI process. You will be exposed to deep learning fundamentals, its applications, as well as elements of managing a AI project. There is a good mixture of theories and practical applications to keep you engaged and practicing what you learned.

Furthermore, the use of tensorflow in this course exposed learners to one of the most popular deep learning framework. Learning tensorflow would never go wrong as it is one of the mostly used deep learning framework in the industries. Programming assignments were also on real life applications such as car detection and style transfer, allowing learners to work on real problems. Being a intermediate course, prior python experience is needed before you can work on these assignments.

The Deep Learning Specialization is highly recommended for beginners and practitioners alike. No matter if you have zero exposure to deep learning or already building some AI application, this specialization will benefits you and get you ready to work on real AI projects.

2. Practical Deep Learning for Coders v3 []

Takes on a practical approach to learn deep learning
Introduces best practices in industry
Lacks structured programming assignment
Uses fastai library

Reviews: Jeremy Howard is an excellent tutor and the ‘Practical Deep Learning for Coder’ is already on its third iteration. is known for their method of teaching data science concepts as contrary to popular approaches, Jeremy uses a code-first principle. This means that you will be building a deep learning model in the very first lesson! Jeremy believes that having hands-on before any theories help students learn and understand concepts faster. This might not be true for everyone but it definitely brought a huge crowd to his course.

Apart from this unique teaching principle, Jeremy’s sharing of industries best practices and recent researches also make the course stood out. AI research is very active and it is almost impossible to get yourself updated with new advancements in AI. Hence, Jeremy filled a very crucial role in updating his students on the most recent breakthrough in AI techniques or algorithms.

Although this course is not as comprehensive as Andrew Ng’s Deep Learning Specialization, it still got a spot in this list due to the algorithms that Jeremy taught. Jeremy focuses on state-of-the-art algorithms for each category of AI and teaches with the goal of creating breakthrough models. Many of his students actually use his techniques and made new records in leaderboards and made new applications.

The only drawback is the use of fastai library in this course. Admittedly, fastai is the most optimized deep learning framework I had seen due to its built-in parameters. However, it still has not gained enough popularity and is not commonly used in the industry.

Another great course not to be missed. Beginners and practitioners alike will appreciate this content with nothing to lose. A good tip is to take Andrew Ng’s Deep Learning Specialization first before this if you are not used to the code-first approach like me.

3. Deep Learning A-Z: Hands-on Artificial Neural Networks [Udemy]

Highly organised contents
Step-by-step coding
Uses both keras and pytorch framework
Lacks proper assignment process

Reviews: A course that contains almost too much. 23 hours of videos carefully organized into 188 lectures. This course is by Kirill Eremenko and Hadelin de Ponteves, both common instructors in their A-Z data science series in Udemy. Kirill and Hadelin did a wonderful job at introducing students to the world of deep learning and AI as Kirill brings students through the tedious theories while Hadelin walks us through the step-by-step coding. Both complement each other roles and gave us this amazing course that is as comprehensive as it can get.

Some of the concepts introduced includes;

  1. Artificial Neural Networks
  2. Convolutional Neural Networks
  3. Recurrent Neural Networks
  4. Self Organizing Maps
  5. Boltzmann Machines
  6. AutoEncoders
  7. Machine Learning Basics

Each of these consists of in-depth theories, codings, model evaluation and optimization, and some homework in-between to drive understanding. Homework is in jupyter notebooks which is another great implementation as it is also one of the most common IDEs used by data scientists.

Furthermore, it is almost unheard of that a course teaches more than one deep learning frameworks. Hadelin codes in both tensorflow/keras and pytorch throughout the course, exposing students to multiple framework at once.

However, the Udemy platform does not have any integrated grading system such as those in edX or Coursera. Hence, homework or assignments are solely based on students’ motivation to do.

Do you need to take all three deep learning courses before working on AI projects? Not really. But I highly recommend you to do so. Taking multiple courses exposed you to different perspectives of each AI instructors and their own best practices. This helps you build your own practice and implement the best habits in an AI project. If this post overwhelm you and you need to start lower, go over to ‘15 Best Courses to learn Data Science in 2020‘ and work on your fundamentals.

Are you going to replace or be replaced? You decide.