Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Evidence-Gated Research (EGR), a novel statistical adoption layer for adaptive model search designed to mitigate path dependency and control false discovery rates. EGR freezes challengers before evidence reveal, builds anytime-valid evidence across environments, and composes a persistent candidate e-value to an online controller, ensuring statistical validity even as incumbents change. Benchmarks demonstrate EGR's effectiveness in controlling false discoveries and improving evidence efficiency compared to traditional methods.
Adaptive model search can now achieve statistically controlled adoption without sacrificing evidence efficiency, even with path-dependent challenger generation.
Adaptive model search is path dependent: once a challenger is adopted, it becomes the reference from which later candidates are generated. A statistically unsupported replacement can therefore alter hypotheses that have not yet been proposed. We introduce Evidence-Gated Research (EGR), a statistical adoption layer for moving-incumbent search. EGR freezes each challenger before decision evidence is revealed, builds anytime-valid evidence across a predeclared set of environments, routes evidence predictably toward unresolved components, composes a persistent candidate e-value, and passes that e-value to an online controller. Under explicit conditional-validity and predictability conditions, the resulting procedure controls false discovery rate for the declared all-environment adoption target even though earlier adoptions change later challengers. In a 5,000-trajectory closed-loop benchmark, development-only e-LOND attains persistent FDR 0.621, whereas no persistent false-adoption path is observed for the audited EGR variants in that finite run. In matched replay over 600 challenger--incumbent pairs, Stagewise EGR preserves fixed-anytime alternative crossing decisions while using 56.1% less decision evidence at the representative threshold. A three-environment public-data study and a 40,000-sample controlled neural benchmark reproduce the evidence-efficiency pattern. These results identify model replacement as a distinct statistical control point in adaptive model development.