SVM can be thought as a generalized perceptron with a kernel that computes the inner product on transformed input vector.