I just completed Stanford U’s Machine Learning Specialization course series over Spring break, and overall, I really enjoyed it.
I found it to be very nicely structured, with each week going over a different topic of Machine Learning, highlighting its current achievements and its major pitfalls.
Andrew Ng is a fantastic teacher that explains every concept in Machine Learning from the ground up, taking special care into making sure even non-coders or people with virtually no math background can understand the basics of how each algorithm works.
Personally, I would say that the aversion to “get into” math topics goes a bit too far in the opposite direction, as I deeply enjoyed the few optional, more mathematical “deep dive” videos that helped me better grasp some of the concepts, and wished there were more of them.
The labs sections are very fun and focused on “real-world” examples which I also enjoyed. Although I do wish they were a bit more hands-on — most of the code is simply given to you at the start and you’re supposed to “fill in the gaps” in the code, which while I admit is very modular and easier to grade, is definitely not my favorite method to learn from. Ironically, this meant that I struggled a bit with the latter course labs since I was writing so few code that I couldn’t grok the syntax for Tensorflow and had to google the documentation most of the time.
I wish the assignments would build up to eventually allowing us to just build our own networks from the ground up using our own code, rather than relying on external files and code snippets that hide from us most of the “backstage magic”, so to speak. But still, the feeling you get when the code just works at the end is wonderful :)
Overall I would highly recommend the course to anyone even slightly interested in Machine Learning!