Atom2Vec: learning atoms for materials discovery

TL;DR

Atom2Vec is an unsupervised machine learning approach that learns atomic properties from materials data, enabling improved prediction of material properties through vector representations of atoms.

Contribution

This work introduces Atom2Vec, a novel unsupervised method to learn atomic features from materials data, enhancing materials property prediction accuracy.

Findings

Atom2Vec successfully learns meaningful atomic representations.

Clustering of atom vectors aligns with human atomic classifications.

Using atom vectors improves materials property prediction accuracy.

Abstract

Exciting advances have been made in artificial intelligence (AI) during the past decades. Among them, applications of machine learning (ML) and deep learning techniques brought human-competitive performances in various tasks of fields, including image recognition, speech recognition and natural language understanding. Even in Go, the ancient game of profound complexity, the AI player already beat human world champions convincingly with and without learning from human. In this work, we show that our unsupervised machines (Atom2Vec) can learn the basic properties of atoms by themselves from the extensive database of known compounds and materials. These learned properties are represented in terms of high dimensional vectors, and clustering of atoms in vector space classifies them into meaningful groups in consistent with human knowledge. We use the atom vectors as basic input units for neural networks and other ML models designed and trained to predict materials properties, which demonstrate significant accuracy.