Redundant Semantic Environment Filling via Misleading-Learning for Fair Deepfake Detection

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: He, Xinan, Zhou, Yue, Hu, Shu, Li, Bin, Huang, Jiwu, Ding, Feng
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916997451743232
author He, Xinan
Zhou, Yue
Hu, Shu
Li, Bin
Huang, Jiwu
Ding, Feng
author_facet He, Xinan
Zhou, Yue
Hu, Shu
Li, Bin
Huang, Jiwu
Ding, Feng
contents Detecting falsified faces generated by Deepfake technology is essential for safeguarding trust in digital communication and protecting individuals. However, current detectors often suffer from a dual-overfitting: they become overly specialized in both specific forgery fingerprints and particular demographic attributes. Critically, most existing methods overlook the latter issue, which results in poor fairness: faces from certain demographic groups, such as different genders or ethnicities, are consequently more difficult to reliably detect. To address this challenge, we propose a novel strategy called misleading-learning, which populates the latent space with a multitude of redundant environments. By exposing the detector to a sufficiently rich and balanced variety of high-level information for demographic fairness, our approach mitigates demographic bias while maintaining a high detection performance level. We conduct extensive evaluations on fairness, intra-domain detection, cross-domain generalization, and robustness. Experimental results demonstrate that our framework achieves superior fairness and generalization compared to state-of-the-art approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2405_15173
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Redundant Semantic Environment Filling via Misleading-Learning for Fair Deepfake Detection
He, Xinan
Zhou, Yue
Hu, Shu
Li, Bin
Huang, Jiwu
Ding, Feng
Computer Vision and Pattern Recognition
Detecting falsified faces generated by Deepfake technology is essential for safeguarding trust in digital communication and protecting individuals. However, current detectors often suffer from a dual-overfitting: they become overly specialized in both specific forgery fingerprints and particular demographic attributes. Critically, most existing methods overlook the latter issue, which results in poor fairness: faces from certain demographic groups, such as different genders or ethnicities, are consequently more difficult to reliably detect. To address this challenge, we propose a novel strategy called misleading-learning, which populates the latent space with a multitude of redundant environments. By exposing the detector to a sufficiently rich and balanced variety of high-level information for demographic fairness, our approach mitigates demographic bias while maintaining a high detection performance level. We conduct extensive evaluations on fairness, intra-domain detection, cross-domain generalization, and robustness. Experimental results demonstrate that our framework achieves superior fairness and generalization compared to state-of-the-art approaches.
title Redundant Semantic Environment Filling via Misleading-Learning for Fair Deepfake Detection
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2405.15173