Feature Map ϕ(x), Basis Function h(x)
hθ(x)=Σl=1Lθlϕl(x)=θTϕ(x)we can extend the linear regression into a linear combination of basis functions
Rd→R(x∈Rd)- gaussian basis exp(−2σ2∣∣x−μl∣∣2)
Using nonlinear basis functions, we allow the function to be nonlinear