arXiv cs.AI· Jie Huang, Cuiling Lan, Qize Yu, Xiulian Peng, Yan Lu·· 14 小时前AI 评分38
MemSteer：将用户历史转化为智能体记忆优化的监督信号
MemSteer: Turning User Histories into Supervision for Agent Memory Optimization
AI 导读
MemSteer 能把无标注的用户历史转化为带证据支撑的问答监督，用于优化智能体记忆设计。基于该监督的 MemSteer-Opt 在不更新模型参数的情况下修改记忆存储、检索与上下文构建，在 ATM-Bench 上相较 Base-BM25 将 Question Score 提升 13.4 分，在 RHELM 上无外部数据与有外部数据分别提升 4.1 和 6.6 分。
正文 · 原文
Abstract:An effective agent memory system should adapt to a user's history and task requirements. Human-annotated supervision can guide this adaptation by providing feedback for evaluating and improving memory designs. Yet suitable labels are rarely available for new user histories, and manually constructing QA pairs and supporting evidence is costly. We introduce MemSteer, which turns unlabeled user histories into question-answer (QA) supervision with supporting evidence for memory-design optimization. MemSteer constructs an evidence-grounded answer plan that specifies how facts across related records jointly support a reference answer. The plan guides question generation, while source-grounded auditing and blind answer recovery check the resulting annotations. To study the utility of this supervision, we develop MemSteer-Opt, which combines QA-based design evaluation with evidence-guided failure diagnosis to revise memory store, retrieval, and context construction without updating model parameters. Experiments on ATM-Bench and RHELM show that this supervision is competitive with official-QA-guided optimization. Relative to Base-BM25, MemSteer-Opt improves ATM-Bench Question Score by 13.4 points and RHELM accuracy by 4.1 and 6.6 points without and with external data, respectively. The supervision also benefits two existing optimizers on ATM-Bench, while new-user experiments demonstrate continued design optimization using supervision generated from the target histories.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.39276 [cs.AI] |
| (or arXiv:2610.39276v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.39276 arXiv-issued DOI via DataCite |
Submission history
From: Jie Huang [view email]
[v1]
Thu, 29 Oct 2026 03:28:49 UTC (446 KB)
来源：arXiv cs.AI · arxiv.org