Improving Deepfake Detection with Reinforcement Learning-Based Adaptive Data Augmentation

TL;DR

This paper introduces CRDA, a reinforcement learning-based adaptive data augmentation framework that enhances deepfake detector generalization by dynamically generating diverse, challenging forgery samples tailored to the detector's learning progress.

Contribution

The paper presents a novel RL-guided, causally-aware data augmentation method that adaptively synthesizes forgery samples to improve deepfake detection across domains.

Findings

CRDA outperforms state-of-the-art methods on multiple datasets.

Dynamic augmentation improves generalization over fixed strategies.

Integration of reinforcement learning and causal inference enhances robustness.

Abstract

The generalization capability of deepfake detectors is critical for real-world use. Data augmentation via synthetic fake face generation effectively enhances generalization, yet current SoTA methods rely on fixed strategies-raising a key question: Is a single static augmentation sufficient, or does the diversity of forgery features demand dynamic approaches? We argue existing methods overlook the evolving complexity of real-world forgeries (e.g., facial warping, expression manipulation), which fixed policies cannot fully simulate. To address this, we propose CRDA (Curriculum Reinforcement-Learning Data Augmentation), a novel framework guiding detectors to progressively master multi-domain forgery features from simple to complex. CRDA synthesizes augmented samples via a configurable pool of forgery operations and dynamically generates adversarial samples tailored to the detector's current learning state. Central to our approach is integrating reinforcement learning (RL) and causal inference. An RL agent dynamically selects augmentation actions based on detector performance to efficiently explore the vast augmentation space, adapting to increasingly challenging forgeries. Simultaneously, the agent introduces action space variations to generate heterogeneous forgery patterns, guided by causal inference to mitigate spurious correlations-suppressing task-irrelevant biases and focusing on causally invariant features. This integration ensures robust generalization by decoupling synthetic augmentation patterns from the model's learned representations. Extensive experiments show our method significantly improves detector generalizability, outperforming SOTA methods across multiple cross-domain datasets.