Unfortunately (IMHO) today machine learning is almost exclusively associated with data analysis, recognition, and classification, specifically using ANNs (this would  be software ANNs that are basically simulations of Neural Networks).  I say that this is unfortunate because this perspective tends to ignore other methods of creating machines that can learn.   But this is where everyone in the A.I. community is heading and so this is what the term "Machine Learning" basically means today.

Here's a video that offers "5 Beginner Friendly Steps to Learn Machine Learning"

https://www.youtube.com/watch?v=IMMDPzECrf0

I have personally been going down this particular path.   The only difference is that I do not view this path as being "exclusive" in terms of other methods of creating machines that can learn.   In fact, I see this entire  approach as merely one small tool in the toolbox.

None the less, this is what most people who are in the field of A.I. will tell you Machine Learning is.   It's basically learning how to construct various types of software ANNs using the tools referenced in the above video.

I would like to add here that these presentations make it appear to be far more difficult than it actually is.  Although I agree with the host of this video that just starting to write programs along these lines is the way to go.   That's certainly how I've been learning.

I'm only a few months down the road of learning the methods described in the above video.   Most of this time has been spent learning about the software tools mentioned.   Fortunately for me this is quite enjoyable to study so it doesn't seem like work to me. ? 

In the video he suggests an approximate time from to learn all of this is two years.  But that seems quite arbitrary to me.   Without specifying exactly what a person will know in those two years  it doesn't make a lot of sense to put a timeline on it.    You simply keep learning more over time. 

Anyway I just thought I'd post my comments on machine learning and the above video to demonstrate what the A.I. community currently sees "Machine Learning" as meaning.

As I have already stated.  I see the concept of machine learning covering a lot more than just software ANNs that are used to classify and recognize data.   In fact, I personally prefer to study the concepts of algorithmic approaches to machine learning (i.e. old-fashioned programming).  This is not to say that these software ANNs wouldn't also be useful in that context, but rather to suggest that the software ANNs alone are not the whole picture.   Again, just my thoughts for whatever they might be worth.