AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection

TL;DR

This paper introduces AWARE-NET, a novel adaptive weighted ensemble framework for deepfake detection that hierarchically combines multiple deep learning architectures with learnable weights, achieving state-of-the-art accuracy and robust cross-dataset performance.

Contribution

It proposes a two-tier ensemble method with dynamic weighting and multiple initializations, improving deepfake detection accuracy and generalization across datasets.

Findings

Achieved state-of-the-art intra-dataset AUC scores of 99.22% and 100%

Demonstrated strong cross-dataset generalization with AUC scores above 72%

Enhanced robustness without data augmentation

Abstract

Deepfake detection has become increasingly important due to the rise of synthetic media, which poses significant risks to digital identity and cyber presence for security and trust. While multiple approaches have improved detection accuracy, challenges remain in achieving consistent performance across diverse datasets and manipulation types. In response, we propose a novel two-tier ensemble framework for deepfake detection based on deep learning that hierarchically combines multiple instances of three state-of-the-art architectures: Xception, Res2Net101, and EfficientNet-B7. Our framework employs a unique approach where each architecture is instantiated three times with different initializations to enhance model diversity, followed by a learnable weighting mechanism that dynamically combines their predictions. Unlike traditional fixed-weight ensembles, our first-tier averages predictions within each architecture family to reduce model variance, while the second tier learns optimal contribution weights through backpropagation, automatically adjusting each architecture's influence based on their detection reliability. Our experiments achieved state-of-the-art intra-dataset performance with AUC scores of 99.22% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.06% (FF++) and 99.94% (CelebDF-v2) without augmentation. With augmentation, we achieve AUC scores of 99.47% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.43% (FF++) and 99.95% (CelebDF-v2). The framework demonstrates robust cross-dataset generalization, achieving AUC scores of 88.20% and 72.52%, and F1 scores of 93.16% and 80.62% in cross-dataset evaluations.