Towards Generalizable Deepfake Detection via Real Distribution Bias Correction

2026-03-14

AI Summary

This work addresses the limited generalization of existing deepfake detection methods, which rely heavily on known forgery samples and struggle with unseen manipulation types. To overcome this, the authors propose the Real Distribution Bias Correction (RDBC) framework, which enhances generalization without requiring any forged samples by leveraging two complementary perspectives on authentic data: the global class distribution and the intrinsic Gaussianity of individual samples. RDBC employs a real population distribution estimation module and a distribution-aware feature whitening module to model distribution parameters using the i.i.d. property of genuine samples and amplify the discrepancy in Gaussianity between real and fake instances. Extensive experiments demonstrate that RDBC achieves state-of-the-art performance under both in-domain and cross-domain settings, significantly improving detection robustness against previously unseen forgery techniques.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Deep Generative Models & Autoencoders
• Application Domains: Misinformation & Fake News

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

To generalize deepfake detectors to future unseen forgeries, most existing methods attempt to simulate the dynamically evolving forgery types using available source domain data. However, predicting an unbounded set of future manipulations from limited prior examples is infeasible. To overcome this limitation, we propose to exploit the invariance of \textbf{real data} from two complementary perspectives: the fixed population distribution of the entire real class and the inherent Gaussianity of individual real images. Building on these properties, we introduce the Real Distribution Bias Correction (RDBC) framework, which consists of two key components: the Real Population Distribution Estimation module and the Distribution-Sampled Feature Whitening module. The former utilizes the independent and identically distributed (\iid) property of real samples to derive the normal distribution form of their statistics, from which the distribution parameters can be estimated using limited source domain data. Based on the learned population distribution, the latter utilizes the inherent Gaussianity of real data as a discriminative prior and performs a sampling-based whitening operation to amplify the Gaussianity gap between real and fake samples. Through synergistic coupling of the two modules, our model captures the real-world properties of real samples, thereby enhancing its generalizability to unseen target domains. Extensive experiments demonstrate that RDBC achieves state-of-the-art performance in both in-domain and cross-domain deepfake detection.

Problem

• deepfake detection
• generalization
• real distribution
• unseen forgeries
• domain generalization

Innovation

• Real Distribution Bias Correction
• Gaussianity
• Generalizable Deepfake Detection
• Feature Whitening
• Population Distribution Estimation