Scalable Variable Selection under Predictor Dependence with Adaptive Virtual Dummies
Abstract
Reliable high-dimensional variable selection requires scalable error-controlling methods. The Terminating-Random Experiments (T-Rex) selector estimates the false discovery rate (FDR) by aggregating early-terminated forward-selection paths in which predictors compete with synthetic dummies. We address two remaining challenges: i) predictor dependence can bias dummy-predictor competition; ii) computation is wasted on recomputing terms shared across experiments. Building on memory-efficient virtual dummies, which sequentially sample projections from their exact conditional law, we estimate the conditional Gaussian law of inactive predictors and map draws onto the remaining sphere radius. A shared lazy Gram cache computes the response product and each requested Gram column once across experiments. Simulations across three covariance structures show that uniform spherical dummies can exceed the target FDR, whereas the proposed method empirically controls FDR. Caching yields more than a sixfold speedup, and the method remains feasible with 100000 predictors, where competing FDR-controlling methods become computationally impractical.
Related
- Variational Bayes for Variable Selection with Adaptive Sharing across Populations and Responses
- L-PASSO: Adaptive Log-penalized Variable Selector, with Sparse Selection and Robustness to the Pilot Estimator
- lepidoptera: A Domain-Agnostic Engine for Causal Simulation with Mixed Variable Types, Vine Copula Dependence, and Individual-Level Counterfactual Propagation
- Prior-data fitted networks for variable selection
- Consistent Order Selection under Non-Identifiability and Dependence
BibTeX
@misc{koka2026scalable,
title = {{Scalable Variable Selection under Predictor Dependence with Adaptive Virtual Dummies}},
author = {Taulant Koka and Michael Muma},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.31257},
eprint = {2609.31257},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.31257},
}
RIS
TY - UNPB
TI - Scalable Variable Selection under Predictor Dependence with Adaptive Virtual Dummies
AU - Taulant Koka
AU - Michael Muma
PY - 2026
DA - 2026/09/25
PB - arXiv
DO - 10.48550/arxiv.2609.31257
UR - https://arxiv.org/abs/2609.31257
AB - Reliable high-dimensional variable selection requires scalable error-controlling methods. The Terminating-Random Experiments (T-Rex) selector estimates the false discovery rate (FDR) by aggregating early-terminated forward-selection paths in which predictors compete with synthetic dummies. We address two remaining challenges: i) predictor dependence can bias dummy-predictor competition; ii) computation is wasted on recomputing terms shared across experiments. Building on memory-efficient virtual dummies, which sequentially sample projections from their exact conditional law, we estimate the conditional Gaussian law of inactive predictors and map draws onto the remaining sphere radius. A shared lazy Gram cache computes the response product and each requested Gram column once across experiments. Simulations across three covariance structures show that uniform spherical dummies can exceed the target FDR, whereas the proposed method empirically controls FDR. Caching yields more than a sixfold speedup, and the method remains feasible with 100000 predictors, where competing FDR-controlling methods become computationally impractical.
ER -