Pioneering Reliable Assessment in Text-to-Image Knowledge Editing: Leveraging a Fine-Grained Dataset and an Innovative Criterion

TL;DR

This paper introduces a comprehensive framework for T2I knowledge editing, including a fine-grained dataset, a reliable evaluation criterion, and an effective editing method, to improve the accuracy and generalization of knowledge updates in diffusion models.

Contribution

It presents CAKE dataset, adaptive CLIP threshold criterion, and MPE editing approach, advancing the evaluation and effectiveness of T2I knowledge editing techniques.

Findings

MPE outperforms previous model editors in accuracy.

CAKE enables more fine-grained assessment of knowledge generalization.

Adaptive CLIP threshold improves reliability of editing evaluation.

Abstract

During pre-training, the Text-to-Image (T2I) diffusion models encode factual knowledge into their parameters. These parameterized facts enable realistic image generation, but they may become obsolete over time, thereby misrepresenting the current state of the world. Knowledge editing techniques aim to update model knowledge in a targeted way. However, facing the dual challenges posed by inadequate editing datasets and unreliable evaluation criterion, the development of T2I knowledge editing encounter difficulties in effectively generalizing injected knowledge. In this work, we design a T2I knowledge editing framework by comprehensively spanning on three phases: First, we curate a dataset \textbf{CAKE}, comprising paraphrase and multi-object test, to enable more fine-grained assessment on knowledge generalization. Second, we propose a novel criterion, \textbf{adaptive CLIP threshold}, to effectively filter out false successful images under the current criterion and achieve reliable editing evaluation. Finally, we introduce \textbf{MPE}, a simple but effective approach for T2I knowledge editing. Instead of tuning parameters, MPE precisely recognizes and edits the outdated part of the conditioning text-prompt to accommodate the up-to-date knowledge. A straightforward implementation of MPE (Based on in-context learning) exhibits better overall performance than previous model editors. We hope these efforts can further promote faithful evaluation of T2I knowledge editing methods.