Artificial Intelligence in the Design of Innovative Metamaterials: A Comprehensive Review

Abstract

Artificial intelligence-based algorithms are becoming essential tools in materials science-related fields because of their excellent functionality in reflecting physics in the training database and predicting the properties of unexplored materials with outstanding accuracy. Designing novel materials with engineered properties, such as metamaterials, is the key to revolutionizing material discovery, and machine learning (ML) and deep learning (DL) can be powerful and indispensable tools for acceleration. This review focuses on the implementation of ML/DL-based approaches for designing metamaterials. Quantum-mechanical, atomistic, and macroscale simulation methods are also assessed as database construction processes. Forward and inverse design methods are summarized in detail, and breakthroughs in generative models are particularly introduced. Moreover, applications in fundamental property prediction and material structural design are reviewed. Finally, the remaining challenging tasks for future related work are presented.