Optical property dataset of inorganic phosphor.

Developing inorganic phosphor with desired properties for light-emitting diode application has traditionally relied on time-consuming and labor-intensive material development processes. Moreover, the results of material development research depend significantly on individual researchers' intuition and experience. Thus, to improve the efficiency and reliability of materials discovery, machine learning has been widely applied to various materials science applications in recent years. However, the prediction capabilities of machine learning methods fundamentally depend on the quality of the training datasets. In this work, we constructed a high-quality and reliable dataset that contains experimentally validated inorganic phosphors and their optical properties, sourced from the literature on inorganic phosphors. Our dataset includes 3952 combinations of 21 dopant elements in 2238 host materials from 553 articles. The dataset provides material information, optical properties, measurement conditions for inorganic phosphors, and meta-information. Among the preliminary machine learning results, the essential properties of inorganic phosphors, such as maximum Photoluminescence (PL) emission wavelength and PL decay time, show overall satisfactory prediction performance with coefficient of determination ( R 2 ) scores of 0.7 or more. We also confirmed that the measurement conditions significantly improved prediction performance.

Contrastive representation learning of inorganic materials to overcome lack of training datasets.

Data representation forms a feature space where forms data distribution that is one of the key factors determining the prediction accuracy of machine learning (ML). In particular, the data representation is crucial to handle small and biased training datasets, which is the main challenge of ML in chemical applications. In this paper, we propose a data-agnostic representation method that automatically and universally generates a vector-shaped and target-specified representation of crystal structures. By employing the new materials representation of the proposed method, the prediction capabilities of ML algorithms were highly improved on small training datasets and transfer learning tasks. Moreover, the prediction accuracies of ML algorithms were improved by 28.89-30.87% in extrapolation problems to predict the physical properties of the materials in unknown material groups. The source code of EMRL is publicly available at https://github.com/ngs00/emrl/tree/master/EMRL.

Efficient learning rate adaptation based on hierarchical optimization approach.

This paper proposes a new hierarchical approach to learning rate adaptation in gradient methods, called learning rate optimization (LRO). LRO formulates the learning rate adaption problem as a hierarchical optimization problem that minimizes the loss function with respect to the learning rate for current model parameters and gradients. Then, LRO optimizes the learning rate based on the alternating direction method of multipliers (ADMM). In the process of this learning rate optimization, LRO does not require any second-order information and probabilistic model, so it is highly efficient. Furthermore, LRO does not require any additional hyperparameters when compared to the vanilla gradient method with the simple exponential learning rate decay. In the experiments, we integrated LRO with vanilla SGD and Adam. Then, we compared their optimization performance with the state-of-the-art learning rate adaptation methods and also the most commonly-used adaptive gradient methods. The SGD and Adam with LRO outperformed all the competitors on the benchmark datasets in image classification tasks.

Nonlinearity encoding to improve extrapolation capabilities for unobserved physical states.

The fundamental goal of machine learning (ML) in physical science is to predict the physical properties of unobserved states. However, an accurate prediction for input data outside of training distributions is a challenging problem in ML due to the nonlinearities in input and target dynamics. For an accurate extrapolation of ML algorithms, we propose a new data-driven method that encodes the nonlinearities of physical systems into input representations. Based on the proposed encoder, a given physical system is described as linear-like functions that are easy to extrapolate. By applying the proposed encoder, the extrapolation errors were significantly reduced by 48.39% and 40.04% in n-body problem and materials property prediction, respectively.

Reverse graph self-attention for target-directed atomic importance estimation.

Estimating the importance of each atom in a molecule is one of the most appealing and challenging problems in chemistry, physics, and materials science. The most common way to estimate the atomic importance is to compute the electronic structure using density functional theory (DFT), and then to interpret it using domain knowledge of human experts. However, this conventional approach is impractical to the large molecular database because DFT calculation requires large computation, specifically, O(n4) time complexity w.r.t. the number of electronic basis functions. Furthermore, the calculation results should be manually interpreted by human experts to estimate the atomic importance in terms of the target molecular property. To tackle this problem, we first exploit the machine learning-based approach for the atomic importance estimation based on the reverse self-attention on graph neural networks and integrating it with graph-based molecular description. Our method provides an efficiently-automated and target-directed way to estimate the atomic importance without any domain knowledge of chemistry and physics.

Tuplewise Material Representation Based Machine Learning for Accurate Band Gap Prediction.

The open-access material databases allowed us to approach scientific questions from a completely new perspective with machine learning methods. Here, on the basis of open-access databases, we focus on the classical band gap problem for predicting accurately the band gap of a crystalline compound using a machine learning approach with newly developed tuplewise graph neural networks (TGNN), which is devised to automatically generate input representation of crystal structures in tuple types and to exploit crystal-level properties as one of the input features. Our method brings about a highly accurate prediction of the band gaps at hybrid functionals and GW approximation levels for multiple material data sets without heavy computational cost. Furthermore, to demonstrate the applicability of our prediction model, we provide a data set of GW band gaps for 45835 materials predicted by TGNN posing higher accuracy than standard density functional theory calculations.

Machine-guided representation for accurate graph-based molecular machine learning.

In chemistry-related fields, graph-based machine learning has received significant attention as atoms and their chemical bonds in a molecule can be represented as a mathematical graph. However, many molecular properties are sensitive to changes in the molecular structure. For this reason, molecules have a mixed distribution for their molecular properties in molecular space, and it consequently makes molecular machine learning difficult. However, this problem has not been investigated in either chemistry or computer science. To tackle this problem, we propose a robust and machine-guided molecular representation based on deep metric learning (DML), which automatically generates an optimal representation for a given dataset. To this end, we first adopt DML for molecular machine learning by integrating it with graph neural networks (GNNs) and devising a new objective function for representation learning. In experimental evaluations, machine learning algorithms with the proposed method achieved better prediction accuracy than state-of-the-art GNNs. Furthermore, the proposed method was also effective on extremely small datasets, and this result is impressive because many real world applications suffer from a lack of training data.

Costless Performance Improvement in Machine Learning for Graph-Based Molecular Analysis.

Graph neural networks (GNNs) have attracted significant attention from the chemical science community because molecules can be represented as a featured graph. In particular, graph convolutional network (GCN) and its variants have been widely used and have shown a state-of-the-art performance in analyzing molecules, such as molecular label classification, drug discovery, and molecular property prediction. However, in molecular analysis, existing GCNs have two fundamental limitations: (1) information of the molecular scale is distorted and (2) global structures in a molecule are ignored. These limitations can seriously degrade the performance in the machine learning-based molecular analysis because the scale and global structure information of a molecule occasionally have a significant effect on the molecular properties. To overcome the limitations of existing GCNs, we comprehensively analyzed the structure of GCNs and developed a costless solution for the limitations of GCNs. To demonstrate the effectiveness of our solution, extensive experiments were conducted on various benchmark datasets.