Nowadays everyone is into Artificial Intelligence and Machine Learning. Although they have become very popular, there are still many of us uncertain of what they mean. And moreover how to use the machine learning to achieve a particular goal.
Course covers these topics from machine learning: linear regression, logistic regression, neural networks, support vector machines (SVM), dimensionality reduction, anomaly detection, recommenders, application of machine learning, and working with large datasets.
Each topic is presented trough slides, which are narrated by professor Andrew Ng. A major advantage of these slides is that they are interactive — after group of slides about particular topic a quick test appears to verify, whether you understand the lecture. These tests are optional, though I strongly recommend to give them a try and do not skip them. I have also one more advice — take notes (ideally handwritten, if you are oldschool like me), while you are watching. They will be helpful later and you can always look at them if you need. And by “later”, I mean even after the course ends!
Most of weeks is ended by graded quizzes and programming assignments. Quizzes are focused on theoretical knowledge, whereas assignments are problems you need to solve by submitting code. Course supports two programming languages: Octave and Matlab. This is another advantage — you can learn basics of new programming languages (if you have not met them before). But don’t worry, you do not have to write tons of lines, each exam comes with some prepared skeleton of code and you have to just write few lines in order to pass it.
All quizzes and programming assignments can be submitted automatically (quizzes right from your web browser, assignments by using your command line), and moreover they are also automatically scored. You are also allowed to submit more than once, so you do not need to worry if something does not work for the first time. There is however a mechanism to prevent abuse — quizzes can be submitted at most 3 times per 8 hours.
If you do not have enough time to finish a week, you can give it try another week. Deadlines are not strict and Cousera use them just to show you an ideal plan.
In my opinion, the course should be renamed to /Introduction to Machine Learning/, because it is rather introductory than complex. However, this is certainly not a disadvantage. One cannot dive deeply into the all methods of machine learning in 11 weeks and especially for the beginners, it is definitely better approach to have at first a big overview.
This course has been my first (at Cousera) and also my best yet. Although it was not difficult, on the one hand it helped me to understand concepts of machine learning, and on the other hand it was really well organised. If your research needs to use machine learning, I recommend you to take this course. And it does not matter whether you are beginner or professional — you will not regret it.