Multi-fidelity machine-learning with uncertainty quantification and\n Bayesian optimization for materials design: Application to ternary random\n alloys

2020/05/29 by Anh Tran, Tran, Anh, Julien Tranchida +5 · 2 citations
Materials Science · #Atomic Physics (physics.atom-ph) #Computational Engineering #Computational Physics (physics.comp-ph) #Electron and X-Ray Spectroscopy Techniques #FOS: Computer and information sciences #FOS: Physical sciences #Finance #Machine Learning in Materials Science #X-ray Diffraction in Crystallography #and Science (cs.CE)

paper · pdf · doi:10.48550/arxiv.2006.00139

openalex publication_date 2020/05/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We present a scale-bridging approach based on a multi-fidelity (MF)\nmachine-learning (ML) framework leveraging Gaussian processes (GP) to fuse\natomistic computational model predictions across multiple levels of fidelity.\nThrough the posterior variance of the MFGP, our framework naturally enables\nuncertainty quantification, providing estimates of confidence in the\npredictions. We used Density Functional Theory as high-fidelity prediction,\nwhile a ML interatomic potential is used as the low-fidelity prediction.\nPractical materials design efficiency is demonstrated by reproducing the\nternary composition dependence of a quantity of interest (bulk modulus) across\nthe full aluminum-niobium-titanium ternary random alloy composition space. The\nMFGP is then coupled to a Bayesian optimization procedure and the computational\nefficiency of this approach is demonstrated by performing an on-the-fly search\nfor the global optimum of bulk modulus in the ternary composition space. The\nframework presented in this manuscript is the first application of MFGP to\natomistic materials simulations fusing predictions between Density Functional\nTheory and classical interatomic potential calculations.\n

Cited by

• Bayesian Neural Networks versus deep ensembles for uncertainty quantification in machine learning interatomic potentials
• Enhancing Corrosion Resistance of Aluminum Alloys Through AI and ML Modeling

Related

• Electronic Structure Prediction of Multi-million Atom Systems Through Uncertainty Quantification Enabled Transfer Learning
• Multi-Objective Bayesian Materials Discovery: Application on the\n Discovery of Precipitation Strengthened NiTi Shape Memory Alloys through\n Micromechanical Modeling
• Machine Learning for Multi-fidelity Scale Bridging and Dynamical\n Simulations of Materials
• Approaches for Uncertainty Quantification of AI-predicted Material Properties: A Comparison
• Machine learning interatomic potential for high throughput screening and optimization of high-entropy alloys