Graph Neural Networks (GNNs) have recently achieved significant success in processing non-Euclidean datasets, such as social and protein-protein interaction networks. However, these datasets often ...

Intelligent Cyber-Physical Systems typically utilize sensors to gather a significant amount of raw data which often possesses temporal and high-dimensional characteristics. Ensuring the data ...

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 ...