Bias-Averse Learning for Mitigating Source Dataset Bias in Avatar Fingerprinting
IEEE Access · 2026 · 14 · 34790-34801
DOI: 10.1109/access.2026.3669873 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Marcellino Sahadewa und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Institute of Information and Communications Technology Planning and Evaluation (IITP) under the Leading Generative AI Human Resources Development, Korean Government, National Research Foundation of Korea.
Abstract (Originalsprache)
Talking-head generation produces avatars from a source (the driver) and a target (the identity). In the context media forensics, standard deepfake detection can only distinguish between “real” vs. “fake” content, identifying the specific driver for privacy purposes remains a less explored task. Some prior work has already addressed this task, called ‘avatar fingerprinting’. However, existing methods face two significant problems: 1) The proposed model contained strong dataset bias. 2) The evaluation dataset failed to penalizes those biases. Based on our investigation, the dataset bias occurred due to a lacks of pixel information. As a result, it suffers from poor performance in same driver dataset settings. To this end, we introduce a bias-averse learning framework. We design a method to learn a dynamic signature from pixel-level dynamic information and use a domain orthogonality loss that drives the learned features to be data-agnostic, introducing data bias free. Then, we introduce a more rigorous ‘single-source’ evaluation protocol that reveals the reliance on shortcuts in prior methods. Our experiments show that our method outperforms prior work by up to 20% in both area under the curve (AUC) and average precision (AP) with this protocol.
Förderung
- Institute of Information and Communications Technology Planning and Evaluation (IITP) under the Leading Generative AI Human Resources Development
- Korean Government
- National Research Foundation of Korea
- Inha University Research Grant
Zitieren
APA
Sahadewa, M., Marchellus, M., & Park, I. K. (2026). Bias-Averse Learning for Mitigating Source Dataset Bias in Avatar Fingerprinting. IEEE Access, 14, 34790-34801. https://doi.org/10.1109/access.2026.3669873
Vancouver
Sahadewa M, Marchellus M, Park IK. Bias-Averse Learning for Mitigating Source Dataset Bias in Avatar Fingerprinting. IEEE Access. 2026;14:34790-34801. doi:10.1109/access.2026.3669873
BibTeX
@article{Sahadewa2026,
author = {Sahadewa, Marcellino and Marchellus, Matthew and Park, In Kyu},
title = {Bias-Averse Learning for Mitigating Source Dataset Bias in Avatar Fingerprinting},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {34790--34801},
doi = {10.1109/access.2026.3669873}
}
EUVIMED