7. Introducing: Support Vector Machines

Welcome to the second section of the course. In this section we shall be introducing a highly effective machine learning algorithm: the Support Vector Machine (SVM). Applicable to both classification and regression problems, these robust algorithms also provide novel of explainations of the decision process via their supporting vectors.

We shall cover:

• Maximising the margin between data classes (the descision surface margin).
• The role of support vectors in application to data problems.
• Non-parametric forms of SVMs via the kernel trick.
• Overlapping class boundaries of noisey data.
• SVM regression: sparse regression by allowing an error margin around fit.