Causal Memory Policy: Making Memory Utility Identifiable by Intervening on Retrieval
作者: Arman Behnam, Binghui Wang
分类: cs.AI
发布日期: 2026-10-01 (更新: 2026-10-02)
💡 一句话要点
提出因果记忆策略以解决记忆效用识别问题
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 因果推断 记忆增强 语言模型 效用识别 机器学习
📋 核心要点
- 现有的记忆增强模型在识别未检索记忆的效用时存在显著不足,导致无法有效决策。
- 本文提出因果记忆策略(CMP),通过干预检索过程来识别记忆的效用,解决了传统方法的局限性。
- CMP在LongMemEval和LoCoMo数据集上显著提升了必要记忆的识别能力,AUC从0.54提升至0.66。
📝 摘要(中文)
记忆增强的大型语言模型必须决定保留哪些记忆,现有系统通过估计每个记忆对任务性能的影响来实现。然而,这些估计完全依赖于已检索的记忆。当记忆从未被检索时,存储级干预产生相同结果,导致其效用无法识别。为此,本文提出因果记忆策略(CMP),通过干预检索本身来恢复识别能力。CMP通过自归一化逆倾向加权在平衡分配设计下估计记忆效用。我们证明了通过检索的记忆效用的因果分解、估计器的无偏性和精确方差,以及不可逆操作下的最优决策规则。实验证明,在LongMemEval上54%的必要记忆和在LoCoMo上67%的必要记忆识别失败,CMP将必要与非必要记忆的区分度从0.54提升至0.66 AUC。
🔬 方法详解
问题定义:本文旨在解决记忆增强模型中未检索记忆效用无法识别的问题。现有方法仅依赖已检索记忆的效果评估,导致存储级干预无法揭示未检索记忆的真实效用。
核心思路:因果记忆策略(CMP)通过直接干预检索过程,保留固定数量的上下文槽以采样已知倾向的记忆,从而实现对记忆效用的有效识别。
技术框架:CMP的整体架构包括记忆采样、倾向估计和效用评估三个主要模块。首先,通过干预检索过程采样记忆;其次,使用自归一化逆倾向加权方法估计记忆效用;最后,基于平衡分配设计进行决策。
关键创新:CMP的核心创新在于通过检索干预实现记忆效用的因果识别,克服了传统方法的局限性,使得未检索记忆的效用得以评估。
关键设计:CMP采用自归一化逆倾向加权作为效用估计的核心技术,确保了估计的无偏性和精确性。此外,设计了平衡分配策略以优化决策规则,适应不可逆操作的需求。
🖼️ 关键图片
📊 实验亮点
实验结果显示,CMP在LongMemEval和LoCoMo数据集上,必要记忆的识别能力显著提升,AUC从0.54提高至0.66,表明该方法在记忆效用识别方面的有效性和优势。
🎯 应用场景
该研究的潜在应用领域包括自然语言处理、智能对话系统和记忆增强的机器学习模型。通过有效识别记忆的效用,CMP能够提升模型在复杂任务中的表现,具有重要的实际价值和未来影响。
📄 摘要(原文)
Memory-augmented large language models must decide which memories to retain, and recent systems do so by estimating each memory's effect on task performance. However, these estimates rely entirely on retrieved memories. When a memory is never retrieved, store-level interventions produce identical outcomes, leaving its utility unidentified. This is a retrieval-level positivity violation, invisible to diagnostics that examine only memory operations. We introduce Causal Memory Policy (CMP), a causal framework that restores identification by intervening on retrieval itself, reserving a fixed number of context slots for memories sampled with known propensities. CMP estimates memory utility by self-normalized inverse propensity weighting under a balanced assignment design. We prove the causal factorization of memory utility through retrieval, the unbiasedness and exact variance of the estimator, and the optimal decision rule under irreversible operations. Empirically, identification fails for 54% of required memories on LongMemEval and 67% on LoCoMo, and the failure persists in a deployed memory system. CMP improves discrimination between required and non-required memories from 0.54 to 0.66 AUC. Finally, we show that identified memory utility alone is insufficient for retention decisions: per-query utility reaches 0.78 AUC on the query for which it is estimated, yet no aggregation available to a retention policy predicts a memory's value on unseen queries. Code is available at: https://anonymous.4open.science/r/cmp-release-D0C3/.