Not too easy. Not too difficult.
Course Overview:
Machine Learning is an important subject because it helps you cope more easily with uncertainty inside your area of work. Suppose for instance you are a CEO of a big company and it happens that you have to decide very quickly whether your company should keep selling a certain product or not. From first hand you would not have enough information to make a sound decision. That's when Machine Learning comes into play.
Course highlights:
The first part of this course covers Supervised Learning, an area in Machine Learning that can enable you to write an application capable of recognising handwritten digits for instance. This was a very fascinating thing to learn, especially the many projects it enabled me to do (handwritten digits recognition, made lot of different types of predictions...). The second part is about Unsupervised learning, a more difficult subject that enables you to force information out of your data through different clustering techniques. It is hugely helpful too especially when you don't seem to know anything about your data which can often happen in practice. The third part of this course is actually the coolest one, it enabled me to train a self driving car; this is Reinforcement Learning. One particular algorithm I used for this is the Q-Learning which is discussed extensively in the course.
Hours per week:
9-11 hours
Advice for students:
The course has a great deal of theory in it but if you get over your fear of odd looking mathematical formulas, you will actually be able to grasp the whole subject much more easily and use the concepts much more efficiently. One particular advice I have, is to always try to implement the algorithms you learn in the course and understand how the formulas you will be implementing in those algorithms actually work.