A neuro-educational approach to taking Andrew Ng’s Machine Learning Course

2 minute read

I recently finished Andrew Ng’s fantastic and well-known Machine Learning course through Coursera. As I progress into my data science journey, I felt that taking and completing his course was one of the rites-of-passage in this field. Since the course has been around for some time, there are already fantastic resources and supplemental material to facilitate taking the course which have been described elsewhere such as here, here, and open source notes from Andrew Ng himself here. Instead, the tips I provide below are for newcomers to effectively learn the material and not necessarily about the course content. Some of this is inspired from my participation in neuro-education activities that I did while I was a post-doc and from taking Peter Newbury’s classes while he was at UCSD.

  1. Space learning over time. This refers to the delayed re-exposure of material. This approach is covered in one of my favorite guides, Organizing Instruction and Study to Improve Student Learning. Instead of spending four consecutive hours on a lesson, devote two sessions of two hours each with the sessions spaced several weeks apart. As a concrete example in the ML course, there are some concepts or practices that are taught early on then are re-exposed later through course assignments. One example of this is practicing vectorization in MATLAB notation, which was taught explicitly in Lecture 5 but is applied several times in course assignments. However, other ideas could use more self-directed re-exposure.

  2. Don’t get too hung up on using MATLAB/Octave versus a more data science-y language like R or Python. I thought using MATLAB/Octave was a bit peculiar, but it made sense to me as I was doing the lessons. Most of the code is very easy to pick up and he even does a brief tutorial as a part of the course.

  3. Before watching the video, I’d recommend looking through the lecture notes or summary first. While things might not make sense immediately, previewing the lecture notes provided a mental roadmap of the lesson while watching the videos. It’s also a good idea to glance through the FAQs on the discussion forum for that week. The FAQs could alert you to errata in the lectures so that you’re aware of them right away.

  4. Write things out. I pretended like I was in school and wrote notes in a dedicated notebook. This was admittedly time-consuming, however, I think the lessons became a little more solidified. When you write things out, you have internal dialogues with yourself like “Is this an ‘i’ or a ‘j’? Does this term go inside or outside the parentheses?” Writing forces you to take a few extra seconds to marinate on the information a little longer and be able to answer small questions like this. My experience is consistent with research showing the benefits of writing versus typing notes. When writing, have different colored pens or pencils. Professor Ng uses several colors when writing or drawing to note emphasis or distinctions. However, the lecture notes will sometimes lose these color distinctions. You can also use different colors to write important things that he says versus what he shows.

  5. Participate in the discussion forums as an implementation of active learning. Don’t be shy about asking questions. There were a few times when asking questions lead me to an answer on my own. There’s no such thing as dumb questions and people are very helpful. (Tom is fantastic!) Return the favor and try to help others too.