Note I took this during summer semester; YMML. This subject covers quite a few fundamentals of machine learning, along with some applications. The simple theory stuff like model complexity, linear models for both regression and classification, probabilistic learners, latent variable models and of course some intro to neural networks are covered. There were some more interesting topics covered like map-reduce, document clustering, autoencoders and self taught learning. The assignments involved both theory and practical questions, and were decently enjoyable though were a bit basic. Lecturer was good, explained stuff clearly at times though wasn't too engaging. The exam was quite fair, though my main gripe is that practise resources are very lacking. There were only four total practise questions for the exam. Not four past/practise exams, four questions. I guess that's what you expect at post-grad level. Can't comment on labs themselves, though the questions were just a showcase of the content. Overall it was an OK subject, you won't hate it but it won't be your most favourite either.