Currently, several convolutional neural network (CNN)-based methods have been proposed for computer-aided COVID-19 diagnosis based on lung computed tomography (CT) scans. However, the lesions of ...

We propose a Crystal Diffusion Variational Autoencoder (CDVAE) that captures the physical inductive bias of material stability. By learning from the data distribution of stable materials, the decoder ...

In this article, in order to better handle this problem, a novel generative model named the conditional variational autoencoder with an adversarial training process (CVA 2 E) is proposed for ...