Several weeks ago I signed up at ml-class.org. Machine learning is of big interest to me and I highly recommend taking this course. I wish I knew earlier how fun and educational it is. I finally managed to get into it this weekend. I am slightly behind the due dates, which results in 20% penalty on your score. Overall it is a well organized and presented class. Hopefully one day people can get a respectable degree from the comfort of their home using similar technology. TIP: Instead of entering your email and password every time, just modify the basicPrompt function in submit.m: function [login password] = basicPrompt() login = 'your@email.com' password = '<your-web-submission-password>';endI'm currently trying to figure out a better way to calculate the logistic regression cost function (vectorization among other things)  I haven't used Octave before but I am getting the hang of it. It seems easy to prototype mathematical algorithms. You can easily do in a few lines of code what could take 30 or more to do in C++/Java, etc. |
This actually wasn't really that hard...
Once you realize that h can easily be vectorized like this:
h = sigmoid(X * theta);
the rest is trivial.