Selective Knowledge Edit Reversal via Gated Singular Vector Shrinkage

• URL: http://arxiv.org/abs/2609.02091v1
• Date: Wed, 02 Sep 2026 04:32:07 GMT
• Title: Selective Knowledge Edit Reversal via Gated Singular Vector Shrinkage
• Abstract summary: We study selective reversal of edited knowledge, where the goal is to reverse targeted edited facts while preserving the remaining edited facts.<n>Based on the hypothesis that each edit is sparsely encoded within the dominant subspace of the edited matrix, we propose a spectral-based reversal framework.<n> Experiments across multiple settings demonstrate the effectiveness of our method in reversing selected edits while preserving unrelated edited facts.
• Score: 29.783345718823487
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Knowledge editing provides an efficient way to update factual knowledge in large language models. However, malicious edits may introduce safety risks, making it necessary to reverse undesirable editing effects. Existing reversal methods for parameter-modifying edits mainly focus on global removal, which may also erase beneficial edits that should be preserved. In this paper, we study selective reversal of edited knowledge, where the goal is to reverse targeted edited facts while preserving the remaining edited facts. Based on the hypothesis that each edit is sparsely encoded within the dominant subspace of the edited matrix, we propose a spectral-based reversal framework that locates edit-sensitive components within the dominant singular subspace of edited weights. Experiments across multiple settings demonstrate the effectiveness of our method in reversing selected edits while preserving unrelated edited facts. These results suggest that different edits are sparsely encoded within dominant singular components and can be separable when the number of edits is moderate, making selective spectral reversal a promising direction for locating edit-specific components and repairing edited language models.

Related papers

• Edit Where You Mean: Region-Aware Adapter Injection for Mask-Free Local Image Editing [28.295031615161136]
  Large diffusion transformers (DiTs) follow global editing instructions well but consistently leak local edits into unrelated regions.<n>We introduce AdaptEdit, a co-trained, instruction- and region-aware adapter framework.<n> AdaptEdit achieves state-of-the-art results, simultaneously outperforming mask-free and oracle-mask baselines.
  arXiv  Detail & Related papers  (2026-04-26T15:28:02Z)
• One Mask to Rule Them All: On Hidden Facts after Editing and How to Find Them [7.195629836132288]
  ROME and MEMIT update factual associations in transformer models by modifying weights.<n>We investigate whether edits rely on a common mechanism, regardless of which fact is modified.<n>Our finding that edits suppress rather than overwrite knowledge explains why ROME and MEMIT fail to propagate changes to related facts.
  arXiv  Detail & Related papers  (2026-04-18T16:18:08Z)
• CoEditor++: Instruction-based Visual Editing via Cognitive Reasoning [98.98349220451216]
  CoEditor++ is a training-free framework that decomposes editing into "what to edit" and "how to edit"<n>We show that CoEditor++ achieves state-of-the-art performance in both general editing and responsible editing tasks.<n>Our findings suggest the potential toward cognitive-centric instruction-based image editing.
  arXiv  Detail & Related papers  (2026-01-31T12:20:46Z)
• SpotEdit: Selective Region Editing in Diffusion Transformers [66.44912649206553]
  SpotEdit is a training-free diffusion editing framework that selectively updates only the modified regions.<n>By reducing unnecessary computation and maintaining high fidelity in unmodified areas, SpotEdit achieves efficient and precise image editing.
  arXiv  Detail & Related papers  (2025-12-26T14:59:41Z)
• FlowDC: Flow-Based Decoupling-Decay for Complex Image Editing [52.54102743380658]
  We propose FlowDC, which decouples the complex editing into multiple sub-editing effects and superposes them in parallel during the editing process.<n>FlowDC shows superior results compared with existing methods.
  arXiv  Detail & Related papers  (2025-12-12T09:08:39Z)
• Quantifying Edits Decay in Fine-tuned LLMs [17.377278510871843]
  This study investigates how fine-tuning affects knowledge editing.<n>We evaluate two state-of-the-art editing methods (MEMIT, AlphaEdit) and three fine-tuning approaches.<n>Our results show that edits decay after fine-tuning, with survival varying across configurations.
  arXiv  Detail & Related papers  (2025-11-08T04:58:03Z)
• CoEdPilot: Recommending Code Edits with Learned Prior Edit Relevance, Project-wise Awareness, and Interactive Nature [15.209899925736751]
  We propose CoEdPilot, an LLM-driven solution to recommend code edits by discriminating the relevant edits. CoEdPilot orchestrates multiple neural transformers to identify what and how to edit in the project regarding both edit location and edit content. Our experiments show that CoEdPilot can well predict the edits (i.e., predicting edit location with an accuracy of 70.8%-85.3%, and the edit content with an exact match rate of 41.8% and BLEU4 score of 60.7).
  arXiv  Detail & Related papers  (2024-08-03T10:23:05Z)
• Optimisation-Based Multi-Modal Semantic Image Editing [58.496064583110694]
  We propose an inference-time editing optimisation to accommodate multiple editing instruction types. By allowing to adjust the influence of each loss function, we build a flexible editing solution that can be adjusted to user preferences. We evaluate our method using text, pose and scribble edit conditions, and highlight our ability to achieve complex edits.
  arXiv  Detail & Related papers  (2023-11-28T15:31:11Z)
• Object-aware Inversion and Reassembly for Image Editing [61.19822563737121]
  We propose Object-aware Inversion and Reassembly (OIR) to enable object-level fine-grained editing. We use our search metric to find the optimal inversion step for each editing pair when editing an image. Our method achieves superior performance in editing object shapes, colors, materials, categories, etc., especially in multi-object editing scenarios.
  arXiv  Detail & Related papers  (2023-10-18T17:59:02Z)
• Learning Structural Edits via Incremental Tree Transformations [102.64394890816178]
  We present a generic model for incremental editing of structured data (i.e., "structural edits") Our editor learns to iteratively generate tree edits (e.g., deleting or adding a subtree) and applies them to the partially edited data. We evaluate our proposed editor on two source code edit datasets, where results show that, with the proposed edit encoder, our editor significantly improves accuracy over previous approaches.
  arXiv  Detail & Related papers  (2021-01-28T16:11:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.