Progressive Open Space Expansion for Open-Set Model Attribution
This work addresses the problem of intellectual property protection and malicious content supervision for synthetic images in real-world scenarios where unknown models are involved, representing an incremental advancement in open-set recognition for model attribution.
The paper tackles the challenge of Open-Set Model Attribution (OSMA), which involves attributing synthetic images to known source models while identifying those from unknown models, by proposing a Progressive Open Space Expansion (POSE) solution that simulates open-set samples with imperceptible traces. The method demonstrates superior performance over existing approaches in experiments on a constructed benchmark dataset.
Despite the remarkable progress in generative technology, the Janus-faced issues of intellectual property protection and malicious content supervision have arisen. Efforts have been paid to manage synthetic images by attributing them to a set of potential source models. However, the closed-set classification setting limits the application in real-world scenarios for handling contents generated by arbitrary models. In this study, we focus on a challenging task, namely Open-Set Model Attribution (OSMA), to simultaneously attribute images to known models and identify those from unknown ones. Compared to existing open-set recognition (OSR) tasks focusing on semantic novelty, OSMA is more challenging as the distinction between images from known and unknown models may only lie in visually imperceptible traces. To this end, we propose a Progressive Open Space Expansion (POSE) solution, which simulates open-set samples that maintain the same semantics as closed-set samples but embedded with different imperceptible traces. Guided by a diversity constraint, the open space is simulated progressively by a set of lightweight augmentation models. We consider three real-world scenarios and construct an OSMA benchmark dataset, including unknown models trained with different random seeds, architectures, and datasets from known ones. Extensive experiments on the dataset demonstrate POSE is superior to both existing model attribution methods and off-the-shelf OSR methods.