Screening Collapse under Recursive Synthetic Data
Abstract
Recursive synthetic-data training can degrade a fitted model, but its effects on the variable-screening stage that precedes model fitting are less understood. When each generation labels data for the next, a marginal screener ranks variables against the current pseudo-label target rather than the original response. Across generations, this target drift can erode or reverse the active--inactive score margin and cause screening to discard variables that clean-label screening would retain. We call this process \emph{screening collapse}. In a Gaussian linear model with minimum-norm recursive generators, we establish finite-sample screening bounds, characterize a late-generation information-loss limit, and derive a minimax impossibility result. Screening collapse can precede and compound ordinary model collapse: omitted variables incur a covariance-weighted loss that no refit restricted to the selected variables can eliminate, while recursive labels further degrade estimation on the retained variables. We also examine two approaches to mitigating screening collapse. Fixed-fraction real-label retention preserves a clean-label component but leaves resolution fixed, whereas data accumulation can avoid screening collapse if the pooled target preserves a positive margin and the growing sample resolves it. Numerical experiments on gene-expression data support the theoretical findings.
Related
- Statistical-computational trade-offs for recursive adaptive partitioning estimators
- Forward recursive aggregator systems and the forward Epstein-Zin recursive paradigm
- Recursive construction of spectral basis and its application to matrices
- Recursive overlap Bernoulli distributions and an entropy concavity conjecture
- Well-ordering principles over recursive comprehension
BibTeX
@misc{zhang2026screening,
title = {{Screening Collapse under Recursive Synthetic Data}},
author = {Jia Zhang and Xin Chen},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.15617},
eprint = {2610.15617},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.15617},
}
RIS
TY - UNPB
TI - Screening Collapse under Recursive Synthetic Data
AU - Jia Zhang
AU - Xin Chen
PY - 2026
DA - 2026/10/11
PB - arXiv
DO - 10.48550/arxiv.2610.15617
UR - https://arxiv.org/abs/2610.15617
AB - Recursive synthetic-data training can degrade a fitted model, but its effects on the variable-screening stage that precedes model fitting are less understood. When each generation labels data for the next, a marginal screener ranks variables against the current pseudo-label target rather than the original response. Across generations, this target drift can erode or reverse the active--inactive score margin and cause screening to discard variables that clean-label screening would retain. We call this process \emph{screening collapse}. In a Gaussian linear model with minimum-norm recursive generators, we establish finite-sample screening bounds, characterize a late-generation information-loss limit, and derive a minimax impossibility result. Screening collapse can precede and compound ordinary model collapse: omitted variables incur a covariance-weighted loss that no refit restricted to the selected variables can eliminate, while recursive labels further degrade estimation on the retained variables. We also examine two approaches to mitigating screening collapse. Fixed-fraction real-label retention preserves a clean-label component but leaves resolution fixed, whereas data accumulation can avoid screening collapse if the pooled target preserves a positive margin and the growing sample resolves it. Numerical experiments on gene-expression data support the theoretical findings.
ER -