跳到正文
arXiv cs.AI· Qingcan Kang, Mingyang Liu, Shixiong Kai, Tao Zhong, Mingxuan Yuan·· 4 小时前AI 评分34

RegimeCouncil:为 LLM 智能体记忆冲突治理设计的配对证据路由

RegimeCouncil: Paired-Evidence Routing For Memory Conflict Governance In LLM Agents

AI 导读

RegimeCouncil 是一个轻量级配对证据路由器,让 LLM 智能体在本地与全局两种记忆抑制范围之间在线仲裁,无需重新训练模型。它通过 Local Advisor 跟踪单个事实的最新结果、Global Advisor 汇总所有事实的近期结果,并由 Meta Governor 默认跟随本地、仅在全局池化累积严格正近期效用边际时切换。

正文 · 原文

View PDF HTML (experimental)

Abstract:Language agents rely on long-term memory to personalize responses across extended interactions, yet stored facts quietly go stale as the world changes, and an agent often learns that using memory hurt only through delayed feedback after acting on an outdated record. Before the memory store is updated, how far should that negative feedback propagate: should it suppress memory use only on future queries for that same fact (a local policy), or also across other stored facts (a global policy)? Existing memory systems implicitly fix one of these two sharing scopes at design time. We formulate this online scope choice under hidden validity and delayed feedback as memory conflict governance, and prove that neither fixed policy is sufficient on its own. We then propose RegimeCouncil, a lightweight paired-evidence router that retains both scopes as advisors and arbitrates between them online without retraining the LLM. Using revealed comparisons between answering with memory and answering without memory (the no-memory anchor), a Local Advisor tracks the latest outcome on the queried fact alone, while a Global Advisor pools recent outcomes across all stored facts. A Meta Governor follows the Local Advisor by default to avoid cross-fact interference and switches to the Global Advisor only when global pooling accumulates a strictly positive recent utility margin, guaranteeing finite-time switching under delayed feedback. Across LongMemEval update streams and real-world historical drift on TempLAMA and TAQA with Qwen3-30B-A3B and GPT-4o, RegimeCouncil outperforms the better fixed-scope policy by 5.2-7.3 points on primary disagreement episodes and across all 30 held-out settings, while ablations show that removing any of its three modules reduces accuracy by 2-7 points.
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.22113 [cs.AI]
  (or arXiv:2611.22113v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.22113

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Qingcan Kang [view email]
[v1] Tue, 17 Nov 2026 02:56:43 UTC (1,494 KB)

来源:arXiv cs.AI · arxiv.org