尹志 (2023-02-28 21:51):
#paper https://doi.org/10.48550/arXiv.2203.17003 ICML, 2022, Equivariant Diffusion for Molecule Generation in 3D。扩散模型在各个领域发展极其迅速。除了图形图像,其触角已经扩展到生物制药、材料科学领域。本文就是一篇使用扩散模型进行3D分子生成的文章。作者提出了一种等变扩散模型,其中的等变网络能够很好的同时处理原子坐标这样的连续变量和原子类型这样的离散变量。该工作在QM9和GEOM两个典型的数据集上取得了sota的性能,是将等变性引入扩散模型的开篇工作之一。
Equivariant Diffusion for Molecule Generation in 3D
翻译
Abstract:
This work introduces a diffusion model for molecule generation in 3D that is equivariant to Euclidean transformations. Our E(3) Equivariant Diffusion Model (EDM) learns to denoise a diffusion process with an equivariant network that jointly operates on both continuous (atom coordinates) and categorical features (atom types). In addition, we provide a probabilistic analysis which admits likelihood computation of molecules using our model. Experimentally, the proposed method significantly outperforms previous 3D molecular generative methods regarding the quality of generated samples and efficiency at training time.
翻译
回到顶部