MissionHD: Hyperdimensional Refinement of Distribution-Deficient Reasoning Graphs for Video Anomaly Detection

TL;DR

This paper introduces MissionHD, a hyperdimensional computing-based method for refining LLM-generated reasoning graphs in video anomaly detection, improving their structure without relying on traditional distribution modeling.

Contribution

It proposes HDC-GSR, a novel hyperdimensional refinement framework that optimizes graph representations directly in a high-dimensional space, tailored for video anomaly detection tasks.

Findings

Achieves consistent performance improvements on benchmark datasets.

Effectively refines reasoning graphs without distribution modeling.

Demonstrates the utility of hyperdimensional computing in graph refinement.

Abstract

LLM-generated reasoning graphs, referred to as mission-specific graphs (MSGs), are increasingly used for video anomaly detection (VAD) and recognition (VAR). However, they are typically treated as fixed despite being generic and distribution-deficient. Conventional graph structure refinement (GSR) methods are ill-suited to this setting, as they rely on learning structural distributions that are absent in LLM-generated graphs. We propose HDC-constrained Graph Structure Refinement (HDC-GSR), a new paradigm that directly optimizes a decodable, task-aligned graph representation in a single hyperdimensional space without distribution modeling. Leveraging Hyperdimensional Computing (HDC), our framework encodes graphs via binding and bundling operations, aligns the resulting graph code with downstream loss, and decodes edge contributions to refine the structure. We instantiate this approach as MissionHD for weakly supervised VAD/VAR and demonstrate consistent performance gains on benchmark datasets.