Denoising Diffusion
Denoising diffusion models are a class of generative models that learn to reverse a diffusion process, gradually removing noise from a data sample to generate new data points. Current research focuses on applying these models to diverse tasks, including image generation, enhancement, and editing, as well as drug design, medical image analysis, and time series generation, often integrating them with other architectures like transformers and GANs to improve efficiency and performance. This approach offers a powerful tool for generating high-quality data in various domains, leading to advancements in fields ranging from materials science to healthcare through improved data augmentation, anomaly detection, and image analysis. The inherent stochasticity of these models is also being investigated for its role in robustness against adversarial attacks.
Papers
Baking Gaussian Splatting into Diffusion Denoiser for Fast and Scalable Single-stage Image-to-3D Generation
Yuanhao Cai, He Zhang, Kai Zhang, Yixun Liang, Mengwei Ren, Fujun Luan, Qing Liu, Soo Ye Kim, Jianming Zhang, Zhifei Zhang, Yuqian Zhou, Zhe Lin, Alan Yuille
Test-Time Adaptation of 3D Point Clouds via Denoising Diffusion Models
Hamidreza Dastmalchi, Aijun An, Ali Cheraghian, Shafin Rahman, Sameera Ramasinghe
Understanding Generalizability of Diffusion Models Requires Rethinking the Hidden Gaussian Structure
Xiang Li, Yixiang Dai, Qing Qu
Counterfactual MRI Data Augmentation using Conditional Denoising Diffusion Generative Models
Pedro Morão, Joao Santinha, Yasna Forghani, Nuno Loução, Pedro Gouveia, Mario A. T. Figueiredo