How to practice DL after graduating the specialization?

Dear mentors, fellow students…

  1. I’ve just started course 4 after graduating courses 1-3. I’m a little concerned, because apart from theory, which is important per se, what I learned to do in up til now is to fill out some chunks of code. It’s not as if I am now able to write the full code of a DL project, or to use a framework such as TensorFlow, as we only had a short glimpse of it at the end of course two. I am starting to get concerned about the question: how to practice in order to become a real DL practitioner as soon as I graduate courses 4-5? Maybe you address the issue later in the speciaization, but I would like to get an answer that will calm my concerns. Where to start?

  2. Personally I would like to work in the sphere of NLP, as I have a PhD in Linguistics (from the faculty of Humanities). I know there are a few courses in coursera about NLP, but which course would you recommend me to take after graduating this speciaization?
    Thank you,

Hi, @Doron_Modan !

The best advice I can give you is to practice with Kaggle competitions. That’s probably one of the best platforms out there for ML/DL projects with real world data.


You will get lots of practice using TF in courses 4 and 5.

1 Like

As Tom says, you’ll get plenty more exposure to TF/Keras as you work your way through DLS C4 and C5. If your primary interest is NLP, you’ll definitely want to get all the way through DLS C5: that covers the types of models that are the most powerful for NLP applications. Once you finish DLS, then you have a number of options. If you feel the need for more TF experience, there are a number of specializations which focus on that. Then there is also the NLP specialization. That gives a lot more of a survey of NLP techniques than you get in DLS C5. NLP covers a lot of other techniques before it gets to Sequence Models and Attention, which are the subjects of DLS C5.

1 Like

In my personal experience, the moment I found myself really remember how to use (some of ) tensorflow is when I worked on my projects, during which I researched a lot on how others implemented their models. It is a process that I had to keep reading others’ code, keep polishing mine, and keep trying something new. A single or two courses never got me there.

My suggestion is that you can search Github where you’ll find thousands of projects that you can study, reproduce and apply what you have learned here.