Robust AI-Generated Face Detection with Imbalanced Data

TL;DR

This paper introduces a robust deepfake detection framework that effectively handles distribution shifts and class imbalance, significantly improving detection accuracy and generalization across diverse datasets.

Contribution

The paper proposes a novel framework combining dynamic loss reweighting and ranking-based optimization to enhance deepfake detection robustness under imbalanced data conditions.

Findings

Improved detection accuracy on imbalanced datasets

Enhanced generalization across different generative models

Superior robustness against distribution shifts

Abstract

Deepfakes, created using advanced AI techniques such as Variational Autoencoder and Generative Adversarial Networks, have evolved from research and entertainment applications into tools for malicious activities, posing significant threats to digital trust. Current deepfake detection techniques have evolved from CNN-based methods focused on local artifacts to more advanced approaches using vision transformers and multimodal models like CLIP, which capture global anomalies and improve cross-domain generalization. Despite recent progress, state-of-the-art deepfake detectors still face major challenges in handling distribution shifts from emerging generative models and addressing severe class imbalance between authentic and fake samples in deepfake datasets, which limits their robustness and detection accuracy. To address these challenges, we propose a framework that combines dynamic loss reweighting and ranking-based optimization, which achieves superior generalization and performance under imbalanced dataset conditions. The code is available at https://github.com/Purdue-M2/SP_CUP.