让多模态大模型编辑更通用，通过识别不变因果轨迹提升准确性
Generalizable Multimodal Large Language Model Editing via Invariant Trajectory Learning
- 将多模态模型编辑视为分布外泛化问题，寻找跨模态提示下的不变因果路径
- 提出ODEdit框架，在3个方向上同时优化编辑可靠性、局部性和泛化能力
- 引入总变差惩罚稳定编辑轨迹，对环境变化更具鲁棒性，适合实际应用
知识编辑已成为高效修正大型语言模型（LLM）中错误或过时知识的关键技术。现有编辑方法依赖于参数或模块修改到输出的刚性映射，导致多模态大模型（MLLM）在泛化上存在局限。本文将MLLM编辑重新定义为分布外（OOD）泛化问题，目标是区分语义变化与事实变化，从而实现多样跨模态提示下的稳健编辑。该OOD问题的核心挑战在于识别能准确泛化的不变因果轨迹，同时抑制虚假相关。为此，我们提出ODEdit——一种即插即用的基于不变学习的框架，通过优化三元分布外风险目标，同时提升编辑的可靠性、局部性和泛化性。进一步提出一种编辑轨迹不变学习方法，将总变差正则项融入风险最小化目标，以稳定编辑轨迹免受环境变化影响。理论分析与大量实验验证了ODEdit的有效性。
原文摘要 · Abstract (English)
Knowledge editing emerges as a crucial technique for efficiently correcting incorrect or outdated knowledge in large language models (LLM). Existing editing methods rely on a rigid mapping from parameter or module modifications to output, which causes the generalization limitation in Multimodal LLM (MLLM). In this paper, we reformulate MLLM editing as an out-of-distribution (OOD) generalization problem, where the goal is to discern semantic shift with factual shift and thus achieve robust editing among diverse cross-modal prompting. The key challenge of this OOD problem lies in identifying invariant causal trajectories that generalize accurately while suppressing spurious correlations. To address it, we propose ODEdit, a plug-and-play invariant learning based framework that optimizes the tripartite OOD risk objective to simultaneously enhance editing reliability, locality, and generality.We further introduce an edit trajectory invariant learning method, which integrates a total variation penalty into the risk minimization objective to stabilize edit trajectories against environmental variations. Theoretical analysis and extensive experiments demonstrate the effectiveness of ODEdit.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。