Exploring Active Data Selection Strategies for Continuous Training in Deepfake Detection
Yoshihiko Furuhashi, Junichi Yamagishi, Xin Wang, Huy H. Nguyen, Isao Echizen
TL;DR
The paper tackles the challenge of keeping deepfake detectors effective as new spoofing methods emerge by proposing a continual-training framework that actively selects a small, informative set of new data from a large pool. It uses a negative energy-based confidence score to guide data selection, combining selected samples with the existing seed data for iterative fine-tuning over K rounds. Experiments on ForgeryNet and diverse pool datasets show the approach can achieve an EER as low as $2.5\%$ using only $15\%$ of the pool data, outperforming random selection and standard baselines. The results demonstrate practical, data-efficient continual learning for up-to-date deepfake detection, with implications for real-world deployment and defense against evolving spoofing methods.
Abstract
In deepfake detection, it is essential to maintain high performance by adjusting the parameters of the detector as new deepfake methods emerge. In this paper, we propose a method to automatically and actively select the small amount of additional data required for the continuous training of deepfake detection models in situations where deepfake detection models are regularly updated. The proposed method automatically selects new training data from a \textit{redundant} pool set containing a large number of images generated by new deepfake methods and real images, using the confidence score of the deepfake detection model as a metric. Experimental results show that the deepfake detection model, continuously trained with a small amount of additional data automatically selected and added to the original training set, significantly and efficiently improved the detection performance, achieving an EER of 2.5% with only 15% of the amount of data in the pool set.