1 paper · 1 filter
Zhanhe Lei, Zhongyuan Wang, Jikang Cheng +5
Standard supervised training for deepfake detection treats all samples with uniform importance, which can be suboptimal for learning robust and generalizable features. In this work…