Automated Deepfake Detection

Abstract

In this paper, we propose to utilize Automated Machine Learning to automatically search architecture for deepfake detection. Unlike previous works, our method benefits from the superior capability of deep learning while relieving us from the high labor cost in the manual network design process. It is experimentally proved that our proposed method not only outperforms previous non-deep learning methods but achieves comparable or even better prediction accuracy compared to previous deep learning methods. To improve the generality of our method, especially when training data and testing data are manipulated by different methods, we propose a multi-task strategy in our network learning process, making it estimate potential manipulation regions in given samples as well as predict whether the samples are real. Comparing to previous works using similar strategies, our method depends much less on prior knowledge, such as no need to know which manipulation method is utilized and whether it is utilized already. Extensive experimental results on two benchmark datasets demonstrate the effectiveness of our proposed method on deepfake detection.

An architecture entropy regularizer for differentiable neural architecture search

“…Celeb-DF WildDeepfake DFDC-pre Acc AUC Acc AUC Acc AUC Meso4 [1] [WIFS18] 67.53 66.17 64.47 66.5 75.39 76.47 RNN [8] [AVSS18] 71. 20 tion4 [1], Xception [25], Multi-Task [20], Capsule [21], EfficientNet-B4 [30], Multi-Attention [41], RNN [8], LTW [28], FT-TS [11], ADD [17], F3-Net [24], FInfer [10], and RECCE [5]. We directly report the quantitative results in the published papers if accessible.…”

“…Third, increasing the number of operations in the candidate space causes an inefficient search process. Liu et al [17] develop an automated deepfake detection framework using common convolutional layers as the search space. However, ordinary convolution is not specially designed to extract forgery traces, and there is still room for performance improvement if the search space is appropriately selected.…”