Least Absolute Deviation Support Vector Regression

Figure 1

Squared loss function 𝑙₁(𝑟), absolute deviation loss function 𝑙₂(𝑟), and smooth approximation function 𝑙₃(𝑟) with ℎ=0.3.