Unsupervised Synthetic Image Attribution: Alignment and Disentanglement
Liu, Zongfang
Liu, Zongfang
Author
Supervisor
Department
Computer Vision
Embargo End Date
2025-05-30
Type
Thesis
Date
2025
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
As the quality of synthetic images improves, identifying the underlying concepts of model generated images is becoming increasingly crucial for copyright protection and ensuring model transparency. Existing methods achieve this attribution goal by training models using annotated pairs of synthetic images and their original training sources. However, obtaining such paired supervision is challenging, as it requires either well-designed synthetic concepts or precise annotations from millions of training sources. To eliminate the need for costly paired annotations, in this thesis, we explore the possibility of unsupervised synthetic image attribution. We propose a simple yet effective unsupervised method called Alignment and Disentanglement. Specifically, we begin by performing basic concept alignment using contrastive self-supervised learning. Next, we enhance the model’s attribution ability by promoting representation disentanglement with the Infomax loss. This approach is motivated by an interesting observation: contrastive self-supervised models, such as MoCo and DINO, are inherently capable of simple cross-domain alignment. We offer a theoretical explanation of how alignment and disentanglement can approximate the concept-matching process by decomposing the targets of canonical correlation analysis. In real-world benchmarks, AbC, we show that our unsupervised method surprisingly outperforms the supervised methods. As a starting point, we expect our intuitive insights and
experimental findings can provide a fresh perspective on this challenging task.
Citation
Zongfang Liu, “Unsupervised Synthetic Image Attribution: Alignment and Disentanglement,” Master of Science thesis, Computer Vision, MBZUAI, 2025.
Source
Conference
Keywords
Unsupervised Image Attribution, Canonical Correlation Analysis, Disentanglement