Multimodal Knowledge Edit-Scoped Generalization for Online Recursive MLLM Editing

2026-07-02

AI Summary

Existing online multimodal knowledge editing methods struggle to precisely control the semantic boundaries of edits, often resulting in insufficient cross-modal generalization or unintended interference with unrelated knowledge. To address this, this work proposes the Edit-Scoped Generalization framework, featuring the novel ScopeEdit editor. ScopeEdit employs a dual-branch mechanism—modal-local absorption and evidence-gated shared generalization—to achieve scope-disentangled knowledge injection within an orthogonal low-rank subspace. By integrating a Sherman–Morrison recursively maintained preconditioner with vision–text alignment evidence, the method effectively regulates cross-modal propagation at constant, single-edit computational cost. Experiments demonstrate that the framework significantly improves the trade-off between in-scope cross-modal transfer and out-of-scope locality across diverse benchmarks, long editing streams, and real-world scenarios, while maintaining both reliability and efficiency.

Abstract

Online multimodal knowledge editing requires injecting a continual stream of visual-textual corrections into multimodal large language models (MLLMs) with bounded overhead and minimal disruption to unrelated behaviors. Existing editors mainly emphasize edit reliability and long-horizon stability, but rarely control the semantic boundary of each edit. Our pilot analyses of post-edit behaviors and internal neuronal activities reveal a scope gap behind reliable edits: instance-level success neither guarantees transfer to valid cross-modal variants nor prevents leakage to unrelated inputs, while edit-related cross-modal responses concentrate in deeper semantic layers. Therefore, we formulate Edit-Scoped Generalization, reframing online MLLM editing from merely correcting an instance to controlling the propagation boundary of each edit. To this end, we propose ScopeEdit, a scope-aware online editor that decomposes each update into a modality-local absorption branch and an evidence-gated shared generalization branch. The local branch supports stable edit absorption, whereas the shared branch enables cross-modal propagation only when visual and textual evidence are sufficiently aligned. Both branches perform scope-separated write geometries in orthogonal low-rank spaces and maintain branch-wise preconditioners via Sherman--Morrison recursions, yielding constant per-edit overhead. Extensive experiments across diverse benchmarks, long-horizon edit streams, MLLM backbones, real-world VLKEB scenarios, and complex vision-language architectures show that ScopeEdit consistently improves the trade-off between in-scope cross-modal transfer and out-of-scope locality, while preserving edit reliability, stability and online efficiency. Our code is available at https://github.com/lab-klc/ScopeEdit.

Problem

• multimodal knowledge editing
• edit-scoped generalization
• cross-modal transfer
• semantic boundary
• online MLLM editing

Innovation

• Edit-Scoped Generalization
• ScopeEdit
• Multimodal Knowledge Editing
• Cross-Modal Propagation
• Low-Rank Orthogonal Update