Efficient recommendation tool of materials by an executable file based on machine learning

Abstract

To accelerate the discoveries of novel materials, an easy-to-use materials informatics tool is essential. We develop materials informatics applications, which can be executed on a Windows computer without any special settings. Our applications efficiently perform Bayesian optimization to optimize materials properties and uncertainty sampling to complete a new phase diagram. We introduce the usage of these applications and show the sampling results for a ternary phase diagram.

Pushing property limits in materials discoveryviaboundless objective-free exploration

“…Among ML-based exploration approaches, efficient material searches based on black-box optimization, 7 which is a problem that nds the maximum of an unknown (black-box) function with a limited number of evaluations, such as Bayesian optimization have been applied in various elds, and many successful examples have been reported. [8][9][10][11][12][13] Drug-like molecule generation methods combining deep learning and Bayesian optimization have also been proposed. [14][15][16] However, black-box optimization generally requires an appropriate optimization target (objective) in advance.…”

Exploring Successful Parameter Region for Coarse-Grained Simulation of Biomolecules by Bayesian Optimization and Active Learning

“…Figure 1b shows the procedure of successful region search based on US. We implemented the method based on a combination of the label propagation (LP) [28] and the least confidence method [29], which showed the best performance among algorithms described in previous studies [22,23]. We assigned a successful or failed label to each parameter based on a success rate.…”

“…However, it is not necessarily appropriate to efficiently search for parameters beyond a certain criterion. In contrast, one of the authors recently proposed an effective sampling method [22,23] for constructing phase diagrams based on uncertainty sampling (US), a type of active learning technique. The method based on US can efficiently determine phases to examine the phase boundary preferentially when two or more phases are sampled.…”

Experimental design for the highly accurate prediction of material properties using descriptors obtained by measurement

“…Here, new data is selected from the candidate materials, which have not been synthesized, prepared in advance. Uncertainty sampling in active learning is useful: the datapoint with the highest uncertainties is selected to improve the prediction accuracy [23][24][25][26][27]. For example, these uncertainties correspond to the deviations of the predicted values, evaluated by such as Gaussian process regression.…”