Kernel-based Nonlinear Fit with Total Least Square(TLS) Method

In this paper, on the basis of linear fit in the total least square(TLS) method sense, we proposed a method of nonlinear fit in the TLS method sense via kernel representation. Namely, by using an appropriate kernel function, the problems of nonlinear fit can be transformed to the problems of linear fit without paying the computational penalty and without the precondition that the fitting function type of the data points is known. The experimental results show that the algorithm presented in this paper is available.