I got my start by getting a PhD, but that's perhaps not a practical recommendation. In reality though, you might say I started learning ML by reading Mitchell in class: https://www.cs.cmu.edu/afs/cs.cmu.edu/user/mitchell/ftp/mlbo... It's dated, but it's quite approachable and does a great job explaining a lot of the fundamentals.

If you want to approach machine learning from a more statistical perspective, you could also have a look at An Introduction to Statistical Learning to start: http://www-bcf.usc.edu/~gareth/ISL/ Or if you're more mathematically inclined than the average bear, you could jump directly into The Elements of Statistical Learning: https://web.stanford.edu/~hastie/ElemStatLearn/

If you want something a little more interactive than a book though, you might have a look at Google's free crash course on machine learning: https://developers.google.com/machine-learning/crash-course/... I checked it out briefly maybe six months ago, and it seemed pretty good. It seemed a bit focused on Tensor Flow and some other tools, but that's okay.