AI RESEARCH
UNATE: UNsupervised ATomic Embedding for crystal structures property prediction
arXiv CS.LG
•
ArXi:2605.25866v1 Announce Type: new Accurately predicting crystal properties is critical for accelerating materials discovery, but it is often limited by scarce labeled data and costly theoretical calculations. To alleviate this, we propose UNATE (Unsupervised Atomic Embedding), a framework that leverages structural information extracted from unlabeled crystal structures. UNATE integrates an unsupervised denoising autoencoder with self-supervised contrastive learning to learn robust atomic representations, which are then used as input features for downstream property prediction.