ExplainabilityTransparencyTrustworthinesscond-mat.mtrl-sci
Explainable AI for Curie Temperature Prediction in Magnetic Materials
M. Adeel Ajaib, Fariha Nasir, Abdul Rehman
August 9, 2025
arXiv.org
2 citations
Key Takeaways
-
Machine learning can reliably predict material properties: The Extra Trees Regressor achieved an R² score of 0.85 ± 0.01, demonstrating that AI models can make accurate predictions about complex physical phenomena like Curie temperatures in magnetic materials, which has practical applications in materials science.
-
Explainability reveals the "why" behind predictions: Using SHAP analysis, researchers identified that specific physicochemical factors—such as average atomic number and magnetic moment—are key drivers of Curie behavior, showing how explainable AI techniques can uncover the reasoning behind model decisions rather than treating them as "black boxes."
-
Different material groups require tailored understanding: By applying k-means clustering to analyze chemically distinct material groups separately, the study demonstrates that responsible AI implementation should account for how models perform differently across subgroups, rather than relying on single overall performance metrics.
-
Domain knowledge improves AI trustworthiness: The researchers augmented their dataset with composition-based and domain-aware descriptors, illustrating that incorporating expert knowledge into AI systems—rather than relying solely on raw data—enhances both model performance and scientific interpretability.
Abstract
We explore machine learning techniques for predicting Curie temperatures of magnetic materials using the NEMAD database. By augmenting the dataset with composition-based and domain-aware descriptors, we evaluate the performance of several machine learning models. We find that the Extra Trees Regressor delivers the best performance reaching an R^2 score of up to 0.85 $\pm$ 0.01 (cross-validated) for a balanced dataset. We employ the k-means clustering algorithm to gain insights into the performance of chemically distinct material groups. Furthermore, we perform the SHAP analysis to identify key physicochemical drivers of Curie behavior, such as average atomic number and magnetic moment. By employing explainable AI techniques, this analysis offers insights into the model's predictive behavior, thereby advancing scientific interpretability.