The rising use of deepfakes in criminal activities presents a significant issue, inciting widespread controversy. While numerous studies have tackled this problem, most primarily focus on deepfake detection. These reactive solutions are insufficient as a fundamental approach for crimes where authenticity is disregarded. Existing proactive defenses also have limitations, as they are effective only for deepfake models based on specific Generative Adversarial Networks (GANs), making them less applicable in light of recent advancements in diffusion-based models. In this paper, we propose a proactive defense method named FaceShield, which introduces novel defense strategies targeting deepfakes generated by Diffusion Models (DMs) and facilitates defenses on various existing GAN-based deepfake models through facial feature extractor manipulations. Our approach consists of three main components: (i) manipulating the attention mechanism of DMs to exclude protected facial features during the denoising process, (ii) targeting prominent facial feature extraction models to enhance the robustness of our adversarial perturbation, and (iii) employing Gaussian blur and low-pass filtering techniques to improve imperceptibility while enhancing robustness against JPEG compression. Experimental results on the CelebA-HQ and VGGFace2-HQ datasets demonstrate that our method achieves state-of-the-art performance against the latest deepfake models based on DMs, while also exhibiting transferability to GANs and showcasing greater imperceptibility of noise along with enhanced robustness. Code is available here: https://github.com/kuai-lab/iccv25_faceshield

We present efficient squaring formulae based on the Toom-Cook multiplication algorithm. The latter always requires at least one non-trivial constant division in the interpolation step. We show such non-trivial divisions are not needed in the case two operands are equal for three, four and five-way squarings. Our analysis shows that our 3-way squaring algorithms have much less overhead than the best known 3-way Toom-Cook algorithm. Our experimental results show that one of our new 3-way squaring methods performs faster than mpz_mul ( ) in GNU multiple precision library (GMP) for squaring integers of approximately 2400-6700 bits on Pentium IV Prescott 3.2 GHz. For squaring in Z[x], our 3-way squaring algorithms are much superior to other known squaring algorithms for small input size. In addition, we present 4-way and 5-way squaring formulae which do not require any constant divisions by integers other than a power of 2. Under some reasonable assumptions, our 5-way squaring formula is faster than the recently proposed Montgomery's 5-way Karatsuba-like formulae.

In this paper, we extend a recent piece of work on low-weight polynomial form integers (LWPFIs). We present a new coefficient reduction algorithm based on the Montgomery reduction algorithm and provide its detailed analysis results. We give a condition for eliminating the final subtractions at the end of our Montgomery reduction algorithm adapted to perform the coefficient reduction. Our experimental results show that a new coefficient reduction algorithm is indeed more efficient than the one presented in [1].