arXiv cs.CL· Lingyun Song, Ziyao Chen, Kang Pan, Xiaolin Han, Yifei Zhang, Xiaoqi Wang, Yudai Pan, Xuequn Shang·· 11 小时前AI 评分29
ROSE：面向多模态大模型终身知识编辑的风险感知正交子空间导航框架
ROSE: Risk-Aware Orthogonal Subspace Navigation for Lifelong Knowledge Editing in Multimodal Large Language Models
AI 导读
研究者提出 ROSE，一个面向多模态大模型终身知识编辑的风险感知正交子空间导航框架，通过将参数更新限制在已编辑子空间的正交补空间来抑制灾难性遗忘。该框架引入动态计算的风险掩码保护关键事实参数子空间，并在推理时用原型锚定的语义门控与扰动驱动的子空间融合激活编辑知识。在两个主流多模态基准的 5 项指标上，ROSE 与 SOTA 方法相当，并在超过 1,000 次连续更新中保持稳定，代码已开源。
正文 · 原文
Abstract:Multimodal Large Language Models (MLLMs) require continuous factual updates to stay current, yet lifelong knowledge editing remains a daunting challenge. The primary obstacle is the severe Locality Erosion and Catastrophic Forgetting, where sequential edits inevitably interfere with pre-trained capabilities and prior updates. In this paper, we propose \textbf{ROSE}, a \textbf{R}isk-aware \textbf{O}rthogonal \textbf{S}ubspace navigation framework for lifelong knowledge \textbf{E}diting. ROSE addresses forgetting by aiming to restrict parameter updates to the orthogonal complement of previously edited subspaces. Crucially, it introduces a risk-guided mechanism that utilizes a dynamically computed risk mask to safeguard specific factual parameter subspaces essential for model stability. During inference, ROSE employs a multi-granularity knowledge integration strategy, featuring prototype-anchored semantic gating and perturbation-driven subspace fusion to precisely activate edited knowledge for relevant queries while limiting interference on unrelated inputs. Comprehensive experiments on two prominent multimodal benchmarks show that ROSE is competitive with state-of-the-art approaches across five key metrics, while maintaining outstanding stability over $1,000$ sequential updates. Our code is available at this https URL.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2611.09770 [cs.CL] |
| (or arXiv:2611.09770v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.09770 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ziyao Chen [view email]
[v1]
Fri, 6 Nov 2026 17:32:49 UTC (480 KB)
来源：arXiv cs.CL · arxiv.org