Data integration for accelerated materials design via preference learning

Abstract

Machine learning applications in materials science are often hampered by shortage of experimental data. Integration with external datasets from past experiments is a viable way to solve the problem. But complex calibration is often necessary to use the data obtained under different conditions. In this paper, we present a novel calibration-free strategy to enhance the performance of Bayesian optimization with preference learning. The entire learning process is solely based on pairwise comparison of quantities (i.e., higher or lower) in the same dataset, and experimental design can be done without comparing quantities in different datasets. We demonstrate that Bayesian optimization is significantly enhanced via data integration for organic molecules and inorganic solid-state materials. Our method increases the chance that public datasets are reused and may encourage data sharing in various fields of physics.

Deep learning enabled inorganic material generator

“…[12][13][14][15][16][17][18] In the past decade, machine learning techniques have been widely used by researchers to address many challenges in the field of material science and engineering. 2,[19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] Ward et al. 27 proposed a set of 145 hand engineered features based on stoichiometric attributes, elemental property statistics, electronic structure attributes, ionic compound attributes which can be used for a broad range of datasets.…”

Deep Learning Enabled Inorganic Material Generator

“…[12][13][14][15][16][17][18] In the past decade, machine learning techniques have been widely used by researchers to address many challenges in the field of material science and engineering. 2,[19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] Ward et al 27 proposed a set of 145 hand engineered features based on stoichiometric attributes, elemental property statistics, electronic structure attributes, ionic compound attributes which can be used for a broad range of datasets. Jha et al in their work 20 introduced a deep learning framework called ElemNet to predict formation enthalpy of materials directly from its elemental composition, eliminating the need of domain knowledge to hand engineer features.…”

Deep Learning Enabled Inorganic Material Generator

“…[12][13][14][15][16][17][18] In the past decade, machine learning techniques have been widely used by researchers to address many challenges in the field of material science and engineering. 2,[19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35] Ward et al 27 proposed a set of 145 hand engineered features based on stoichiometric attributes, elemental property statistics, electronic structure attributes, ionic compound attributes which can be used for a broad range of datasets. Jha et al in their work 20 introduced a deep learning framework called ElemNet to predict formation enthalpy of materials directly from its elemental composition, eliminating the need of domain knowledge to hand engineer features.…”