Score: 2

Demystifying Foreground-Background Memorization in Diffusion Models

Published: August 16, 2025 | arXiv ID: 2508.12148v1

By: Jimmy Z. Di , Yiwei Lu , Yaoliang Yu and more

BigTech Affiliations: Massachusetts Institute of Technology

Potential Business Impact:

Finds if AI copied parts of its training pictures.

Diffusion models (DMs) memorize training images and can reproduce near-duplicates during generation. Current detection methods identify verbatim memorization but fail to capture two critical aspects: quantifying partial memorization occurring in small image regions, and memorization patterns beyond specific prompt-image pairs. To address these limitations, we propose Foreground Background Memorization (FB-Mem), a novel segmentation-based metric that classifies and quantifies memorized regions within generated images. Our method reveals that memorization is more pervasive than previously understood: (1) individual generations from single prompts may be linked to clusters of similar training images, revealing complex memorization patterns that extend beyond one-to-one correspondences; and (2) existing model-level mitigation methods, such as neuron deactivation and pruning, fail to eliminate local memorization, which persists particularly in foreground regions. Our work establishes an effective framework for measuring memorization in diffusion models, demonstrates the inadequacy of current mitigation approaches, and proposes a stronger mitigation method using a clustering approach.

Country of Origin
πŸ‡¨πŸ‡¦ πŸ‡ΊπŸ‡Έ Canada, United States

Page Count
20 pages

Category
Computer Science:
CV and Pattern Recognition