What’s “next” after Deep Learning specialization from deeplearning.ai?

I thought it may be a good idea that I summarize the “next thing” after this deep learning specialization as there is n number of discussion around this by folks that completed or about to complete 5-course specialization.

I think this might be the next several paths that can be taken to continue:
1. Go to CS231n and CS244(d) if you haven’t completed already. Watch videos, take notes, go to slides and reading suggestions and if possible try to complete assignments.
2. fast.ai is a great place to go after. Preferably start with their part-1 and then move on to part-2. Watch Jerome’s video, spend time in the forum, read wikis and if possible experiment with given notebooks with different datasets. Search or ask answers in the forum if you are stuck one something. There are plenty of great materials in fast.ai that will supplement your understanding and grip on many complex ML concepts, take full advantage of those besides notebooks. Look at their github, pickup one and reproduce that for another use case. Don’t copy and paste code type every line (pain I know) and ask yourself can you explain this to others now?
3. More advanced learners can go deep and go fundamentals such as the theory of deep learning https://stats385.github.io/ and understand how masters of the master researchers and scientist who spent their lifetime understanding intelligence and now explaining deep learning which sometime may seem like a miracle. There is a video series of statistical learning from MIT in youtube and http://www.mit.edu/~9.520/fall17/ where you will find Thomas Poggio and team (CBMM) created object detector using SVM for autonomous vehicle 20 years back and how it progressed today with millions time improvement. Read textbooks such as deeplearning book, ISLR, ESLR and PRML. I heard Yashua Bengio said more people bought this book than people can understand. There is only handful of people in this world who can understand deeplearning book.
4. Practical hands-on experience can only get you a job, so folks who aren’t working in AI already go for solving some Kaggle competitions and make sure you write about it, blog and MUST create a nice GitHub repo for showing off. Note that those are your resume.

Okay, enough said. Feel free to add here. Happy learning.

Source: Deep Learning on Medium