Interpretable and Trustworthy Deepfake Detection via Dynamic Prototypes

Abstract

In this paper we propose a novel human-centered approach for detecting forgery in face images, using dynamic prototypes as a form of visual explanations. Currently, most state-of-the-art deepfake detections are based on black-box models that process videos frame-by-frame for inference, and few closely examine their temporal inconsistencies. However, the existence of such temporal artifacts within deepfake videos is key in detecting and explaining deepfakes to a supervising human. To this end, we propose Dynamic Prototype Network (DPNet) -- an interpretable and effective solution that utilizes dynamic representations (i.e., prototypes) to explain deepfake temporal artifacts. Extensive experimental results show that DPNet achieves competitive predictive performance, even on unseen testing datasets such as Google's DeepFakeDetection, DeeperForensics, and Celeb-DF, while providing easy referential explanations of deepfake dynamics. On top of DPNet's prototypical framework, we further formulate temporal logic specifications based on these dynamics to check our model's compliance to desired temporal behaviors, hence providing trustworthiness for such critical detection systems.

To appear in the 2021 IEEE Winter Conference on Applications of Computer Vision (WACV 21')

References in corpus (17)

• Methods for Interpreting and Understanding Deep Neural Networks
• This Looks Like That: Deep Learning for Interpretable Image Recognition
• DeepFakes: a New Threat to Face Recognition? Assessment and Detection
• Visualizing and Understanding Convolutional Networks
• Towards Robust Interpretability with Self-Explaining Neural Networks
• Temporal Segment Networks: Towards Good Practices for Deep Action Recognition
• On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Models
• Neural-Symbolic Learning and Reasoning: A Survey and Interpretation
• ForensicTransfer: Weakly-supervised Domain Adaptation for Forgery Detection
• Efficient Formal Safety Analysis of Neural Networks
• SlowFast Networks for Video Recognition
• DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection
• Two-Stream Neural Networks for Tampered Face Detection
• Capsule-Forensics: Using Capsule Networks to Detect Forged Images and Videos
• Thinking in Frequency: Face Forgery Detection by Mining Frequency-aware Clues
• Interpretable Adversarial Perturbation in Input Embedding Space for Text
• Towards Generalizable Deepfake Detection with Locality-aware AutoEncoder