An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency
- ID: dd91425d
- 原文链接: https://arxiv.org/abs/2609.22043
- PDF: https://arxiv.org/pdf/2609.22043v1
- 作者: Yiming Zhang, Jinghong Zhang, Haoran Zhao, Yiren Ma, Chunlei Zhao
- 日期: 2026-09-18
- 更新: 2026-09-18
- 分类: agents
- arXiv 分类: c, s, ., C, L
- 来源类型: paper
- 标签: memory-decision, rag, hallucination, interpretability, zero-parameter
- 质量评分: 4/5
- 抓取时间: 2026-09-22T04:22:47Z
中文导读
LLM 记忆系统研究集中在高效检索，对'检索到的记忆是否该信'关注很少当记忆库中存在冲突立场时，标准 RAG 会盲目注入记忆并放大幻觉：对易受记忆注入影响的模型，冲突记忆下 RAG 幻觉率显著高于无记忆基线受前额叶皮层记忆信号机制启发，论文提出记忆决策层 MDL：一个位于检索与生成之间的零参数决策控制器，核心是三信号互补编码器，经 QR 正交子空间投影融合相关性可靠性与任务风险，加上元工作记忆信号，形成可解释的决策表示来量化记忆可信度；并显式解耦置信度与一致性论文在冲突记忆场景下验证其对幻觉放大的抑制
为什么值得关注
零参数记忆决策层 MDL 融合相关性/可靠性/任务风险三信号量化记忆可信度，冲突记忆下抑制 RAG 幻觉放大
关键信息
- 论文标题：An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency
- 作者：Yiming Zhang, Jinghong Zhang, Haoran Zhao, Yiren Ma, Chunlei Zhao
- arXiv：https://arxiv.org/abs/2609.22043
- PDF：https://arxiv.org/pdf/2609.22043v1
- 发布时间：2026-09-18
- 最近更新：2026-09-18
- arXiv 主分类：cs.CL
- arXiv 全部分类：c, s, ., C, L
- 评论：17 pages, 6 figures, 10 tables
- 关联标签：memory-decision, rag, hallucination, interpretability, zero-parameter
English Abstract
Memory systems for large language models have focused predominantly on efficient retrieval, whereas the decision of whether retrieved memories should be trusted has received comparatively little attention. When the memory store contains conflicting positions, standard retrieval-augmented generation (RAG) blindly injects memories and amplifies hallucinations: in models susceptible to memory injection, the RAG hallucination rate under conflicting memories is markedly higher than that of a memory-free baseline. Inspired by memory signaling mechanisms in the prefrontal cortex, we propose the Memory Decision Layer (MDL), a zero-parameter memory decision controller situated between the retrieval and generation stages. Its core is a three-signal complementary encoder that fuses relevance, reliability, and task risk through QR-based orthogonal subspace projection and a meta-working-memory signal into an interpretable decision representation that quantifies the trustworthiness of retrieved memories. Building on this encoder, MDL explicitly decouples confidence from consistency and introduces risk inversion and explicit abstention. Evaluations on mainstream large language models and multiple open-source datasets show that MDL reduces the hallucination rate under conflicting memories by about 56.04% in general scenarios and approaches zero hallucination in high-risk scenarios. The controller is fully white-box: it relies purely on geometric operations, requires no trained parameters, and adds only about 0.14 ms per decision -- roughly 50x faster than the embedding-retrieval step that precedes it and four to five orders of magnitude faster than an LLM self-evaluation call.
English Summary
Memory systems for large language models have focused predominantly on efficient retrieval, whereas the decision of whether retrieved memories should be trusted has received comparatively little attention. When the memory store contains conflicting positions, standard retrieval-augmented generation (RAG) blindly injects memories and amplifies hallucinations: in models susceptible to memory injection, the RAG hallucination rate under conflicting memories is markedly higher than that of a memory-free baseline. Inspired by memory signaling mechanisms in the prefrontal cortex, we propose the Memory Decision Layer (MDL), a zero-parameter memory decision controller situated between the retrieval and generation stages.
Obsidian Notes
- 内容由
opencli arxiv paper拉取 arXiv 元数据与摘要生成。 - 中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。