Machine Learning for Elastic Properties of Materials: A predictive benchmarking study in a domain-segmented feature Space

Insights into the unique characteristics across different classes of materials are crucial for Machine Learning (ML) tools and reveal the physics behind the studied process. Traditional predictive modeling of elastic properties of materials is limited to only a few classes of materials and a small set of ML tools despite the broad applications of these materials. While in recent years, Graph neural networks (GNNs) have outshined traditional ML models in terms of predictability, their intensive data requirement and lack of interpretability may limit practical applicability. In this work, we developed a domain-segmented feature space using a diverse set of material attributes and performed a predictive analysis of elastic properties of materials using state-of-the-art ML tools. By deducing the model-independent overall ranking of the features based on feature importance learned by each model, the knowledge is then transferred to GNNs. The findings indicate a saturation limit in the predictability of traditional ML models, but the transfer of task-specific feature importance knowledge to the GNNs can enhance their performance by reducing their data requirement while retaining considerable accuracy.