Proto-LeakNet: Towards Signal-Leak Aware Attribution in Synthetic Human Face Imagery

2025-11-06

AI Summary

Addressing the challenges of source attribution and authenticity verification for diffusion-generated facial images, this paper proposes Proto-LeakNet—a signal-leakage-aware framework that enables interpretable, retraining-free identification of both known and unknown generators. Methodologically, it integrates partial latent-space forward resimulation, a temporal attention encoder, and a feature-weighted prototype head to construct discriminative latent representations. A unified pipeline combines closed-set classification with density-based open-set evaluation to jointly tackle source identification and unknown-generator detection. On standard closed-set benchmarks, Proto-LeakNet achieves a Macro AUC of 98.13%, significantly outperforming state-of-the-art methods, while demonstrating strong robustness against common post-processing operations. This work establishes a novel paradigm for provenance tracing of generative content.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

The growing sophistication of synthetic image and deepfake generation models has turned source attribution and authenticity verification into a critical challenge for modern computer vision systems. Recent studies suggest that diffusion pipelines unintentionally imprint persistent statistical traces, known as signal leaks, within their outputs, particularly in latent representations. Building on this observation, we propose Proto-LeakNet, a signal-leak-aware and interpretable attribution framework that integrates closed-set classification with a density-based open-set evaluation on the learned embeddings, enabling analysis of unseen generators without retraining. Operating in the latent domain of diffusion models, our method re-simulates partial forward diffusion to expose residual generator-specific cues. A temporal attention encoder aggregates multi-step latent features, while a feature-weighted prototype head structures the embedding space and enables transparent attribution. Trained solely on closed data and achieving a Macro AUC of 98.13%, Proto-LeakNet learns a latent geometry that remains robust under post-processing, surpassing state-of-the-art methods, and achieves strong separability between known and unseen generators. These results demonstrate that modeling signal-leak bias in latent space enables reliable and interpretable AI-image and deepfake forensics. The code for the whole work will be available upon submission.

Problem

• Attribution of synthetic face images to their source generators
• Detecting signal leaks in diffusion model outputs
• Robust attribution without retraining for unseen generators

Innovation

• Signal-leak-aware attribution framework for synthetic images
• Re-simulates partial diffusion to expose generator cues
• Uses temporal attention and prototype-based embedding space