Quantization-based 3D-CNNs through Circular Gradual Unfreezing for DeepFake detection

In the dynamic domain of synthetic media, deepfakes challenge the trust in digital communication. The identification of manipulated content is essential to ensure the authenticity of shared information. Recent advances in deepfake detection focused on developing sophisticated CNN-based approaches. However, these approaches remain anchored within the continuous feature space, potentially missing manipulative signatures that might be more salient in a discrete domain. For this task, we propose a new strategy which combines insights from both continuous and discrete spaces for enhanced deepfake detection. Our hypothesis is that deepfakes may lie closer to a discrete space, potentially revealing hidden patterns that are not evident in continuous representations. In addition, we propose a new gradual-unfreezing technique, employed into the proposed framework in order to slowly adapt the network parameters to align with the new combined representation. Via comprehensive experimentation, it is highlighted the efficiency of the proposed approach, in comparison to state-of-the-art deepfake detection strategies.