| Title | A Generalizable Deepfake Detector based on Neural Conditional Distribution Modelling |
| Publication Type | Conference Paper |
| Year of Publication | 2020 |
| Authors | Khodabakhsh, A., Busch, C. |
| Conference Name | 2020 International Conference of the Biometrics Special Interest Group (BIOSIG) |
| Date Published | sep |
| Keywords | anomaly detector, conditional probabilities, deep neural networks, DeepFake, Face detection, face recognition, faces, feature extraction, generalizable Deepfake detector, generative adversarial networks, Human Behavior, human factors, Metrics, neural conditional distribution modelling, neural nets, photo-realistic generation techniques, PixelCNN, probability, Probability distribution, pubcrawl, resilience, Resiliency, Scalability, synthetic face generation methods, synthetic images, Task Analysis, Training, two-step synthetic face image detection method, Universal Background Model, Video Forensics, video-realistic generation techniques |
| Abstract | Photo- and video-realistic generation techniques have become a reality following the advent of deep neural networks. Consequently, there are immense concerns regarding the difficulty in differentiating what content is real from what is synthetic. An example of video-realistic generation techniques is the infamous Deepfakes, which exploit the main modality by which humans identify each other. Deepfakes are a category of synthetic face generation methods and are commonly based on generative adversarial networks. In this article, we propose a novel two-step synthetic face image detection method in which general-purpose features are extracted in a first step, trivializing the task of detecting synthetic images. The anomaly detector predicts the conditional probabilities for observing every individual pixel in the image and is trained on pristine data only. The extracted anomaly features demonstrate true generalization capacity across widely different unknown synthesis methods while showing a minimal loss in performance with regard to the detection of known synthetic samples. |
| Citation Key | khodabakhsh_generalizable_2020 |
A Generalizable Deepfake Detector based on Neural Conditional Distribution Modelling