Teaching: You're Doin' It Right

Self-note, or note for whoever wants to teach (or is a teacher).

I'm taking the ML-Classes by Andrew Ng + team from Stanford University.
I'm also taking the AI-Classes by Norvig & Thun... But the experience is totally pretty different and actually I'm loving the Machine Learning course much more than the other.

So, I'm resuming here what really love about these lessons, which I personally consider to be very-high quality (even if there are few glitches, but are negligible).

1. Video lessons can be downloaded. In AI course this is not possible, since they use youtube as platform. This is important, because determine when and how people can watch these videos. Also, videos are short (from 5 to 15 minutes, rarely less than 20), which improve usage over the classic "whole lesson" of 1+ hours.
2. Explanations are precise and they go straight to the point. I feel it's not the same with the AI class, where are a bit too slow and fuzzy.
3. Digitally-annotated presentations is a great way of explaining: teacher prepares the slides and, using a tablet device, writes things on screen using multiple colours. This improves readability and speed. In AI classes, paper and marker is used, and it's not as good as the first. Khan academy also uses the former technique and, I think, it's the best.
4. FLOSS tools are used, Octave in this case, which is pretty nice. Tools are also introduced properly, and this is very important. I followed some courses at university were we used Matlab, but the explanation has been done "en passant" and the using experience was awful.
5. Validation tools are pervasive: doing quizzes and tests often is a very good strategy to make students learn quickly. This is done both in ML and IA classes, luckily, but I appreciated also the programming exercises validation in ML class: *from octave*, a function is made available to send your exercises to server, and you get immediate feedback in an interactive feedback. This is great and actually the first time I see something like this. Very clever idea.

In general, in the ML class I see a well-developed framework for e-learning which could (and should) be replicated and used often.

If you organize courses, are a teacher or you're interested in giving students a good e-learning framework, take a look at how ML classes are done.

Stay --sync