Materials representation and transfer learning for multi-property prediction

Abstract

The adoption of machine learning in materials science has rapidly transformed materials property prediction. Hurdles limiting full capitalization of recent advancements in machine learning include the limited development of methods to learn the underlying interactions of multiple elements as well as the relationships among multiple properties to facilitate property prediction in new composition spaces. To address these issues, we introduce the Hierarchical Correlation Learning for Multi-property Prediction (H-CLMP) framework that seamlessly integrates: (i) prediction using only a material's composition, (ii) learning and exploitation of correlations among target properties in multi-target regression, and (iii) leveraging training data from tangential domains via generative transfer learning. The model is demonstrated for prediction of spectral optical absorption of complex metal oxides spanning 69 three-cation metal oxide composition spaces. H-CLMP accurately predicts non-linear composition-property relationships in composition spaces for which no training data are available, which broadens the purview of machine learning to the discovery of materials with exceptional properties. This achievement results from the principled integration of latent embedding learning, property correlation learning, generative transfer learning, and attention models. The best performance is obtained using H-CLMP with transfer learning [H-CLMP(T)] wherein a generative adversarial network is trained on computational density of states data and deployed in the target domain to augment prediction of optical absorption from composition. H-CLMP(T) aggregates multiple knowledge sources with a framework that is well suited for multi-target regression across the physical sciences.

Formula Graph Self‐Attention Network for Representation‐Domain Independent Materials Discovery

“…This renders a multi‐output regression problem. Recently, GNNs have been successfully applied to predict the absorption spectra of three‐cation metal oxides [ 42 ] and phonon density of states. [ 8 ] In a similar vein, a multi‐class classification GNN is implemented to predict protein functions.…”

“…[38] The attention mechanism has been adapted in several ML architectures for materials property prediction with improved accuracy. [28,30,33,37,[39][40][41][42] We show that Finder can outperform some state-of-the-art stoichiometry-only models such as Roost and compete with crystal graph models such as MEGNet and CGCNN on diverse benchmark databases curated from the Materials Project (MP) repository. Compared to other models revisited in this work, our model displays faster convergence and achieves lower errors at all training set sizes explored.…”

New Opportunity: Machine Learning for Polymer Materials Design and Discovery

“…Under the circumstance of dealing with the limited training set, transfer learning is required to obtain the pretrained model of the features and relative properties, which could produce a more outstanding performance in the prediction accuracy with the comparison of the direct ML model with the small data set. [63][64][65] Lee et al [66] employed the GNNs with transfer learning to predict the lowest excited state energy of poly(3hexylthiophene) in single crystal and solution phases. DFT calcu-lation data from short oligomers of different lengths were used to pre-train a GNNs model data from short oligomers of various lengths whose repeating unit is the same as that of the target long oligomers.…”

Density of States Prediction for Materials Discovery via Contrastive Learning from Probabilistic Embeddings

“…[4][5][6] The recent advent of machine learning prediction of materials properties has introduced the possibility of even higher throughput primary screening due to the minuscule expense of making a prediction for a candidate material using an already-trained model. [7][8][9][10] Toward this vision, we introduce the materials to spectrum (Mat2Spec) framework for predicting spectral properties of crystalline materials, demonstrated herein for the prediction of the ab initio phonon and electronic densities of state.…”

“…We recently reported a multi-property prediction model H-CLMP 7 for prediction of experimental optical absorption spectra from only materials composition. H-CLMP implements hierarchical correlation learning by coupling multivariate Gaussian representation learning in the encoder with graph attention in the decoder.…”