Suppressing Gradient Conflict for Generalizable Deepfake Detection

View on arXiv ← Back to list

Authors: Ming-Hui Liu, Harry Cheng, Xin Luo, Xin-Shun Xu

Published: 2025-07-29 06:48:22+00:00

AI Summary

This paper addresses the problem of poor generalization in deepfake detection models when training with both original and online synthesized fake images. It proposes a Conflict-Suppressed Deepfake Detection (CS-DFD) framework that mitigates gradient conflicts between these data sources, improving both in-domain accuracy and cross-domain generalization.

Abstract

Robust deepfake detection models must be capable of generalizing to ever-evolving manipulation techniques beyond training data. A promising strategy is to augment the training data with online synthesized fake images containing broadly generalizable artifacts. However, in the context of deepfake detection, it is surprising that jointly training on both original and online synthesized forgeries may result in degraded performance. This contradicts the common belief that incorporating more source-domain data should enhance detection accuracy. Through empirical analysis, we trace this degradation to gradient conflicts during backpropagation which force a trade-off between source domain accuracy and target domain generalization. To overcome this issue, we propose a Conflict-Suppressed Deepfake Detection (CS-DFD) framework that explicitly mitigates the gradient conflict via two synergistic modules. First, an Update Vector Search (UVS) module searches for an alternative update vector near the initial gradient vector to reconcile the disparities of the original and online synthesized forgeries. By further transforming the search process into an extremum optimization problem, UVS yields the uniquely update vector, which maximizes the simultaneous loss reductions for each data type. Second, a Conflict Gradient Reduction (CGR) module enforces a low-conflict feature embedding space through a novel Conflict Descent Loss. This loss penalizes misaligned gradient directions and guides the learning of representations with aligned, non-conflicting gradients. The synergy of UVS and CGR alleviates gradient interference in both parameter optimization and representation learning. Experiments on multiple deepfake benchmarks demonstrate that CS-DFD achieves state-of-the-art performance in both in-domain detection accuracy and cross-domain generalization.

Key findings

CS-DFD achieves state-of-the-art performance on multiple deepfake datasets, outperforming baselines in both in-domain and cross-domain settings. The method effectively addresses the '1+1<2' problem, demonstrating the importance of mitigating gradient conflicts. The approach is robust and adaptable to different backbone architectures.

Approach

CS-DFD uses two modules: Update Vector Search (UVS) finds an alternative update vector to reconcile conflicting gradients from original and synthesized fakes; Conflict Gradient Reduction (CGR) uses a novel loss to learn a low-conflict feature embedding space.

Datasets

FF++, Celeb-DF, DFDC, DFDCp, UADFV

Model(s)

EfficientNet, Xception, ViT-L, ViT-B

Author countries

China, Singapore

← Previous