I\'d like to perform a nonlinear regression for dimensionality reduction with a dataset that has more predictors than observations, and predictors can also be m