ScholarPulse 日报 2026-08-08

2026-08-08 学术简报：2 篇。合成临床基准的真实度需在效用约束下显式优化，而非仅依赖效用检查作为真实度充分证据，尤其在医疗隐私敏感场景中。

今日速览

序号	标题	来源	日期	主题	推荐等级
1	Improving the Realism of Synthetic Clinical Benchmarks Under Utility Constraints	arXiv	2026-08-06	AI-Agent	高
2	EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning	arXiv	2026-08-06	AI-Agent	高

重点论文与技术动态

1. Improving the Realism of Synthetic Clinical Benchmarks Under Utility Constraints

来源：arXiv
日期：2026-08-06
作者/机构：Omid Bazgir, Md Nasir, Jacob Hoffman, Yang Yang, Manu Agrawal, Anusua Trivedi
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.AI, cs.DB, cs.LG

一句话结论

合成临床基准的真实度需在效用约束下显式优化，而非仅依赖效用检查作为真实度充分证据，尤其在医疗隐私敏感场景中。

核心内容

基线基准结构不真实：缺失率79.44%，可操作行比例仅12.75%，38.94%患者无可操作措施，前三词集中度100%，真实度通过缺失结构、简单性、结构合理性及人口一致性测量。
两个确定性修订方法在保持效用下限前提下有效提升真实度指标，而朴素密集化控制保留不真实模板。
内部基准真实度与源保真度相关但不同，需分别优化而非视为同一目标。

方法与数据

基于Synthea生成患者数据，经电子健康记录工作流处理并模拟操作数据流水线。
摘要未明确数据规模。

价值判断

值得关注：医疗AI合成基准的真实度问题在隐私敏感环境中尤为关键，需避免仅通过效用检查掩盖结构缺陷。
可复用点：效用约束的真实度提升框架可推广至其他隐私敏感数据合成场景。
局限/待核查：摘要未提及其他局限，如计算成本或跨领域泛化性。

摘要

Synthetic clinical benchmarks for enterprise AI agents can pass existing utility checks and still remain structurally unrealistic, especially in privacy-sensitive healthcare settings where operational data are hard to access. We study how to improve such benchmarks without breaking the downstream utility checks already used in practice. We formulate benchmark revision as utility-constrained realism improvement: dataset changes should increase realism while staying above an operational utility floor. We instantiate this idea on a care-gap benchmark derived from Synthea-generated patients exercised through demonstration electronic health record workflows and then processed by the same downstream pipeline as operational data. Realism is measured through missingness structure, simplicity, structural plausibility, and population alignment. The baseline benchmark is extremely thin: sampled-pair missingness is 79.44%, only 12.75% of rows are actionable, 38.94% of patients have zero actionable measures, and top-three token concentration reaches 100.0%. Two deterministic revisions improve these panels while remaining above the current utility floor, whereas a naive densification control preserves unrealistic templating. We further show that internal benchmark realism and source fidelity to an aggregate operational reference are related but distinct objectives. These results suggest that synthetic benchmark quality should be optimized explicitly, with utility treated as one constraint rather than as sufficient evidence of realism.

2. EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning

来源：arXiv
日期：2026-08-06
作者/机构：Zishan Xu, Zhiyuan Yao, Yuxin Chen, Yifu Guo, Zhengxi Lu, Yuquan Lu
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.AI

一句话结论

EnvACE通过世界回放机制实现了无需外部环境交互的LLM智能体训练。

核心内容

该方法在训练中交替执行动作生成与环境响应回放，内部化动作与环境响应的动态关系，形成世界模型直接支持决策，在BFCL-v4等基准测试中显著优于环境扩展基线。

用世界回放替代外部环境交互，避免构建和验证成本。
策略交替生成工具调用并扮演环境生成响应，两者联合端到端优化。
内部化环境动态形成世界模型，支持决策无需外部交互。

方法与数据

方法：世界回放机制，策略交替动作与回放。
数据：在BFCL-v4、tau^2-Bench等基准测试集上验证，摘要未明确。

价值判断

值得关注：世界回放机制有效突破外部环境约束，为LLM代理训练提供新路径。
可复用点：内部化环境动态的框架可推广至其他智能体训练任务。
局限/待核查：摘要未明确提及局限。

摘要

Training large language model agents for long-horizon tool use typically relies on interactions with real or synthesized executable environments, whose construction and verification are costly, or on external simulators that are difficult to ground. We introduce EnvACE, an agentic reinforcement learning method that replaces external environment interaction during training with world rehearsal. The policy alternates between acting and rehearsal: it first generates a tool call, then plays the role of the environment to produce the response induced by that action, and conditions subsequent decisions on the rehearsed response. Both roles are jointly optimized end-to-end using task-success rewards. Through world rehearsal, the policy internalizes the relationship between actions and their environment responses in its parameters, yielding an agent world model that directly supports decision making. Across BFCL-v4, tau^2-Bench, VitaBench, and FinMCP-Bench, EnvACE achieves strong and transferable performance, outperforming environment-scaling baselines in the overall evaluation. Controlled studies further show that world rehearsal consistently improves policy learning across model scales. At test time, the internalized world model enables private rehearsal before committed execution, yielding further gains under a moderate rehearsal budget without additional external interaction. Our findings establish world rehearsal as a new path toward scaling LLM agent training beyond the constraints of external environments. Our code is publicly available at https://github.com/Within-yao/EnvACE.