[skyo]

I started watching the lectures at ml-class.org this weekend, and I found myself noticing the same thing. It's wonderful how quickly the professor goes through the introductory material so you can start learning right away. After only half an hour of watching, I had learned things I didn't know before (supervised learning vs. unsupervised learning, regression vs. classification, etc.). And after another hour he had covered all the way through a whole algorithm for solving regression problems. I'm finding this class to be much more efficient at teaching than a lot of my lectures in college were.

[throw_away]

The biggest differences in my mind between the online ml-class and the lectures I attended in college are:

1) ability to watch the videos at 1.5x speed. this is great because I usually find speech to be such a low-bandwidth transport.

2) ability to have captions on the videos. I don't know why, but I understand what people are saying better if I have captioning. I do this with regular broadcast tv as well.

3) ability to pause or go back when I don't quite understand something. I can look at the formula for a little bit longer until it clicks, or I can jump back a minute or two to figure out where I went off the track.

4) broken up into 10 minute chunks with memory of what you've already watched. this way I can watch the lecture when I have time, but without having to go searching for where I left off.

5) the interstitial and end quizzes. the interstitials make it so that I can be sure that I understand the material before I move on. the end quizzes are awesome. I forgot how much I liked tests in school & how life in industry is pretty much devoid of such things.