跳到正文
arXiv cs.AI· Noah Flynn·· 56 分钟前AI 评分43

保真度遗憾:在可验证设计中度量模拟器失配

Fidelity Regret: Measuring Simulator Misspecification in Verifiable Design

AI 导读

一项针对 RLVF(可验证奖励微调)的研究提出“保真度遗憾”指标,衡量用廉价模拟器替代验证器选设计时造成的性能差距。研究在覆盖 5 个实测门库、3 种生物的 21,362 个设计组合子空间上测试,72 个单元的平均归一化遗憾为 0.369,仅 6 个单元为零。

正文 · 原文

View PDF HTML (experimental)

Abstract:RLVF assumes the verifier evaluates the quantity of interest. In genetic-circuit design, a topological verifier can establish that a program is a valid circuit computing the target function, whereas ranking its physical performance requires a simulator. We measure fidelity regret, the reference simple regret incurred when a cheap simulator selects the design, over a 21,362-design combination subspace spanning 5 measured gate libraries and 3 organisms. Exact ordered-assignment checks retain positive regret in every tested cell while shifting magnitudes in both directions. Mean normalized regret is 0.369 over 72 cells and only 6 have zero regret. The discrepancy depends strongly on the reference axis: the yeast library SC1C1G1T1 has mean regret 0.113 for cytometry quality and 0.801 for growth burden, and every evaluated library--circuit-size pair has positive burden regret. Topological verification ties all valid gate assignments, while an oracle whitelist subject to retaining at least 25% of the design space leaves positive regret in 30 of 52 cells. Design-space optimization and feedback-guided search with frozen language models reach the high-regret regions; on a matched zero-regret control, additional optimization instead reduces regret. At a matched modeled-cost budget, mitigation is budget- and library-dependent. An in-sample fixed rule spending $\alpha{=}0.34$ of the budget re-scoring top cheap-simulator designs achieves mean excess regret 0.016, compared with 0.305 for cheap-only and 0.034 for reference-only selection, improving on cheap-only in 330 of 360 cells. More broadly, any workflow optimized through a simulator or digital twin can satisfy its verifier yet perform poorly on the intended physical objective. Fidelity regret provides a transferable template for detecting this gap before optimization and comparing budget-aware remedies across scientific domains.
Comments: NeurIPS 2026 Workshop "Sim2Science: ML with Imperfect Scientific Models"
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.22263 [cs.AI]
  (or arXiv:2611.22263v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.22263

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Noah Flynn [view email]
[v1] Tue, 17 Nov 2026 06:11:07 UTC (480 KB)

来源:arXiv cs.AI · arxiv.org