Tuning-free ridge estimators for high-dimensional generalized linear models
This work addresses a computational bottleneck for researchers and practitioners using ridge regression in high-dimensional settings, though it appears incremental as it modifies existing methods rather than introducing a new paradigm.
The paper tackled the problem of tuning parameter calibration in ridge estimators for high-dimensional generalized linear models by proposing modified versions that avoid tuning altogether, resulting in improved empirical prediction accuracies compared to standard ridge estimators with cross-validation.
Ridge estimators regularize the squared Euclidean lengths of parameters. Such estimators are mathematically and computationally attractive but involve tuning parameters that can be difficult to calibrate. In this paper, we show that ridge estimators can be modified such that tuning parameters can be avoided altogether. We also show that these modified versions can improve on the empirical prediction accuracies of standard ridge estimators combined with cross-validation, and we provide first theoretical guarantees.