Applications of machine learning in computational nanotechnology

Abstract

Machine learning (ML) has gained extensive attentions in recent years due to its powerful data analysis capabilities. It has been successfully applied to many fields and helped the researchers to achieve several major theoretical and applied breakthroughs. Some of the notable applications in the field of computational nanotechnology are machine learning potentials, property prediction and material discovery. This review summarizes of the state-of-the-art research progress in these three fields. Machine learning potentials bridge the efficiency vs. accuracy gap between density functional calculations (DFT) and classical molecular dynamics (MD). For property predictions, machine learning provides a robust method that eliminate the needs of repetitive calculations for different simulation setup. Material design and drug discovery assisted by machine learning greatly reduces the capital and time investment by orders of magnitude. In this perspective, several common machine learning potentials and machine learning models are firstly introduced. Using these state-of-the-art models, developments in property predictions and material discovery are overviewed, respectively. Finally, this paper was concluded with an outlook on future directions of data-driven research activities in computational nanotechnology.

Machine Learning and Nanotecnology: How do they contribute to each other

“…This challenge resides in the physical limitations and the manifestation of complex quantum phenomena on this scale, which are not yet been completely elucidated. In this context, artificial intelligence techniques, such as machine learning, have helped the development of nanotechnology by allowing both the overcoming of physical limitations associated with experimental research and the progressive elucidation of complex physical phenomena [29].…”

Machine Learning for Shape Memory Graphene Nanoribbons and Applications in Biomedical Engineering

“…Concurrently with this, data-driven approaches, including those based on ML, have experienced remarkable advances over recent years in many applications. Nanotechnology and materials science are among those fields where such advances have been largely due to the development of ML potentials, bridging the gap between the efficiency and accuracy in DFT and (classical) MD calculations (e.g., [90]). Depending on the ML models and descriptors used, some of the most common methodologies in this group can be differentiated by the ways we control the degree of freedoms, which can be done efficiently through the MLIP functional forms already mentioned at the beginning of this section.…”