arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.31436cs.AIcs.LGcs.MA
收藏

保守型元智能体搜索:非平稳环境下的架构自适应

Conservative Meta-Agent Search: Architecture Adaptation under Non-Stationarity

Ali Atiah Alzahrani

AI总结:

针对非平稳环境下智能体架构自适应问题,提出CMAS方法,在96个金融流测试中,其归一化平均代理动态-基准遗憾降至0.104,保留效用升至0.48,性能优于最强自适应基准。

摘要:

自动智能体系统设计通常在具有即时标量奖励的静态任务套件上进行评估,而部署的系统面临持续变化、延迟依赖反馈、组件相关故障以及架构变更成本高昂等问题。我们将元智能体架构搜索形式化为带有切换成本和硬风险约束的序列现任替换问题,并提出保守型元智能体搜索(Conservative Meta-Agent Search, CMAS)。CMAS检索已存档的架构并提出类型化图编辑,随后通过公共随机数下的配对影子重放,结合干预性组件信用对每个架构与现任架构进行评分。架构升级要求经验校准的、带方差惩罚的同时评分高于预先指定的阈值,之后进入模拟金丝雀阶段并执行预先指定的回退。由于回测无法证明元级技能,我们引入MetaInvest-Bench,这是一个 regime-switching 模拟器,其指定的工作者能力可计算出代理最优架构,具备泄漏控制重放、对抗性治理轨迹及五个阴性对照。金融流作为元级泛化的风险敏感测试平台;我们不主张任何超额收益(alpha)。在96个流中,CMAS将归一化平均代理动态-基准遗憾从0.116降至0.104,将保留效用从0.44提升至0.48,优于最强的自适应基准;当额外工作者无价值时,它会收缩至接近单智能体团队。其部署推理成本降低约6%,但包含搜索的 token 成本是所比较方法中最高的。观察到的每次更新的假升级率为3.4%,而名义目标为5%,密封重跑中为4.8%。该校准是经验性的:它涉及作者设计的基准上的更新级重放统计,既不保证全效用提升,也不保证模拟金丝雀阶段无损害,更不保证序列级错误控制。

英文摘要:

Automated agentic-system design is typically evaluated on static task suites with immediate scalar rewards. Deployed systems face persistent shift, delayed dependent feedback, correlated component failures, and costly architecture changes. We formalize meta-agent architecture search as sequential incumbent replacement with switching costs and hard risk constraints and introduce Conservative Meta-Agent Search (CMAS). CMAS retrieves archived architectures and proposes typed graph edits, then scores each against the incumbent by paired shadow replay under common random numbers, with interventional component credit. Promotion requires an empirically calibrated, variance-penalized simultaneous score above a pre-specified margin, and is followed by a simulated canary stage and pre-specified rollback. Since backtests cannot certify meta-level skill, we introduce MetaInvest-Bench, a regime-switching simulator whose specified worker competence makes a surrogate-optimal architecture computable, with leakage-controlled replay, adversarial governance tracks, and five negative controls. Financial streams serve as a risk-sensitive testbed for meta-level generalization; we claim no alpha. Across 96 streams, CMAS cuts normalized mean surrogate dynamic-oracle regret from 0.116 to 0.104 and raises held-out utility from 0.44 to 0.48 over the strongest adaptive baseline; when extra workers add no value, it contracts to near-single-agent teams. Its deployment inference cost is ~6% lower, though its search-inclusive token cost is the highest of the methods compared. The observed per-update false-promotion rate is 3.4% against a nominal 5% target, and 4.8% in a sealed rerun. The calibration is empirical: it concerns an update-level replay statistic on an author-designed benchmark, and guarantees neither full-utility improvement, nor absence of simulated canary-stage harm, nor sequence-level error control.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑