Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection

The generation of facial images via generative models has gained significant popularity, while the task of discriminating between authentic and synthetic faces has proven to be increasingly challenging. This challenge is exacerbated when novel generative models emerge, as it is difficult to obtain a...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiaoyong Liu, Pengcheng Song, Pei Lu, Yanjun Wang
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10754643/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846128577159888896
author Xiaoyong Liu
Pengcheng Song
Pei Lu
Yanjun Wang
author_facet Xiaoyong Liu
Pengcheng Song
Pei Lu
Yanjun Wang
author_sort Xiaoyong Liu
collection DOAJ
description The generation of facial images via generative models has gained significant popularity, while the task of discriminating between authentic and synthetic faces has proven to be increasingly challenging. This challenge is exacerbated when novel generative models emerge, as it is difficult to obtain a substantial number of images from these new models and the limited number of samples can undermine the accuracy of training. To tackle these issues, we introduce a few-shot deepfake detection approach based on meta-learning with relation embedding. Initially, we employ an embedding function to generate feature representations of the images. Subsequently, we convert the basic representations of feature maps into their corresponding self-correlation tensors, enabling us to learn the structural patterns inherent in these tensors. Finally, we utilize a learnable metric to classify the self-correlation tensors. Our model is trained using an initialization parameter meta-learning strategy, extracting generalizable knowledge through training on multiple interrelated tasks, thereby enhancing model performance. The effectiveness of our approach has been validated through experiments on the miniImageNet, Stanford-Dogs, and CUB-200-2011 datasets. Additionally, we conducted tests on a self-constructed deepfake face dataset, and the results indicated that the proposed method exhibits strong performance compared with other methods.
format Article
id doaj-art-eeb409b6c52c4bf2a73d321b6e2d7422
institution Kabale University
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-eeb409b6c52c4bf2a73d321b6e2d74222024-12-11T00:04:56ZengIEEEIEEE Access2169-35362024-01-011218013518014510.1109/ACCESS.2024.349935310754643Meta-Learning With Relation Embedding for Few-Shot Deepfake DetectionXiaoyong Liu0Pengcheng Song1Pei Lu2https://orcid.org/0000-0002-6284-5805Yanjun Wang3College of Physics and Electronic Information Engineering, Guilin University of Technology, Guilin, ChinaGuangxi Key Laboratory of Embedded Technology and Intelligent System, Guilin University of Technology, Guilin, ChinaGuangxi Key Laboratory of Embedded Technology and Intelligent System, Guilin University of Technology, Guilin, ChinaGuangxi Key Laboratory of Embedded Technology and Intelligent System, Guilin University of Technology, Guilin, ChinaThe generation of facial images via generative models has gained significant popularity, while the task of discriminating between authentic and synthetic faces has proven to be increasingly challenging. This challenge is exacerbated when novel generative models emerge, as it is difficult to obtain a substantial number of images from these new models and the limited number of samples can undermine the accuracy of training. To tackle these issues, we introduce a few-shot deepfake detection approach based on meta-learning with relation embedding. Initially, we employ an embedding function to generate feature representations of the images. Subsequently, we convert the basic representations of feature maps into their corresponding self-correlation tensors, enabling us to learn the structural patterns inherent in these tensors. Finally, we utilize a learnable metric to classify the self-correlation tensors. Our model is trained using an initialization parameter meta-learning strategy, extracting generalizable knowledge through training on multiple interrelated tasks, thereby enhancing model performance. The effectiveness of our approach has been validated through experiments on the miniImageNet, Stanford-Dogs, and CUB-200-2011 datasets. Additionally, we conducted tests on a self-constructed deepfake face dataset, and the results indicated that the proposed method exhibits strong performance compared with other methods.https://ieeexplore.ieee.org/document/10754643/Deepfake detectiondiffusion modelfew-shot learninggenerative adversarial networkmeta-learningself-correlation
spellingShingle Xiaoyong Liu
Pengcheng Song
Pei Lu
Yanjun Wang
Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
IEEE Access
Deepfake detection
diffusion model
few-shot learning
generative adversarial network
meta-learning
self-correlation
title Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
title_full Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
title_fullStr Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
title_full_unstemmed Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
title_short Meta-Learning With Relation Embedding for Few-Shot Deepfake Detection
title_sort meta learning with relation embedding for few shot deepfake detection
topic Deepfake detection
diffusion model
few-shot learning
generative adversarial network
meta-learning
self-correlation
url https://ieeexplore.ieee.org/document/10754643/
work_keys_str_mv AT xiaoyongliu metalearningwithrelationembeddingforfewshotdeepfakedetection
AT pengchengsong metalearningwithrelationembeddingforfewshotdeepfakedetection
AT peilu metalearningwithrelationembeddingforfewshotdeepfakedetection
AT yanjunwang metalearningwithrelationembeddingforfewshotdeepfakedetection