In order to make accurate predictions of material properties, current machine-learning approaches generally require large amounts of data, which are often not available in practice. In this work, MODNet, an all-round framework, is presented which relies on a feedforward neural network, the selection of physically meaningful features, and when applicable, joint-learning. Next to being faster in terms of training time, this approach is shown to outperform current graph-network models on small datasets. In particular, the vibrational entropy at 305 K of crystals is predicted with a mean absolute test error of 0.009 meV/K/atom (four times lower than previous studies). Furthermore, joint learning reduces the test error compared to single-target learning and enables the prediction of multiple properties at once, such as temperature functions. Finally, the selection algorithm highlights the most important features and thus helps to understand the underlying physics.

As the number of novel data-driven approaches to material science continues to grow, it is crucial to perform consistent quality, reliability and applicability assessments of model performance. In this paper, we benchmark the Materials Optimal Descriptor Network (MODNet) method and architecture against the recently released MatBench v0.1, a curated test suite of materials datasets. MODNet is shown to outperform current leaders on 4 of the 13 tasks, whilst closely matching the current leaders on a further 3 tasks; MODNet performs particularly well when the number of samples is below 10,000. Attention is paid to two topics of concern when benchmarking models. First, we encourage the reporting of a more diverse set of metrics as it leads to a more comprehensive and holistic comparison of model performance. Second, an equally important task is the uncertainty assessment of a model towards a target domain. By applying a distance metric in feature space, we found that significant variations in validation errors can be observed, depending on the imbalance and bias in the training set (i.e., similarity between training and application space). Both issues are often overlooked, yet important for successful real-world applications of machine learning in materials science and condensed matter.

Machine-learning models have recently encountered enormous success for predicting the properties of materials. These are often trained based on data that present various levels of accuracy, with typically much less high-than low-fidelity data. In order to extract as much information as possible from all available data, we here introduce an approach which aims to improve the quality of the data through denoising. We investigate the possibilities that it offers in the case of the prediction of the band gap relying on both limited experimental data and density-functional theory relying different exchange-correlation functionals (with an increasing amount of data as the accuracy of the functional decreases). We explore different ways to combine the data into training sequences and analyze the effect of the chosen denoiser. Finally, we analyze the effect of applying the denoising procedure several times until convergence. Our approach provides an improvement over existing methods to exploit multi-fidelity data.

Machine-learning models have recently encountered enormous success for predicting the properties of materials. These are often trained based on data that present various levels of accuracy, with typically much less high- than low-fidelity data. In order to extract as much information as possible from all available data, we here introduce an approach which aims to improve the quality of the data through denoising. We investigate the possibilities that it offers in the case of the prediction of the band gap using both limited experimental data and density-functional theory relying on different exchange-correlation functionals. After analyzing the raw data thoroughly, we explore different ways to combine the data into training sequences and analyze the effect of the chosen denoiser. We also study the effect of applying the denoising procedure several times until convergence. Finally, we compare our approach with various existing methods to exploit multi-fidelity data and show that it provides an interesting improvement.

To improve the precision of machine-learning predictions, we investigate various techniques that combine multiple quality sources for the same property. In particular, focusing on the electronic band gap, we aim at having the lowest error by taking advantage of all available experimental measurements and density-functional theory calculations. We show that learning about the difference between high- and low-quality values, considered a correction, significantly improves the results compared to learning on the sole high-quality experimental data. As a preliminary step, we also introduce an extension of the MODNet model, which consists of using a genetic algorithm for hyperparameter optimization. Thanks to this, MODNet is shown to achieve excellent performance on the Matbench test suite.