I first learned Machine Learning online.
I fell for it ?! We'll, it is beautiful.
You get a very basic idea of the application of Machine Learning.
You understand how to use some existing libraries to solve some real-world problems.
You get to implement some basic algorithms (at least when I did the courses).
The depth to which you understand Machine Learning is amazing.
It becomes very easy to tweak some of the algorithms to suit the requirement.
There's a lot of concepts that I feel are very important, I didn't get online but in a class by a very knowledgeable professor. Some people teaching online are like the best in the field, but, there's a lot of limitations when it comes to brainstorming or many such things.
I wouldn't say it is hard to get access to resources like research papers online, but, what to look for when you need something is not very streamlined and you end up getting lost. In an offline course though, you have a streamlined syllabus that is altered according to current trends and industry requirements. Basically, you know where to look and what exactly to look for when you want to build something.
End of the day, if you are learning to get a job, I feel it is better if you do it offline as the companies(in general) are not very open to online degrees or certificates, not yet.
You gotta learn it in class to understand the beauty of ML algorithms!
Online, if you just wanna use it to target some problem sets.
See this Machine Learning Course for more information :