Support Vector Machine
This chapter explores a classic algorithm that one must keep in one’s machine learning arsenal called the support vector machine (SVM), which is mainly used for classification problems rather than regression problems. Since its inception in the 1990s, it was commonly used to recognize patterns and outliers in data. Its popularity declined after the emergence of boosting algorithms such as extreme gradient boost (XGB). However, it prevails as one of the most commonly used supervised learning algorithms.
In the 1990s, efficient learning algorithms based on computational learning were developed for non-linear functions. Algorithms such as linear learning algorithms have well-defined theoretical properties. With this development, efficient separability (decision surfaces) of nonlinear regions that use kernel functions was established. Nonlinear SVMs are quite frequently used for the classification of real (nonlinear) data.
SVM was initially known as...