VStress：面向重复验证器的相关性感知审计与自适应预算分配

View PDF HTML (experimental)

Abstract:Repeated verifier calls are useful only when they contribute conditional information. We introduce VStress, an auditable replay contract, and VStress-CA, a correlation-aware allocation policy that estimates the conditional marginal information of an unqueried verifier on a sealed calibration split, discounts uncertainty, normalizes by call cost, and stops or abstains when the next call is not informative. The controller freezes its decision and cost ledger before joining the clean oracle; a dependence-shift alarm disables channel preference and falls back to exact-stop. The controlled audit gives the mechanism boundary: at 35% symmetric corruption, majority-5 improves balanced accuracy from 0.6578 to 0.7739, whereas at 65% it loses 0.1226 points. In the matched fixed-budget comparison, breadth, redundancy, and adaptive allocation obtain balanced accuracies 0.6048, 0.6375, and 0.6538, with 3.4216 calls per item and an RLVR score of 0.6417 for VStress-CA. Dependence diagnostics also increase from same-model repeats to cross-family channels, with conditional marginal gains of 0.0126, 0.0462, and 0.0913. These measurements turn correlation from a post-hoc warning into an auditable allocation decision.

Comments:	27 pages, 5 figures
Subjects:	Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR)
Cite as:	arXiv:2609.36958 [cs.LG]
	(or arXiv:2609.36958v2 [cs.LG] for this version)
	https://doi.org/10.48550/arXiv.2609.36958 arXiv-issued DOI via DataCite

Submission history

From: Miaobo Hu [view email]
[v1] Tue, 29 Sep 2026 07:59:03 UTC (1,492 KB)
[v2] Fri, 16 Oct 2026 12:02:53 UTC (1,492 KB)