[xtracto]

I respectfully disagree. A PhD is to apply knowledge you already know. If you want to go the academic route, go for a Masters. That's what you use to learn. I have a PhD in CompSci, and I can tell you, that's NOT the path to follow to learn a subject.

Now, if you want to go the independent learning route, I can recommend the Udemy course: https://www.udemy.com/course/python-for-data-science-and-mac... It will give you a good hands on overview of all the current topics in Machine Learning.

You first will be learning to USE ML methods, then you can start extending the ML field itself (if you want).

[bitL]

I agree with you in principle, however it's about perception. For some reason a PhD and even better articles at top conferences are the minimal qualification for great ML jobs. So just having a MS in ML is often insufficient. If you want to start your own company doing ML or work on entry-level ML things, then yes, alternatives are there but if you want to work on important things for a lot of money, non-PhD route is rare. I've only once met a person who dropped out of a PhD at a top 5 university to work on a top-end ML. And unfortunately due to the pace in the industry, not working on a top-end stuff means working with significantly worse models/ideas somebody already used and discarded for something better. I have some private info about what the best ML folks are doing and many techniques even taught at Stanford grad courses are already obsolete (not just architectures, but the whole categories of how certain things are modeled in ML).