Machine Learning & Pattern Recognition (MLPR)

Response to 2015/16 survey feedback.

Many people appreciated this course, and did well. Many others struggled, and I also received extensive feedback and critical comments. I also ran my own survey asking for opinions on specific choices about how this course is delivered. In response, I am completely overhauling the way this course is delivered in 2016/17.

I have changed the style and number of the lectures, so I can go through the material in more detail. I have rearranged the course so that the mathematics that may need revising is introduced more gradually. As a result I am having to rewrite most if not all of the tutorial sheets. I am also providing extensive class notes, which are also being written from scratch with this class in mind. These contain code snippets in Matlab and Python. I have swapped to a forum system that allows easier annotation and posting of maths and code.

As a result of all these changes, there will inevitably be some rough edges. The forum annotation system allows students to flag these to me with minimal effort, and in most cases I will update the notes on the same or next day.

However, I would like to emphasise that no matter what I do, students who do not have the maths background to take this class will still struggle and should not take MLPR. I gave this warning in the first class. I have also rewritten the guidance on the background requirements, and prominently posted a self-administered "pre-test" to make these requirements clearer, and help those missing only manageable parts.

Finally, to the strongest and most prepared students in the class: I haven't forgotten you. I will attempt to link to optional outside reading from some of the lecture notes, and the forum system allows you to comment on and question these. Where I don't, challenge me to. Also, although I am covering some topics more slowly than before, I am also adding more detail. I hope to share (and test) a more nuanced understanding than before for those at the top of the class.

Iain Murray, October 2016