As I said before when someone asked about learning stuff online, the best source is Udemy.com . It has great teachers and probably the best structured classes. And the best thing is that classes are constantly being updated by their authors. Plus, you can ask the author of the class a question if you don't understand something.
Simply - Udemy.com is the best source to learn almost any topic online.
I used to search about that because i tried to learn something online. Here are the best choices:
MIT Open Course Ware
Google Code University
I think the best way to self learn machine learning is to code while learning the theories, so that you will have a deeper understanding of the theories and applications. I would suggest that you first familiarize yourself with programming languages such as Matlab and Python.
First, I would recommend that you have some basic knowledge about mathematics, especially statistics.
Second, you have learn more about machine learning itself. As for intro books, I would recommend:
Machine Learning in Action (this book combines code and machine learning theories in a very clear way. I would suggest you to try coding according to this book in order to familiarize yourself with these knowledge)
Elements of Machine Learning (this one is more difficult and has more requirements. As for intro, the book above is more recommended)
In the mean time, if you can take some online courses to strength your understanding.
I would recommend courses from Experfy, a Harvard based company that provides various online courses related to IT and Tech. To address your request (which is more about the applications of machine learning), I would recommend Machine Learning for Predictive Analytics from Experfy if you want to broadly learn about how machine learning can be applied to different areas, as it has a lot of real world cases and demos.
Third, to further understand machine learning, you can read:
Bishop’s Elements of Statistical Learning (you can read this book first)
Hastie, Tibshirani, and Friedman’s The Elements of Statistical Learning (this book requires more mathematical background and you can read it after finishing the first one)
I will put my two cents and it will be mainly focused on Deep Learning. Deep Learning has greater learning ability due to non-linear activation functions . In simple words typical what typical ML model does is it tries to find a linearly separable boundary. The linearly separable boundary is a line in 2-d or a plane in 3-d or hyperplane in higher dimension.
On other hand due to non-linear activation functions, DL model is able to classify data that are not even linearly separable. That's why I recommend DL. Also the concepts are more or less same in both ML and DL.
Here are few of the resources I would definitely recommend. These are simple and clear most of your basics:
- MOOC of fast.ai : It has two parts of video series for DL and one for ML. Jeremy jumps directly into implementation and then explains many thing which I find quite awesome. For time being, I would say to avoid fast.ai library and use Keras. fast.ai library hides many thing which can be avoided if you are a beginner.
- MOOC from Coursera of DL by Andrew NG : It starts with basics and goes to a higher level. Its slow paced though and has 5 courses.
Also keep practising to sharpen your skills on kaggle.com or a problem on which you would like to work.
Language of choice: Python [No war on this. Please!!]
These are the best resources I have collected and should be sufficient themselves (and they are extensive)
It takes time to learn all these things depending upon your pace. Practise is the only way to get better and intuitive understanding and don't diversify on reading too many things. It does nothing except creating confusion.
Udacity - Geeks for Geeks - Wikibooks - Stack Overflow - Code Academy - GitHub
Coursera & Google
Both free of cost & provide structured material for someone, willing to learn ML
I personally think that internet is a great source to learn anything.
But in addition we have to know that whether we are on right way or wrong.
Along with this we have to get all necessary basic information about AI and related industries, Because without knowing basic no can be expert.