It has many advantages, such as using kernel function to avoid local minimal point, sparse nature of solutions, limit used to control capacity or the number of support vectors, etc.