Intuition with Evidence: Adaptive Dual Memory for Multimodal Agents

As multimodal agents engage in sustained interactions, long-term memory serves two distinct but complementary functions: fine-grained evidence recall and internalized interpretation for global understanding.

随着多模态智能体进行持续的交互，长期记忆发挥着两种截然不同但又互补的功能：细粒度的证据回溯和用于全局理解的内化解释。

Although episodic and latent memory address these demands through explicit historical records and internalized experience, neither paradigm alone can accommodate the diverse memory affinities across tasks, owing to limitations in episodic evidence coverage and integration, and in latent factual grounding.

尽管情景记忆和潜在记忆通过明确的历史记录和内化经验满足了这些需求，但由于情景记忆在证据覆盖和整合方面的局限性，以及潜在记忆在事实基础方面的不足，单一范式都无法适应不同任务中多样的记忆亲和性。

To bridge this divide, we propose a dual-memory framework, unifying a model-native latent memory as an intuitive prior with a multi-indexed episodic memory for exact evidence reconstruction via retrieval-conditioned joint alignment.

为了弥合这一鸿沟，我们提出了一个双重记忆框架，将模型原生的潜在记忆作为直觉先验，与多索引情景记忆相结合，通过检索条件下的联合对齐实现精确的证据重构。

At inference, the framework adaptively composes these memories through a training-free routing mechanism driven by Jensen-Shannon Divergence (JSD) sensitivity and confidence proxies for query-dependent memory affinity.

在推理阶段，该框架通过一种无需训练的路由机制自适应地组合这些记忆，该机制由 Jensen-Shannon 散度 (JSD) 敏感度和置信度代理驱动，以实现针对查询的记忆亲和性选择。

Extensive evaluations across multimodal benchmarks demonstrate state-of-the-art performance, outperforming single-paradigm baselines overall (+4.67% over vanilla Episodic), with further analysis highlighting episodic strengths in evidence-localized tasks and latent contributions to integrating dispersed information and compensating for imperfect retrieval.

在多模态基准测试上的广泛评估证明了该方法达到了最先进的性能，总体上优于单一范式的基线模型（比原始情景记忆提升了 4.67%）。进一步分析强调了情景记忆在证据定位任务中的优势，以及潜在记忆在整合分散信息和补偿不完美检索方面的贡献。