Machine learning for inverse design of acoustic and elastic metamaterials
Krupali Donda, Pankit Brahmkhatri, Yifan Zhu, Bishwajit Dey, Viacheslav Slesarenko
Abstract
Recent rapid developments in machine learning (ML) models have revolutionized the generation of images and texts. Simultaneously, generative models are beginning to permeate other fields, where they are being applied to the effective design of various structures. In the field of metamaterials, in particular, machine learning has enabled the creation of sophisticated architectures with unconventional behavior and unique properties. In this article, we review recent advancements in the ML-driven design of a particular class of artificial materials — phononic metamaterials — that are capable of programming the propagation of acoustic and elastic waves. This review includes an in-depth discussion of the challenges and future prospects, aiming to inspire the phononic community to advance this research field collectively. We hope this article will help readers understand the recent developments in generative design and build a solid foundation for addressing specific research problems that could benefit from the application of machine learning models. • Inverse design of acoustic and elastic metamaterials via machine learning is discussed. • The pros and cons of the specific models are analyzed. • The potential of physics-aware models is highlighted. • Future challenges are outlined and discussed.