MG-BERT: leveraging unsupervised atomic representation learning for molecular property prediction

January 2021 Xiaochen Zhang, Chengkun Wu, Zhijiang Yang, Zhenhua Wu, Jiacai Yi, Chang-Yu Hsieh, Tingjun Hou, Dongsheng Cao Briefings in Bioinformatics

Molecular Representation Learning BERT Property Prediction

MG-BERT introduces a BERT-style framework for unsupervised atomic representation learning from molecular structures. The learned representations improve downstream molecular property prediction and demonstrate the value of pretraining strategies in cheminformatics.