I am looking for a way to do nonlinear partial least squares in R or matlab. I thought kernel pls was a way to do it but it is not directly related to nonlinear pls. Do I have to calculate my own nonlinear kernel matrix and pass it to kernel pls? Any hint would be appreciated.
Update: I don't think there is any direct way to do nonlinear pls in R or matlab. But I found a package kernlab that can be useful when you deal with kernels for nonlinearity. I think this can be a good start point.