跳到正文
arXiv cs.AI· Kangjun Noh, Soyu Kim, Kyungwoo Song·· 2 小时前AI 评分33

CISE:用不完美代理奖励实现可靠的自演化搜索

Reliable Self-Evolution with Imperfect Proxy Rewards

AI 导读

一项 arXiv 预印本研究提出 Conformal Interval-Driven Self-Evolution(CISE),用条件 conformal inference 和逐迭代在线密度比估计构建候选特定的奖励区间,降低不完美代理奖励带来的假阳性污染。

正文 · 原文

View PDF HTML (experimental)

Abstract:Large language model (LLM)-based self-evolving search is a promising approach to scientific discovery. However, high-fidelity evaluation of every candidate is prohibitively expensive in some domains. Self-evolving systems in such settings therefore rely on low-cost but imperfect proxy rewards, which may assign high scores to infeasible candidates. These false positives may contaminate both the final output and the feedback used to guide subsequent generations. This motivates statistically calibrated reward intervals for more reliable self-evolving search. We propose Conformal Interval-Driven Self-Evolution (CISE), which constructs candidate-specific reward intervals using conditional conformal inference and iteration-wise online density-ratio estimation. CISE uses conservative interval-based rewards for evolutionary feedback and returns candidates only when all required property intervals lie entirely within their respective feasible regions. We derive fixed-iteration coverage results under explicit assumptions of independence and covariate shift. We evaluate CISE on three self-evolving search tasks in materials science. In our experiments, all candidates returned by CISE are true positives under high-fidelity evaluation, whereas the baselines return more candidates but include false positives. These results highlight the value of a smaller, more precise shortlist when downstream validation budgets are limited. Our repository is available at this https URL.
Comments: Preprint
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2610.02975 [cs.AI]
  (or arXiv:2610.02975v2 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2610.02975

arXiv-issued DOI via DataCite

Submission history

From: Kangjun Noh [view email]
[v1] Fri, 2 Oct 2026 08:07:27 UTC (3,516 KB)
[v2] Tue, 17 Nov 2026 02:39:14 UTC (1,569 KB)

来源:arXiv cs.AI · arxiv.org