Atomtransmachine: An atomic feature representation model for machine learning.

[Display omitted] • Atomtransmachine: an atomic feature representation model for machine learning. • A spatial convolution layer is proposed to extract the spatial features of atoms. • A multi-attention mechanism is used to screen important features of atoms in forming new crystal structures. • Atomtransmachine is effective and can improve the efficiency of machine learning algorithms. In this study, a self-monitoring model is proposed to extract the atomic characteristics of the main group elements and transition metals from several molecular structures. Different from previous studies, we use a spatial convolution layer to extract the spatial features of atoms and a multi-attention mechanism to screen their important features in forming new crystal structures. Extensive numerical analyses show that the features extracted using the proposed model are effective and can improve the efficiency of machine learning algorithms. [ABSTRACT FROM AUTHOR]