Deep Learning Roadmap – Part II of the NLP Roadmap


In this second part of our three part series, we will see a roadmap to go from Machine Learning to Deep Learning, as we continue our journey Towards NLP.

The tortuous Path to the Force

Hello hello hello! Hopefully you enjoyed the first part of our Complete Natural Language Processing Roadmap! In part one, we saw a complete Data Science Roadmap. Congratulations! You took your first steps on the road to the force, and you are now so much closer to being an NLP Jedi Master! Now, we’ll dive into a Deep Learning Roadmap. I must admit, it was a bit harder to create this one. Although I had so many great ideas for courses and resources you could check out, I didn’t know how much detail I should go into the actual roadmap. I have to give credit where credit is due, and thank these two articles for the inspo:

I tried to condense all the information I could find and think of in the most organic way. I hope you find this article useful! It is going to be a little shorter than the previous one in the series, since I find it a bit harder to find courses on specific topics. Deep Learning courses, at least the ones I tried, are all well-rounded and encompass a large variety of sub-topics. So if you have any more resources to suggest, please feel free to share in the comments! I am always looking for ways to improve my Deep Learning roadmap.

The Deep Learning Roadmap

I divided the roadmap into three sections (you can see them from left to right), but you would actually advance in parallel for most of the topics. Once you get familiar with Neural Networks, you’ll start to experiment with different architectures, and you’ll learn all about the different training tools and techniques. As I said, most courses on Deep Learning are pretty well-rounded. I thought I would list them right away so that you can get an accurate picture of the resources available.

  1. Practical Deep Learning For Coders
  2. Deep Learning Specialization on Coursera
  3. Deep Learning Book by Ian Goodfellow
  4. NYU Deep Learning
  5. Building Advanced Deep Learning and NLP Projects


Where to Start: Practical Deep Learning For Coders

I recommend starting with the course “Practical Deep Learning for Coders”. I found this course as the “prerequisites” for a mini HuggingFace course. I personally decided to try it out because (for once!) the documentation was pretty clear about how to proceed: follow the course, follow the mini HuggingFace course about the Transformers library, and then, move on to the Coursera NLP specialization. Usually, when blogs, articles and other platforms recommend several courses, they do not provide an exact order, which can be pretty stressful. Especially if, like me, you like to follow a roadmap 😉 . 

I must admit I was not disappointed. The course is not only information-packed, but it also provides great tips on how to improve. The instructor will bring to your attention not-so-small details (e.g., why should we choose one framework over the other, what common misconceptions we should be aware of, and much more). This course will cover the basics through a hands-on approach. It is definitively my favorite course so far, so give it a try!

Deep Learning Specialization on Coursera

The Deep Learning Specialization on Coursera will go a bit further in terms of explanations. You get five courses neatly packed in one package, and you will be able to test out the theory you learn about in each lesson. The first course about Deep Learning and Neural Network is a perfect follow-up from the course. The second course, Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization, is what really makes this course a valuable resource. It covers important topics such as Dropout and BatchNorm. These are typical interview questions, so make sure to follow this course if you are looking for a job in the field.

 The other courses will take you further down the path of Deep Learning, and you will be confronted with topics such as Structuring ML problems. The last two courses explain in detail Convolutional Neural Networks and Sequence Models – pivotal architectures in the field of Deep Learning, especially in NLP! I think it is a really well-designed course, and one that will help you transition into further specializations in Deep Learning.

Deep Learning Book by Ian Goodfellow

Ok listen. I am not saying you need to follow all the resources I am suggesting. But, I highly recommend you look at this book and try to follow along. Don’t get me wrong, online courses are great (I should know, I took like 5 billion of them, and I am not going to stop anytime soon!). But sometimes nothing hits just as right as a good book. And this one is actually a great resource. 

My PhD advisor recommended this one to me, and once I started reading, I could definitely see why. The book goes from the mathematics and statistics essentials you need to build a solid foundation, all the way to more advanced deep learning research topics. While this is a more theory-oriented approach, the explanations are clear and truly enriching. 

The book is well written and is considered the Bible of Deep Learning theory. I frankly believe it is hard to find books of such great quality, no matter the subject. The text is educational and exhaustive, as it covers all the Deep Learning basic techniques and more. You will find this book easy to follow even if you are a novice, since it provides you all the basis you need to advance. 

I would highly suggest you follow the FastAI course and then try to solidify your acquired hands-on knowledge with the more theoretical approach this book provides. These two resources pair really well together, like prosecco and Parmesan cheese. 

NYU Deep Learning Course

This is another great resource you can try out. What I like about this course is that it is really well-rounded. It has not only college-style lectures about many of the different topics you’ll find in the roadmap, but also practical examples you can follow along to and try out for yourself. I would say this is the perfect course for you if you are in a pinch and really want to go through most of the roadmap as fast as possible. I still would prefer the FastAI course + Deep Learning book, but this remains a valid resource for you to learn most of the important topics.

Building Advanced Deep Learning and NLP Projects provides this great course, perfect if you prefer to learn by doing. You’ll have a hands-on coding environment to help you practice as you learn. What I really love about this course is that it helps you build your portfolio as you are learning. As you may know – I talk a lot about it in this post – building your portfolio of projects is a must if you want to pursue a career in Deep Learning. And, even if you just want to build personal projects, having a portfolio ready always helps. By the end of the course, you should be able to use some of the algorithms that are widely used in industry and academia. I would say this is a great resource if you want to get ready for the job market, since it really focuses on how to code machine learning models.

A few more resources you might find useful once you are done with the basics

Ok, so now that we have covered the basics, it might be good to list a few extra resources you might find useful. I am not going to detail each course/tutorial, since I believe most of them are pretty self-explanatory. Furthermore, they are not really necessary. This is more elective material that might come in handy once you start exploring for your own, and you might need a little tip or a little refresher. The use you make of these resources really depends on how far you want to go in deep learning.

Of course, this list is not exhaustive, and I will update it periodically as I advance myself in the field. So make sure to stick around and look out for updates on this post (or other similar posts)! I will also try to provide more information and more sources for specific and/or advanced topics. As always, your contribution is more than welcome. If you have anything you’d like to suggest, please do so in the comment section! I do my best to always improve the content you will find on this blog.


Advancing on the Deep Learning Roadmap

I hope you enjoyed the content and your journey so far. Learning Data Science and Deep Learning mind seem hard and intimidating at the beginning, and it is! But it is also really fun once you get over the fear. Don’t lose motivation and keep up the good work! I’ll be here to cheer you on, and to share with you my journey Towards NLP. 

See you soon for the third and last piece of our Natural Language Processing Roadmap! You are almost there, may the force be with you!