FakePolisher: Making DeepFakes More Detection-Evasive by Shallow Reconstruction

Huang, Yihao; Juefei-Xu, Felix; Wang, Run; Guo, Qing; Ma, Lei; Xie, Xiaofei; Li, Jianwen; Miao, Weikai; Liu, Yang; Pu, Geguang

doi:10.1145/3394171.3413732

Cited by 52 publications

(33 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the future, we will extend the proposed attack against other tasks, e.g., visual object tracking [28,29,30] and Deep-Fake evasion [31,32], and also in tandem with other natural modalities such as [33,34,35]. In addition, we can regard the adversarial bias field as a new kind of mutation for DNN testing [36,37,38,39].…”

Section: Discussionmentioning

confidence: 99%

Bias Field Poses a Threat to DNN-Based X-Ray Recognition

Tian

Guo

Juefei-Xu

et al. 2021

2021 IEEE International Conference on Multimedia and Expo (ICME)

Self Cite

View full text Add to dashboard Cite

Chest X-ray plays a key role in screening and diagnosis of many lung diseases including the COVID-19. Many works construct deep neural networks (DNNs) for chest X-ray images to realize automated and efficient diagnosis of lung diseases. However, bias field caused by the improper medical image acquisition process widely exists in the chest X-ray images while the robustness of DNNs to the bias field is rarely explored, posing a threat to the X-ray-based automated diagnosis system. In this paper, we study this problem based on the adversarial attack and propose a brand new attack, i.e., adversarial bias field attack where the bias field instead of the additive noise works as the adversarial perturbations for fooling DNNs. This novel attack poses a key problem: how to locally tune the bias field to realize high attack success rate while maintaining its spatial smoothness to guarantee high realisticity. These two goals contradict each other and thus has made the attack significantly challenging. To overcome this challenge, we propose the adversarial-smooth bias field attack that can locally tune the bias field with joint smooth & adversarial constraints. As a result, the adversarial X-ray images can not only fool the DNNs effectively but also retain very high level of realisticity. We validate our method on real chest X-ray datasets with powerful DNNs, e.g., ResNet50, DenseNet121, and MobileNet, and show different properties to the state-of-the-art attacks in both image realisticity and attack transferability. Our method reveals the potential threat to the DNN-based X-ray automated diagnosis and can definitely benefit the development of biasfield-robust automated diagnosis system.

show abstract

Section: Discussionmentioning

confidence: 99%

Bias Field Poses a Threat to DNN-Based X-Ray Recognition

Tian

Guo

Juefei-Xu

et al. 2021

2021 IEEE International Conference on Multimedia and Expo (ICME)

Self Cite

View full text Add to dashboard Cite

show abstract

“…More powerful weapons should be continuously developed for fighting AI aided fakes as new techniques for producing various fakes will emerge inadvertently. One of the future research directions could be to investigate how the proposed DeepSonar method can be extended to or work in tandem with various detectors [3,13,14,46,55] on other modalities of the 'fakes' such as AI-generated / forged images, and DeepFake videos, etc.…”

Section: Discussionmentioning

confidence: 99%

DeepSonar: Towards Effective and Robust Detection of AI-Synthesized Fake Voices

Wang

Juefei-Xu

Huang

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

With the recent advances in voice synthesis such as WaveNet, AIsynthesized fake voices are indistinguishable to human ears and widely applied to producing realistic and natural DeepFakes which are real threats to everyone. However, effective and robust detectors for synthesized fake voices are still in their infancy and are not ready to fully tackle this emerging threat. In this paper, we devise a novel approach, named DeepSonar, based on monitoring neuron behaviors of speaker recognition (SR) system, a deep neural network (DNN), to discern AI-synthesized fake voices. Layer-wise neuron behaviors provide an important insight to meticulously catch the differences among inputs, which are widely employed for building safety, robust and interpretable DNNs. In this work, we leverage the power of layer-wise neuron activation patterns with a conjecture that they can capture the subtle differences between real and AI-synthesized fake voices and provide a cleaner signal to classifiers than raw inputs. Experiments are conducted in three datasets (including commercial products from Google, Baidu, etc.) containing both English and Chinese languages to corroborate the high detection rates (98.1% average accuracy) and low false alarm rates (0.02 equal error rate) of DeepSonar in discerning fake voices. Furthermore, extensive experiment results show its robustness against manipulation attacks (e.g., voice conversion and additive real-world noises). Our work also poses a new insight into adopting neuron behaviors for effective and robust AI aided multimedia fakes forensics as an inside-out approach instead of being motivated and swayed

show abstract

“…The authors of [ 31 ] modified the loss function and added a frequency loss term. The authors of [ 32 ] performed a shallow reconstruction of fake images by learning a linear dictionary and aimed to reduce the artifacts introduced in the process of image synthesis. Therefore, a frequency-based detection method is not ideal for the newest deepfake dataset.…”

Section: Related Workmentioning

confidence: 99%

MFF-Net: Deepfake Detection Network Based on Multi-Feature Fusion

Zhao

Zhang

Ding

et al. 2021

Entropy

View full text Add to dashboard Cite

Significant progress has been made in generating counterfeit images and videos. Forged videos generated by deepfaking have been widely spread and have caused severe societal impacts, which stir up public concern about automatic deepfake detection technology. Recently, many deepfake detection methods based on forged features have been proposed. Among the popular forged features, textural features are widely used. However, most of the current texture-based detection methods extract textures directly from RGB images, ignoring the mature spectral analysis methods. Therefore, this research proposes a deepfake detection network fusing RGB features and textural information extracted by neural networks and signal processing methods, namely, MFF-Net. Specifically, it consists of four key components: (1) a feature extraction module to further extract textural and frequency information using the Gabor convolution and residual attention blocks; (2) a texture enhancement module to zoom into the subtle textural features in shallow layers; (3) an attention module to force the classifier to focus on the forged part; (4) two instances of feature fusion to firstly fuse textural features from the shallow RGB branch and feature extraction module and then to fuse the textural features and semantic information. Moreover, we further introduce a new diversity loss to force the feature extraction module to learn features of different scales and directions. The experimental results show that MFF-Net has excellent generalization and has achieved state-of-the-art performance on various deepfake datasets.

show abstract

FakePolisher: Making DeepFakes More Detection-Evasive by Shallow Reconstruction

Cited by 52 publications

References 44 publications

Bias Field Poses a Threat to DNN-Based X-Ray Recognition

Bias Field Poses a Threat to DNN-Based X-Ray Recognition

DeepSonar: Towards Effective and Robust Detection of AI-Synthesized Fake Voices

MFF-Net: Deepfake Detection Network Based on Multi-Feature Fusion

Contact Info

Product

Resources

About