Sparse Regression Distilled from a Single Robust Fit

Wooyoung Shin, Seunghwan Park

Robust linear fits can resist response contamination yet remain too dense or unstable for useful global explanations. We propose penalized distillation, which fits a smoothly clipp…