Beyond False Discovery Rate: A Stepdown Group SLOPE Approach for Grouped Variable Selection

Researchers introduce Group Stepdown SLOPE, a new statistical method for high-dimensional feature selection that improves upon existing frameworks by controlling multiple error metrics and exploiting group structure in data. The method provides better statistical power while maintaining strict error control in machine learning applications.

• →Group Stepdown SLOPE overcomes limitations of existing methods like SLOPE by controlling k-FWER and FDP metrics beyond just False Discovery Rate.
• →The method provides closed-form regularization sequences under orthogonal designs with finite-sample guarantees.
• →Extensions gk-SLOPE and gF-SLOPE enable group-level error control for grouped variable selection problems.
• →A data-driven calibration approach maintains convexity and scalability for non-orthogonal general designs.
• →Empirical results show the method achieves nominal error control with markedly higher statistical power than competing approaches.