What maths courses are needed for Machine Learning

Question

I may sound dump. But I really like to know what maths courses are needed for Machine Learning. I am not computer science graduate but seriously interested in AI, ML, Neural network etc, and I know Java and familiar with Python as well. Can somebody let me know what maths courses are taught in computer science universities year by year? I am a fast learner and will be doing self-study on websites like Coursera etc.

Would appreciate if somebody can also suggest any basic CS maths book and online videos.

Many thanks Max

Answer 1

I don't know if this will help you, but my university course on machine intelligence didn't require a whole lot of maths, but having some decent mathematical intuition would help. For example, it helps understand the shortcomings of artificial neural networks to be able to look at them as multivariable functions, and have some rudimentary mental picture their graphs.

Support Vector Machines, which are like a more sophisticated neural net, rely on optimisation and linear algebra techniques. My university's course didn't go into the inner workings much, but those areas might be handy there.

Answer 2

You would need similar prerequisites as for most probability theory and statistics courses, i.e. a solid grasp on matrix algebra in particular and linear algebra in general, (differential) calculus, maybe some optimization theory and numerical analysis. You also want a good grasp of essential statistical concepts. A good reference for the latter is The Elements of Statistical Learning, which can be downloaded for free.

Answer 3

Baysian statistics in particular are important to learn, rather than just frequentist stats

Answer 4

As a side note / addition to Stijn's answer, you probably won't really need any "traditional" clean computer science courses, like knowing the best way to find a median of a list of numbers, or knowing what a Fibonacci heap is, but rather some of the messier stuff like optimization theory and numerical methods and numerical analysis.

Also, unless your ML course is strictly about the concepts, you'll probably need to know programming in some language. In our basic ML courses here at Carnegie Mellon, students are expected to complete an empirical final project, which usually means doing some coding.

And depending on the level of depth you are expected to understand things at, like Naive Bayes classifier and Gaussian processes, you will probably need a strong foundation in mostly Bayesian statistics.