# Support Vector Machines ### Intuition Given a classification problem, some observations in our training data get the distinction of being **support vectors**.  Above, we see that there is a green point and blue point that are each given the distinction of being support vectors. The support vectors are what *support* the **margin**. All other data points are not directly contributing to where the margin is. In this case, every other point in our data set is *irrelevant*. This means that we could move all other green points (other than green support vector) and all other blue points (other than blue support vector), and our decision boundary would not be changed (unless of course a new point was moved such that it became the support vector). Note in practice we generally use a **soft margin**. ### Math --- Date: 20211222 Links to: Tags: References: * [Intuition Video (ritvikmath)](https://www.youtube.com/watch?v=iEQ0e-WLgkQ&t=0s) * [Math behind SVM(ritvikmath)](https://www.youtube.com/watch?v=bM4_AstaBZo)