SVM optimizes the decision boundary so that it falls equally between the separation in the data because it has the least risk of causing an error

Although SVM uses a linear model, the kernel trick can be used to map the data to a higher dimensional space and then non-linearly classify it

Plotting two features with the decision boundary can help visualize how SVM is working

Some common non-linear kernels used for SVM include polynomial and radial basis function (RBF)

Non-linear kernels have parameters which need to be adjusted to get the best result, and the more we increase nonlinearity the more increase overfitting risk