- creates optimal decision boundaries in a feature space
- Linear decision boundary: Create a n-1 dimensional decision boundary in a n-dimensional feature space
- margin: width that the boundary can be increased by before hitting a point
- find a decision boundary with the maximum margin
- the data points on the margin are called support vectors