Multimodal Agent MemorySubmit a paper

EventMemAgent

EventMemAgent: Hierarchical Event-Centric Memory for Online Video Understanding with Adaptive Tool Use

Siwei Wen, Zhangcheng Wang, Xingjian Zhang, Lei Huang, Wenjun Wu · 2026-02-17

Read the paperPDF

Management
T2 Dynamic Memory
Memory changes during deployment through a stable management policy.
Primary modality
Video & streaming
Also: text
Representation
Hybrid
Distinct operative representation paths are explicitly linked for downstream retrieval, reasoning, or control.

How experience becomes memory

Each path is one memory component, traced from what is stored to how it is found again.

  1. event-granular short-term frame buffer

    1. Represented asRetained perceptual source evidenceSource- and modality-specific
    2. Stored inFlat sequential records
    3. Retrieved byMetadata / spatiotemporal retrieval
    4. Source evidenceSource directly available
    Short-term memory detects event boundaries and reservoir-samples frames in a bounded buffer.Full paper: 3.2.2. Structured Long-Term Memory · PDF p. 4 · Figure 1

  2. long-term structured event archive

    1. Represented asSpatial, temporal & cross-modal structured memoryCross-represented explicit
    2. Stored inHierarchical multi-tier store
    3. Retrieved byMetadata / spatiotemporal retrieval, Planner / tool-assisted retrieval
    4. Source evidenceSource linked & recoverable
    Past observations are archived event by event for later active evidence capture.Full-paper method / memory representation and retrieval · PDF p. 2 · Figure 1

How memory changes (T2 operations)

  • Forgetting · Discrete eviction

    Event-granular reservoir sampling bounds the short-term frame buffer. For EventMemAgent, this specifically supports explicit deletion or eviction of stored items of event-granular short-term frame buffer.

  • Consolidation · Episodic / hierarchical summarization

    Detected events are consolidated into the long-term archive. For EventMemAgent, this specifically supports episodic or hierarchical summarization of event-granular short-term frame buffer.

Related systems