However, methods relying on deep learning are prone to producing erroneous results under adversarial sample attacks. To address this issue, we investigated the generation of adversarial samples for quality assessment, aiming to test, evaluate, and enhance deep learning-based Image Quality Assessment (...
Diffusion-Based Adversarial Sample Generation for Improved Stealthiness and Controllability [NeurIPS'2023] Haotian Xue1,Alexandre Araujo2,Bin Hu3, andYongxin Chen1 1GaTech,2NYU,3UIUC Introduction Diff-PGD utilizes strong prior knowledge of Diffusion Model to generate adversarial samples with higher steat...
our prosody conditional discriminator significantly improves the quality of the generated speech by accurately emulating prosody. We use denoising diffusion generative adversarial networks to improve the prosody generation speed. Consequently, DiffProsody is capable of generating prosody 16 times faster than ...
To address this problem, we present TSDiff, an ML model that learns a direct mapping between TS conformations and 2D molecular graphs. Thus, one can skip the proper selection of conformations and orientations. Moreover, TSDiff can generate various TS conformations possible from the 2D graph wit...
Generative Adversarial Networks (NIPS 2014) 2 2023.04.05 Kwangsu Mun Jisu Kim Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks (ICCV 2017) A Style-Based Generator Architecture for Generative Adversarial Networks (CVPR 2019) 3 2023.04.12 Beomsoo Park Seunghwan Ji Denoisi...
Toward realistic image compositing with adversarial learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8415–8424, 2019. 2 [13] Jooyoung Choi, Jungbeom Lee, Chaehun Shin, Sungwon Kim, Hyunwoo Kim, and Sungroh Yoon...
Data Generation. In order to generate latent causal variables, we adopt random graphs, where each edge in a fixed topological order is sampled from a Bernoulli distribution with a parameter that is equal to 0.5. We consider the SCM to be linear Gaussian and we sample the weights from a mult...
(VAE)-based models8, generative adversarial network (GAN)9, normalizing flows10,11,12,13and diffusion models14,15. By adopting generative models, current machine learning methods10,11,16,17,18,19start from learning the underlying distribution of molecules and yield candidate molecules from ...
The proposed novel approach to topology optimization employs improved diffusion generative adversarial networks to address the limitations of traditional algorithms in computational efficiency and generation quality. The method reduces computational complexity by compressing high-dimensional parameters into low-dimen...
Adversarial purification with Score-based generative models Jongmin Yoon, Sung Ju Hwang, Juho Lee ICML 2021. [Paper] [Github] 11 Jun 2021 Natural Language Diffusion-LM Improves Controllable Text Generation Xiang Lisa Li, John Thickstun, Ishaan Gulrajani, Percy Liang, Tatsunori B. Hashimoto...