Posted on

NUS Computing Review: CS3244 Machine Learning

The basic module information is available at his website: https://www.comp.nus.edu.sg… – you may want to take a look at it to see the topics that will be taught. The first half of the module generally focuses on simple techniques used for learning and some fundamentals like bias, variance and regularization, while the second half focuses on current topics like Neural Networks, Random Forests, MDPs. For the project teams you are expected to form groups of size 2-3 and you will be matched with another group of size 2-3. (Generally forming groups of size 5-6) The idea is to be able to work with some of your friends but still have to meet and work with strangers as well.

The Good:

– The module topics are selected to been more relevant in today’s world. I have some friends who took the module before under different Profs and they found the topics not as useful with current AI landscape.

– The project is very open ended (Apply ML to anything you find interesting).

The bad:

– Module takes up a lot of time – you have flipped classrooms, so every week there are videos to watch as the current topic, there is a colab notebook to complete (Some of them can easily take up 3 hours) and there are tutorials to work on. Then, there is the usual midterm, project and finals.

The ugly:

– The tutorial & midterm questions were actually set by TAs – in my opinion this makes it feel like the Prof was unprepared and thus passed this load on to the TAs instead.

– The Prof doesn’t typically respond to emails – so if you generally email the prof for clarification/consultation be prepared to just get ghosted for the semester.

If you do take this mod I would recommend trying to understand the formulas and what they are trying to do. Within the colab assignments you will be asked to implement certain formulas and if you just write the code to do the calculation blindly you will not be learning anything. Try to dissect the formula and see what it is trying to do. Additionally, it would be very helpful to be familiar with using python. You do not need to know how to use Tensorflow/keras, but hopefully you have written some scripts in python before and are familiar working with it.

If you already know what a Neural Network is and how it works, how backpropagation works, activation functions, this module will likely be easy for you.

Head over to our Shop for more module content!