Machine-learning (ML) approaches have proven to be of great utility in modern materials innovation pipelines. Generally, ML models are trained on predetermined past data and then used to make predictions for new test cases. Active-learning, however, is a paradigm in which ML models can direct the learning process itself through providing dynamic suggestions/queries for the “next-best experiment.” In this work, the authors demonstrate how an active-learning framework can aid in the discovery of polymers possessing high glass transition temperatures (Tg). Starting from an initial small dataset of polymer Tg measurements, the authors use Gaussian process regression in conjunction with an active-learning framework to iteratively add Tg measurements of candidate polymers to the training dataset. The active-learning framework employs one of three decision making strategies (exploitation, exploration, or balanced exploitation/exploration) for selection of the “next-best experiment.” The active-learning workflow terminates once 10 polymers possessing a Tg greater than a certain threshold temperature are selected. The authors statistically benchmark the performance of the aforementioned three strategies (against a random selection approach) with respect to the discovery of high-Tg polymers for this particular demonstrative materials design challenge.

The authors developed a Behler–Parrinello-type neural network (NN) to improve the density-functional tight-binding (DFTB) energy and force prediction. The Δ-machine learning approach was adopted and the NN was designed to predict the energy differences between the density functional theory (DFT) quantum chemical potential and DFTB for a given molecular structure. Most notably, the DFTB-NN method is capable of improving the energetics of intramolecular hydrogen bonds and torsional potentials without modifying the framework of DFTB itself. This improvement enables considerably larger simulations of complex chemical systems that currently could not easily been accomplished using DFT or higher level ab initio quantum chemistry methods alone.

Our ability to describe crystal structure features is of crucial importance when attempting to understand structure–property relationships in the solid state. In this paper, the authors introduce robocrystallographer, an open-source toolkit for analyzing crystal structures. This package combines new and existing open-source analysis tools to provide structural information, including the local coordination and polyhedral type, polyhedral connectivity, octahedral tilt angles, component-dimensionality, and molecule-within-crystal and fuzzy prototype identification. Using this information, robocrystallographer can generate text-based descriptions of crystal structures that resemble descriptions written by human crystallographers. The authors use robocrystallographer to investigate the dimensionalities of all compounds in the Materials Project database and highlight its potential in machine learning studies.

The authors develop a computational approach that integrates machine learning (ML) and density functional theory (DFT) with experimental data to predict formable and thermodynamically stable iodine-containing apatites. This is an important problem because radioactive iodine is toxic and capturing it in solid waste forms have implications in remediation treatments. The authors train ML models using 336 compositions and screen 54 iodine-containing compounds in apatite stoichiometry. ML models predict 18 as formable and 24 as nonformable in the apatite structure; 12 compounds were identified to be uncertain. DFT convex hull predicted two to be thermodynamically stable, one as metastable, and nine as unstable.

Recent studies illustrate how machine learning (ML) can be used to bypass a core challenge of molecular modeling: the trade-off between accuracy and computational cost. Here, we assess multiple ML approaches for predicting the atomization energy of organic molecules. Our resulting models learn the difference between low-fidelity, B3LYP, and high-accuracy, G4MP2, atomization energies and predict the G4MP2 atomization energy to 0.005 eV (mean absolute error) for molecules with less than nine heavy atoms (training set of 117,232 entries, test set 13,026) and 0.012 eV for a small set of 66 molecules with between 10 and 14 heavy atoms. Our two best models, which have different accuracy/speed trade-offs, enable the efficient prediction of G4MP2-level energies for large molecules and are available through a simple web interface.

Facilitating the application of machine learning (ML) to materials science problems requires enhancing the data ecosystem to enable discovery and collection of data from many sources, automated dissemination of new data across the ecosystem, and the connecting of data with materials-specific ML models. Here, we present two projects, the Materials Data Facility (MDF) and the Data and Learning Hub for Science (DLHub), that address these needs. We use examples to show how MDF and DLHub capabilities can be leveraged to link data with ML models and how users can access those capabilities through web and programmatic interfaces.

This paper demonstrates the application of Natural Language Processing (NLP) tools to explore large libraries of documents and to correlate heuristic associations between text descriptions in figure captions with interpretations of images and figures. The use of visualization tools based on NLP methods permits one to quickly assess the extent of the research described in the literature related to a specific topic. The authors demonstrate how the use of NLP methods on only the figure captions without having to navigate the entire text of a document can provide an accelerated assessment of the literature in a given domain.