Classifying in high dim.

• The kernel allows us to calculate dot products in the higher dim space

• Thus we can learn the boundary in higher dim space, without transforming the data

• The classifier becomes

• This gives a non-linear decision boundary in the input space

Notes: