ABSTRACT
This chapter focuses on one particular algorithm, the Support Vector Machine (SVM), which is one of the most popular algorithms in modern machine learning. It was introduced by V. Vapnik in 1992 and has taken off radically since then, principally because it often provides very impressive classification performance on reasonably sized datasets. There is rather more to the SVM than the kernel method; the algorithm also reformulates the classification problem in such a way that we can tell a good classifier from a bad one, even if they both give the same results on a particular dataset. There is a lot of advanced work on kernel methods and SVMs. This includes lots of work on the optimisation, including Sequential Minimal Optimisation, and extensions to compute posterior probabilities instead of hard decisions, such as the Relevance Vector Machine.
